A global survey of 908 environmental scientists across eight countries reveals that scientific productivity is influenced more by identity than by merit. Women publish up to 45% fewer English-language papers than men, a gap that persists across career stages. For female non-native English speakers from lower-income nations, productivity falls by as much as 70% compared with male Anglophone peers from wealthy countries. These disparities raise questions about the fairness of research assessment and call for systemic reform.

Gender at the Core of Inequality

Science has long presented itself as a meritocracy, but women continue to face barriers that blunt their contributions. The survey, conducted between June and October 2021, gathered responses from researchers in Bangladesh, Bolivia, Britain, Japan, Nepal, Nigeria, Spain, and Ukraine. Participants had published at least one first-authored peer-reviewed English-language paper, making them an active subset of the scientific workforce.

The gendered disparity in productivity was evident across contexts. On average, women published 45% fewer English-language papers than men at early career stages. The disadvantage narrowed with time but did not disappear. By late career, female researchers had accumulated more than a dozen fewer publications than male colleagues with comparable years of experience.

These findings echo decades of evidence that women are underrepresented in virtually every metric of academic recognition. They publish fewer articles, receive fewer citations, are less successful in securing grants, and are underrepresented in editorial boards and senior positions. They perform more teaching and administrative service tasks essential to universities but undervalued in promotion systems. Parenthood and caring responsibilities, which are more commonly shouldered by women, further slow research output.

The study underscores how disadvantages accumulate early. For women in their first decade of research, the productivity gap was most pronounced. At precisely the point when hiring committees and funding panels scrutinise publication counts, women appear less competitive. This disadvantage then cascades into reduced opportunities, smaller networks, and weaker future productivity—a textbook example of the Matthew Effect, in which initial inequities amplify over time.

Unlike linguistic and economic disparities, the gender gap remained persistent even when non-English outputs were counted. Women published less than men, regardless of language. This robustness suggests that structural gender inequality is harder to mitigate through broader metrics alone.

When Language and Economy Multiply Disadvantage

Although gender was the most consistent predictor of lower productivity, language and economic background also played significant roles. Non-native English speakers faced a productivity penalty of roughly 15% compared with native speakers, while researchers from lower-income countries experienced a 10% reduction.

The burden is not merely statistical. Non-native speakers reported spending 51% more time writing papers in English. They experienced rejection due to language issues 2.6 times more often than native speakers and were 12.5 times more likely to be asked to revise their English during peer review. For scientists in lower-income contexts, underfunding, limited travel opportunities, and systemic biases in review processes compounded these challenges.

When combined, these disadvantages became staggering. A female scientist who is also a non-native English speaker produced 60% fewer English-language publications than a male Anglophone peer from a wealthy country. Adding a lower-income background raised the deficit to 70%. In absolute terms, such researchers produced up to 20 fewer publications by late career than their privileged peers.

This finding reveals a cumulative disadvantage: gender, language, and economy do not act independently but multiply one another. The system thus creates near-insurmountable barriers for those at the intersection of these identities.

Yet the analysis also uncovered a paradox. When total publications—including non-English papers—were counted, the language and economic gaps narrowed or even reversed. Non-native English speakers from lower-income nations often produced more total papers than their Anglophone counterparts, suggesting that their contributions are substantial but undercounted by global metrics. The one gap that persisted across measures was gender. Women continued to publish less, whether in English or other languages.

Box 1. Estimated productivity gaps

Scientist category	Reduction in English-language output
Female vs male peers	−45%
Female + Non-native English	−60%
Female + Non-native English + Low-income background	−70%

Source: Amano et al., PLOS Biology (2025).

Why Productivity Metrics Distort Science?

The reliance on English-language publication counts as the primary measure of productivity has become increasingly questioned. Yet such metrics still dominate decisions in hiring, promotion, and funding. The study’s authors argue that this practice systematically underrepresents the contributions of women, non-native English speakers, and scientists from poorer countries.

Australia’s Research Opportunity and Performance Evidence (ROPE) policy offers a partial precedent by allowing researchers to declare career interruptions such as illness or parental leave. But linguistic and economic barriers are not recognised as legitimate factors, even though the study demonstrates their profound impact.

Ignoring non-English outputs further skews the record. Research in biodiversity, climate change, and public health often appears in local languages, yet databases like Web of Science and Scopus rarely index such work. As a result, evidence critical to global challenges is systematically marginalised. Recognising these publications would not erase inequality, but it would better reflect actual productivity.

Broader Implications and Reform

The implications of undercounting are not limited to individual careers. Science thrives on diversity of perspectives. Teams with greater gender and linguistic diversity consistently produce more novel and higher-impact findings. Female inventors are more likely to develop patents addressing women’s health needs, while non-English research often captures ecological data unavailable elsewhere.

By failing to recognise these contributions, the current system distorts not only who succeeds but also what questions are asked and what knowledge is valued. The underrepresentation of women and non-Anglophone scientists narrows the scientific agenda, perpetuating blind spots in both fundamental and applied research.

The findings also highlight generational shifts. Younger non-native English speakers sometimes outproduce their seniors in combined outputs, partly because of increasing pressure to publish in English and partly because early-career scientists adopt AI-based writing tools more readily. But these tools cannot compensate for structural inequities in funding, collaboration, and recognition.

The study concludes with a call for explicit recognition of linguistic, economic, and gender disadvantages in research assessment. Reform could take several forms:

• Inclusion of non-English publications as legitimate outputs in hiring and funding.
• Support for language editing and translation services as standard research costs.
• Adjustment for structural burdens such as unequal teaching and service loads.
• Training for evaluation committees to recognise compounded disadvantages.

Such measures would align with the principles of the Declaration on Research Assessment (DORA), which urges evaluation of research on content rather than venue or language. Resistance is likely, given the convenience of English as a lingua franca. But universality built on a single language excludes as much as it includes. Treating English dominance as neutral ignores the hidden costs borne by those who must translate themselves to be heard.

The evidence from 908 scientists across diverse contexts is unambiguous: women publish less than men, and the gap widens dramatically when compounded by language and economic barriers. Productivity metrics centred on English-language output not only misrepresent contributions but also entrench inequality.

Science cannot claim universality while silencing half its talent and marginalising large parts of the globe. Recognising non-English outputs, adjusting metrics for structural disadvantage, and reforming evaluation practices are essential steps. Until then, the ledger of science will remain unequal, counting less of what matters and more of what entrenches privilege.