“Everything Depends on Yourself”: Freedom, Hidden Competition and Student Self-Management in Sino-British Transnational Higher Education

Abstract

Freedom and choice are often presented as benefits of transnational higher education, but participants in this study also described freedom as something they had to learn to manage. This article examines that tension through a qualitative case study of one Sino-British university. It draws on 12 Chinese-language semi-structured interviews analysed using reflexive thematic analysis, together with a separate qualitative analysis of public institutional materials. Participants valued having room to make their own decisions, although some recalled early uncertainty when direct supervision was reduced. Competition did not disappear; it became harder to read and extended beyond grades to internships, research experience, postgraduate applications and employment. Participants generally knew that support existed, but it often became useful only after they recognised a problem and asked for help. Their accounts therefore described academic, linguistic, relational and career self-management. Several participants also described using generative artificial intelligence (AI) to understand course content, translate material and prepare for assessment, although AI-related material was uneven across the interviews and included concerns about accuracy, disciplinary fit and independent judgement. We use institutionalised autonomy to describe how arrangements that widen discretion can also shift planning, help seeking and risk judgement towards students. The concept is offered as a context-specific specification of responsibilisation rather than a general theory. This single case is not statistically representative, but it identifies a mechanism that can be examined in other transnational settings.

1. Introduction

Sino-British transnational universities are often presented as spaces of educational freedom. They offer English-medium curricula, flexible mobility routes and a more independent learning culture than the examination-centred schooling from which many students come. The empirical puzzle is why arrangements publicly presented and often experienced as freedom can also generate pressure, self-surveillance and uneven burdens of planning, help seeking and judgement. This article examines how student freedom is coupled with individual responsibility at a Sino-British university, and how participants experience and manage the academic, linguistic, relational and career demands that follow. Generative artificial intelligence (AI) enters the analysis as one emerging resource described by a subset of participants, not as a uniformly experienced phenomenon or a separate focus of the study. The interviews analysed in this article suggest that freedom is not experienced only as empowerment. Participants repeatedly described the university as a place where “nobody forces you”, “nobody manages you” and “everything depends on yourself”. These accounts were often appreciative, but they also carried uncertainty, loneliness, invisible comparison, fear of falling behind and pressure to plan. One participant described the shift as moving “from others forcing you to you driving yourself”. Another called university learning “cultivation by yourself”, noting that classmates may already have reviewed the same material several times before their effort becomes visible. Xi’an Jiaotong-Liverpool University (XJTLU) provides an analytically generative case for examining how contemporary transnational higher education (TNE) organises freedom, linguistic adaptation, pathway choice and student self-management. Rather than treating 4+0 and 2+2 as separate identity categories, the analysis approaches the pathway system as part of a wider institutional ecology that requires students to manage choice, uncertainty and responsibility.

Therefore, the article’s main contribution is to use institutionalised autonomy as a context-sensitive, mid-range specification of responsibilisation. This concept situates autonomy within institutional arrangements and examines how choice becomes consequential as responsibility for planning, monitoring, help seeking and risk judgement shifts towards students. This perspective connects learner-level accounts of self-regulation with the broader literature on governmentality by showing how institutions make self-management an expected part of student conduct. The analysis examines how participants respond through academic, linguistic, relational and career practices. Generative AI is considered only as an emerging resource described in a subset of interviews. The public institutional texts and interview accounts are analysed separately before being brought together for interpretive comparison. As a single-case study, the article does not estimate the prevalence of these experiences across transnational higher education; instead, it identifies a mechanism for further comparative investigation.

On this basis, this article addresses two research questions:

RQ1. How are autonomy, choice and support framed in the public institutional discourse of a Sino-British transnational university?

RQ2. How do students in this case experience and manage the responsibilities associated with institutional freedom?

2. Literature Review and Theoretical Framework

2.1. Transnational Higher Education and the Everyday Work of Internationalisation

Transnational higher education refers to educational provision in which programmes, providers or credentials cross national borders while students may remain partly or wholly in their home country (Knight, 2016). Existing research has mapped the organisational forms, risks and rationales of transnational provision, including international branch campuses, joint institutes, dual degrees and cross-border partnerships (Altbach & Knight, 2007; Healey, 2015; Knight, 2008, 2016; Lane, 2011). This scholarship has moved the field beyond questions of programme delivery and cross-border mobility. It shows that transnational education is not a simple extension of a home curriculum into another national context, but an arrangement assembled through policy agendas, market pressures, institutional reputation, quality assurance regimes and uneven relations between sending and receiving systems. Transnational education is therefore better understood as a contested institutional space in which global higher education is organised, branded, regulated and made governable.

Institutional-level accounts, however, do not fully explain how transnational education is lived by students. A degree route may be advertised as flexible, but students must decide whether and when to transfer. A curriculum may be internationalised, but students must turn that curriculum into daily study practices. A university may provide career resources, but students must know how to find, interpret and act on them. Research on international branch campuses has examined student satisfaction and perceptions of educational quality (Wilkins et al., 2012), but the everyday work through which students manage autonomy, language, peer comparison and employability remains less visible. This gap is especially important in Sino-foreign cooperative universities. These institutions are not simply “foreign universities in China”. Rather than functioning as straightforward extensions of overseas universities, these institutions are hybrid educational spaces shaped by the intersection of students’ prior experiences of domestic schooling, international curricula, English-medium instruction, local labour-market opportunities, and aspirations for transnational mobility. Students often move from a structured secondary system into a university culture that presents independence as both a promise and an expectation. Thus, the present study shifts attention from the formal provision of transnational education to the daily work of becoming the kind of learner such provision requires. A student-experience perspective also avoids treating transnational education as a single institutional product. Educational demands may be mediated by route, discipline, stage, language confidence and prior academic-literacy experience (Lea & Street, 1998; Piller, 2016). In this case, route is treated as one part of the institutional ecology rather than as a fixed student type. The study neither compares the quality of 4+0 and 2+2 nor infers systematic route effects from the small, heterogeneous sample.

2.2. English-Medium Instruction as Linguistic and Academic Labour

English-medium instruction (EMI) is central to many internationalised universities. A major review of EMI in higher education notes that the rapid global expansion of EMI has outpaced evidence about its effects on student learning, language development and disciplinary understanding (Macaro et al., 2018). Dearden (2015) similarly describes EMI as a growing global phenomenon linked to internationalisation, mobility and employability, while Galloway et al. (2017) show that EMI in China and Japan is shaped by both global aspirations and local constraints. In this article, EMI is understood not merely as a language of instruction, but as a pedagogical and social condition that shapes how students participate in class, formulate questions, engage with academic reading and writing, negotiate confidence, and envisage future mobility.

