A case study of Chinese master’s students’ perceptions of supervisors’ use of evaluative language in feedback

A master’s thesis requires students to carry out a research project under the supervisor’s guidance, which is a challenging undertaking of a master’s program (Paran et al., 2017). Master’s students, who are often regarded as novices in the academic discourse community, frequently encounter difficulties during the thesis writing process (Li, 2006). These range from macro-level issues, such as deciding on a research topic and identifying a research gap, to micro-level concerns, such as formatting, spelling, grammar and referencing (Allison et al., 1998). In this regard, supervisory feedback is essential in ensuring the success and quality of students’ completion of their research projects. This type of feedback, whether through written comments or annotations on thesis proposals or drafts, or verbal feedback during supervisory meetings, is highly valued by students as it helps to maintain research standards and guides them towards producing logical and coherent work (Abdulkhaleq and Abdullah, 2013; Hyland, 2013).

However, the degree to which supervisory feedback can achieve its desired outcomes is affected by the language used in communicating it (Hyland and Hyland, 2019). How language is used in supervisory feedback tends to affect how students perceive, value and engage with it (Hyland and Hyland, 2019). For instance, feedback that is overly critical or poorly worded can trigger vulnerability and helplessness among students (Boud and Brew, 1995), disrupt student–teacher relationships (Basturkmen et al., 2014), and even jeopardize students’ research process (Hyatt, 2005). Consequently, it is crucial to scrutinize the language employed in both written and oral supervisory feedback to evaluate whether the comments provided are appropriate and beneficial (Starfield et al., 2015). While research on supervisory feedback has tended to focus on its academic content, less attention has been paid to the specific use of language such as evaluative language (Bastola, 2022). Therefore, this study seeks to fill this gap by examining master’s students’ perceptions of using evaluative language in supervisor feedback.

The feedback examined in this paper concerns supervisors’ use of evaluative language in feedback during the master’s thesis writing process. It encompasses written annotations and comments on the thesis’s research proposals and drafts, as well as verbal dialogues in supervisory meetings, excluding any written or oral feedback after the final submission of the thesis. According to the theoretical framework of interpreting feedback established by Carless (2006), this study is conducted as a case study at a Sino-foreign university in China. Aiming to investigate master’s students’ perceptions of their supervisors’ use of evaluative language in feedback on their thesis proposals and drafts, this study mainly addresses one research question:

• (1)

  How do master’s students perceive the use of evaluative language in supervisory feedback?

At the outset, it is important to define feedback, as supervisory feedback falls into this category. In this study, the definition of feedback proposed by Carless (2016) is adopted, which refers to feedback as “dialogical processes in which learners make sense of information from different sources and use it to improve their work or learning strategies” (p. 1). This emphasizes the two-way nature of feedback, with students acting as active agents who seek, interpret and respond to feedback to close the feedback loops.

Overall, despite the breadth of research on master’s thesis supervision, including aspects such as the supervisor-student relationship (e.g. de Kleijn et al., 2013; Liang et al., 2021), the balance between the “supporting” and “gatekeeping” supervision objectives for thesis’s supervisors (e.g. Anderson et al., 2006), and adequate supervision strategies such the proposal of multi-voiced supervision that combines students in colloquia with individual supervision (e.g. Vehviläinen, 2009; Pilcher, 2011), the focus on supervisory feedback on master’s theses remains relatively sparse (Bastola and Hu, 2021a), which warrants further investigation. Of the number of research on supervisory feedback, a systematic review conducted by Chugh et al. (2022) revealed that most studies are primarily concerned with the postgraduate level without distinguishing between master’s and doctoral students. This can be seen in the qualitative research conducted by Abdulkhaleq (2021) among 24 postgraduate students at a Malaysian public university. Aiming to examine how ESL postgraduate students perceive the effectiveness of both written and oral feedback, Abdulkhaleq (2021) found that students perceived feedback that concentrated on content, research methodologies and argument constructions as effective. In contrast, feedback on grammar and punctuation is perceived as the least helpful. However, the study did not distinguish between feedback given to master’s students and PhD students. This lack of distinction might be a concern since the findings derived from supervisory feedback on doctoral theses might not be generalizable to master’s theses. PhD students often differ from their master’s counterparts regarding “disciplinary socialization, previous research experience, mastery of academic discourse, feedback needs and expectations” (Bastola and Hu, 2021a, p. 410). Hence, this study specifically focuses on master’s students.

In addition, Chugh et al. (2022) also noted a geographical bias in supervisory feedback studies, focusing on Western contexts such as the UK. Due to the cultural and contextual factors, findings from one educational or cultural setting may not directly apply to other educational settings. Given this Western-centric focus, Chugh et al. (2022) suggested that further research in the postgraduate supervision area could be conducted in non-Western contexts, particularly in China, to broaden the understanding of supervisory feedback in diverse educational settings. Thus, this study is situated within the context of higher education in China.

