Understanding doctoral students’ needs for thesis discussion writing and supervisory curriculum development: a sociocultural theory perspective

ABSTRACT

The global blossoming of doctoral education calls for a better understanding of doctoral candidates’ needs concerning thesis writing. Although scholars proclaim the particular difficulty of writing a thesis discussion, empirical studies on doctoral students’ needs for the discussion chapter are insufficient. To fill the gap, using individual interviews, we investigated seven candidates’ views on challenges and helpful resources, noting L1/L2 students’ differences regarding writing the discussion. We found that students perceived challenges differently: challenges (or absence of challenge) are connected with individual situations. We also found that students relied mainly on supervisors to help them with thesis discussion writing and yet supervisors sometimes were not able to provide sufficient guidance. Lastly, we found L1 candidates had ideologically related reasons for writing choices whereas L2 candidates’ reasons were to fit in. Another L1/L2 difference is that L2 students reported using more resources for language difficulty than L1 students did. We draw on sociocultural theory to make sense of these findings to offer insights into pedagogy. We conclude the paper with a discussion on the implications for supervisors, writing instructors and students.

KEYWORDS:

• ESP/EAP
• needs analysis
• doctoral thesis discussion writing
• supervisory curriculum
• sociocultural theory

Introduction

In the context of internationalised higher education and knowledge economy, an increasing number of candidates are undertaking doctoral research (Bitchener, 2018). Researchers have discussed the difficulties of doing a doctoral study through the medium of formal academic English, especially for international students whose first language is another language (Manathunga, 2014; Zhang, 2022). When it comes to writing a thesis discussion, scholars proclaim the particular difficulty: It must showcase the author’s understanding of research expectations, putting own findings in the context of other studies and explaining what they mean and why they matter (Geng & Wharton, 2016; Paltridge & Starfield, 2020; Shen, 2020), which often poses challenges to student writers. The linguistic challenges facing English as a second language (L2)¹ writers (Bitchener & Basturkmen, 2006; Deng, 2009) make it even harder.

Given that the number of students enrolled in doctoral programmes is increasing, ‘doctoral students are in desperate need of help with their dissertations’ (Paré et al., 2009, p. 183), and a thesis discussion is particularly hard to write, more studies on doctoral students’ needs for the discussion chapter are needed. We interviewed both L1 and L2 doctoral students’ views of challenges of thesis discussion writing and addressed this need. We investigated what resources participants used for support and what differs between L1/L2 students too. We further developed the interpretation of students’ needs in alignment with sociocultural learning theory (Vygotsky, 1978) – that alignment gives direction for improved practice. The study adds to two earlier studies on postgraduates’ views of thesis discussion writing, one of which examined masters (Bitchener & Basturkmen, 2006) and the other doctoral students (Deng, 2009), with both focused on L2 students’ challenges. To enrich the understanding of students' needs for the discussion writing, our two research questions are:

1. What do doctoral students perceive as challenges concerning writing the thesis discussion? What are L1/L2 students’ differences, if any, relating to challenges?

2. What are the resources used by doctoral students with their thesis discussion writing? What are L1/L2 students’ differences, if any, relating to the use of resources?

Theoretical framework

The theoretical frameworks underpinning this study are English for Specific Purposes (ESP) needs analysis and sociocultural theory. The literature has documented the foundational role of needs analysis in ESP and the core role of the learner in needs analysis (Basturkmen, 2005; Brown, 2016). Brown (2016) puts this: ‘Notice that ESP is fundamentally linked to “the specific needs of a particular group of learners,” or put another way, if there is no needs analysis, there is no ESP’ (p. 5). Basturkmen (2005) notes that ‘A key feature of ESP course design is that the syllabus is based on an analysis of the needs of the students’ (pp. 17–18). In their definition of ESP, Dudley-Evans and St John (1998) claim that the need ‘to meet specific needs of the learner’ is an absolute characteristic of ESP (p. 4). Given the setting of this study is ESP, the fundamental characteristic of ESP – learner needs – drove us to examine doctoral students’ perspective of discussion writing.

