Non-native English speaking scientists work much harder just to keep up
theconversation.com
theconversation.com
I grant that it’s incredibly easy for me to sound dismissive of this issue as a native English speaker, but I don’t think throwing AI-as-a-buzzword at the problem is much of a solution.
That said, I did really like this potential solutions described by the article; sounds like a great tactic not only to address this issue, but to build relationships and cross-border research networks.
“Conference organisers also have myriad opportunities to support non-native English-speaking participants. For example, last year’s Animal Behaviour Society conference incorporated a multilingual buddy program to improve inclusivity.”
The struggle is real, while I was in academia, I got to know brilliant people who didn't make it far enough because they had it in their own native language for too long and it became too late to invest a year or so working on their English..
That said, this article and paper don't really seem to provide any kind of solution, they just identify a problem. I don't really see what the scientific community can do about it? Aside from inventing a babel fish, this seems like a problem for education ministers around the globe.
As a non-native English speaking scientist, I don't think this is a problem that can or should be "solved" (as by, I don't know, switching to Esperanto or something like that); but awareness is important. For example, reviewers not being jerks about papers by non-native speakers would already be helpful. In this sense, I think articles like this are useful because many native speakers greatly underestimate the extra effort required.
Another example: in my particular research community, there was discussion about forbidding ChatGPT as a paper writing assistant. But every person I saw taking that position was a native English speaker. No non-native would ever defend that, because for us it's obvious that it's probably the single best technological advance of all time to bridge the paper writing gap. Native speakers were all fussed about writing being "fake", soulless, etc. whereas the main worry of many non-native is being able to express their ideas in a fluent enough form to not cause the paper to be rejected.
My comment above is translated from my broken English to simple native-ish English by GPT-4.
This is not a point that's come up in my native-English-speaking bubble. Thanks for pointing this out!
I’ve used for different languages and native speakers didn’t tell it was written by a non-native neither an IA.
---
Pretty sure I would recognize this sentence as fishy because no one would use non-native like that.
But this probably would indeed be helpful and seemingly valid to use this to improve papers. Although it did alright on this one sentence, I'd still be worried that it would change the meaning of something somewhere that a less fluent speaker wouldn't notice.
E.g. the preprint server that allows alternate languages states that:
> The reliance on English language publications to communicate scientific findings is clearly a hindrance to the open sharing of scientific knowledge
But wouldn’t the hindrance be much stronger if you’re writing a paper exclusively in a language like Portuguese instead of English?
Everyone needs to know the top 10 languages instead of settling on the one most people are already using.
Also if you're going to try dunking on me, at least get it right, English is the most spoken language in the world
https://www.berlitz.com/blog/most-spoken-languages-world
https://en.wikipedia.org/wiki/List_of_languages_by_total_num...
Especially since there are plenty of people who aren't native speakers, but got to the native level of fluency (usually, due to moving to an english-speaking country at some point and spending 99% of their time in english-only environments for many years). How do we account for those?
I always had trouble with the whole "native speaker" categorization, because I definitley don't fall under it by definition, but English is legitimately my most comfortable and most frequently used language by far, and has been this way for over a decade.
Like, the only thing that gives away my non-native status is that sometimes my accent comes out accidentally, but that's just stupid muscle memory. I am way more comfortable with english in every aspect (speaking, writing, etc.) than my actual native language, despite often having an accent in english, but never in my native language (which i have exactly zero foreign accent in).
I'm also kind of shocked at the idea that language barriers cause greater barriers to reading a paper (91% time increase) than to writing one (51%) in a non-native language.
> That said, this article and paper don't really seem to provide any kind of solution, they just identify a problem. I don't really see what the scientific community can do about it?
Guessing at the authors' motivations—for me, although it's obvious in retrospect, if you had asked me yesterday "what are some barriers to non-native speakers of English entering the sciences?", I wouldn't necessarily have thought of the extra time to read and write papers, and just this spreading of awareness is a reasonable end in itself.