Evaluative language is defined as the use of language to convey “the speaker’s or writer’s attitude or stance towards, viewpoint on, or feelings about an entity or proposition” (Hunston and Thompson, 2000, p. 5), which is often used in providing feedback (Bastola and Hu, 2021b). It is primarily concerned with the semantic meaning of lexical choices to convey people’s attitudes or emotions (Martin and White, 2005). As emphasized by Hu and Choo (2016), the use of evaluative language in teacher feedback is not only crucial for mediating the effectiveness of feedback but also plays an important role in facilitating the achievement of learning goals during the educational process. For example, excessive criticism expressed through evaluative language can make students vulnerable, undermine their confidence and negatively affect their engagement in the learning process (Bastola and Hu, 2021b). Conversely, by incorporating informal language, inclusive pronouns and hedging in evaluative language, supervisors can mitigate the negative impact on students’ self-esteem and foster a sense of shared academic identity between supervisors and students (Bastola and Hu, 2021b). Therefore, it is important to analyze the evaluative language in supervisory feedback to “determine whether or not the comments are appropriate, helpful and insightful” to facilitate learning and improvement (Starfield et al., 2015, p. 132).

According to the appraisal model of evaluative language established by Martin and White (2005), there are three major categories in evaluative language: attitude, engagement and graduation. The first category, “attitude”, encompasses the emotional aspects of communication, such as the individual’s emotional responses, judgments and appreciation (Bastola and Hu, 2021b). The second category is “engagement”, which refers to the way in which the speaker or writer positions themselves in relation to other voices. Subcategories within “engagement” include “expand” and “contract”: the former is heteroglossic devices showing openness to involve other opinions, whereas the latter allows much less dialogical space that is regarded as a device delivering factual ideas rather than welcoming different opinions. Lastly, the final category, “graduation”, concerns the strength and intensity of evaluative language, exemplified by intensifiers such as “at all” in “We are not satisfied with your work at all” (Jin and Ruan, 2023, p. 5). This category includes “force” – which refers to “the up-scaling or down-scaling that adjusts the intensification of a value such as from important to very important” (p. 3), and “focus”, which sharpens or blurs semantic boundaries, such as the use of “sort of” (Jin and Ruan, 2023).

While evaluative language has been extensively studied in academic discourse, such as the examination of the role of evaluative language in rapport building with university students (Vinagre and Corral Esteban, 2018), its use in supervisory feedback remains in the margin of assessment research. Only one quantitative study, conducted by Bastola and Hu (2021b), has examined evaluative language in supervisory feedback provided by master thesis supervisors at a comprehensive university in Nepal. Their findings suggested that Nepalese supervisors frequently use evaluative language to express appreciation and judgments, whereas affective comments are less common (Bastola and Hu, 2021b). Based on this, Bastola and Hu (2021b) recommend that supervisors could mitigate harsh expressions of negative affect when providing supervisory feedback, avoid direct negative judgments on students’ social esteem, and integrate instances of joyous appreciation of students’ efforts to support them. However, this study primarily reflects the supervisor’s perspective, with limited consideration of the student’s perspective, which might be a potential drawback. As Bastola and Hu (2021b) pinpointed, understanding what constitutes effective supervisory feedback practices does not necessarily translate into implementing such practices because students and teachers may have different perceptions of feedback. Carless (2006) argued that students perceived feedback differently from teachers regarding various aspects, such as the amount of detail and the degrees of usefulness. Similarly, Bastola (2022) has conducted an empirical study on supervisory feedback, aiming to examine whether there are differences in perceptions of feedback between master’s students and their supervisors. Findings indicate that students’ understanding of the feedback differs from the supervisors’ intended meaning due to linguistic difficulties, and disparities in academic background and research focus between students and their supervisors (Bastola, 2022). Therefore, it is imperative to investigate how feedback is perceived from the student’s perspective to improve supervisory feedback and align it more closely with students’ needs. In light of this, this study explores master’s students’ perceptions of evaluative language in supervisory feedback.

With regard to students’ perception of evaluative language in feedback, research in this area has been somewhat overlooked. Van der Kleij and Lipnevich (2021) highlight the importance of investigating students’ perceptions of feedback, noting that these perceptions act as a precursor that determines student engagement in the feedback process. Only one research, conducted by Jin and Ruan (2023), has investigated the perceptions of undergraduates at an EMI university in China about the use of evaluative language by a lecturer when delivering oral feedback. Like Bastola and Hu’s (2021b) research, Jin and Ruan’s (2023) study also adopted the appraisal framework for analysis. Using mixed-method research that included a small scope of questionnaires and student interviews, they discovered that the use of evaluative language that conveys a positive attitude toward students’ efforts and achievements is highly appreciated by students (Jin and Ruan, 2023). However, the scope of their research was confined to a presentation assignment of a single course over the 2020 academic year, which might limit its data representativeness. Therefore, further research is needed to explore students’ perceptions of evaluative language in feedback to tailor feedback to students’ needs better. Building on Van der Kleij and Lipnevich’s (2021) definition, which considers students’ perceptions as “how students comprehend, perceive, and value a feedback message from both cognitive and affective perspectives” (p. 349), this study aims to address this gap by examining how students cognitively and emotionally perceive the use of evaluative language in supervisory feedback.

