EFL Supervisors’ Written Feedback Focus and Language Functions on Masters Degree Students Thesis

The objective of this study was to examine EFL supervisors’ written feedback focus, and language functions at some selected public universities in Ethiopia. To address the objective of the study, a descriptive exploratory design was adopted. 55 supervisees were randomly selected of 255 for the questionnaire survey. Besides, four supervisors were selected for interviews. Moreover, feedback comments on ve theses drafts were utilized as data sources. To decide the language functions and feedback foci, frequency counts and percentage were used. Nonetheless, the interview data were analyzed qualitatively as per their themes. The results showed thesis supervisors used the directive feedback language function most prevailingly over the expressive in their feedback provision. There was no balance of praise, criticism, and suggestions in the provision of feedback comments. The supervisors largely focused on content knowledge, followed by genre knowledge, and then linguistic accuracy and appropriateness (forms), respectively.


Introduction
Tertiary education, particularly postgraduate studies, requires a lot of writing as the courses require numerous writing. As well, thesis writing demands a high academic writing calibre. Since English is a foreign language in Ethiopia, writing in English is a daunting task. To develop the students' academic writing, teachers have to provide adequate and effective feedback on their students' writing (Kumar & Stracke, 2007). Thesis supervisees may not attain the academic writing expected of them without their supervisors' written feedback provision on their thesis works since feedback provides learning experiences in writing (Kumar & Stracke, 2007). Accordingly, continuous written feedback on supervisees' theses drafts is the major source of learning and improvement upon which supervises revise their drafts and close the gaps between their actual level and the standards expected of them ( Concerning language functions of written feedback, a couple of studies showed the prevalence of direct feedback provision on students' errors (Gul et al., 2016;Lucero, Fernández & Montanero, 2018). Besides, Yenus (2018) carried a study on language functions of supervisors' written feedback to students' thesis at Bahir Dar University, Ethiopia. The data were collected from eight supervisors' written feedback comments. The result showed that the directive clari cation language function was dominantly employed while expressive approval was ignored or rarely used. Besides, the nding revealed that directive instruction and expressive disapproval were almost equally applied next to directive clari cation.
This shows that supervisors ignored expressive approval while applying expressive disapproval from the directive function category. In other words, the feedback is dominantly criticism devoid of positive comments (praise). This implies that there is no balance of praise, criticism, and suggestions in the provision of feedback comments.
Similarly, Ghazal et al. (2014) showed that teachers critiqued students' work without offering suggestions and hence the comments lacked a balance of praise, criticism, and suggestions. Moreover, the feedback comments were found overcrowded with several symbols for criticism with no elaboration or direction to the students for improvement. In the same vein, Lee (2009) disclosed that teachers mostly provide critical feedback on weaknesses ignoring the strengths of the work. Furthermore, it was indicated that teachers used to capitalize on students' errors without offering supportive suggestions on how to improve their works. Nevertheless, studies suggest effective written feedback should balance both the strengths and weaknesses of the students' writings (Ghazal et Lee (2009) also showed that the feedback teachers usually give to their students focus on form.
On the same issue of concern, Bitchener and Basturkmen (2010) investigated the feedback focus of supervisors on 35 supervisors in the elds of Humanities, Mathematics, and Commerce at six New Zealand universities. The data were gathered through questionnaires, interviews, and samples of the feedback given on thesis drafts. The nding showed feedback on gaps in the content covered were the most frequent areas with a particular focus on theoretical understanding and coverage. Besides, feedback on building an argument coherently and cohesively was also found in the area of attention in the feedback provision.
The studies reviewed above ought to be re-reviewed in terms of their perspectives, study focus, and tools employed to identify and justify the felt gaps. Accordingly, some of the studies were done on written feedback from teachers' perspectives (Gul et.al, 2016;Lucero, Fernández, & Montanero, 2018;Hyland & Hyland, 2001). Others viewed written feedback on students' assignments from both teachers' and students' perspectives (Ghazal et.al, 2014;Mulliner & Tucker, 2015). The study by Yenus (2018) emphasized on focus and language functions of supervisors' written feedback to M.A students' thesis while Kumar and Stracke (2007) stressed on language functions of written feedback on the Ph.D. thesis. Concerning tools employed, most studies used a questionnaire and focus group discussions or interviews. However, some studies applied written feedback comments as the only tool (Iqbal, et al., 2014;Kumar & Stracke, 2007;Lucero, Fernández & Montanero, 2018;Yenus, 2018). On the other hand, Ghazal et.al (2014) employed feedback comments and student interviews while Hyland and Hyland (2001) utilized feedback comments along with teacher interviews and think-aloud protocol.
Limited studies were done on feedback provision on students' thesis work (Bitchener & Basturkmen, 2010;Kumar & Stracke, 2007;Yenus, 2018). Besides, both Kumar and Stracke (2007) and Yenus (2018) used only feedback comments as a tool while multiple tools and perspectives could be used on large participants. Moreover, thesis supervisors' written feedback focus and language functions in feedback provision have not yet been explored through multiple tools and various perspectives. Thus, it seems that there is a felt niche on EFL supervisors' written feedback focus, and language functions at some selected public universities in Ethiopia. To this effect, this study sought to: (1) examine the focus of supervisors' written feedback on supervisees' thesis, and (2) inspect the language functions of written feedback comments the thesis supervisors provide on their supervisees' thesis.