To define what we mean by needs and needs analysis, Hutchinson and Waters (1987) notion of ‘learning situation’ is relevant, referring to learners’ difficult situations (challenges), and assistance situations in this study. Our use of learner needs is not limited to Hutchinson and Waters (1987) original definition that learning needs is ‘what the learner needs to do in order to learn’ (p. 54). We established learner needs, by examining two aspects: Challenges and use of resources.

Sociocultural theory views human mental activity as essentially socioculturally constructed and human mental processes as social in origin. This has been particularly illustrated by the concept of zone of proximal development (ZPD, Vygotsky, 1978): A fertile learning space exists between a learner’s actual development level and potential development level, where achieving potential is conditioned by the support from others. Influenced by ZPD, many researchers study the mechanisms of effective support. Silalahi (2019) itemises factors that should be present within the ZPD: Assistance, mediation, cooperation, imitation, target and crises. Some other researchers (Aljaafreh & Lantolf, 1994; Wood et al., 1978) propose contingency, dialogue and intersubjectivity as effective learning support. The idea of social sources of cognitive development and the concept of ZPD match our experience that social support is indispensable to thesis discussion writing, which involves higher cognitive performance (Badenhorst, 2023; Geng & Wharton, 2016) and awareness of discourse community expectations (Carter et al., 2020). Doctoral candidates’ thesis writing is a social negotiation as they seek to establish themselves as credible researchers, and they undertake this task within a web of support and critique from others. This assumption drove our investigation of students’ use of resources in writing thesis discussions. Further, the mechanisms of effective support – contingency, dialogue and intersubjectivity – provide insights to interpreting and explaining findings of students’ needs, particularly, students’ use of resources. Sociocultural theory anchors our empirically based implications on the theoretical ground.

Thesis writing challenges

Numerous studies discuss the difficulties of doing postgraduate research through the medium of English, especially for international students whose first language is another language (i.e. Aitchison et al., 2012; Badenhorst & Guerin, 2016). These difficulties include mental and cultural challenges beyond the text and linguistic challenges in producing the text. Our study focuses on linguistic challenges. Of the researchers studying on linguistic challenges, some highlight the communication problems on the macrolevel of audience, citation and overall structure (Allison et al., 1998; Angélil-Carter, 2000; Pecorari, 2006; Petrić, 2007). Some note local problems of grammar, tense and language structure (Bunton, 1998; Sheridan, 2011; Strauss, 2012).

Bitchener and Basturkmen (2006) observed a particular problem in thesis discussion writing. Their interview of four pairs of postgraduates and supervisors shows that students often have limited understanding of the function of the thesis discussion. They also find that students tend to give more weight to their own interpretations rather than using other academic texts to contextualise claims. Paltridge and Starfield (2020) comment on this problem that ‘students are often not aware of the need, in the discussion section, to show the relationship between the results of their study and the results of similar studies and related arguments in the published literature’ (p. 175). Deng (2009) examines discussion texts written by six L2 Chinese social science doctoral students and identifies students’ rhetorical mismatch between what is promised in the introduction and what is delivered in the discussion and conclusion.

The above studies are meaningful in that they located thesis writing challenges and increased instructors' and supervisors’ attention to them. However, the diversity and complexity of learners implies more studies are needed for a full understanding of postgraduates’ challenges. These two studies focus only on weaknesses and challenges – both studies interpret needs as challenges – and give little attention to other aspects relating to students’ needs, such as what resources participants used for support and what differs between L1 and L2 students’ experiences. Further, neither of these two studies is able to develop the interpretation of students’ needs in alignment with sociocultural theory (Vygotsky, 1978) to give direction for improved practice.