For students whose previous education was not primarily in English, EMI can create both opportunities and uneven burdens. It may provide access to global knowledge, postgraduate study and international credentials, but it can also shift language work onto students, who must manage comprehension, academic writing, classroom participation and disciplinary vocabulary. Piller’s (2016) work on linguistic diversity and social justice reminds us that language access is never merely technical; it is tied to unequal distributions of prior preparation. In a transnational university, EMI may therefore operate as both a resource and a sorting mechanism. It is part of the institutional promise of internationalisation, but it also becomes a domain of self-management. This article contributes to EMI research by connecting language adaptation to broader patterns of autonomy and self-governance. Participants did not describe English only as a linguistic obstacle. They treated it as one element in a wider ecology of independent study, resource seeking, teacher communication, AI use and career planning. EMI therefore forms one layer of institutionalised autonomy: students are expected to manage language difficulty as part of managing themselves.

This reading also changes how language support is understood. If EMI difficulty is framed only as a proficiency problem, the solution appears to be more English training. That is necessary, but insufficient. In the interviews, language difficulty was tied to knowing when to ask questions, how to reread lecture slides, whether to use translation or AI tools, how to write in an acceptable academic style and how to participate without feeling exposed. Language support therefore overlaps with academic literacy, confidence, assessment literacy and help seeking. For students in transnational universities, the challenge extends beyond understanding content delivered in English. It also involves learning to participate in the academic practices that English-medium instruction presupposes, including how to read, write, question and engage in classroom discussion.

This reframing aligns EMI difficulty with an academic literacies perspective, which treats academic language not as a neutral and transferable skill but as a set of socially situated practices bound up with disciplinary conventions, identity and power (Lea & Street, 1998). The pertinent question is not only whether students possess sufficient English proficiency, but whether they can acquire the contextual literacies that English-medium study quietly assumes, including genre, citation, argument structure and disciplinary voice. In a transnational university, these literacies are not always taught explicitly; students are often expected to infer them. EMI therefore becomes a linguistic site of institutionalised autonomy. The institution supplies the medium, while students assemble the practices that make the medium usable.

2.3. Learner Autonomy and Self-Regulated Learning

Learner autonomy and self-regulated learning provide an important vocabulary for analysing student independence. Learner autonomy is often understood as the capacity to take charge of one’s own learning (Benson, 2011). Self-regulated learning research examines how learners set goals, monitor progress, select strategies and adapt behaviour in response to feedback (Boekaerts et al., 2000; Pintrich, 2004; Zimmerman, 2002; Zimmerman & Schunk, 2011). In the present case, students regulate time, attention, assessment preparation, language learning, office-hour use, AI tools, internship planning and postgraduate applications.

These theories, however, can make autonomy appear primarily psychological: some students are self-regulated, while others lack motivation, strategy or discipline. The interviews point to a more institutional reading. Students do not simply bring different degrees of autonomy into the university. They enter an environment that makes autonomy a condition of success. Flexible schedules, optional resources, weak supervision and multiple career pathways do not merely allow self-regulation; they demand it. The problem is therefore not only whether students can regulate themselves, but how the institution makes self-regulation necessary and how unevenly students are prepared for that demand.

This distinction matters for both analysis and practice. If autonomy is treated as an individual trait, students who struggle can easily be read as deficient. If autonomy is treated as an institutional demand, attention shifts to the environment that asks students to plan, seek, compare and decide. It also becomes possible to ask whether the institution teaches students how to manage freedom or simply assumes that they will learn it. The distinction matters because autonomy is an attractive educational value. Universities rarely present independence, flexibility or student-centred learning as problems. They are more often presented as signs of maturity, international quality and learner empowerment. This article does not reject those values. It asks what happens when they are embedded in institutional routines before students have fully learned how to practise them. Autonomy can be enabling and demanding at the same time; a student can value freedom while still feeling that freedom has become a burden. This ambivalence is the empirical object of the article.

2.4. Governmentality, Responsibilisation and the Self-Managing Student

To capture the institutional dimension of autonomy, this article draws on governmentality. Foucault’s (1991) concept of governmentality directs attention to forms of power that operate not only through command and discipline, but through freedom, responsibility and self-monitoring. Rose (1999) and Dean (2010) further show how modern subjects are governed by being asked to govern themselves. In higher education, this perspective has been used to analyse the marketised and neoliberal university, where students are addressed as responsible, employable, entrepreneurial and self-improving subjects (Macfarlane, 2017; Olssen & Peters, 2005). Within this tradition, the contemporary student is increasingly addressed as an entrepreneurial self who treats education as an investment in personal human capital (Brown, 2015; Bröckling, 2016). The injunction is not simply to learn, but to manage, accumulate and signal value through grades, internships, languages, networks and credentials as assets in a competitive graduate market (Brown et al., 2003; Tomlinson, 2017). This human-capital rationality matters for the present case because it reframes ordinary academic decisions as calculated investments. When students plan their studies by working backwards from job requirements, they are not only preparing for employment. They are enacting a mode of subjectivity in which the self is governed as an enterprise to be continuously improved.

This perspective is useful because the students in this study are not primarily governed by direct control. They are governed through a setting in which freedom has consequences. They are free to attend or not attend, ask or not ask, plan or not plan, transfer or not transfer, use resources or ignore them. These choices carry academic and career consequences, and students know this. The resulting pressure is not the older pressure of constant external supervision. It is the pressure of having to supervise oneself.

2.5. Institutionalised Autonomy: Conceptual Contribution

To analyse these dynamics, we use institutionalised autonomy as a context-sensitive, mid-range concept. It describes institutional arrangements that expand students’ discretion while placing greater weight on their ability to plan, interpret expectations, seek support and manage consequences. Four features define the concept: greater choice or less direct supervision; success criteria that retain their force even when partly implicit or dispersed across academic and career domains; the transfer of responsibility to students for planning, help seeking, time management and risk judgement; and the potential for uneven consequences arising from differences in linguistic, informational, social and family resources. This institutional focus distinguishes the concept from learner autonomy, which centres on the learner’s capacity to take charge of learning (Benson, 2011), and from self-regulated learning, which examines goal setting, strategy use, monitoring and adjustment (Pintrich, 2004). Institutionalised autonomy shifts attention to how these practices become institutionally expected and consequential. It builds on governmentality and responsibilisation (Dean, 2010) by specifying how freedom, responsibility and self-management are organised in the transnational university examined here. Table 1 summarises how institutionalised autonomy relates to, and differs from, these adjacent concepts.

Table 1. Conceptual boundaries of institutionalised autonomy.