Research Methodology
Context of the study This study was conducted on written feedback the thesis supervisors' offer to their supervisees in Teaching English as a Foreign Language (TEFL) while conducting their M.A thesis. The study took place in four public universities in Ethiopia. Among the universities of Ethiopia, Arbaminch University, Wolayeta Sodo University, Hawassa University, and Dilla University were selected for this study due to their relative proximity to the researchers' workplace, Arbaminch. The data were collected in 2020 between January and October.

Research design
The objective of this study was to examine EFL supervisors' written feedback focus, and language functions at some selected public universities in Ethiopia. To address the objective of the study, a descriptive exploratory design with a qualitative approach was adopted. Since the experience of written feedback at the graduate level is less studied, a descriptive exploratory design was considered appropriate.
Participants of the study Selected EFL thesis supervisors in the four universities and their supervisees were the participants of the study. From the total of 255 supervisees, 55 were selected through a simple random sampling technique for a questionnaire survey. However, four supervisors were also selected based on their willingness for indepth interviews.

Data collection instruments
This study aimed to explore the feedback focus and language functions of supervisors' written feedback on supervisees' thesis at selected universities in Ethiopia. To address these objectives, a questionnaire, interviews with selected supervisors, and in-text feedback comments on thesis drafts were utilized. A detailed description of each tool is provided below.

Questionnaire for supervisees
The purpose of the supervisees' questionnaire was to obtain quantitative data on their supervisors' feedback focus. A close-ended questionnaire was prepared based on the literature on a ve-point Likert scale that ranged from never (1), rarely (2), sometimes (3), often (4), and to always (5). The questionnaire constituted of 15 items in three sections. The rst section covered (n=6) items on linguistic accuracy and appropriateness (form), and the second and the third sections formed content knowledge (n=5) items and genre knowledge (n=4) items, respectively. The reliabilities of the sections were also checked. The Cronbach alpha value of 0.91 was obtained for linguistic accuracy and appropriateness while Cronbach alpha values of 0.88 and 0.89 were gained for content knowledge and genre knowledge, respectively.
These indicated that the dimensions for feedback focus are highly reliable for data collection.

Interview with supervisors
Semi-structured interview questions were utilized to intensively probe supervisors' view of feedback language functions in their feedback practices. Thus, the purpose of the interview was to generate elaborated in-depth information on the issue mentioned. To this effect, face-to-face individual interviews were held.

In-text feedback comments on thesis drafts
To generate valuable data about supervisors' feedback focus and language functions, the in-text feedback comments on supervisees' theses drafts were used. To this effect, supervisees' in-text feedback comments on ve theses were selected randomly from each university assuming adequate data would be obtained from them.
The in-text written feedback comprised of all comments offered by the supervisor provided in the margin of the text most of the time. The in-text feedback comments were transcribed word for word and are available in the appendix section. Each comment, phrase, or word that communicated a single message to the supervisee was taken as one piece of communication.