Methods

Rationale of interviews

This study takes a constructivism perspective and case study approach to understanding students’ needs for thesis discussion writing. In line with this paradigm, individual interviews were used because they could provide depth and detail of students’ needs (our interview questions are provided in supplementary material A). Observations (Angrosino & Rosenberg, 2011) were not very practicable because doctoral writing occurs outside the classroom, usually in private. An audience may be disruptive. The focus of group discussions is on the group rather than the individual (Galloway, 2020). Similarly, think aloud was not suitable for this study because it is a method appropriate for enquiring into cognitive processes, such as memory and noticing (Zhang & Zhang, 2020). Despite the personal accounts of diaries, diary methods seem to be classroom based (Rose et al., 2020). It was impractical to ask doctoral students to write journals for our research purpose. Given that our purpose was to obtain insider accounts and the setting was not classroom based, an interview seemed to be the best option to capture participants’ attitudes, opinions and experiences (McKay, 1990; Rose et al., 2020). Our practical consideration was that the Covid-19 pandemic and lockdowns meant that we would have had difficulty in accessing participants if we had chosen observation, focus group discussions, or think-aloud.

Purposeful sampling and participants

Our research purpose is to investigate doctoral students’ needs for discussion chapter writing and enrich the understanding of the topic. Interview sampling purposively focuses on individuals who could fit for the purpose, and best inform us about the research questions. We chose doctoral participants who had experience in discussion writing, targeting doctoral students who were then writing a discussion or had recently finished their discussion section. The participants came from language education and bioengineering disciplines. The reason was that this study is one of the three studies of the first author’s PhD project. The participants came from these two disciplines in the other two studies. We chose the same two disciplines for consistency. Lastly, both L1 and L2 students participated in the study because they would enable us to consider L1/L2 differences, long a topic of interest in second language learning and relevant to support doctoral writing.

Finally, we recruited seven doctoral students as participants. Of the seven participants, one chose face-to-face interviews and others chose online interviews. Four participants came from Language Education (two L1 and two L2). Three were from Bioengineering (one L1 and two L2). Here they are labelled P1, P2, P3, etc. to mask their identity (see Table 1).

Table 1. Interview participants’ profile.

Participants	Language background	Discipline
P1	L2	Bioengineering
P2	L2	Bioengineering
P3	L2	Language Education
P4	L2	Language Education
P5	L1	Bioengineering
P6	L1	Language Education
P7	L1	Language Education

Data collection and analysis

The interviews were conducted by the first author. One interview was conducted with each participant. Each interview took 40–50 min. Depending on participant’s choice, either English or Chinese was used during the interview. The first author recorded and transcribed the interviews and used the structural coding strategy (coding the data according to research questions (Miles et al., 2014; Saldaña, 2009)) to analyse interview data. While coding was a recursive and iterative process in practise, we described it in a linear way in supplementary material B for an easier understanding. Member checking was used to ensure the overall trustworthiness of the analysis. We emailed participants transcripts so that they could check any mistakes. We also shared the major findings with participants so that they could assess whether the interpretation accurately represented them (see supplementary materials C for member checking records). The discussion and agreement among us three authors also verify the analysis. Peer debriefing and reflexivity were employed to guarantee the quality of the analysis too.

Results and discussion

Students’ perceived challenges

A variety of challenges were reported. The challenges column in Table 2 holds aspects that candidates thought were needed in a discussion, but could not see how to produce them.

Table 2. PhD students’ perceptions concerning difficulties in writing the discussion.

Participant ID	Challenges
P1	Deep level commentary on results
P2	Understanding of the discussion’s expectations, high level of literacy, sufficient commentary on results, evidence of critical thinking
P3	Deep-level interpretation, structure, connections between the discussion and literature review, integration of quantitative and qualitative findings
P4	Moving on from the results and showing reflection on the results
P5	Structuring
P6	Making claims appropriately, regulating for objectivity while presenting findings
P7	No perceived challenges

Some of these challenges are fairly common in our experience, for example, demonstrating a high level of literacy and deep level of understanding while contextualising own work within the context of the field. Our findings here demonstrate the high ‘cognitive demands of the discussion’ (Bitchener & Basturkmen, 2006, p. 7). Particularly, the difficulty of commenting on results deeply, of making claims appropriately, and of structuring, confirms that dauntingly high cognitive and meta-cognitive abilities and skills are needed to write a thesis discussion (Geng & Wharton, 2016).

P1, P2 and P3 had the same challenge of commenting on results sufficiently and deeply. However, their reasons were different. P1 connected this challenge to his prior knowledge background and the target journal’s requirement,

I am not confident in using the terms because of my electronic engineering background … and for the Physiology [focused] journal, you need to express some idea, some hypothesis, to explain why this kind of thing happened. This is the most difficult part.