Adjacent Concept	Usual Emphasis	How Institutionalised Autonomy Differs
Learner autonomy	Capacity or willingness to take charge of learning	Autonomy is analysed as an institutional requirement before it is evaluated as an individual capacity.
Self-regulated learning	Goal setting, strategy use, monitoring and adaptation	Self-regulation is located inside a system that makes planning, monitoring and resource seeking necessary.
Governmentality	Broad analysis of how conduct is governed through freedom, norms and self-regulation	Institutionalised autonomy specifies a bounded educational architecture through which governing through freedom became observable in this case.
Responsibilisation	Transfer of responsibility and risk from institutions to individuals	Institutionalised autonomy identifies which arrangements transfer navigational work and how its consequences vary among students.

Operationally, institutionalised autonomy links institutional arrangements, the work expected of students and the consequences attached to that work. Drawing on governmentality and responsibilisation (Dean, 2010; Shamir, 2008), the analysis treats English-medium instruction, pathway choice, modular curricula, office hours, digital platforms and career services as arrangements that structure students’ options and responsibilities. It then examines how participants turn these possibilities into routines, decisions, help seeking and plans, and where they locate the consequences of delayed or ineffective navigation in assessment, mobility and career planning. The framework keeps both the enabling and demanding aspects of autonomy in view, including opportunities for exploration as well as disorientation, hidden comparison and pressure when expectations remain difficult to discern. Attention is also given to variation, including positive experiences of freedom and continued reliance on teachers and peers.

In this study, institutionalised autonomy offers a focused way to examine how responsibility is organised through language, pathway choice, optional support, time and mobility. It identifies a specific mechanism and unit of analysis while directing attention to observable institutional conditions and variation in participants’ responses. The concept also provides a basis for comparison with other settings. In the analysis that follows, it functions as a sensitising framework, and the interpretation is developed and refined through participants’ accounts.

2.6. Generative AI as an Emerging Resource for Self-Management

Generative AI has entered higher education as a potential resource for explanation, translation, brainstorming, feedback and writing support. Its educational significance, however, cannot be inferred from technical availability alone. Bearman et al. (2023) show that higher-education discussions of artificial intelligence are organised through competing discourses of enhancement, disruption and risk, while Selwyn (2022) cautions against treating AI as a neutral solution detached from institutional purposes and unequal conditions of use. These perspectives direct attention to how AI is framed, used and governed rather than assuming a uniform effect on learning. For this article, generative AI is relevant as a possible resource within a broader ecology of self-management. It may provide cognitive and language support in an English-medium environment, including explanation, translation and assessment preparation, but it can also create dependence and shift responsibility for accuracy, academic integrity, disclosure, disciplinary fit and appropriate reliance to learners. Because AI was not covered uniformly in the interviews, the analysis treats these practices as an inductively emerging subtheme. It does not estimate prevalence or claim that AI defines transnational higher education.

3. Institutional Setting: XJTLU as a Transnational University

XJTLU is a Sino-British transnational university established through a partnership between Xi’an Jiaotong University and the University of Liverpool and is based in Suzhou, China. The institution is named because its partnership structure and public programme information are matters of public record; individual participants are anonymised. Public institutional materials indicate that, depending on their programme, students may complete all four stages of study at XJTLU, commonly referred to as 4+0, or complete the first two stages at XJTLU before transferring to the University of Liverpool for the final two stages, commonly referred to as 2+2 (University of Liverpool, n.d.-a, n.d.-b; Xi’an Jiaotong-Liverpool University, n.d.-a). For selected undergraduate programmes at XJTLU Entrepreneur College (Taicang), a 2+1+1 route is also available (Xi’an Jiaotong-Liverpool University, n.d.-a, n.d.-d). Institutional materials also present Year One as a transition structure that includes academic modules, English-language classes and core skills development. English for Academic Purposes (EAP) is described as subject-focused and connected to academic skills, discipline-specific vocabulary, authentic texts, lectures, student-centred activities, discussions and tutorials (Xi’an Jiaotong-Liverpool University, n.d.-e). Learning and teaching materials describe students as actively engaged, working in teams, developing and testing ideas, questioning received opinion and undertaking independent projects (Xi’an Jiaotong-Liverpool University, n.d.-b). Resources and support pages describe development advisers, class tutors, student services, learning platforms, online systems and career-development support (Xi’an Jiaotong-Liverpool University, n.d.-c).

This article analyses these materials as public institutional texts with a partly promotional function, not as evidence of internal policy or implementation. Collected after the interviews, they show how XJTLU and the University of Liverpool publicly present pathways, learning, support and the ideal student. They cannot establish whether the same wording and provision were applied during data collection, how services operated in programmes or whether participants encountered specific pages. The documents and interviews were therefore examined as separate evidence strands: the former illuminate the public construction of autonomy, while the latter show how participants described their own experiences.

4. Research Methodology

4.1. A Qualitative Case Study of Institutional Discourse and Student Experience

The study adopts a qualitative, single-institution case design to examine public institutional constructions of autonomy alongside students’ accounts of everyday university life. Autonomy is treated as situated within an institutional context. Public texts and interviews form two analytically distinct evidence strands: the former show how choice, support and responsibility are publicly framed, while the latter show how participants describe navigating these conditions. XJTLU offers an analytically useful setting in which English-medium instruction, mobility choices, student-centred learning, optional support and employability planning intersect. The analysis uses this case to formulate a mechanism that can be tested in other settings. The empirical claims are bounded by this institutional setting and participant group.

4.2. Participants and Interview Data

4.2.1. Participants, Recruitment and Interview Procedures

The primary dataset comprised 12 Chinese-language semi-structured interviews with eight current undergraduates and four recent alumni from business, humanities and social sciences, mathematics, information management and engineering. Five participants followed the 4+0 route and seven followed the 2+2 route. Further participant characteristics are summarised in Table 2. Recruitment combined purposive and snowball sampling. Initial notices were circulated through online platforms, including Xiaohongshu (RedNote), and subsequent recruitment drew on referrals from existing participants and introductions through personal contacts. Recruitment sought variation in discipline, study stage and pathway. Participation was voluntary, and no compensation or other incentive was offered. The interviewer had no prior acquaintance with the participants and held no teaching, assessment, advisory, supervisory or other institutional authority over them. All 12 interviews were conducted online in Chinese during May 2026 through videoconferencing platforms, including Tencent Meeting; none were conducted face to face. With participants’ consent, the interviews were audio-recorded. The first author prepared the Chinese transcripts, and a co-author checked them against the audio recordings. The recorded interviews lasted approximately 20 to 156 min, with a mean duration of approximately 68 min.

Table 2. Participant profiles.