Beyond that, I think part of the point is to de-externalize the cost. Currently, native English speakers bear, or at least know that they are bearing, little to none of the cost of the difficulty others have accessing their work, and I think the hope is to force people to think consciously about what they could do. For example, in math, at least, there are journals and series that publish translations of non-English works into English, but I know of few that work the other way; that could be a major good, even if it is a daunting undertaking. If that's too much—as it surely is as an immediate solution—one thing that occurs to me is that perhaps there could be some sort of training for scientific writers in how to write more accessible English—where jargon is OK (since it's to be read by specialists), but "advanced" English constructs should be avoided when possible. tetris11 mentions elsethread (https://news.ycombinator.com/item?id=36787311) that Nature apparently does something like this.
This actually makes sense to me. Reading a paper takes time on the scale of hours, and having to stop to parse difficult sentences or look up unfamiliar words can add to that substantially. Writing a paper takes time on the scale of weeks to months, and there's a lot that goes into it other than putting words down on paper. Making figures, figuring out how to argue a certain point, thinking about what examples to provide to best illustrate what you're trying to get across--these are big parts of the writing process that are more or less language agnostic in the sense that the time-intensive part can be carried out in the writer's native language and the result translated into English.
If I were to try to read a paper in my field in Russian (a language I have a decent but far from fluent grasp on, and I don't know the jargon of my field) I would expect it to take at least twice as long as reading it in English. Translating one of my papers into Russian would take a long time, but it would not double the total time spent on writing the paper.
When writing people needs just a small subset of the language. Lets say that with the 10% of the language you can express most ideas in a simple way, and be understood. When you read you face a different subset, maybe a much more wide vocabulary, a lot of phrasal verbs, not familiar constructions, local jargon or ambiguous cultural context... plus the research theme, that can be difficult to grasp by itself.
A funny and not uncommon fact when you are in an international congress is to find that everybody understands the other, until the English native researcher starts to speak.
Journals often look for this symptoms as a seal of quality, but it is not. The sad fact is that speaking English well doesn't made you an expert on science, or a genius, you are just an English-native.
Good point (although the last issue surely faces a reader of any linguistic background, it can doubtless be complicated by the others).
> Journals often look for this symptoms as a seal of quality, but it is not.
I'm not sure I understand—what are journals looking for?
The solution is to push even more English education if international scientists want to keep up.
I would have the exact same opinion if Chinese was doing the same.
I must say though, that at least when it comes to medical journal articles, some native chinese sources I've found were better written. Less verbose, flowery, jargony. Much more concise. Curious to know if this applies to other fields too.
Almost certainly the world will converge onto fewer and fewer languages, likely just one being commonly used in the end.
Sad to lose the edge that language gives to cultural uniqueness, but hard to argue the costs of communication barriers are worth it.
Another way it could go is for widely available real time translating tools. The technology is already pretty much there... but this still adds overhead to communication
I could see something like Apple Airpods (or any smart headphone equivalent) automatically detecting language and translating in real time. That would be a pretty minimal overhead.
One thing is that we probably won't lose any languages anymore. There will be plenty of books, texts and videos in the Internet, so everyone who will want to learn his mother language, can do it.
Another thing is that even today AI is perfectly able to translate both text and voice. And it's not going to get any worse with time. So even if yesterday you were forced to learn English to keep up with recent Rust developments, today you can use AI to translate articles. Popular websites like stackoverflow are lagging behind, but it's only a matter of time before they'll add universal translator and enable everyone to interact with everyone else with their preferred language.
People don't learn their mother tongues from books, they learn it from their parents. Languages begin to die out not when people can no longer learn them, but when their native speakers cease to find it useful, because a larger culture has assimilated them and they get more out of using the majority language instead.
Consider a future when a Spaniard spends most of their day talking in English to people online, and everyone else they talk to also does the same. There's an inflection point at which everyone just decides to speak in English all the time.
Oh no, no, no, no.
If I had to guess I'd say your native language is one of world's top 10 and translation works pretty well between those top 10 languages. My native language has less native speakers than there are people living in the borough of Bronx in NYC. Machine/AI translation is laughably bad. And we're in a good position because we've had written word and high literacy for long time now. For languages lacking those there's no hope basically, these languages will be dead in next 100-200 years. Even Irish is struggling despite having state support thrown at it for quite some time now, for unofficial languages that don't have neither prestige nor incentives to be learned there is zero hope left. Zero.