Given the absence of a theoretical framework tailored specifically for supervisory feedback, this study will be conceptually grounded in Carless’s (2006) framework for interpreting feedback. This framework, adapted from Higgins (2000), was originally developed to elucidate how teachers and students perceive assessment feedback in the context of undergraduate education in Hong Kong. It consists of three interconnected elements: discourse, power and emotion.

Higgins (2000) asserts that many students struggle to understand and interpret feedback accurately, largely because it is presented in an academic discourse that may not be fully accessible to them. In this context, the first element “discourse” refers to the specific language encoded in tutor comments. As noted by Baynham (2000), this issue is further complicated by the variability of discourse across disciplines, subjects and even individual tutors. Here, discourse serves as a means by which tutors, perhaps unintentionally, exercise power over students. This introduces the second element, “power”. The dominant ideology within universities often distinguishes between novice students and authoritative academics, a distinction typically reinforced by the role of teachers, who are more focused on correcting and judging academic work than encouraging dialogue about it (Ivanic et al., 2000; Lea and Street, 1998). As Boud (2012) observed, “We judge too much and too powerfully, not realizing the extent to which students experience our power over them” (p. 43). This imbalance in the power dynamic between teacher and student can prevent students from becoming active participants in the feedback processes (Yang and Carless, 2013). The third element, “emotion”, refers to the inherently emotional nature of the assessment process (Boud, 2012). Students invest both their emotions and their time in assessment tasks, and teachers’ feedback can elicit emotional responses, both positive and negative. These emotional responses can influence students’ self-esteem, their trust in the teacher and, consequently, their learning strategies and their responses to feedback (Carless, 2006).

This study employed a qualitative case study methodology, which took place at a Sino-foreign university in mainland China. This university cooperates with a British university to provide Chinese students access to British higher education without needing to study abroad. Two master’s programs are offered at this university: one-year full-time and two-year part-time taught postgraduate programs, and two-year research postgraduate programs. This study mainly focused on the taught postgraduate programs, which include 13 disciplines across the Humanities and Social Sciences, Business, and Science and Engineering, as opposed to the research postgraduate programs, which exclusively pertain to Science and Engineering. Among these taught postgraduate programs, a thesis is compulsory for disciplines within the Humanities and Social Sciences and Science and the Faculty of Economics and Finances. Disciplines such as Applied Linguistics, International Communication Studies and International Management require the thesis to be written only in English. A thesis is usually about 15,000 words and takes four to six months to complete, unless a student requests an extension due to mitigating circumstances. During this process, each student is chosen or assigned to a supervisor within the student’s department, and they usually have regular face-to-face or online supervisory meetings and frequent email communication.

This study has employed purposive sampling to select 12 participants from the Faculty of Humanities and Social Sciences and the Faculty of Economics and Finance. The gender and the study mode (full-time or part-time postgraduate program) of the participants are not restricted. As Lareau (2021) emphasized, the purposive sampling method is instrumental in efficiently using limited resources and ensuring the inclusion of best-informed participants. Thus, this sampling method is well-suited for this study. It selects participants with direct experience of master’s thesis supervision, which provides rich insights into how master’s students perceive their supervisors’ use of evaluative language. Moreover, due to time and resource constraints, this approach is also advantageous as it allows for an in-depth exploration of people’s perceptions without the need for a large sample size (Lareau, 2021). Drawing from Guest et al.’s research (2006), which found that 12 interviews are sufficient for achieving data saturation in homogeneous studies using purposive sampling, this study has recruited 12 participants to ensure the reliability of the findings. One summary table of the students’ information is shown below (Table 1).

To address the research question, individual semi-structured interviews were conducted for data collection; each was lasted between 45 and 60 min. Prior to each interview, students were asked to recall their thesis supervision experience by reviewing previous thesis drafts, documents and emails with their supervisors to better prepare for the interview questions. Given that all participants were native Chinese speakers, interviews were conducted in Chinese, with only one participant specifically requesting English. All interviews were audio-recorded, transcribed and translated into English for subsequent coding and analysis. To verify translation fidelity (Thompson and Dooley, 2019), all Chinese text has undergone a two-stage translation process. It was first translated into English using Youdao Translator, and then the resulting English text was further translated back into Chinese using DeepL in order to ensure the accuracy of translation.

This study employed a thematic analysis method as proposed by Braun and Clarke (2006). Using this data analysis method, themes that are relevant to the research question have been identified by analyzing the participants’ responses to all interview questions. The data analysis in this study followed a six-step procedure based on Braun and Clarke (2006). The steps included “(1) familiarizing with the data, (2) systematically coding the data, (3) generating initial themes, (4) developing and reviewing themes, (5) refining, defining, and naming the themes, and (6) writing up” (Braun and Clarke, 2006, p. 16). The three stages above were data-driven, meaning that the themes were developed directly from the data. The themes then underwent a process of revision and refinement in accordance with Carless’ (2006) theoretical framework for interpreting students’ perception of feedback.

Before proceeding, it should be acknowledged that there are some limitations to this study. The perspectives presented in this study are a limited representation of Chinese students. The extent to which they are representative of students in other contexts is arguable, but given the internationalization of higher education, their views are worth considering. The qualitative data obviously cannot be used as a basis for statistical generalizations, but it does enable the Chinese master’s students’ views to be articulated and analyzed.