Code	Broad Programme Area	Route/Status at Interview
P1	Communication/humanities and social sciences	4+0 alumna
P2	Information management and information systems	4+0 final-year student
P3	Economics	4+0 undergraduate student
P4	Financial English/humanities and social sciences	4+0 Year 3 undergraduate student
P5	International business	4+0 final-year student
P6	Applied mathematics	2+2 final-year student
P7	Electrical and electronic engineering	2+2 final-year student in the United Kingdom
P8	Financial mathematics	2+2 alumna
P9	Marketing	2+2 Year 3 undergraduate student in the United Kingdom
P10	Mathematics with finance	2+2 Year 3 undergraduate student in the United Kingdom
P11	Accounting and finance	2+2 alumna
P12	Applied mathematics	2+2 alumnus

Sample adequacy was assessed using the concept of information power (Malterud et al., 2016). The focused single-case aim, participants’ direct experience of the institutional setting, the theoretically specific concern with autonomy and the substantive detail available across the interview corpus strengthened the information power of the study. Its scope was constrained by variation across disciplines, pathways and stages, as well as by the uneven coverage of AI. The sample therefore supports a case-bounded interpretive account of institutionalised autonomy. It does not provide a basis for estimating prevalence, conducting systematic institutional comparison or claiming that the AI-related practices reported here are typical.

4.2.2. Interview Guide and Scope of AI-Related Material

The interview guide covered academic adaptation, English-medium learning, route choice, teacher and peer interaction, institutional resources, career planning, definitions of excellence and experiences of pressure. The original project examined student identity and comparisons across pathways. Generative AI was not included as a stand-alone topic or as a core module administered uniformly across the sample. Some participants raised AI while discussing independent learning, English-medium study, help seeking, classroom participation and assessment preparation. AI-specific follow-up prompts were used in some interviews, but not across the full sample. AI is therefore treated as an inductively emerging subtheme in the present analysis. The full interview guide is provided in File S1 of the Supplementary Materials.

AI-related material appeared in nine of the 12 interviews, although its depth varied. Seven participants (P1, P2, P3, P5, P6, P7 and P10) provided accounts sufficiently developed to inform the AI subtheme. P4 and P9 mentioned AI only briefly, while P8, P11 and P12 provided no substantive AI material. Classification depended on whether an interview contained sufficient detail about practices, purposes, consequences or limitations to support thematic interpretation. Simple keyword occurrence was not used as the criterion. These categories define the evidential scope of the subtheme. They do not measure the prevalence or intensity of AI use, and non-mention is not interpreted as evidence of non-use.

4.3. Public Institutional Document Corpus

The secondary strand is a bounded, purposively selected corpus of eight official public webpages and documents (Table 3). The search boundary was limited to the official XJTLU and University of Liverpool domains and to materials whose substantive content addressed degree pathways, 2+2 mobility and transfer, Year One, learning and teaching, support resources or the courses and guidance attached to the 2+2 route. News items, social-media posts, third-party pages, internal materials and general promotional or programme pages unrelated to these themes were excluded. All items were accessed on 11 June 2026, after the interviews conducted in May 2026. They are analysed only as evidence of how the institutions present and construct autonomy, choice, support and the self-directing student. They cannot establish how arrangements were implemented, the quality of services, whether participants encountered any specific page or whether institutional conditions were unchanged at the time of the interviews; the later access date is therefore retained as an explicit temporal boundary.

Table 3. Public institutional document corpus and analytical use.

Document/Webpage Title	Publishing Institution	URL	Access Date	Selection Rationale	Corresponding Analytical Theme
Degree types	XJTLU	https://www.xjtlu.edu.cn/en/study/why-study-at-xjtlu/degree-types	11 June 2026	Describes degree-completion and mobility options.	Choice, 4+0/2+2 pathways and mobility responsibility
Learning and teaching	XJTLU	https://www.xjtlu.edu.cn/en/study/why-study-at-xjtlu/learning-and-teaching	11 June 2026	States the publicly promoted approach to learning and teaching.	Active learning and the ideal self-directing student
Resources and support	XJTLU	https://www.xjtlu.edu.cn/en/study/why-study-at-xjtlu/resources-and-support	11 June 2026	Presents academic and student-support resources.	Support availability, access and help seeking
Transfer to University of Liverpool	XJTLU	https://www.xjtlu.edu.cn/en/study/international-mobility/transfer-to-university-of-liverpool	11 June 2026	Describes transfer opportunities and advertised guidance.	Mobility transition and route planning
Year One	XJTLU	https://www.xjtlu.edu.cn/en/study/why-study-at-xjtlu/year-one	11 June 2026	Describes the first-year structure and transition provision.	Year One, EMI transition and scaffolding
The 2+2 programme	University of Liverpool	https://www.liverpool.ac.uk/xjtlu/study-in-liverpool/2-2-programme/	11 June 2026	Presents the purpose and structure of the 2+2 route.	Pathway choice, transfer and mobility
2+2 courses	University of Liverpool	https://www.liverpool.ac.uk/xjtlu/courses/	11 June 2026	Lists courses available within the transfer route.	Curricular choice and disciplinary planning
2+2 guide (in Chinese)	XJTLU & University of Liverpool	https://www.xjtlu.edu.cn/wp-content/uploads/2026/04/UoL_22_Guide_2025_zh.pdf	11 June 2026	Consolidates publicly stated route information and expectations; publication year not inferred from filename.	Advance planning, transfer and student responsibility

4.4. Data Analysis

Interview data were analysed using reflexive thematic analysis (Braun & Clarke, 2006, 2019, 2022). NVivo 14 (Lumivero, Denver, CO, USA) was used to organise the Chinese-language transcripts and retrieve coded material. Coding decisions, theme development and interpretation were undertaken manually by the researchers. The coding combined deductive and inductive elements. Broad sensitising categories were informed by the research questions, while more specific codes were developed from participants’ accounts. The first author conducted the initial coding. Candidate themes were developed and refined through repeated comparison of coded extracts, complete transcripts and patterns across participants. Attention was given to divergent and limiting cases, including positive experiences of independence, continued reliance on teachers or peers, disciplinary limits to AI assistance and interviews containing little or no AI material. Institutionalised autonomy was developed through iterative engagement with the data and relevant theory; it was not used as a fixed coding framework from the outset.

Public institutional texts were analysed separately following Bowen’s (2009) approach to document analysis. Initial coding addressed active learning, independence, choice, self-reliance, employability, support availability and student responsibility. The interview and document strands were brought together only after each had been analysed separately. Interpretive comparison examined convergence, tension and silence between public representations and participants’ accounts. The two strands were not treated as equivalent or mutually validating evidence. Public institutional texts were not used to establish how services operated, whether participants encountered particular materials or whether institutional provisions functioned as described in the public materials. Because the documents were collected after the interviews, similarities between the two strands were not interpreted as evidence of contemporaneous implementation.