Methods of data analysis
The objective of this study was to examine EFL supervisors' written feedback focus, and language functions at some selected public universities in Ethiopia. The data from the written text was arranged and coded into categories. To this effect, the focus of written feedback comments were grouped as content knowledge (if comments are on the accuracy, completeness, and relevance of thoughts and clarity of ideas), linguistic accuracy, and appropriateness (if comments highlighting grammar, punctuation, tenses, surface structure errors, and word choices), and writing genre (if comments are on the functions of different parts of a thesis which includes literature synthesis, argument building, re ective writing, and referencing style, etc) according to Bitchener and Basturkmen (2010) categorization of feedback focus.
The categorization and organization of feedback language functions into the directive and expressive functions and the subcategories were adapted from earlier studies (Kumar & Stracke, 2007). As to Kumar and Stracke (2007), directive feedback is ordering the supervisees to do something and is subcategorized into clari cations and instructions. On the other hand, the expressive function of feedback is conveying feelings and is comprised of praise, criticism, and opinion /suggestion. It is comprised of praise/approval, criticism/disapproval, and opinion/suggestion (Hyland & Hyland, 2001;Kumar & Stracke, 2007). Consequently, feedback comments which appreciate or credit student work was coded as praise, while comments that show of dissatisfaction or negative were coded as criticism, and comments that give ways to do the work were coded as a suggestion (Hyland & Hyland, 2001).
To develop an appropriate categorization and coding, the researchers tried the categorization numerous rounds individually and then in pairs. The comments were, therefore, double-coded to ensure the credibility of the coding. The organized data obtained through supervisees' questionnaire and in-text written feedback comments on the thesis were quanti ed and organized into frequency counts and percentages.
The data obtained through interviews from the supervisors were also coded and organized into the directive and expressive functions of language and their sub-categories. The organized data were then analyzed qualitatively as per their themes to address the language functions of written feedback.

Data analysis and discussion
This section introduced data analysis and discussion on thesis supervisors' feedback language functions and focus on their feedback comments to the supervisees Supervisors' feedback functions This subsection examines thesis supervisors' feedback language functions in their feedback comments to the supervisees. To this effect, on-script feedback comments, and interviews with the supervisors were employed. The data have been analyzed and presented below.
On-scripts feedback functions As shown in table1, the written feedback students received on their theses are directive and expressive. Sums of 532 feedback comments were found in the written drafts of the students. Among which, 397 (74.6) written feedback is directive which urges the receiver for action while the expressive feedback category account 135 (25.4). Thus, it seems that the supervisors use the directive function most dominantly in their feedback provision. Total 532 100 Table 2 shows the sub-categories of directive and expressive feedback language functions. Fix feedback sub-categories were obtained from feedback comments on theses of the students which included directive-instruction, directive-clari cation, expressive-approval, expressive-disapproval, and expressivesuggestion.
As to feedback language functions sub-categories (table2), the results indicated that supervisors' feedback on students' theses was dominantly directive-clari cation 221 (41.5 %) followed by directiveinstruction 135 (25.4%). Directive-clari cation feedback requests the students for the clear elucidation of ideas in the paper for direction and revision. Some of the directive-clari cation common in the comments comprise: 'how?', 'what do you mean?', and 'evidence?? How do you relate these (teachers' traditional teaching and students' reading) to VLS? '.
The second most commonly offered feedback was directive-instruction feedback. This feedback type gives direction to the students to make necessary amendments in the texts. Among the many feedback comments of directive-instruction provided in the theses, some of them which may exemplify the entire are: 'please check spelling', 'follow SGS format for the cover page and title page', 'put this in its right place, not here', and 'consistently (in all) capitalize initial letters of content words in titles and sub-titles.' Expressive-disapproval feedback was the third frequently provided feedback which accounts for 88 (16.5%) of the total comments. Some of the expressive-disapproval feedback comments the students received includes: 'this can't be a sound justi cation', 'faulty parallelism', 'this is inappropriate citation; no two names should be used like this', 'your background looks like literature review, not study background', 'it is also too long and lacks focus', and 'your questionnaire as an instrument lacks a description.' The expressive-suggestion was the fourth that took 85 (16%) of the total comments. For instance, some of the expressive-suggestion comments provided in the theses of the students are: 'you need to relate your research nd with previous research ndings. You should show the differences and the similarities between the nding of your study and the previous studies', 'try to shorten the background by focusing on only the pertinent issues which have a direct bearing on your title', and 'you need to speci cally state what methods you employed to ensure the validity of each tool's data; then state the same about reliability.' On the contrary, expressive-approval feedback was insigni cant 3 (0.6%) in the written feedback comments the students received. These include: 'this is a good objective'; 'you already started talking about similar local studies. Keep on doing this', and 'the background seems good.' This implies that the provision of negative comments (criticisms) seems to be double and is overriding in the feedback comments as compared to opinions for improvement and acknowledgment of the strength of students' attempts.

