How do researchers write Narrative CVs? Does language matter?
Do underrepresented researchers write narrative CVs differently? And which features of their language influence how these CVs are judged?
Narrative CVs offer researchers the opportunity to present their careers through their own stories, focusing on achievements, contributions, and broader impact rather than just traditional metrics. One of their key aims is to reduce bias in evaluation. However, as DORA has highlighted, not all researchers find it equally easy to write these CVs. Language skills and expression can pose real challenges – especially for researchers from underrepresented backgrounds and whose first language is not English – and need to be considered when designing support mechanisms.
This raises important questions. Do underrepresented researchers write narrative CVs differently? And which features of their language influence how these CVs are judged?
To explore these questions, we analysed a sample of narrative CVs written by researchers, which were then evaluated by reviewers. We looked closely at the language used in these CVs using two main tools: LIWC-22, a linguistic software that captures a wide range of psychological and structural features, and BERTAgent, which identifies expressions of agency – the capacity to act intentionally and achieve goals – or lack of it in text. For more detailed information about sample and analyses, see our online report.
The findings were revealing. Women tended to use more agentic language compared to men, while also placing greater emphasis on social aspects – suggesting a capacity to highlight both individual achievements and collaborative contributions. Researchers from ethnic minority backgrounds were generally less likely to use leadership-related words. Ethnic and sexual minority researchers were less likely to include language that conveyed certainty.
We also checked whether receiving feedback on a narrative CV made a difference. The answer was yes. Those who received feedback included more agentic and cause-effect language. This suggests that feedback plays a meaningful role in shaping more effective and credible narratives.
Next, we examined which linguistic features were associated with the way both the CV and the researcher were evaluated, including perceived clarity of the CV and perception of the researchers in terms of competence, contribution, and authenticity. CVs that reflected a lack of agency, relied heavily on negations, or used language highlighting discrepancies were consistently rated less favourably. These patterns affected not just the evaluation of the CV itself, but also how the researcher behind it was perceived.
So what does this mean for researchers preparing their narrative CVs? The way you write matters. Language that clearly conveys your ability to achieve goals and affirmative language, while avoiding language that may convey uncertainty, can make a significant difference.
Our project provides initial evidence that differences in language use exist across groups of researchers and that specific linguistic patterns are linked to less favourable evaluations. Understanding these dynamics is a key step toward making narrative CVs a more equitable and inclusive tool for recognising excellence in research.
Report: https://doi.org/10.15126/901716
Data of the study are available here: https://osf.io/2qnaf/#:~:text=%2010.17605/OSF.IO/2QNAF