4.5. Ethics and Positionality

Ethical approval was granted by the Ethics Committee of the Key Laboratory of Modern Teaching Technology, Ministry of Education (approval number L20260301-01). All participants received information about the study and provided informed consent to participate and to be audio-recorded. Participation was voluntary, and participants could refuse to answer any question or withdraw without consequence. Transcripts were anonymised and identifying details were generalised where necessary. Interviews were conducted in Chinese. The lead author translated the extracts selected for quotation. After translation, a co-author reviewed the coding associated with those extracts with reference to the relevant Chinese transcript segments. The Chinese transcripts remained the reference point for interpretation. Adjustments for English readability were limited to syntax and did not alter the substantive meaning of the extracts. This process constituted a collaborative review of source material and coding. It was not a formal independent back-translation.

The lead author’s familiarity with XJTLU’s pathway terminology and institutional context supported context-sensitive questioning and interpretation. This familiarity also created a risk that locally specific assumptions could be taken for granted. Such assumptions were considered during theme refinement, and co-author review of quoted extracts and their associated coding provided an additional check on interpretations shaped by institutional familiarity.

4.6. Trustworthiness

The credibility of the analysis was strengthened by reviewing candidate themes against complete transcripts, analysing interview and document data separately, and retaining variation and limiting cases throughout the interpretive process, following guidance on quality practice in reflexive thematic analysis (Braun & Clarke, 2021). These cases included positive experiences of autonomy, continued reliance on teachers and peers, disciplinary limits to AI assistance and interviews containing little or no substantive AI material. They helped define the scope and boundaries of the interpretation and were not excluded as anomalies.

5. Results

5.1. Institutional Framing of the Self-Directing Student

Within the bounded public-text corpus, the institutions construct a desirable student as active, self-reliant, mobile and resource seeking. The learning-and-teaching page describes students as “actively engaged in the learning process”, working in teams, developing and testing their own ideas and being ready to question received opinion, including the opinion of a teacher (Xi’an Jiaotong-Liverpool University, n.d.-b). The same public framing foregrounds confidence, self-reliance, independent goal setting and judgement. It is evidence of how autonomy is promoted as the expected mode of study, not evidence that every student has experienced learning in this way. The public materials present students as able to complete their degree at the China-based campus or transfer to the University of Liverpool for the final stages of study. Within this representation, flexibility positions the student as a route planner who weighs mobility, finances, adaptation, postgraduate plans, domestic internships and future employability. Advisers, tutors, learning platforms, student services and career support are described as resources that students can access, use or seek. The grammar of availability therefore constructs an ideal student who can recognise a need and activate support. This is a claim about institutional presentation; it does not show that participants saw the pages, that services operated as represented during the May 2026 interviews or that access produced equivalent outcomes.

The analytical significance of this public framing does not depend on treating it as a complete account of institutional practice. The texts make active learning, independent judgement, pathway choice and support availability visible, but say less about what students must do to act on these ideals. This work includes deciding which opportunities matter, recognising when one is falling behind, translating general support into discipline-specific action and managing uncertainty. We interpret these silences as a possible hidden curriculum within the broader mechanism of institutionalised autonomy, through which students may be expected to infer and assemble the conduct of autonomy for themselves (Margolis, 2001; Snyder, 1971). What remains implicit is not module content but expected conduct, including when to seek help, how to read selectively, how to turn a flexible timetable into a routine and how to recognise difficulties before they become visible in grades. This is an authorial interpretation of the public documents. The webpages are not treated as evidence of student behaviour or institutional implementation.

Only after the two evidence strands had been coded separately were they compared interpretively. In the interviews, P1 described autonomy developing “in a subtle way”, P2 characterised the transition as moving from being pushed by others to driving himself, and P4 stressed that resources would not be placed in front of students unless they sought them out. P9 and P10 linked pathway decisions to personal goals and future positioning. These accounts provide points of convergence and variation across disciplines, stages and routes; they do not verify the webpages, nor do the webpages verify them. Table 4 makes this interpretive relationship explicit.

Table 4. Interpretive comparison of public institutional framing and interview accounts.

Institutional Framing	Representative Public-Text Emphasis	Responsibility Transferred to Students	Interview Strand: Convergence or Tension
Active learner	Students develop ideas, work in teams and question received opinion.	Participate, speak, ask, interpret and act.	P1 learned autonomy “in a subtle way” through the environment.
Self-reliant learner	Students develop confidence, self-reliance and independent judgement.	Set goals and monitor progress without constant supervision.	P2 described a shift from being pushed by others to driving himself.
Mobile route planner	Students may complete 4+0, 2+2 or selected 2+1+1 routes.	Weigh mobility, adaptation, cost, career timing and postgraduate plans.	P9 and P10 treated route choice as linked to personal goals and future positioning.
Resource-seeking student	Advisers, services, platforms and career support are available.	Recognise needs, locate services, ask for help and use feedback.	P4 stressed that resources do not appear automatically.

5.2. Freedom, Early Disorientation and Positive Autonomy

Most participants described the university as a space of considerable freedom compared with their earlier schooling, including fewer direct controls over attendance, timetable use, study routines and teacher supervision. Many valued this freedom because it enabled exploration, independent time management and a less constrained identity. Several also described early disorientation, while others emphasised gradual adaptation and the development of independence. The theme is therefore not that freedom is inherently harmful, but that its benefits depend partly on students learning how to organise and use it.

“At XJTLU, I gradually learned how to study autonomously in a subtle way. This learning was not only academic; it also included daily self-management and a new understanding of interpersonal relationships.”

(P1)

P1 emphasised that autonomy was not conveyed through explicit instruction. The institution did not explicitly say “you must be autonomous”. Instead, students were placed in an environment that made independent action necessary:

“It did not directly tell me, ‘You must be autonomous’ or ‘you must explore independently’. But through many influences at XJTLU, I slowly developed a relatively independent ability.”

(P1)

The same environment could feel “too loose” at first. P1 linked academic adaptation to social and emotional adaptation: learning to read material and use office hours was inseparable from learning how to manage daily life and relationships. P3 described the adjustment more sharply as “growing pains”. Students did not always choose adaptation in a deliberate way; demands “suddenly arrive” and students have to respond. The weakening of fixed class structures also carried a social cost. Several participants noted that classmates changed across modules and that a stable class community was weaker than in earlier schooling. Freedom over association therefore came with responsibility for constructing one’s own belonging.