P2 referred to being challenged because of different academic cultural influences. He commented that he was trained under an Asian educational system which is characterised as ‘learn by heart’. In P2’s view, this educational background caused him to lack the critical thinking skills needed for discussion writing. P3 referred to the difficulty of integrating quantitative and qualitative findings in the discussion, an instantiate of challenge of commenting on results sufficiently and deeply, ‘as it is mixed-methods, and there was a big challenge methodologically. I needed to integrate to make it really mixed-methods … that was [what] the discussion chapter [was] for’.

Two participants added other points for consideration. P6 (L1)’s challenge seems to be specifically connected to his tremendous enthusiasm for his research topic.

I think the only difficulty was ensuring that I did not make unsubstantiated claims. I’m really passionate about my research topic … . staying within the boundaries of what the data can actually imply, I had to go back and … check my enthusiasm as I wrote the discussion … I think that you have to regulate for objectivity as you write, which can be challenging. (P6)

P4 (L2) reported that she was stuck in the results for one year because her research topic demanded extensive statistics knowledge which was outside of her specialised knowledge. She worried about her results which affected her discussion writing.

I didn’t write the discussion but kept reflecting on the results … . The thoughts were different every day … I suffered a lot from the analysis and the process of thinking for preparing the discussion. (P4)

Then one participant (L1) explained why she did not report any specific difficulty in writing the discussion: Good planning, great supervisors, and that her research was grounded in her teaching practice were related to P7’s ease in writing the discussion.

In summary, some challenges were identified in this study. While some are specific and haven’t been reported before (problem of integrating quantitative and qualitative findings, difficulty in connecting the literature section to the discussion section), they demonstrate the cognitive challenges of discussion writing and echo previous studies (Bitchener & Basturkmen, 2006; Geng & Wharton, 2016). Beyond the specific challenges, we observed the relevance and impact of the broader social-cultural factors on challenges. We found that challenges (or absence of challenge) are connected with individual situations, or ‘socio-contextual elements’ (Al Hilali & McKinley, 2021, p. 86), such as research topic, research method, thesis structure, planning, relationship with supervisors, enthusiasm and prior cultural and knowledge background. This added to the literature on what is known about students' challenges in thesis discussion writing.

L1/L2 students’ differences relating to challenges

We found two differences between L1 and L2 students relating to challenges. There are differences in the reason for using or not using the first person pronouns and active voice, and differences due to the multiple languages involved in L2 thinking. We report these differences under the challenge theme because first, no participant made the decision easily on active or passive voice, and second, the interference of the first language reported by L2 participants below means a language challenge distinguishes L2 from L1 students.

Different reasons for using or not using first-person pronouns and active voice

Our data showed that L1 students had ideological-related reasons for using the first person pronoun and active voice, and L2 students’ reason was to fit in – they followed what other community members did or recommended (We hypothesised before the interview began that L2 students would use more the first person pronouns than native speaker students. This hypothesis was developed based on a previous study’s finding that writers who are learning English tend to use more personal metadiscourse, including the first person pronouns, than native speakers (Adel, 2001)).

L1 participants showed reflective engagement with active voice. P5 (L1, Bioengineering) explained that he knew that most scientists love using passive voice, but he personally disliked using passive tense, and thought it added confusion,

Say, you have a result, if you say something like that, we found that A causes B, it’s very clear that you are the one found that. If you say, it was found that A causes B, you could have found it yourself. But, you could be mentioning something that in the past, and it just makes things confusing. I think it’s completely unnecessary. (P5)

P5, on passive voice, noted ‘I think it’s a way to distant yourself from your research which I don’t like. You have to make it clear that this is what you think’. This comment indicated that P5’s style preference was connected to both his valuing of clarity and to his desire to voice his attitude. P5 was sure of his own judgement, resisting authority, ‘She [his supervisor] tried to remove [active voice], but I am quite stubborn’.