Supervisors' interview data analysis on feedback functions
Teachers were interviewed on the focus and language functions of feedback comments. Concerning the focus of feedback comments, one of the interviewed supervisors underscored that he pays attention to content, genre, and linguistic accuracy and appropriateness though he gives more attention to content and noted the following: I mainly focus on organization, the relevance of concepts included (content), and formatting issues. Sometimes, I also consider mechanics and diction as well. I often give due emphasis on content and organization as they are the most important pillars of the learners' written research report.
Describing the focus of feedback comments he offers to his supervisees, the other supervisor noted: I focus on all parts of the candidates' research work but I give special attention to the links among the research title, the stamen of the problem, the objectives, the methodology section, and the analysis section with the major ndings. My reason for doing this is that focusing on these parts may give me the picture of the research; how things are woven together in a thread-the ow of the entire research work.
The third interviewed supervisor believes that he focuses only on essential aspects. The speci c aspects he focuses on include adequacy and relevance of contents, word choices and organizational issues, writing styles, and conventions. This supervisor seems that he is eclectic in his focus on feedback comments.
The last interviewed supervisor emphasized that effective feedback should indicate where the problems lie and how these problems should be addressed. He further noted that supervisors' written feedback comments are meant for improving supervisees' work and hence they should use them as much as possible. Accordingly, he mentioned that he focuses on content, genre, and linguistic accuracy and appropriateness although he pays more attention to content in his feedback comments. The interviewee additionally noted the following: I focus on all aspects of a thesis while giving feedback comments. Truly speaking most of the time I focus on content. But this does not mean I ignore other aspects. For instance, I suggest students adhere to the school of graduate studies (SGS) guidelines of my university. Besides, I give due attention to the language aspect as well though I give more attention to that of contents.
On language functions of feedback comments, the interviewed supervisors replied that they use directive language in their feedback comments. Speci cally, they mentioned that they apply directive clari cation most of the time. Besides, they claimed that they use directive instruction feedback comments sometimes as well. In the elaboration of these, one of the interviewees noted the reason behind his choice for directive clari cation over directive instruction, and noted: I use both [directive clari cation and directive instruction] though I tend to use directive clari cation in my written feedback most of the time. The reason for this is that this [directive clari cation] kind of feedback comments is very much important to encourage students to search for knowledge or to create insight learning (which important to discover solutions to problems). Besides, it gives the candidates to see their works and to revisit them in light of the comments.
The other interviewed supervisor replied that his feedback language most of the time is that of directive clari cation which requests the supervisees to elaborate and substantiate what they are writing. Also, he reported he uses expressive suggestions to give them room to revisiting their work as well rather than praise or criticisms. Moreover, he mentioned that he sometimes suggests the resources useful to address the comments given.
However, the interviewees mentioned that they do not often use expressive approval and suggestions though they abundantly apply criticisms in their feedback comments. As to his use of expressive suggestion feedback comments, one of the interviewees stated he uses suggestions rarely when he feels that the learner is somehow confused.