P2 captured the central paradox:

“It changed from others pushing you to having to drive yourself. XJTLU is so relaxed and so free. It all depends on yourself. If you want to compete, you can compete very hard; if you want to slack off, no one will manage you.”

(P2)

For P2, freedom was double-edged: it reduced external coercion but made the consequences of self-monitoring more visible. He recalled playing extensively in his first year and receiving lower grades because no one clearly told him what to do; once his goals became clearer, his learning plan became more systematic. P1, by contrast, placed greater emphasis on the independence gradually enabled by the environment. Taken together, these accounts show variation in pace and affect. Autonomy could be valued as development while still exposing students to the risk of drifting before expectations became legible.

A cross-route counter-case came from P8, a 2+2 financial mathematics alumna, who recalled a rapid and positive adjustment:

“XJTLU was itself a very open and diverse environment, and it felt quite relaxed. I accepted it very quickly.”

(P8)

P8’s account qualifies any suggestion that reduced supervision necessarily produced distress. Together with P1’s positive account of developing independence, it shows that early disorientation varied according to prior preparation, discipline, stage and personal circumstances rather than forming a uniform sequence.

Disorientation was not limited to students who lacked ability. Several participants were academically capable and eventually successful, yet still described a period of uncertainty in which the rules of success were less explicit than before. This matters because it prevents a deficit interpretation. The issue was not simply that some students were insufficiently disciplined. Rather, students were learning a new educational grammar: how to interpret a lecture that did not cover everything, how to use feedback that did not prescribe a single correct answer, how to decide whether missing class was harmless or risky and how to turn a flexible timetable into a routine. Early disorientation was therefore a normal effect of moving into an institutional culture that assumes self-direction.

5.3. Hidden Competition and Multidimensional Planning

A second theme is hidden competition. Participants did not describe the transnational university as pressure-free. Instead, pressure became less visible and more distributed. Students were no longer constantly ranked within a fixed class, but they remained exposed to grade point average (GPA) thresholds, scholarships, postgraduate applications, internships, research opportunities, English performance and career preparation.

“In high school, you know what your classmates are doing. But at university it is like cultivation by yourself. Other people may already have studied the material two or three times and finished the past papers. Without realising it, everyone has quietly competed.”

(P3)

P3 called this “invisible competition”. The absence of visible comparison did not mean that comparison disappeared. Students could lose their sense of crisis until grades, internships or applications made the gap visible. This is a distinctive form of pressure. It is not the pressure of being constantly watched; it is the pressure of not knowing enough about what others are doing.

The pressure was intensified because excellence was fragmented across academic and future-oriented domains. P4 associated excellence with proactive communication with teachers and self-learning, not only scholarships or grades. P12, a 2+2 applied mathematics alumnus, reported that research and genuine intellectual interest eventually felt more challenging and meaningful than simply maintaining a high GPA. P5, a 4+0 international business student, made the changing standard explicit:

“Previously, people might have thought that the student with the best grades was the most excellent. Now everyone wants all-round development: you need this, that and everything else.”

(P5)

We interpret this fragmentation as a shift in the form of competition. In an examination-centred system, students know the ladder on which they are competing. In the accounts here, several ladders operated at once: a student might be ahead in GPA but behind in internships, active in research but less prepared for employment, or confident in English discussion but less strategic about postgraduate applications. The result was multidimensional planning as participants decided which profile of excellence to pursue while knowing that other profiles might also matter.

In the separately coded public-text strand, programme and career materials present disciplinary knowledge, research projects, final-year projects, employability skills, internships, placements, career planning, curriculum vitae (CV) preparation and mentoring as available or desirable. These materials show how the public institutional discourse broadens the field of valued activity; they do not establish how opportunities operated or whether participants used particular services. When placed alongside the interview strand only at the interpretive stage, the public framing helps illuminate why participants’ planning extended beyond the question “How do I pass this module?” towards the more diffuse problem “How do I build a competitive self?”

The hiddenness of competition also changes students’ temporal experience. In a visible examination system, students know when competition formally begins and ends. In the accounts here, competition feels continuous because any period of freedom might be used by someone else for internships, research, language preparation or postgraduate planning. This produces diffuse anxiety. Students may not be studying at a given moment, but they know that others could be quietly accumulating advantage. Hidden competition therefore turns free time into morally charged time. Rest, exploration and social life remain possible, but they are shadowed by the thought that time should perhaps have been invested more strategically.

This temporal dimension deserves emphasis because institutionalised autonomy does not only distribute responsibility across domains; it also restructures students’ relationship to time. The accounts suggest a developmental arc that maps onto the architecture of the degree. In the first year, freedom is experienced largely as disorientation, and time feels open but unstructured. As students approach internships, transfer, final-year projects and postgraduate applications, the same freedom is reinterpreted as a scarce resource to be invested strategically. Earlier idle time is retrospectively recoded as lost advantage. The institution does not impose this timetable through direct supervision. Its flexible structure invites students to internalise it. Hidden competition is, in this sense, a temporal regime. It asks students to govern not only what they do, but when they ought to have begun doing it.

5.4. Conditional Support and Varied Self-Management Practices

5.4.1. Academic, Linguistic, Relational and Career Self-Management

The third theme brings together conditional support and varied practices of self-management. Resources were described in institutional materials or experienced by participants as available, but activating them required students to recognise a need, locate an appropriate person or service, formulate a question and follow through. Participants managed this condition across academic work, language, relationships and careers through different combinations of planning, prioritising, help seeking and resource use. These practices were neither solitary nor uniform, and no single tool or strategy organised the whole dataset.

“Students need to be more proactive. It is not passive knowledge feeding. You need to use office hours.”

(P4)

P4 explained that lectures alone were insufficient. Students needed to review slides, practise after class, identify problems and ask teachers or AI tools for help. Her later statement crystallised the institutional logic of conditional support:

“The resources are there, but if you do not go and find them, they will not be placed in front of you.”

(P4)

Office hours illustrate this logic. They may provide individualised academic help, but they require students to recognise a problem, formulate a question, book a meeting and communicate actively. For students accustomed to teacher-led instruction, this is not a minor procedural adjustment. It is a new form of academic agency.

P11, a 2+2 accounting and finance alumna, similarly described human assistance as part of her routine rather than as a failure of autonomy:

“If I did not understand something, I discussed it with classmates in the same programme, and sometimes I asked the teacher.”

(P11)

This account counters an individualised reading of self-management. For some participants, acting autonomously included knowing when to rely on teachers and peers rather than resolving every difficulty alone.

