
Today, most breakthrough science is published in English. But in many parts of the world, English is not a native language, making translations essential to bring news, information, and perspectives to diverse audiences. Moving science-focused content across language boundaries requires more than running articles through Google Translate or flipping through a multilingual dictionary in search of a word-for-word translation. Translations have to be both accurate and culturally sensitive—a tough task with high stakes. Poorly translated science doesn’t just confuse, or fail to communicate. It can actively seed misinformation and distrust.
As a multilingual science writer and translator, I often find myself thinking through these challenges. Around the start of the COVID-19 pandemic, for instance, I volunteered to put together communication resources in Kannada, a language that 43 million people speak in South India. I was also translating a children’s book on COVID-19. But most of my experiences have focused on one Indian language, and I know that other translators working in different languages would certainly encounter their own difficulties in their work—some that overlap with my own, and some that would not.
I recently spoke with five translators and asked them about their experiences, both before and during the pandemic. I was first struck by the commonalities: Translating science content is challenging, irrespective of the languages involved. But there were also subtle differences and nuances. One translator, working in the South Indian language Tamil, described how he removed a reference to yellow finches when translating a text on brood parasitism, wherein birds sneak their eggs into the nests of other species. The better references for his Indian audience were cuckoos and crows, which his audience could better relate with. Another translator noted the challenges of working in South Africa, where Indigenous languages, such as isiZulu, lack the local infrastructure to write and report on science, often requiring him to invent entirely new words in his translations to describe unfamiliar scientific ideas.
Differences in local dialects also presented a hurdle for another translator who speaks Arabic, which varies between the countries that speak it—a rough analogue to the challenges I encountered when translating into Kannada, which also has many regional dialects. Across languages, most of the translators I spoke to said that the most effective translations were rarely word-for-word replicas of the original text. Many of them had their audiences in mind, knowing that the best communication would require sometimes replacing unfamiliar concepts and references with more familiar, culturally sensitive examples, even coining new terms when needed—a process sometimes called transcreation.
All the translators I spoke to found their work highly gratifying. But most of them also told me they are underpaid—sometimes even working for free—and their work often goes unnoticed. Despite the essential nature of their work, translators still don’t always receive bylines in stories. And yet, all of us still do the work with passion and commitment to serve our readers—something that comes through clearly in the dialog I had with my colleagues. (This discussion has been edited for length and clarity.)
The translators who participated in the session are:
- Mohamed Elsonbaty Ramadan, freelance science journalist, founder of the Arab Science-Media Hub, and co-founder of the Arab Forum of Science Media and Communication. Ramadan translates science stories from English to Arabic.
- Daniela Navarro, Spanish translator at Planeteando, a Spanish science-outreach blog that covers earth sciences. Navarro translates science-based stories from English to Spanish for Eos and other outlets.
- Sibusiso Biyela, a digital science communicator at ScienceLink in South Africa. Biyela translates science stories from English to isiZulu, a South African language. He also leads a pan-African initiative to translate science-based content into six African languages.
- T V Venkateswaran (TVV), a senior scientist at Vigyan Prasar, a science communication initiative by the Department of Science and Technology, Government of India. TVV translates science stories from English to Tamil, a South Indian language.
- Zhang Boran, a freelance science writer. Zhang translates science stories from English to Mandarin Chinese.
Spoorthy: Let’s begin with what considerations go into your (or your editors’) decisions about whether to translate a story. Are there times when a good story is just not a good candidate for translation?
Daniela: In my case, the articles are selected by the editor. The translators [then] review the list and select those that they feel more comfortable with translating from English to Latin American Spanish. “Hot topics” or [those] most visited by the Eos audience are preferable. We as translators can also search the Eos website and propose a piece to translate. For example, I proposed translating “If Disney Princesses Were Earth and Environmental Scientists.” The translated article reached a significant Hispanic audience.
Boran: There is one good story that comes immediately to mind. A few years ago Randall Munroe (the author of the webcomic xkcd) wrote a piece [for The New Yorker] using only the most common thousand words [in English] to explain Einstein’s relativity. Chinese works differently from English, [so] it’s very difficult to convey. I had my doubts, but I like xkcd, so I still tried my best. It turned out to be a dud. Our readers didn’t like or understand the piece. So something too language- or culture-specific would not be a good candidate.