The analysis of the data is categorized into four themes derived from the analytical procedures described above: clarity and specificity, effectively promote learning, emotions, and power relations. While these themes are discussed separately for the sake of clarity, it is important to note that they are interrelated, with several points relevant across multiple themes.

As Carless (2006) stated, if students are able to engage successfully with teachers’ feedback, they need a starting point from which they can access the discourses and decode the language used in which the feedback is embedded. This theme here demonstrates how the clarity and specificity of evaluative language influences students’ understanding of feedback. Codes such as clear; direct pointing out the limitation; easy to understand; confusion and doubts were examples grouped under this category. The data indicate that some students perceived that they could easily interpret their supervisor’s evaluative language since it was not only clear about the specific limitations of their theses, but also actionable. However, other students stated that they found it difficult to decode their supervisors’ evaluative language reasonably well for two main reasons: lack of detail and ambiguity. Regarding the lack of detail perspective, two students expressed that their supervisors only used evaluative language without any further explanation or guidance, which resulted in the evaluative language being too general, thus causing doubt and confusion in the comprehension process. An example from Student 1 is given below:

He just threw me one sentence in the annotation: “Your data analysis is too superficial and not deep enough”. This seemed very vague and general to me. I wondered what the definition of “deep” is? What part of my data analysis was not deep? How should I go deep? And how could I go deeper? (S1)

In this case, this student emphasized that this caused her extra waiting time to clarify this issue and “it was very inefficient” as she expressed, “things that could have been resolved by talking about it once had to go back and forth several times. And it usually took a week for my supervisor to reply to my email”. In further follow-up questions, when the student was asked what “detail” she was referring to specifically, the student found further guidance, such as explicit explanations of the reasoning behind the evaluation or actionable steps on how to improve, to be more helpful.

Furthermore, some students also felt that the evaluative language used by their supervisors was vague because the words they chose were semantically ambiguous, which could easily lead to misunderstandings and create additional challenges for them in decoding their supervisors’ feedback. An example of this is the adjective “interesting”, which is often used in evaluative language as an “attitude” device to show appreciation in feedback. From students’ responses, it is evident that the term “interesting” elicited a range of reactions among the students, from very positive to ironic. This ambiguity often left students uncertain about its true meaning, particularly due to the lack of conversational context in the written feedback. For example:

He often wrote “Interesting” in the annotation. I found it ambiguous, to be honest. I don’t know if he really thought my thesis was interesting, or he thought I was weird because what I wrote was culturally different from his background. (S3)

Similar sentiments were echoed in comments such as “It’s OK” and “It looks fine”. Student 11 commented, “It confuses me. Sometimes I felt it indicated my thesis was passable, while other times it seemed to show my work was not well written, like a soften critique … because he didn’t praise me like good job or well done”. When asked how these students would respond to such ambiguous evaluations, they collectively expressed a tendency to disregard the feedback. A common reason given was a concern that contacting their supervisor about such a minor issue might make them appear overly meticulous and trivial. This resulted in students simply ignoring or rejecting the feedback. In this case, some students expressed a preference for polarized evaluations. For example, Student 4 stated, “If my writing is good, I want my supervisor to acknowledge it with comments like ‘well done’ or ‘good job’. Conversely, if it’s not good, I would appreciate direct criticism rather than ambiguous comments”.

The previous theme has reported concerns from students about how the clarity and specificity of evaluative language affect their understanding of feedback. This theme expands on these findings by exploring how students use clear evaluative language to facilitate improvement during the thesis writing process. Codes such as self-reflection and identifying the gaps were examples incorporated in this theme.

The data suggests that clear and direct evaluative language in feedback contributes significantly to deep learning, encouraging students to critically reflect on their previous understanding, identify learning gaps and areas for improvement, and actively engage in the feedback loops. This is evident in the case of Student 8, who described her most unforgettable experience of receiving evaluative language in feedback as follows:

His comment on my introduction: “Your introduction section does not get straight to the point. Without presenting data and figures about the current employment condition in China, the argument doesn’t convince me,” made me feel a little bit sad at first. However, it clearly and directly pointed out my limitations and prompted me to reflect on my current writing approach. (S8)

After revising her introduction part and receiving positive feedback – “Your introduction is very well written now. Much better than last time”, Student 8 added, “I feel like I am really participating to make my thesis better and better during the back-and-forth process”. Based on this example, although it has been argued that this directive, teacher-led feedback may be associated with low levels of student self-regulation (García-Jiménez et al., 2015), it is undeniable that such directive evaluative language plays an important role in promoting reflective learning among students. As described by Rogers (2001), the process of reflective learning in this example began when the supervisor challenged the student’s understanding of writing. It continued as the student actively interpreted the feedback and critically reviewed her work, ultimately gaining new insights that led to improved writing techniques. Boud and Molloy (2013) further emphasize that this kind of reflective learning signifies deep learning by encouraging higher-order thinking skills, such as self-reflection and self-evaluation. Consequently, it empowers students to become active participants who effectively decode and apply feedback, thereby closing the feedback loop, continuously improving and enhancing their learning experience. Furthermore, the student added that her supervisor’s evaluations often referred back to her performance in previous drafts. This made her feel that her supervisor was responsible and took her work seriously, potentially building a trusting relationship between her and her supervisor. In line with Carless (2012), such a trustful relationship facilitates the dialogic feedback process, as the student reported that she was more motivated to seek feedback from her supervisor and more active in responding to feedback.