Supervisors'feedback focus
This subsection examines thesis supervisors' feedback focus on supervisees' theses. To this effect, onscript feedback comments, and interviews with the supervisors, and a questionnaire to the supervisees were employed. The data have been analyzed and presented below.
Supervisees' views of their supervisors' feedback focus  Table 3 depicts that six attributes were used to measure thesis supervisees' view of their supervisors' feedback focus on their theses. The majority of the supervisees (40.4% and 27.5%, on average) replied that their supervisors' feedback focuses are on linguistic accuracy and appropriateness as frequently as 'often' and 'always', respectively. Speci c to the attributes of linguistic accuracy and appropriateness, a signi cant number of the supervisees indicated that their supervisors were less concerned to focus on grammatical accuracy, and coherence, and cohesion as compared to the four attributes. Supervisors' feedback focus on the contents of supervisees' theses is presented below.  As shown in table5, four indicators were used to measure supervisees' view of their supervisors' feedback focus on their thesis works. The majority of the supervisees (40 % and 20%, on average) revealed that their supervisors' feedback focuses are on the genre as regularly as 'often' and 'always', respectively. Nevertheless, there are still large percentages of supervisees who rated their supervisors' feedback focus on genre knowledge as frequently as sometimes (26%) and rarely (12.3%). On indicators of feedback focus on genre, the supervisees have proven that their supervisors' focus on the rationale to including particular content in a particular part-genre, and functions and contents of different parts of the thesis as often as sometimes and rarely.
In conclusion, the average percentage in the tables above showed that the supervisees perceived that these supervisors focused dominantly on linguistic accuracy and appropriateness, followed by genre, and content respectively as frequently as ' always' and 'often'. However, it seems essential to examine supervisors' feedback focus on supervisees' theses to make an accurate conclusion of the issue. Therefore, supervisors' feedback focus on supervisees' theses is presented in table 6 below.
On-scripts data analyses on feedback focus Supervisors' written feedback focus on students' theses was collected from in-text comments of the theses. The data were collected and organized into content knowledge (its accuracy, completeness, and relevance), genre knowledge (the functions of different parts of a thesis), and linguistic accuracy and appropriateness. Table 6 shows content knowledge (66.5%) was the most frequently observed written feedback which requires students' conceptual understanding, accuracy, completeness, and relevance. A few of these comments, as illustration include: 'indicate how your work is different from others?', 'not clear?', and 'can this be a reason for the sampling?' The second most frequent feedback focus, though it is one-third of content knowledge, was genre knowledge (20.5%) which deals with referencing and citations, the functions of different parts of a thesis, and the relevance and appropriateness of the thesis for scienti c research. Some of these include: 'do you think this discussion is appropriate here?', 'bring the discussion here', and 'it must be the title of the journal which should be bold.' The third frequent feedback focus was on linguistic accuracy and appropriateness (13%). A few exemplars of the written comments asked students to revise, edit, or use the correct and formal language which include: 'this is not an appropriate term in research', 'check the completeness of your sentence', 'check language?', and 'not parallel???' are some to mention.
In conclusion, it seems that the majority of the comments focus on content knowledge, followed by genre knowledge, while a limited percentage of feedback focuses on linguistic accuracy and appropriateness.