P6 (L1, Language Education) expressed his epistemological awareness when making decisions on the use of the first person pronouns and active/passive voices,

I remember trying to use the passive voice to sound more academic, but I was restructuring [earlier drafts]. I remember restructuring sentences to use the passive voice and to sound very academic, which in ethnographic research is very strange. So, like, disembodying myself from the research felt very, hmm, you know, hmm, almost like I was godlike, so I didn’t really like how that sounded. (P6)

In P6’s field, the passive voice that P6 saw as academic was inappropriate in the context of ethnography, and needed to be changed. This choice demonstrated P6’s critical thinking on a language choice that demonstrates epistemological integrity.

In comparison, L2 students who are working in an unfamiliar culture often look to follow how other academics do things. The following comments capture L2 care with fitting into how others work when it comes to the use or avoidance of the first person pronouns. P4 (L2) turned to exemplars, other theses, to copy the language used in them.

I used the first person pronouns because the previous thesis authors used the first person pronouns. I followed them. I had many doctoral theses in my hand and I would consult them when I had uncertainties for thesis writing. For describing the structure of the discussion chapter, I found that these authors used the first person pronouns. I just followed them. They were my supervisor’s students and had already graduated. My supervisor did not raise any issue on this matter and I saw it as approval. (P4)

P3 (L2) expressed compliance with supervisors overtly: ‘One of my supervisors doesn’t like the first person pronouns. She really likes the passive. So it was like I couldn’t say “I”, “we” or anything like that’. He added,

To be honest, I don’t really have a super big opinion yet how my academic voice should sound like, because I also found doing PhD is very stressful. I found that I always try to do things the way that other people want to do them, you know, always try to please. You know when your supervisor say “don’t do like this”, I just follow instruction. (P3)

P3’s choice may be explained as simply ‘meeting the immediate needs’ (McKinley, 2015, p. 191) of his sociocultural situation without having to slow down and absorb all the rhetorical reasons for choice.

The above result showed that L1 participants trusted their own sense of identity and writing style preference, and the strength of their convictions on the subject. It also suggested L1 students’ confident critical relation (Shor, 1992) to writing and knowledge. By contrast, L2 students were holding back with respect to critical thinking and resistance while relying on learning from following others. The finding would seem an example where L2 students are at a disadvantage in making critical decisions confidently. On the other hand, our finding indicated that L2 students’ use or avoidance of the first person pronoun and active voice is constrained by the awareness that, as non-native speakers, they do well to follow what seems to be standard usage. The finding supports the argument that ideology and inequality are unavoidable in L2 education (Benesch, 2001; Pennycook, 1997).

L2 students' first language interfere

The second difference between L1 and L2 is that more than one language is involved in the thinking behind L2 students’ discussion writing. Generally, L2 students reported the negative influence of their first language on discussion writing in terms of vocabulary, grammar and sentence structure. As P2 said, ‘It has negative impact to a certain extent. Like especially the sentence, grammar and vocabulary’. P1 commented:

My first language is Persian, because you know we are thinking in Persian or in our mother language but we are writing in English here. Then when you are writing in English, sometimes … it doesn’t make sense to the native speaker … and then you try to make some changes and you make it worse. (P1)

P3 also noted the negative influence of his first language on discussion writing in terms of sentence structure, as shown in the episode below.

My first language was kind of an obstacle in that sense. My English is too much Spanish-like. For instance, I’m used to writing long sentences. And some of the sentence components that in English usually go at the end, I tend to put at the beginning. … Like complements in English, time or space, that should go at the end, I put that at the beginning. (P3)

Previous studies reported that L2 students have more problems than L1 students (i.e. Casanave & Hubbard, 1992; Dong, 1998). Our study adds to these studies providing evidence for more challenges relating to language among L2 students.

Resources used for support and (in)effective support

Candidates were asked about the assistance they used for discussion writing. The main sources of help were supervisors and texts, particularly theses and research articles that could be used as exemplars. Of seven participants, six reported supervisors and six reported theses and articles – these were found to be the top two sources of help.