Another situation where we have to pass up a story is when the public discourse [around a particular topic] is unfriendly. If there is a conspiracy going around, like microchips in vaccines, we will refrain from [translating stories that report on] new advances in real nanotech chips, so as to not add fuel to the fire. It’s unfortunate, but the current situation in social media really is not ideal. Political situations are even more problematic.
Sibusiso: There are stories that I would like to translate in isiZulu, but I reconsider when there are too many words I would have to explain—or coin new terms—for the reader to understand what’s going on. This makes me consider stories that would make sense to the lay reader, based on the science either impacting them in some way, or if it references something they can easily identify with. That’s the hook, and without it, I don’t consider a story for translation.
Mohamed: The most important factor is the target audience’s interest. This shifts the translations to focus on certain fields, such as health, water, energy, environment, and technology, [rather than] basic sciences and mathematics. As a result, many good stories can be skipped. On the other hand, many “poor” stories can be translated as they seem more interesting to the audience. State censorship, or even self-censorship, can play a role, too. Some scientific stories will not be translated [or might be translated differently] due to political or cultural reasons. For example, in a cover story about “The War on Science,” published by National Geographic magazine, the word evolution was removed from the cover of the Arabic version of the issue.
TVV: Editorial practices impact my selections. For example, most outlets do not prefer long-form articles anymore. The linguistic abilities of the audience also create a hurdle. Some write-ups address people with higher education. However, as higher education in India is mainly [conducted in] English, stories translated into Indian languages are not welcomed. Further, [while] some of the stories are good to read in English, idioms or allusions to literature or myths are not “translatable.” Ultimately, I choose stories that do not call for too much prior knowledge, are not steeped in the cultural cosmos of the original writer, and are of manageable length.
Spoorthy: What type of training/skills are necessary to equip a person to do high-quality translations of science journalism?
Daniela: I think being fluent in both languages is definitely a must. You could come across slang when the writer is quoting interviewees, or when the writer wants to be informal. For sayings or proverbs, it can be challenging. For instance, in English, you say “it’s raining cats and dogs” when torrential rain is happening. In Latinx Spanish, we don’t have this [idiom]. So I think how a Latinx would refer to it: “It’s pouring rain” (Lloviendo a cántaros). In addition, you need to know technical terms in the discipline or area you’re translating.
Sibusiso: It is always important to understand your audience. For translations of science content, I consider the readers I am writing for to decide the kind of tone I will write the article in, whether it be casual, conversational isiZulu, or something more formal. Understanding the culture you’re translating for can help you avoid certain pitfalls, such as disrespectful language that might seem fine in English but [would be] a faux pas in isiZulu.
The audience I usually write for has antagonized science as being too difficult, or something only associated with English-speaking white people, so I try to explain in detail what science was done, and what the researchers did, to dispel the notion that something magical or clandestine happens in laboratories and research institutions.
TVV: Fluency in both languages is, of course, a must, but not enough. Most engaging articles will have cultural elements that make connections with the readers. Adages, idioms, and allusions to myths or legends abound in popular-science writing. This means the [translator] must be widely read—beyond science.
Mohamed: From my experience, I think fluency in both languages with advanced writing skills is essential. Having a scientific background or translation background can be a huge advantage. In the Arab world, there are no university degrees that specialize in scientific translation. However, there are a few short training courses on scientific translation, especially medical translation. These have enabled many scientific translators to start their career. At the same time, there are many other translators who have learned the craft by practicing and receiving feedback. I have never received any training in translation throughout my career, but I think that having training in my early career might have been an advantage.
Boran: I’m tempted to say you need relevant scientific knowledge on the topic, but if a translator needs prior knowledge to understand a journalistic piece, wouldn’t it mean the piece failed to reach the general audience? Also, it seems to place too much burden on a translator. My current guess is translators need to know how science and journalism work—i.e., how scientists make discoveries, and how journalists usually report on them. Basically, they need to receive similar training to the author of the text.
Understanding the general culture is certainly good, but I don’t think it’s much of an issue. Understanding the current hot topic seems to be more important (the #MeToo movement in academia, or the replication crisis in psychology, or a newly dropped COVID conspiracy, etc.).