Furthermore, in interviews, four students also emphasized the value they placed on their supervisors’ evaluative language, particularly its alignment with the assessment criteria. Examples of such language include statements such as “You should pay more attention to maintaining links between paragraphs”, highlighting the criteria for coherence and cohesion; “Your literature review is comprehensive, but it tends to list all sources rather than critically analyze them”, focusing on the criteria for a critical literature review. These evaluations increase student engagement by being perceived as objective and providing valuable information. They remind students of the importance of referencing the assessment criteria and help them to identify gaps between their current performance and the standards required for a high-quality thesis. This, in turn, leads to the student’s active engagement in the feedback loop. Furthermore, evaluations linked to assessment criteria also have the potential to “feedforward”, influencing future assignments or tasks rather than simply reflecting on current master’s theses. Although a master’s thesis is often the final piece of work required for a master’s degree, this feedback was seen as beneficial for students going on to doctoral study or working in the same field as their master’s programs, as it encouraged self-regulated learning. For example, Student 4 noted:

Such an evaluation (be careful to maintain links between paragraphs) was also very helpful for my writing at PhD level because it helped me understand what constitutes good coherence and cohesion. Now, when I write articles for my PhD, I always refer back to this comment to make sure that each paragraph is clearly linked to the next. (S4)

In this scenario, evaluative language directed at generic issues is perceived as effective, reinforcing a point proposed by Knight and Yorke (2003) that while teachers often provide detailed and specific feedback, it is the broader, general feedback that is more effective in stimulating learning. This is because such feedback has the potential to feedforward, to future tasks, rather than merely reflecting on completed tasks (Carless, 2006). As argued by García-Jiménez et al. (2015), feedforward plays a facilitative role, allowing students to apply what they have learned to upcoming tasks and scenarios. In this respect, self-regulated learning is likely to occur (Hounsell et al., 2008). This is evident in the case of Student 4 as the student learns to generalize previously received feedback to new tasks and situations. He sets goals (e.g. to ensure coherent transitions between paragraphs), monitors his work (e.g. if I find that the logic between sentences or paragraphs is lacking), adjusts his writing approaches as necessary (e.g. to use transitional phrases or repeat key points to improve it) and evaluates his performance.

However, the use of evaluative language in supervisory feedback was not always beneficial for students. Three students reported that some uses of evaluative language, such as a blanket compliment not related to the actual content of the work, were not perceived as helpful. For example, when Student 1 received her supervisor’s comment “That looks hard-working”, although she was pleased that the supervisor had seen and appreciated her efforts, she still felt that it was not effective because “it didn’t relate to any of my content and I don’t know if my supervisor has really read my paragraphs carefully”. She further added, “I don’t want him to just make empty statements about how good I am or mindless statements about how bad I am. I want something constructive that can really help me improve my current work”.

As students are emotionally involved in the feedback process, the theme here illustrates how students perceive their supervisor’s use of evaluative language from an affective perspective (Van der Kleij and Lippevich, 2021). The data show that students pay particular attention to evaluative language, especially words that convey the attitudinal meanings of the supervisor, such as “well done” and “good job”. Students can easily detect the positive or negative attitudes and emotions embedded in the language through rhetorical devices such as adjectives that carry significant evaluative meanings. In interviews, students generally reported positive emotions when they received positive evaluations. These emotions include happiness from recognition, cheerfulness from improvement, and feelings of encouragement and motivation. The data clearly show that positive emotions benefit students in several ways. Not only do positive emotions motivate students to “work harder to continue on the research path” and to have more confidence to take risks, such as “trying out these new ideas in practice”, but they also improve student–teacher relationships and reduce students’ psychological distress. Students realize that their supervisors recognize the strengths in their theses, which makes them “braver and more willing” to discuss and reveal problems encountered during the thesis writing process. This transparency is an important indicator of students’ trust in their supervisors, as they do not resort to “faking good” (i.e. hiding or ignorant mistakes) and feel comfortable revealing their weaknesses and problems to their supervisors (Carless, 2012).

Surprisingly, when it comes to negative evaluations, the data demonstrate that students generally do not show emotional volatility in response to most negative feedback, as they perceive it as a normal part of academic discussion and the language used by supervisors is typically euphemistic, softening the impact. For example, Student 9 noted, “I think it doesn’t involve very negative aspects. It is normal for someone to question what you have written. His way of assessing is acceptable to me, so I don’t see a problem”. Furthermore, students reported that instead of feeling upset or worried, they were more concerned about whether their supervisors provided guidance or suggestions for improvement after evaluations. For example, Student 1 stated, “When I get negative comments, I focus on what’s wrong and discuss with him how to fix that part. Emotions don’t solve problems.”