Supervisors

P1 (L2, Bioengineering) stated the importance of his supervisor’s help in finding references, which helped build his confidence when he was writing the discussion. He also mentioned practical help from another two co-supervisors with rephrasing and content selection. P3 (L2, Language Education) emphasised the insight he obtained from his supervisor’s advice on how to move beyond reiteration of findings. P4 (L2, Language Education) commented that her comparison of her findings with previous studies was improved by her supervisor’s explicit feedback on how to organise the discussion to meet this rhetorical purpose. P6 (L1, Language Education) included that his supervisors’ critical comments on his unsupported claims enabled him to produce a more defensible discussion. P7 (L1, Language Education) spoke highly of her supervisor’s thorough and attentive support for her thesis discussion writing,

… suggestions related to content and the order of content, suggestions related to clarity of meaning and choice of words, suggestions related to research integrity, so things like “hedge your statements” … the smallest group was related to style and formatting, so “this doesn’t need speech marks”, “please don’t end sentences with a possessive apostrophe” … and then there were three points of encouragement … (P7)

On the other hand, doctoral students reported some unhelpful advice that was more of an obstacle than a support. P2 (L2, Bioengineering) stated that his supervisor’s general and implicit comment ‘Your discussion is not enough. Try to improve’ made discussion writing more challenging. He expressed his struggles and elaborated his expectations of supervisor’s support.

I don’t know what to improve if I have already tried my best. What I needed to know [was] a little bit guidance like, okay, you need to include this this and this, you need to explain this idea more, something like that. But sometimes just tell[ing] me, you need to improve this paragraph, and that’s it. I don’t know how to improve … that’s a kind of difficulty for me. (P2)

The comments P2 made suggest that he and the supervisor had different understandings of ‘enough’. It seems that consideration of what is enough according to the supervisor is in an abstract sense while the student needed enough to be explained in a concrete sense. As Vygotsky (1934/2012) comments, there may be different frameworks in which the child–apprentice and adult–expert think of the same object: ‘The child’s framework is purely situational, with the word tied to something concrete, whereas the adult’s framework is conceptual’ (p. 142). This phenomenon confirms Wertsch’s (1985) argument that the participants often bring different perspectives of the situation to the interaction. Further, our data suggest that the implicit feedback provided by the supervisor seems not to be the most appropriate form of mediation to create intersubjectivity in P2’s case.

Exemplar texts

The usefulness of text exemplars was observed fairly frequently. Some participants referred to theses and research articles for learning how to frame their discussion (‘I use them as my main template’). Texts were also useful in helping students familiarise themselves with typical discussion expectations and organise and select contents of the discussion (‘It just gives me an idea of what’s expected. Do people use headings? How much do they put in limitations of the study?”, I looked at other students’ theses and paid attention to how research papers in my field with similar data types for structuring theirs).

Peers and domain academics

Some participants acknowledged practical support from peers. P2 was comforted and encouraged by colleagues’ and friends’ words and settled gradually through asking senior candidates how to write the discussion and talking about his difficulties with them. P7 valued colleagues’ conversations and stated the reciprocal benefits of a healthy social relationship with peers: ‘the lovely colleagues, we always talk and bounce ideas off’. P3 explained that peers’ questions led him to think deeply and thoroughly about the reasons for his findings, contributing to his logic and argument development in his discussion. He said that the arguments developed in the discussion had not been possible without peers’ help because ‘supervisors are so busy and [have] so many students’. Lastly, P6 mentioned the help he obtained from domain experts and academics in his department who were not his supervisors.

To sum up, the above findings endorse individual’s higher levels of cognitive development are dependent on social interactions. Our sociocultural approach led us to notice that talk between people is one major resource and engagement with what expert others have written showing the discipline culture is the other. The two facets of ‘sociocultural’, social and cultural, underpin how learning to write as a fully accepted researcher is learned.

Online resources

Useful online resources were also reported. Many participants used various websites and YouTube videos to learn the purpose of the discussion and the principles of the discussion (‘discussion needs to present new knowledge’, ‘discuss equals debate’, ‘discussion chapter should be from general and known to specific and new, back to general again’). Two L1 and one L2 mentioned that they used academic blogs on writing. Most L2 students used Google for help with vocabulary and one L1 reported the Thesaurus as helpful with the word choices.