Discussion
To examine these supervisors' language functions in their feedback comments to the supervisees, onscript feedback comments and interviews with the supervisors were employed. The results obtained from both on-script feedback comments and interviews disclosed that the supervisors generally used the directive feedback language most dominantly in their feedback provision as compared to expressive feedback. This nding is consonant with previous works in which directive feedback has been found pervasiveness in the provision of feedback (Gul et al., 2016;Leng, 2014;Lucero, Fernández & Montanero, 2018;Yenus, 2018). Conversely, the study done by Kumar and Stracke (2007) showed that the directive and expressive feedback functions shared equal status in the feedback provision of the supervisors.
Speci c to the feedback language functions sub-categories, the results gained from both on-script feedback comments and interviews showed that the supervisors' feedback comments were prevailingly directive-clari cation followed by directive-instruction. Previous works also corroborate this nding that directive clari cation language function was dominantly employed over directive instruction (Yenus, 2018). On the contrary, the study carried out by Leng (2014) showed that directive-instruction feedback comments dictate over directive-clari cation in the provision of supervisors' feedback.
As to the expressive categories, the expressive-disapproval (criticisms) and expressive-suggestion feedback comments were almost equally the third frequently provided feedback comments. In contrast, expressive-approval (praise) was found rare in the written feedback comments the supervisees received.
Thus, it was concluded that there was no balance of praise, criticism, and suggestions in the provision of feedback comments. The current nding is similar to the early works in the area which indicated that the feedback comments are dominantly criticism devoid of positive comments or praises (Leng, 2013;Yenus, 2018). Similarly, Ghazal et al. (2014) showed that teachers critiqued students' work without offering suggestions and hence the comments lacked a balance of praise, criticism, and suggestions. Moreover, the feedback comments were found crowded with several symbols for criticism with no elaboration or direction to the students for improvement. In the same vein, Lee (2009)  To scrutinize thesis supervisors' feedback focus on supervisees' theses, on-scripts feedback comments, interviews with the supervisors, and a questionnaire to the supervisees were used. The outcomes acquired from on-script feedback comments and interviews revealed that the feedback comments supervisors provide to their supervisees were principally focused on content knowledge, followed by genre knowledge, and then on linguistic accuracy and appropriateness, respectively. On the contrary, the result obtained from the student questionnaire has shown quite opposite outcome to the result gained from on-script feedback comments and interviews with the supervisors. The result acquired through student questionnaire indicated the supervisees perceived their supervisors' focused prevailingly on linguistic accuracy and appropriateness, followed by genre, and content knowledge, respectively as frequently as ' always' and 'often'. Since on-script feedback comments are rst-hand and real shreds of evidence to judge the gravity of feedback focus of the supervisors, it seems reasonable to conclude that the feedback comments the supervisors provide to their supervisees were largely focused on content knowledge, followed by genre knowledge, and then linguistic accuracy and appropriateness (forms), respectively.
The present nding is concurrent with the previous works in the area which showed that feedback comments focused on gaps in the content covered were the most frequent followed by a focus on either genre of the thesis or linguistic accuracy and appropriateness (

Conclusions
The results of the study showed that the theses supervisors used to offer directive feedback language function most prevailingly in their feedback provision more than the expressive feedback. In particular to the sub-categories of the feedback language functions, the feedback comments the supervisees received were mainly directive-clari cation followed by directive-instruction. This implies that the supervisors preferred to ask for additional information from the supervisees for a clearer elucidation instead of offering directions on explicit issues for improvement. This further suggests that the supervisors' feedback comments are more re ective questions that provide the supervisees opportunity to re ect as compared to the directive-instruction which requests actions for changes. Besides, directive-clari cation feedback comments are more positive and friendly and may motivate the supervisees to revisit their works.
Besides, the expressive-disapproval (criticisms) and expressive-suggestion feedback comments were almost equally the third frequently provided feedback comments while the expressive-approval (praise) was rare in the written feedback comments the supervisees received. This implies that praise, criticism, and suggestions in the provision of feedback comments are imbalanced.
Concerning thesis supervisors' feedback focus, the study concluded that the feedback comments the supervisors provide to their supervisees were largely focused on content knowledge, followed by genre knowledge, and then linguistic accuracy and appropriateness (forms), respectively.
In summing up, it is reasonable to conclude that the nding of this study may give EFL/ESL thesis supervisors insight on the focus of feedback they provide to their supervisees. Besides, the result of the study may help the supervisors vary the language functions they most commonly use in their feedback provision.

Recommendations
It is recommended that thesis supervisors' should offer feedback that needs to cover all features of supervisees' written texts with a focus on content, genre, linguistic appropriacy, and accuracy, etc. This may assist the supervisees to develop their thoughts completely and present them well. Besides, the supervisors ought to keep up offering directive-clari cation feedback which seeks more information from the supervisees' explanation along with directive-instruction as necessary. Moreover, the supervisors' should balance praises, criticisms, and suggestions in their feedback provision to encourage the supervisees to improve the quality of their written theses.
One of the limitations of this study is that it was based on in-text feedback written in the margin of the text as we could not get the overall feedback comments of the supervisors. Also, the in-text feedback comments were taken once instead of progressive comments on drafts over time. The data used in this work is available for reviewers on request at any time.

Funding
This study was funded by Arba Minch University, College of Social Sciences and Humanities, Ethiopia.
The fund was used for data collection, data encoding to SPSS and analysis.

Competing interests
We declare that there is no competing interests