However, the data also reveal two particular scenarios that trigger negative emotions in students when they receive negative evaluations. Firstly, students react negatively to evaluative language that uses “disclaim” and “force” to directly negate what the students have written. For example, in sentences like “There is no logic behind your literature review at all”, “This doesn’t make any sense” and “The meaning of this category is very very unclear” reported in interviews, the use of disclaim (no, doesn’t) that directly negate what students have written shows less dialogical spaces and are seen as providing students with facts other than opinions, and the use of force (intensifiers: at all, very very) reinforces the negative tones that the quality of student’s writing is evaluated as not logical and unclear. In this case, such lexical items would directly trigger negative feelings such as anger, anxiety and embarrassment in the students. For example, Student 8 remarked, “He commented on my data analysis part in the annotation, ‘The meaning of this category is very very unclear’. I was overwhelmed and embarrassed and very sad because he used two ‘very’. This made me feel like my interpretation of the category was crap.”

In addition to the sole use of evaluative language, some students argued that their supervisors’ use of punctuation in written feedback was also a vital means of potentially reinforcing its negative tone. For instance, Student 8 noted, “He commented on my data analysis section, ‘This makes no good sense’ with the ellipsis ‘ … ’. This ellipsis made me feel like he was speechless with disappointment. This made me think that the situation was serious, and my data analysis part was a piece of rubbish”. Student 10 also had a similar experience when she received a comment on her methodology section – “??? This doesn’t make sense”. These three question marks made the student feel more anxious and worried as she felt that her supervisor thought her writing was nonsense, where she expressed, “I felt the tone of that sentence with the three question marks was serious. It surprised me and I felt like my heart skipped a beat. It made me really upset and worried”.

Secondly, some students reported experiencing negative emotions when they received direct negations without explanation or further guidance. This led them to perceive their supervisor’s evaluations as subjective and irresponsible. In these cases, negative emotions such as anger and dissatisfaction were triggered not only by the direct negation itself, but also by the lack of explanation or guidance. An example from Student 8 can be seen as follows:

When I saw his comment, “I don’t think there’s any logic behind your literature review at all”, I was honestly a bit angry. This made me feel that he was not as responsible for his evaluation. I actually didn’t think it was very objective because it was his personal opinion … without any basis or reason. (S6)

When students were subsequently asked whether their feelings would change if their supervisors provided explanations and guidance after such direct negations, all agreed that although they might initially feel sad, detailed explanations and guidance would alleviate their negative emotions. They emphasized that their main concern was to understand how to rectify the problems and improve their work.

In sum, the data show that such negative emotions caused by the two occasions of direct negation can hinder students’ active engagement in feedback processes. On the one hand, direct negation with intensifiers and punctuation made students feel psychologically unsafe to approach their supervisors for feedback or clarification. As Student 3 described, “I feel under pressure and lack the courage to approach him. I spend a long time preparing an email, hesitating and thinking carefully about my words. I’m afraid I might ask silly questions.” On the other hand, the perception of subjectivity and lack of seriousness in dealing with their work, caused by the lack of explanation and guidance, can damage the supervisor-student relationship, which further inhibits students’ willingness to actively engage in the feedback process. As expressed by Student 6, “I didn’t want to ask for his feedback”, suggesting a withdrawal from potentially beneficial academic interactions and feedback seeking.

As feedback is often described as a two-way dialogue that engages students in closing the feedback loop, rather than a monologue (Carless, 2016), this theme explores how imbalanced power relations between supervisors and students, manifested through evaluative language, can impede students from being active participants in the feedback process. This theme includes codes such as authoritative researcher and evaluator, obeying the evaluations and devaluing students’ own judgments.

On the one hand, the data show that supervisors’ dual roles as authoritative researchers and final markers led some students to accept their evaluations unconditionally and revise their writing accordingly, rather than forming their own judgments. For example:

Since my supervisor is an expert in the field and has many years of academic experience, I naturally assumed the validity of his assessments. Therefore, I trusted his advice and made the necessary changes wherever he suggested. (S8)

It is important that he is the one who assesses my thesis! I believed that following his evaluations would help me to get a higher grade. (S9)

These examples illustrate how students, influenced by the perceived expertise of their supervisors and extrinsically motivated by the prospect of higher grades, become passive recipients of information, completely guided by their supervisors, rather than actively driving their own learning processes. As García-Jiménez et al. (2015) highlight, such feedback dynamics are problematic as they lead to an over-reliance on teacher guidance rather than fostering students’ abilities to make independent judgments and modifications to their work, all of which are crucial for self-regulated learning.

