When geopolitics shapes what Chinese international students feel able to say

By Student Voice AI

Updated Apr 10, 2026

When campus politics turns students into symbols, student feedback changes with it. At Student Voice AI, we see that shift surface quickly in open-text comments about belonging, safety, and whether students feel able to speak honestly. In a recent paper in Higher Education, Gengqi Xiao and Jing Yu examine how U.S.-China geopolitical tensions shape Chinese international students’ language practices and identity negotiations. For UK higher education teams, the takeaway is practical: what students say, and what they leave unsaid, can be shaped by perceived risk and visibility, not only by teaching quality or overall satisfaction. [Paper Source]

Context and research question

Universities are international spaces, but they are not insulated from global politics. When geopolitics becomes part of everyday public discourse, international students can find that their accents, names, and participation are interpreted through assumptions about national identity. For student experience teams, that is not abstract context. It affects who speaks in class, who asks for help, and who trusts personal tutoring and feedback channels enough to use them.

Xiao and Yu use a raciolinguistic perspective to ask how U.S.-China geopolitical tensions influence Chinese international students’ (CISs) understanding of language and identity, and how those tensions shape what happens in academic and social spaces. The study draws on qualitative interviews with 12 CISs at a large Midwestern university in the United States, analysed through thematic analysis. That makes the paper especially useful for teams trying to interpret student feedback more carefully, because it focuses on the conditions that shape whether feedback is voiced at all.

Key findings

First, the paper shows that linguistic features can become status signals. Students described accent- and name-based hierarchies where Western-aligned speech is granted credibility, while Chinese linguistic features are treated as “other”. The practical implication is immediate: the same contribution can be received differently depending on who makes it, and how their language is perceived.

Second, the findings highlight protective adaptation. Students adopted strategies to reduce visibility and risk, including concealing their national identity, selectively code-switching, and performing “Americanness”. These strategies can be rational, but they come with a student voice cost: when participation feels unsafe, students contribute less, ask fewer questions, and may avoid putting specific experiences into written feedback. That means silence in survey data may reflect self-protection, not indifference.

"Students adopt adaptive strategies, such as concealing their national identity, selectively code-switching, and performing “Americanness,” to protect themselves in academic and social spaces."

Third, the authors describe how macro-level tensions become micro-level experiences. Geopolitics shows up in classrooms through tokenisation and in public life through harassment and fear. Tokenisation matters because it turns some students into representatives for a country or a political position, rather than ordinary members of a cohort. Harassment and fear matter because they change day-to-day routines and shape which spaces students avoid. For universities, that is a reminder that belonging risks often surface first in everyday interactions, not formal complaints.

Finally, the study emphasises that politicised identity is not just background context. Language performance is evaluated through a geopolitical lens, which can constrain participation and well-being. The paper also notes moments of critical reflexivity, solidarity, and resilience, but the core implication for institutions is clear: inclusion work has to anticipate these dynamics, not only respond after harm has occurred.

Practical implications

For UK higher education teams using module evaluations, NSS-style surveys, or targeted international student feedback, the value of this paper is not just descriptive. It points to concrete design choices that can make student voice work safer, more trustworthy, and more useful, especially when teams learn from teaching evaluation surveys students and staff help design.

Treat “speaking up” as a safety issue as well as an engagement issue. If students may be judged through accent, name, or geopolitics, build psychologically safe routines for participation in seminars and group work, and state explicitly that students are not expected to “represent” a nation or group.
Use student voice questions that capture mechanism, not only sentiment. Open-text prompts that ask what made a space feel safe or unsafe, and what changed that feeling, give teams evidence they can act on, rather than generic satisfaction scores.
Design feedback channels for trust. Where risk is perceived, anonymity, clear escalation routes, and visible “you said, we did” loops matter. If students expect exposure or inaction, they will self-censor long before a survey closes.
Analyse inclusion signals by cohort where governance allows. Segmenting open-text feedback by international status, programme, and other characteristics helps teams detect patterns that averages hide, so they can target support more precisely.
Connect qualitative and scaled measures. Short belonging or safety items show direction, but what ethnic-minority students mean by belonging usually becomes clearer in free-text comments, whether that is tokenisation in class, peer exclusion, or fear of harassment.

Student Voice Analytics supports this kind of work by classifying belonging, safety, discrimination, and participation themes in open-text feedback, and by helping teams track how those themes shift over time across cohorts.

FAQ

Q: How can we use surveys to understand whether international students feel safe to speak up?

A: Ask about specific contexts, not only overall satisfaction. In addition to a short belonging or safety item, include open-text prompts about seminars, group work, and staff interactions, for example “What makes it easier or harder to contribute in teaching sessions?” and “Have recent events outside the university affected your sense of safety or belonging on campus?”. Make confidentiality clear, and close the loop with visible actions so students can see that sensitive feedback leads to change.

Q: What should we be cautious about when applying this study to a UK institution?

A: The study draws on interviews with a small number of students in a single US setting, so it does not estimate prevalence. Its value is in surfacing mechanisms: how geopolitics and racialised perceptions can shape language, identity, and participation. For UK decision-making, treat the findings as hypotheses to test against your own evidence, including pulse surveys, open-text comments, incident reporting trends, and targeted follow-up conversations.

Q: What does this mean for analysing open-text student comments at scale?

A: It highlights that silence can be a signal. If students feel exposed or tokenised, they may avoid leaving identifiable detail in comments, or stop responding altogether. Analysing open-text at scale helps you spot recurring themes such as belonging, safety, discrimination, and classroom climate, but it works best when paired with segmentation and careful interpretation of who is responding, who is not, and which experiences may be under-reported.

References

[Paper Source]: Gengqi Xiao, Jing Yu "Politicized identity and language practices: Understanding Chinese international students’ language ideologies amid U.S.-China geopolitical tensions" DOI: 10.1007/s10734-026-01623-3

Request a walkthrough

Book a free Student Voice Analytics demo

See all-comment coverage, sector benchmarks, and reporting designed for OfS quality and NSS requirements.

All-comment coverage with HE-tuned taxonomy and sentiment.
Versioned outputs with TEF-ready reporting.
Benchmarks and BI-ready exports for boards and Senate.

Prefer email? info@studentvoice.ai

UK-hosted · No public LLM APIs · Same-day turnaround