Open science is built on a simple promise: knowledge should be available to everyone. We talk a lot about paywalls, open data and open peer review. But there’s another barrier we don’t talk about nearly as much, and it’s hiding in plain sight. It’s the language this article is written in.
Why science speaks english?
Most of the leading journals in each field are in English (we are not going to go into the historical reasons that have led to this situation, as that would take us off topic). So, the rule is simple: if you want your work to be seen, you need to publish in English. A 2020 study by Valeria Ramírez-Castañeda notes that about 98% of science papers are in English.
And, in theory, this doesn’t sound so bad. One shared language makes things easier. A researcher in Brazil can read a paper from Finland. A team in Japan can work with a team in Kenya, with no translator. Science moves faster when everyone speaks the same language.
However… it’s not all good news
Here’s the problem. A shared language only works if everyone can actually use it, and many researchers can’t use English as easily as native speakers can.
A 2023 study by Tatsuya Amano and colleagues surveyed hundreds of environmental scientists from different countries. It found that non-native speakers were at a disadvantage in every activity they looked at, from reading and writing papers to presenting at conferences. They needed a lot more time to read a paper and to write one, and their work was more likely to be rejected because of the writing rather than the science.
And it doesn’t stop at the people producing research. Think about who reads it. Picture a doctor in a small town in Spain, a local policymaker, a teacher, or a patient trying to understand a new treatment. They may have a basic level of English, or none at all. A paper might be open access and free to download, but if they can’t understand it, is it really open? The door is unlocked, but the sign on it is in a language they can’t read.
There’s also a loss running the other way. Plenty of valuable research is published in Spanish, Portuguese, Chinese, French and many other languages. But because the global scientific community mostly pays attention to English, that knowledge often gets overlooked.
Can AI fix this?
It’s tempting to say that artificial intelligence has already solved the problem. Translation tools are better than ever. A researcher can write in their own language and get a decent English version in seconds, and a reader can paste an English paper into a chatbot and get a summary in Spanish.
That’s genuinely helpful, and it’s already changing how many people work. But it raises questions too. Relying on AI can also mean handing your text to commercial tools, which not everyone can afford or wants to use. And that also raises confidentiality issues. Some texts, like unpublished manuscripts or peer review reports, simply shouldn’t be uploaded to these tools. And there’s a deeper question: if AI makes it easier than ever to write in English, does it help people take part in science on their own terms, or does it just make English even more dominant?
There’s also a limit to what AI can do. Writing gives you time to check, edit and translate. Talking doesn’t. Live translation tools exist, but they’re still far from perfect, and they can’t keep up with a fast, informal chat. This is where the language gap hurts the most, and it’s much harder to fix.
Figure 1: AI-generated image with GPT-5.6 illustrating the opportunities and limitations of artificial intelligence in overcoming language barriers in science.
About this article
You might have noticed something by now. This article is about language as a barrier to open science, and, at least for now, it’s only available in English.
That’s not an oversight on our part. It’s the same pressure we’ve been describing. We wrote it in English because that’s how it will reach the widest international audience, and that choice means some of the people we most want to reach, including many of our own neighbours, may struggle to read it. It’s a small example of a big problem…
Some questions to leave you with…
We don’t have the answers, but here are some questions we keep coming back to:
- How do we actually fix this? And whose job is it: researchers, journals, universities or funders?
- Should researchers publish the same paper in two languages? For example, one version in English and one in Spanish, so at least their own community can read it quickly. And if so, who pays for the extra time and work?
- Should journals accept papers in other languages and take care of the translation themselves?
- How much should we trust AI translation in science? And who checks that it got things right?