Spoorthy: After receiving a story to be translated, what are some of the first steps you might take, and what are some of your go-to resources during this process?
TVV: First, I would tease out the analogies, similes, metaphors, and thought experiments the story uses. I will examine if the same would work for my audience. Second, I would examine the “interpretation” or “meaning making” the writer has projected in the story. Third, I would list all the technical words in the story and see which of them have an easy, well-known equivalent in the target language, [and] which of them need to be newly coined. Some of the technical-terms corpuses [that have been] created by language experts come in handy. Fourth, to harmonize the story with the target-language readers, I would look for suitable proverbs, adages, folk tales, or myths that can be alluded to.
Daniela: I read the story, identify new terms if I don’t know them in English, and find their meaning in the Oxford or Cambridge dictionary. Then, [I] identify scientific terms and look for a literal Spanish-translated word. If there is slang, [I] figure out how to express them in Latinx Spanish. Sometimes, I have problems translating when I am not familiar with the science discipline. When this happens, I reach my GeoLatinas community for advice. Some of them even direct me to English-Spanish online dictionaries. For example, in geology, I use Ceramica [an online English-Spanish geology glossary].
Mohamed: First of all, I read the entire story just to know what I should expect. Then, I search for the keywords that I do not know or understand and find the Arabic equivalent. I tend to use online dictionaries, specifically Almaany.com [an online Arabic-English dictionary] for terms in Arabic, and Reverso [a multilingual online translator] for phrases and expressions.
Boran: First step is to read the story thoroughly. I find that some newbie translators read and translate at the same time, thinking it will save trouble. Definitely don’t do that.
OK, the actual first step is to circle the weird part, the part that feels unnatural [or has a] leap of logic, etc. More often than not, it means I misunderstood something. [I then] search the [words] in the dictionary to see if it has some unfamiliar usage.
Spoorthy: What are some of the common mistakes that you’ve seen happen in translations of science stories?
Daniela: As a translator, I always take care when translating. Sure, you can use Google Translate (which I do) or other apps for quick translation. However, you’re responsible for delivering a correct translation, expressing the author’s message. Additionally, I must add that I’m Chilean, and Chilean Spanish is the most complex Latinx Spanish language (a comparison I always use is the English from New Zealand). We use so many slang [expressions] and sayings, that in my case, I have to re-read my translations to make sure I follow the correct grammar and word usage. Fortunately, my editor is Mexican (they use a more formal Spanish) [and] sometimes finds mistakes and lets me know.
Boran: Hoo boy. I’ve got a whole list due to my work as an editor:
- Not reading the whole article first
- Getting lazy and using Google Translate
- Not using the dictionary and Wikipedia often enough
- Neglecting English-specific grammar elements like tense
- Trying too much to preserve the word order in a sentence
- Freestyle translating [translating what the translator thinks is the gist of the sentence without understanding the complete sentence structure in English]
I find many new translators have another difficulty with science pieces: They are not comfortable with complex and weird sentences and are always trying to “bend” them. Basically, they will try to completely rearrange the sentences to make them sound natural, sacrificing the meaning in the process.
What I do is to identify the general grammar structure, find the corresponding Chinese way of expression and use it, so I rearrange the sentence according to rules rather than how it “feels.” It’s not really a well-composed set of rules, but it works for me.
TVV: When common words are used as technical words, it causes confusion to some who are not familiar with the subject. For example, the word force means a specific concept in physics, but can mean police force, violence, etc., in everyday usage. Also, nowadays when people use Google Translate, hilarious mistakes occur. Take the word migration. When it is a change of residence due to a new job or studies, the Tamil word used is iṭampeyarvu; however, it is not appropriate for bird migration. The term used for the seasonal migration of birds or animals is valacai pōtal.
Mohamed: Among the common mistakes for translation into Arabic are: translating scientific terms without considering the story context; translating statistics (confusing relative and absolute risk; average and median); confusing different types of diseases; translating names of researchers and universities (literal translation of written letters rather than how they are pronounced); [translating] without considering Arabic grammar. Finally, depending on Google Translate can be catastrophic for Arabic.
Sibusiso: I’ve read a lot of stories that seem to be direct translations of published English versions by the same publication. To me this feels like a tick-box exercise, as if the translated story was an afterthought. In some instances, the translated story still has English-sounding sentence structure and conventional newspeak not found in the translated language. This makes the reading experience unpleasant and can create the idea that science stories in isiZulu are inherently bad.