On the other hand, influenced by the dual role of supervisors, other students reported that the language used in evaluations discouraged them from approaching their supervisors for further discussion or debate, especially when their opinions differed. Contrary to previous literature, which suggests that students often acquiesce to their supervisors’ views and devalue their own judgments (Boud and Molloy, 2013), participants in my study adopted a different, less proactive approach. In order to avoid conflict, they omitted sentences that could lead to confrontation when they disagreed with their supervisor’s evaluations. This behavior is illustrated by the experience of student 10, who recalled a significant incident during the feedback process for her thesis:

I clearly remember receiving a comment in the annotations where he wrote: “I think maybe the data you used to prove the problem is wrong and outdated. The system of Sino-foreign cooperative universities has existed in China for many years and is not, as you say, in its infancy.” I disagreed because all the literature I reviewed did not suggest that China’s Sino-foreign cooperative education had reached a mature stage of development. They argued that it was still at the beginning stage. However, I didn’t dare to argue with him, so I simply deleted the sentence. (S10)

When asked why she hesitated to argue with her supervisor, she explained:

My supervisor is an authoritative expert who has studied this field for many years, so when he told me directly that my use of data was wrong and outdated, I felt that he was informing me rather than inviting me to debate with him. It was a bit tricky for me because I neither agreed with him nor dared to insist on my point of view because he was also the final examiner of my thesis. So, I just deleted that sentence. (S10)

From this example, it is evident that, unlike the previously mentioned students who uncritically accepted and obeyed their supervisor’s evaluations without making their own independent judgments, Student 10 here attempted to interpret the feedback message, internalize it and use it to make a judgment about whether to modify her introduction part. However, the power imbalance between her as a master’s student and her supervisor as an authoritative researcher with the authority to assess meant that the supervisor’s direct negation, described as “wrong and outdated”, was perceived by the student as an authoritative, almost non-negotiable evaluation. This could only be challenged by a student who was particularly strong-willed to argue with the supervisor (Jin and Ruan, 2023). This perception led her to delete the potentially controversial sentence rather than engage in further discussion with her supervisor, thus failing to fulfill the ideal of reciprocal communication between students and teachers as emphasized by Boud and Molloy (2013). Furthermore, previous research has shown that the inclusion of pronouns such as “I think” and hedges such as “perhaps” and “maybe” can soften the tone of evaluative language and introduce uncertainty that invites dialogue (Jin and Ruan, 2023). So I asked the student whether she perceived her supervisor’s use of “I think” or “maybe” as an indication of a subjective stance that was open to challenge. She replied, “Not really, I think he was just being polite. As it was a direct statement and not a question, he didn’t ask me whether I agreed with him or not”. This statement further reinforces the negative effect of direct negation under the imbalanced power relationships in supervision.

Despite these challenges, there were exceptions among the interviewees. Two students felt that their supervisors’ evaluative language did not represent authority. Rather than conforming to their supervisor’s views, they felt confident in defending their own, which they attributed to a sense of ownership and control over their own learning. For example, Student 7 stated:

I don’t think that all my supervisor’s evaluations are always correct. For example, I disagreed with his evaluation, like his criticism that the analytical model I used to analyze big data was not comprehensive, I argued with him and finally convinced him. (S7)

Under the imbalanced power relations, this student’s experience highlights the possibility of autonomy and active engagement in the feedback process, as the student was empowered to drive his own learning forward.

This study explores master’s students’ perceptions of their supervisors’ use of evaluative language in feedback through cognitive, affective and social lenses, revealing important insights into the feedback process and its impact on student learning.

From the cognitive perspective, students value feedback that is specific, directive and actionable. It helps them bridge the gap between current and desired performance while facilitating deep learning, an aspect that has not been thoroughly analyzed in previous research. Such feedback encourages students to reflect on their approaches, develop metacognitive strategies and engage in self-regulated learning. Importantly, this study also highlights that evaluative language aligned with assessment criteria are perceived as effective. It not only supports immediate performance improvement but also serves as a feedforward function, preparing students for future academic or professional endeavors—a finding not previously suggested in the systematic review by Chugh et al. (2022). Such evaluative language use in feedback enables students to “leap out of the loop” (p. 120) and start to engage in self-regulated learning by setting personal learning goals, monitoring their progress and adjusting their strategies accordingly (García-Jiménez et al., 2015). In contrast to previous studies, this study found that students prioritized feedback focused on content quality over effort, reinforcing the idea that feedback should target work rather than personal attributes. Overall, these findings highlight the importance of evaluative language in promoting lifelong learning by developing students’ metacognitive skills.

From the affective perspective, evaluative language in feedback produces a spectrum of emotional responses. In line with Jin and Ruan (2023) and Bastola and Hu (2021b), positive evaluative language increases motivation and confidence and strengthens the student–teacher relationship. Conversely, direct negative evaluations often evoke negative emotions due to their face-threatening nature and potential damage to students’ self-esteem, especially when linguistic devices (e.g. modality, intensity and punctuation) reinforce the tone. A notable finding is that students priorities follow-up guidance over the type of feedback itself; the absence of such guidance not only exacerbates negative emotions but also undermines trust in supervisors. This highlights the need for supervisors to balance critical feedback with supportive explanations in order to maintain a constructive and empathetic supervisory relationship.

Finally, from a social perspective, feedback is seen as a social practice embedded in an unbalanced supervisor–student relationship. In line with Jin and Ruan’s (2023) findings, the dual role of supervisors as both assessors and facilitators of learning (Yang and Carless, 2013) influences how evaluative language is used and perceived by students. Specifically, the authoritative tone of evaluative language can reduce the space for dialogue and hinder students’ active participation in the feedback processes. In this case, consistent with previous studies, students often become passive recipients of information, accepting their supervisor’s perspective, devaluing their own judgments and sometimes even ignoring feedback to avoid conflict and further discussion. Interestingly, this study also found a small group of students who actively disregarded the unbalanced power dynamic, driven by an awareness of the need to take control and ownership of their learning. This finding contrasts with Jin and Ruan (2023) where such behaviors were not prominently observed.