Writing courses

One L2 student from Bioengineering (P1) reported the effective help he obtained from a genre-based writing course, which was exclusively for L2 students who had not passed an English language diagnostic test that the University of Auckland runs for all doctoral students. He explained that because of this course, he was clearer about how to organise and arrange the elements of specific genres than before (in his master’s study).

On the other hand, P2 (L2, Bioengineering) shared his negative experiences of attending a writing group. His account showed the importance of timeliness in support, ‘unless you start writing discussion, you do not pay much attention. At that time, I hadn’t started my thesis writing. I just went there and sat there and talked about other stuff, and was not really involved’. This comment raises the question ‘What students should I include for my writing group to be maximally effective?’, the issue of targeting participants for a writing group (Aitchison, 2009)

Similarly, P4 (L2, Language Education) reported a negative experience of attending a writing course in her department. She explained that the instructor introduced many possibilities and provided many options, but she did not quite understand what the instructor wanted her to do and did not know what to do herself. She ended up lost when faced with a massive amount of information. P4’s comments suggested supportive programmes should consider student’s disposition and avoid overload. This finding, collaborating with the dialogic negotiation principle proposed by Aljaafreh and Lantolf (1994), highlights the importance of a dialogic activity in order to assess the needs of the novice and adapt help to support growth and development within the ZPD.

Our findings about sources of support give further evidence that supervisors and workshops could be catalysts for positive cognitive change or might not be (Carter & Laurs, 2014; Wei et al., 2019). It seems helpful for supervisors and learning advisors to see that some support exerted little positive influence and even exerted a negative influence on students’ discussion writing when out of time with the writer, or too off-puttingly dense. This aligns with Vygotsky’s (1934/2012) discussion of the relationship between instruction and development – the curve of learner development does not always coincide with the curve of instruction. Our findings endorse some principles of effective support in the literature (contingency, a dialogic activity and intersubjectivity, Aljaafreh & Lantolf, 1994; Wood et al., 1978). Further, our findings imply that the ideal time to provide support for discussion writing is when the student is at the stage of writing the discussion, engaging fully in the task. As suggested by sociocultural researchers (e.g. Wertsch, 1985; Wood, 2005), constructive interaction within ZPD is preconditioned by active participation – candidates need to engage.

L1/L2 students’ differences relating to resource use

The data showed that while L1 students seldom used the dictionary, L2 students used it quite often. Apart from that, L2 students used more resources for learning phrases and collocations. P1 (L2 student) commented that checking for accuracy is essential to work as ‘you need to Google it and you need to see if this phrase is correct or not. Or sometimes you don’t know that this collocation has a different meaning in English’. P2 (L2 student) added that work was needed to get a wide enough vocabulary, ‘I have to search on Google a lot. Because if I write naturally, I don’t have many words’. P3 (L2 student) reported more resources used for learning phrases and collocations apart from the dictionary.

I used the Academic Phrasebank, which is very useful … I also used COCA, which is the corpus of contemporary American English and the corpus for the British English. When I had some doubts on collocations, I wrote the two words together in the COCA or the British corpus and if it was a frequent expression, I used it; if it was not used frequently, I didn’t use it because it was probably wrong. At some points, I used Ngram. in Google. I use Ngrams, which is also related with collocations. (P3)

In comparison, L1 students’ use of resources for language aspects is limited. As L1, P7 reported when asked if she used a dictionary:

A dictionary? No, you know, being a native speaker, I would know the words and if it sounded right. Intuitively I know these words. I would probably use Thesaurus only, so that I’m not using the same word all the time. (P7)

This finding echoes our finding in ‘L2 students’ first language interfere’; : L2 students’ first language has a negative influence on their thesis discussion writing, indicating that L2 students have more challenges relating to language than L1 students.