Spoorthy: How do you as a translator handle quotes (and anything else that needs high fidelity) when they are said in a different language? How do your editors/publications fact-check them? Are there any best practices around translating quotes?
Daniela: I try to translate them literally, especially if they’re using scientific terms. Now, when quoting, [if] there are sayings in Spanish that don’t make sense in the translation, then I try to find another way of expressing. Sometimes, I have doubts about the person’s [gender], so I Google their names and affiliations to make sure. In Spanish, we use many words that end according to the subject’s gender.
Sibusiso: Unfortunately, my editor cannot fact-check the articles I produce, so we usually outsource the task of proofing to someone who is proficient in isiZulu in hopes of spotting any glaring grammatical errors. The best way I have to fact-check the science somewhat is to ask if the article makes sense to a non-scientist and ask them to tell me what they understood about it. If there is an English version [accessible to the editor] that goes with the article, that’s the one the editor can fact-check and proof. The bigger worry for me is that the isiZulu version is readable.
TVV: Most of the work I do is really transcreation. At times, the quotes are not directly translated word by word, but the gist is given without quotes. When quotes are necessary, they are translated and used. In such a situation, I ask a few friends to read the original and translated materials and seek their suggestions.
Spoorthy: Do you, in your translation work, ever engage with the original authors of the piece? Can writers do anything to make translators’ lives easier?
Daniela: I don’t have direct contact with the authors for Eos translations. In GeoLatinas Blogs, we follow a process of contact between the translator and the author. So, if there are doubts or questions about the original piece there, the translator can freely engage the author.
TVV: Until now I have not engaged with the original author(s), but this question sparks some thoughts. The original author must have understood the science behind the popular-science story, to the extent necessary for popular-science writing. They may also be useful in suggesting additional sources that a transcreated work may require.
Mohamed: I have not engaged with the original author of the piece. However, I think that if the author takes into consideration that her/his work may be translated into another language(s), this can help in producing more translatable work. For example, the author may consider avoiding using certain metaphors, idioms, or phrases that are difficult to translate. In addition, the author can even tailor the content to be more universal to target different audiences in different languages.
Spoorthy: How can editors/publications best support translators? Considering the vital role of translators in the Global South, where English literacy is not too high and the pay for their service is abysmal, how can the broader science-journalism community support translation of science-based journalistic content? What do you as a translator expect?
Sibusiso: I think the best support I would appreciate is an open discussion like this one about the challenges we face and the strategies we use to overcome some of them. Of great help would also be efforts by publications to invite readers to take part in improving the translations, to talk about what they like, or stuff they find hard to understand in the content that’s been produced. For some of us, we’re making these strategies up as we go and hoping they work, but by involving the people we are writing for, we can create engagement between the publication and its readers. You might even find that the readers don’t want the translated articles in the first place; unfortunate, but an important data point.
TVV: The current payment for translation is inadequate to attract quality and quantity. Perhaps some mechanism to provide support would be required. One possibility is establishing translation syndicate services, with part subsidy.
Daniela: I think a best-practice guideline or checklist is a very good idea, especially for the new translators. I’ve only been translating science journalism as a volunteer for more than a year. Now, going backward and doing a reflection, I think an [introduction] session with the editing staff can be a great idea to discuss the direction of translations and best practices. The problem with voluntary work is that sometimes you don’t count on training and the parties trust in their experience mostly. Nevertheless, my experience with my editor at Planeteando has been fantastic. Anthony makes sure that each translation is appropriate for our Latinx community and gives me feedback so I can improve.

Spoorthy Raman is a Toronto-based independent science journalist. Her words have appeared in Science, Deccan Herald, The Print, and AGU Blogosphere, among other publications. She is also the author of India’s Adventures in Space, an illustrated children’s book, and writes Tailspin, a monthly column for Deccan Herald on life with pets, with a sprinkle of science. Until 2020, she was the managing editor of Research Matters, a multilingual science-news portal covering science from India. She was a recipient of the 2019 AAAS-EurekAlert! Fellowship for International Science Reporters. She is on Twitter @RamanSpoorthy.