This study contributes to the limited literature on evaluative language in supervisory feedback, particularly in the context of Chinese masters students at a Sino-foreign university. By addressing the cognitive, affective and social dimensions of feedback, the findings underscore the critical role of evaluative language in shaping students’ learning, emotional well-being and agency. The study also highlights unexplored aspects, such as the nuanced emotional impact of linguistic devices and the importance of follow-up guidance, offering new insights into effective feedback practices. These contributions are important for improving the feedback process and promoting student autonomy in academic settings.

Based on the findings, it is evident that how supervisors communicate evaluations can be a significant barrier to students becoming active agents in the feedback process. In this section, I suggest two strategies for supervisors to improve their use of evaluative language when providing feedback during the thesis writing process. Firstly, as discussed earlier, the language used in supervisor feedback can elicit a spectrum of emotional responses in students. It may be beneficial to avoid overly harsh expressions of negative affect, such as direct negative judgments about students’ social esteem. Such overly critical evaluations are likely to alienate students by threatening their faces and demoralizing them (Bastola and Hu, 2021b). However, this does not mean that supervisors should avoid using direct negation or overly soften their criticism. Indeed, critical feedback can be highly effective, particularly for overconfident students who may need direct criticism to recognize their shortcomings (Yang and Carless, 2013). As supervisory feedback should be “characterized by an ethos of care” (Sutton, 2012, p. 39), the key here is for supervisors to build trust with students by showing empathy with students’ emotions, bridging the social distance with students sensitively, and valuing and respecting students’ perspectives. Such actions are likely to increase students’ trust in their supervisors (Carless, 2012). Building this trust is crucial for fostering a dialogic environment where students feel comfortable and open to discussing their challenges and problems encountered rather than faking good, as well as seeking guidance during the thesis writing process. From my findings, I have found that an effective way for supervisors to build trust is to ensure that each evaluation is accompanied by thorough explanations or guidance. This not only helps students to use feedback constructively but also portrays supervisors as responsible and trustworthy, encouraging students to invest their trust in the supervisory relationship.

Secondly, it is better for supervisors to use evaluative language that encourages dialogue when providing feedback. The findings show that most students felt that their supervisors’ evaluative language conveyed an authoritative voice, which they perceived as delivering a definitive truth or fact, thus limiting the space for dialogue. Given that feedback is inherently a two-way process—not just a one-way transfer of information—it is vital that students are actively engaged in dialogue with their supervisors. To facilitate this dialogue, supervisors can effectively use questions as evaluative tools in their feedback. According to Jin and Ruan (2023) and Thompson (2013), questions are an integral part of the “expand” system of evaluative language and serve as important dialogical devices that explicitly invite student participation and help to create a communicative atmosphere between students and teachers. For example, instead of directly negating a student’s argument, supervisors might ask open-ended questions such as, “Why do you think … ?” This approach not only expands the dialogue space for students to express their views, but also bridges the social distance between supervisors and students. By using questions, supervisors demonstrate their respect and empathy for students and soften the challenging tone (Carless, 2012). Such strategies not only encourage students to actively engage in the dialogical processes but also are beneficial to build trust between supervisors and students.

However, as feedback is referred to as a two-way process (Carless, 2006), it is also vital to provide suggestions for students in order to facilitate their roles as active agents in the feedback process. In this case, student feedback literacy established by Carless and Boud (2018) should be taken into account. Building on the concept of student feedback literacy, students’ first step when receiving evaluative comments should be to appreciate this feedback. This appreciation involves recognizing that the purpose of evaluative comments is to improve their work and understanding their active role in this process, which goes beyond simply receiving information to decoding, processing and using these evaluations effectively. Furthermore, managing emotional responses to feedback is also essential, as noted by Carless and Boud (2018). When receiving direct criticisms or negative evaluations, students are advised to maintain emotional equilibrium and avoid becoming defensive. Finally, students need to recognize the need to act on the feedback they receive. As emphasized by Boud and Molloy (2013), active engagement in processing feedback and applying it to subsequent work is essential to effectively closing the feedback loop.

This study investigated Chinese master’s students’ perception of their supervisors’ use of evaluative language in feedback at a Sino-foreign university in mainland China. As a small-scale qualitative research project, the interview-derived data underscore the significance of this study: as few studies have explored how master’s students perceive their supervisors’ use of evaluative language in feedback, this study attempted to address this gap. In future research, studies could explore how linguistic devices in evaluative language affect students’ emotional responses and perceptions of feedback. In addition, studies could also investigate strategies to empower students to challenge power imbalances in the feedback process and become more active participants. Extending this research to diverse cultural and institutional contexts would also provide a more comprehensive understanding of feedback practices across higher education.

I would like to express my gratitude to Professor Hugo Horta, Professor Jisun Jung, Professor David Carless and the anonymous reviewers for their valuable feedback in improving this manuscript.