Conclusions and implications

Our study hones in on doctoral discussion writers, their challenges and resources, and the differences between L1 and L2. We acknowledge our limited data sources: Only two disciplines, one institution and limited participants were involved, whereas doctoral education is context-specific, as Ferris (1998) notes. However, our study adds to the literature on students’ thesis writing challenges by noting both what will be familiar to readers and also the relevance of individual situational variations. It provides empirical evidence as to the important roles of supervisors and textual exemplars in thesis writing. It confirms that instructions outside of the learner’s ZPD and untimely advice exert little positive influence or even some negative influence on students’ writing. Our study also adds to second/foreign language studies by evincing the ideology issues in academic writing with a small amount of empirical evidence of this. Finally, we propose implications based on the findings yielded in this study.

Curriculum implications for writing instruction

This study was conducted in a university in which learning services have been reduced. Within this unfavourable context, our interview data indicate that misalignment exists between the doctoral writing programmes provided and students’ needs. Given students’ challenges, views about (in)effective help, and L1/L2 students’ differences identified in this study, writing programme designers and instructors might ask: What contents and topics should be contained in the course? What materials should be used? How can we best run the course? Should the course be exactly the same for L1 and L2 students? Should L2 students have additional courses tailored for their needs? Discussion of these questions is necessary for a workable course design (Brown, 2016) and for the efficiency of ESP programmes. The following conclusions that emerge from this study may be helpful in addressing the above questions.

Necessity of a specialised corpus of thesis discussions

Considering the high use of exemplar texts among our participants, a specialised corpus of thesis discussions would profit from teaching thesis writing. Flowerdew (2017) values a specialised corpus in teaching academic writing, saying ‘the more specific the better … an academic corpus is better than a general one; a discipline-specific corpus is better than a broader one; and a genre-specific corpus is better than one covering various genres’ (p. 105). Such a corpus of exemplars would enable students to examine which rhetorical purposes are preferred in their given context and foster students’ autonomy and noticing strategies.

Addressing cultural inequality and differences

This study gives evidence of what most supervisors know, namely, although L2 doctoral candidates invest much more time and effort, yet they remain at a disadvantage when writing a thesis. It also captures cultural differences in using or avoiding first-person pronouns and active/passive voice. The thesis instruction curriculum should include overtly explaining reasons for textual choices if there’s a chance that might not be clear.

Implications for supervisory curriculum development

This study confirms that doctoral students rely mainly on supervisors to help them with thesis discussion writing and that supervisors sometimes cannot provide sufficient guidance (Aitchison et al., 2012; Wei et al., 2019). This endorses the huge value of good supervisory work. Supervisors are advised to apply the principles of effective instruction (contingency, dialogue and intersubjectivity) – our data capture the importance of this pedagogical theory. Findings show that rather general feedback (e.g. ‘this needs fixing’) cannot enable students to understand deep level and abstract comments quickly. The implication is that explicit advice is especially essential for L2 writers (Carter & Laurs, 2014; Yeh, 2010).

Secondly, supervisors can expect some discrepancies when instructing students to write or revise the discussion. As our data have shown, it cannot be assumed that students will do what supervisors suggest. Supervisors should accept that sometimes when students disagree with their advice or resist it, that is a sign of their increasing development as an independent researcher (Carter & Kumar, 2016). We suggest that finding ways to be explicit about the demonstration of critical analysis in text is worthwhile, and ethically important as increasing numbers of doctoral students nowadays write across cultures and have to learn to negotiate difference and power.

Implications for students

Students should cultivate as many allies as possible amongst peers, learning advisors and faculty members. This study has shown a central idea of sociocultural theory, that learning has origins in social resources and interactions. The implication is that doctoral students can create opportunities to discuss with supervisors, talk to peers, friends or learning advisors. They may also search library databases, consult completed theses as exemplars and draw advice from digital media (Nguyen, 2022). With effective use of resources, they may grow metacognitively as they learn to write the discussion effectively.

Acknowledgements

We thank the China Scholarship Council and the University of Auckland for their generous financial support to this study in the form of a joint doctoral scholarship awarded to the first author.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by China Scholarship Council; University of Auckland [grant number 201508130058].

Notes

1 L2 students in the two studies cited refer to students for whom English is their foreign language or second language other than their mother tongue. To align with these studies, we take the term L2 to refer to second, third or fourth language users of English. That is, English is not their mother tongue.