Language was another domain of self-management. Because reading, teaching and assessment were conducted in English, participants described work around comprehension, academic writing, presentations and discipline-specific vocabulary. Some initially found English-medium study alienating or difficult; others adapted with much less strain. Across accounts, strategies included translation tools, vocabulary checks, rereading slides, asking teachers and gradually becoming more comfortable with academic expression. EMI therefore generated work, but the intensity and meaning of that work varied with prior language preparation and disciplinary circumstances.

“There was definitely an adjustment process in academic English writing, but it was not too painful if your English foundation was reasonably strong.”

(P1)

P1’s account is a limiting case: it shows that an adjustment demand should not be equated with uniform suffering. The interview evidence supports variation in EMI-related labour, not a claim that English-medium learning burdened all participants in the same way.

Career planning formed a third domain. P2 described working backwards from job requirements, learning programming, databases and other skills because employment advertisements made those skills visible. P5 reported using university emails, career fairs and internship information, but stressed that students had to act on opportunities themselves. In these accounts, the curriculum is only one part of learning. Students become managers of an academic–career portfolio, deciding how coursework, internships, research, applications and future employment should fit together.

Relational self-management was also important. Group projects and presentations required students to negotiate different working styles, manage conflict and handle the risk that other group members might contribute unevenly. P5 described a group project in which members nearly fell out because some wanted to plan everything before acting while others preferred trial and error. Team-based assessment therefore did not simply cultivate collaboration. It also required students to manage interpersonal uncertainty as part of academic work.

5.4.2. Generative AI as an Emerging and Bounded Practice

For several participants, generative AI formed one emerging resource for academic self-management, but AI-related accounts were uneven across the dataset. The substantive analysis draws on P1, P2, P3, P5, P6, P7 and P10, whose accounts describe practices, purposes, consequences or limits in sufficient detail. P4 and P9 made only brief references, while P8, P11 and P12 provided no substantive AI material. The developed accounts concerned explanation, translation, writing or brainstorming, assessment preparation and alternative routes to immediate help. They should not be read as evidence that all participants, all XJTLU students or transnational students generally used AI in the same way.

“For students in Sino-foreign universities, or for students studying overseas, we basically let AI teach us and guide us in how to learn.”

(P3)

“Sometimes, listening to a two-hour lecture is less helpful than having Doubao explain it for a while. The opportunity cost then feels different, and students may become less willing to attend.”

(P5)

“If there is something we do not understand, we ask AI directly. There is less need to ask teachers. But AI sometimes makes mistakes, so when I really cannot understand something, I still ask the teacher or classmates.”

(P6)

For these participants, AI sometimes changed the route through which institutional freedom was converted into study action. P5 used it to preview difficult macroeconomics content and to generate practice questions after reviewing lecture and tutorial materials. She also identified a cost: an immediate answer could displace time spent reasoning through a problem. This account supports a bounded interpretation of AI as one possible self-regulation resource, not a conclusion that AI was central to the academic lives of the whole sample.

Other accounts establish important limits and alternatives. P6 used AI for immediate questions but returned to teachers, classmates and discussion when outputs were wrong or a problem remained unclear. P10 highlighted disclosure expectations when using AI. From an engineering perspective, P7 found AI useful for translating abstracts and assessing the relevance of a paper but sharply limited in hands-on troubleshooting:

“Hardware debugging has many parts: whether the circuit is connected correctly, whether something has burned out, whether the hardware itself is the problem, or whether a parameter is wrong. It requires repeated testing; AI cannot really help. It takes time, and some luck.”

(P7)

The evidence supports three cautious analytical points: AI could assist comprehension, translation, preview and revision; it could alter help seeking by providing an immediate alternative to teachers or peers; and it could add responsibility for verifying accuracy, sustaining independent thought and judging disciplinary limits. These findings identify a possible mechanism for further investigation rather than establishing prevalence. AI is therefore not used as an umbrella label for participants’ career planning, relationship management or other forms of self-management that were not themselves mediated by AI.

5.4.3. Route-Specific Variation in Conditional Support

Participants recognised that institutional support existed, including development advisers, office hours, teacher feedback, EAP, learning platforms, career services and transition guidance. The important variation concerned how that support became usable. Some participants described strong feedback or guidance that made expectations visible; others experienced resources as conditional on already recognising a problem, locating the service and knowing how to ask. Conditional scaffolding therefore does not mean an absence of support, but a gap between formal availability and students’ uneven capacity to activate it.

P1 valued academic writing training, citation norms and detailed teacher feedback. In one instance, a teacher’s extensive written feedback helped her feel seen and better understand academic ethics and originality. This suggests that institutionalised autonomy does not mean an absence of teaching. Strong feedback can scaffold autonomy by making academic expectations visible. Other participants, however, experienced feedback as limited or useful only when they already knew how to act on it. Support therefore still required interpretation and follow-through.

Among participants who transferred through the 2+2 route, accounts placed particular emphasis on combining academic work with housing, independent living, cultural adaptation, English use and future planning. In the public-text strand, the transfer page presents transition guidance and one-to-one support as available (Xi’an Jiaotong-Liverpool University, n.d.-d), and the Chinese-language 2+2 guide consolidates publicly stated route information and expectations for advance planning (Xi’an Jiaotong-Liverpool University & University of Liverpool, n.d.); this wording records advertised provision and does not demonstrate its implementation during the May 2026 interviews. Separately, P7 described studying abroad as a combination of freedom, loneliness and pressure. P8 reported that prior English-medium preparation made the academic forms in Liverpool manageable, while she still adapted to a different rhythm, assessment culture and mode of independent work. These accounts identify emphases within this small group rather than a route effect.

Among 4+0 participants, accounts more often foregrounded continuity in China alongside flexible schedules, domestic internships, career planning, office hours, English-medium learning and hidden comparison. Some regarded remaining in China as strategically useful for local internship timing and career exploration. These contrasts are descriptive patterns in the present interviews, not evidence that one route is inherently more or less autonomous than the other. Taken together, the accounts complicate a simple distinction between “international” and “local” students: 4+0 participants remained within an internationalised curriculum, while 2+2 participants added physical mobility and independent living. Because the sample was not designed for systematic route comparison, the findings do not establish that one route produces greater pressure. They indicate only that pathway and location may shape the domains in which participants narrated self-management and support.

Across the results, the four elements of institutionalised autonomy recur as an empirical pattern rather than remaining only a theoretical definition. Institutional arrangements expanded choice or reduced direct supervision; criteria of success persisted but became distributed across grades, language, research, mobility, internships and careers; students assumed planning, help-seeking, time-management and risk-judgement work; and the consequences varied with participants’ preparation, confidence, and access to information and support. This synthesis provides the empirical basis for the conceptual comparison developed in the Discussion.

6. Discussion

6.1. Research-Question-Led Discussion

Regarding RQ1, the bounded public corpus presents autonomy through three connected features: an active and self-reliant learner, choice among pathways and opportunities, and support represented as available but requiring student activation. This discourse constructs a self-directing student who is expected to interpret options, plan ahead and initiate help seeking. The finding concerns public institutional representation only. Because the materials were accessed on 11 June 2026, after the May 2026 interviews, it does not establish how provision operated during data collection or whether participants encountered any of the pages.

For RQ2, participants experienced institutional freedom as both enabling and demanding. Some valued the independence, flexibility and opportunities for exploration it provided, while others recalled early disorientation as direct supervision diminished. Their accounts also showed that comparison had not disappeared. It became less visible and extended across grades, internships, research experience, transfer decisions, postgraduate applications and career preparation. Several participants described responding through sustained self-monitoring, assessing how effectively they were using their time, whether they were keeping pace and when they should seek help. Empowerment, hidden comparison and pressure to monitor one’s own progress could therefore coexist within this case.

6.2. Evidence-Supported Implications

The data directly support four bounded implications for the case institution. First, Year One provision can explicitly teach time planning, independent reading, effective use of office hours and ways of recognising and articulating a need for help. Second, proactive check-ins can be placed at transitions identified in the accounts: pathway choice, transfer preparation, internships, final-year projects and postgraduate applications. Third, formally available resources can be made more visible and embedded within programmes so that access depends less on students already knowing where and how to search. Fourth, EMI and academic-literacy support can be tailored to disciplinary reading, writing, presentation and assessment practices rather than offered only as generic language assistance. These recommendations target difficulties and support conditions directly described in the interviews and represented in the public-text corpus. The study did not evaluate any intervention, so it supports the proposed focus of scaffolding rather than claims about effectiveness. Any check-ins, embedded support or discipline-specific provision should therefore be piloted and evaluated with students at the relevant transition points. A cautious extension concerns AI literacy. In the disciplines and tasks represented in the interviews, guidance could address verification, disclosure, appropriate reliance and permitted or prohibited uses, alongside continued access to teachers, peers and discipline-specific support. Future research could examine whether such guidance improves students’ judgement and learning. The present case does not support general policy prescriptions for TNE institutions and cannot establish the effectiveness of any intervention.

Across these results, institutionalised autonomy can be specified through four linked steps. Institutional arrangements expand choice or reduce direct supervision. Criteria of success retain their force but may become dispersed or partly implicit. Students consequently assume greater responsibility for planning, help seeking, time management and risk judgement. The effects of these demands may differ according to their linguistic, informational and social resources. Learner autonomy centres on the learner’s capacity to take charge of learning (Benson, 2011), while self-regulated learning examines processes such as goal setting, strategy use, monitoring and adjustment (Pintrich, 2004; Zimmerman, 2002). Institutionalised autonomy shifts the analytical starting point to the conditions that make these capacities and practices institutionally expected and consequential. Governmentality and responsibilisation provide broader accounts of governing through freedom and transferring responsibility (Dean, 2010; Shamir, 2008). The present concept identifies the educational arrangements through which these processes became observable in this case. Its contribution is a mid-range specification of existing theory grounded in institutional conditions. The accounts did not follow a single pattern. P1 and P8 described autonomy and transition positively, while P11 continued to draw on teachers and classmates. Participants also differed by pathway, discipline and stage in the difficulties they associated with English-medium study, mobility and independent work. Their accounts indicated variation in English confidence, access to information and peer connection. These patterns point to uneven preparedness for acting on formally available choices. The study did not systematically collect socioeconomic or family-background data, so it cannot establish class-based inequality. The possibility that formally available choices produce unequal consequences remains a proposition for future investigation.

Generative AI formed a limited part of this broader pattern. The evidence came mainly from seven developed accounts; two participants mentioned AI briefly, and three provided no substantive material. In the developed accounts, AI was described as supporting explanation, translation, preparation and revision. It could also alter routes of help seeking and add responsibility for accuracy, disclosure, disciplinary fit and independent judgement. P6’s continued reliance on teachers and peers and P7’s account of hardware debugging illustrate the task-specific and relational limits of this support. These accounts suggest that AI may expand both individual capacity and judgement responsibility for some students. The study does not establish prevalence and provides no basis for describing AI as a general learning infrastructure across TNE.

6.3. Limitations and Future Research

The study draws on retrospective self-reports from 12 participants at one institution and was not designed to estimate prevalence, causal effects or systematic subgroup differences. AI-related evidence was further limited because AI was not a standardised core interview topic and the relevant accounts varied considerably in depth. Classroom practice and actual AI use were not observed, and no usage records were collected. The public materials were gathered after the interviews, so they illuminate institutional representation but cannot establish contemporaneous implementation or participants’ exposure to particular texts. Translation involved interpretive judgement. A co-author reviewed the coding associated with the quoted extracts with reference to the corresponding Chinese passages, but no formal independent back-translation was undertaken. The lead author’s familiarity with the institutional setting aided contextual interpretation and may also have influenced follow-up questions and theme development. Transcript review and co-author discussion were used to examine this influence, which cannot be fully eliminated.

Comparative and longitudinal research across TNE settings could examine whether and how the proposed four-step mechanism operates across pathways, disciplines, stages of study and different levels of linguistic, informational and social resources. Studies focused on AI should use a consistent interview protocol and, where feasible, complement self-reports with learning diaries, observations or task-based evidence of actual use. Intervention research could assess Year One induction, transition check-ins, embedded support, discipline-specific EMI provision and AI-literacy guidance. Such work would test the propositions generated by this case and clarify the conditions under which they apply.

7. Conclusions

In this case, “everything depends on yourself” encapsulates the central finding. Public materials presented choice, independence and available support as opportunities. Participants’ accounts revealed the practical work required to act on them, including recognising expectations, organising time, seeking help and judging risk when criteria for success were not always visible. Freedom could support exploration and confidence, but it could also bring early disorientation, hidden comparisons and continued self-monitoring. Institutionalised autonomy names the relationship between expanded student discretion and the transfer of navigational responsibility. As a context-sensitive, mid-range specification of responsibilisation, it locates self-management in the institutional arrangements that make choices consequential and directs attention to how students encounter these demands under different linguistic, informational and relational conditions. For a subset of participants, generative AI was one limited resource within this broader pattern, extending support while adding responsibility for verification and judgement. Its relevance beyond this case remains an empirical question: when institutions expand freedom, what work and resources are required to act on it, and how are the consequences distributed?