Thanks, this made me laugh.
No, there are enough things which can mean multiple things in one language and doesn't have the same pairing in the other.
Chances are we'll be back to banging rocks together.
Perhaps it's time to reimagine the scientific paper in a more terse format with more declarative statements and data rather than the mini-novel. That would make machine translation more tractable and may even lead to better tools to check assertions and search related work.
I'm not quite sure what you mean. tetris11's meaning is clear, and I was just making a whimsical observation, but I think it is possible (though clearly wrong in context) to parse:
> Nature is really good at giving you a spec with examples and guidelines you need to adhere to
as telling you something about the behavior of nature, analogous to "nature does nothing in vain" (Google's first auto-complete suggestion when I asked it for a sample sentence).
The italics matter. But neither case, the magazine or actual nature, should have the definite article.
Not sure why they are the gatekeepers of science. It seems like something out of the 1970s before the era of internet.
Its good to see that people are able to write a paper and just post it on a website.
We have way too many anti-intellectual voices already, and people will believe a flashy facebook page and memes more than a peer-reviewed article.
Anecdotal. Here's another: my language teachers from year 3-12 were above average if anything—most of my peers were at least conversational in Spanish (or their selected language) by high school.
At the time, the school system offered Spanish, French, German and Latin. Those who took Latin almost always had medical school ambitions. I asked.
> Compared to a fellow PhD student who happens to be a native English speaker, you’ll need 91% more time to read a paper in English. This equates to an additional three weeks per year for reading the same number of papers.
> The next big hurdle comes when trying to publish your own paper in English.
> First, you’ll need 51% more time to write the paper. Then you’ll likely need someone to proofread your text, such as a professional editor.
This is non-sense and is given without a single source other than a study likely tailor made for this result. I know plenty of ESL scientists and this is the least of their concern.
Or if at some point computer real time verbal and written translation between languages and measuring units is truly flawless, we have sidestepped the entire problem. But any solution that calls for fragmentation of human knowledge by having different papers and different conferences using different languages is tribal, not scientific.
That is not to say that learning multiple languages is bad. It's just cultural, not scientific. With any luck, science will one day give us enough leisure that we have more time for culture.
I think Mandarin will have a hard time increasing its sphere of influence, despite what cyberpunk authors may have predicted 30 years ago.
[1] https://www.cia.gov/the-world-factbook/countries/world/#peop...
PRC academic system that heavily focuses on science is spitting out educated classes that can in aggregate rival the size (and increasingly quality) of the English tertiary network. The current STEM production rate is ~5M or OECD combined. People mocked at PRC academic paper mills a few years ago, but now PRC talent is topping science and innovation indexes (controlled for quality), that's the natural outcome of exploiting talent at PRC scale. Project trend to 2050 (past 26 years of new births already locked in) and PRC can feasibly hammer out 100M+ talent in just STEM, and most importantly, increasingly be able to retain talent without the best being brain drained. Effect going forward is that PRC is going to snowball the generation of high impact scientific research and it will increasingly be written in mandarin - PRC removed mandatory English education. PRC scientific base will likely be large enough to rival English RoW, the dynamic will be more the PRC science sphere will be less inclined to integrate with the Anglo sphere. Especially if translation tech improves. IMO the mandarin science ecosystem will be fine competing if not leading in her own bubble. If anything, the PRC might prefer the gatekeeping science behind language as a competitive advantage. Global science will splinter and mandarin science will be it's own pole.
Shortcoming of India elite demographic dividend argument is, western academic institutes have limited capacity to absorb Indian talent, who let’s be frank, is replacing current cohort or PRC talent due to geopolitics which will make Indian brain drain problem worse and PRC talent retention better. English proficiency among brain drained elites will make spinning up indigenous Indian S&T at scale difficult, which India herself has a poor track record of exploiting. That can change with time obviously, but IMO there's something systemic happening with Indian diaspora doing well abroad for much longer than PRC diaspora but not meaningfully converting that into success for their birth country. Without India actually raising the ceiling for western academic ecosystem, you're looking at anglo and mandarin ecosystems competing closely. Unless anglos massively boost skill immigration, but that's politically difficult and the opposite is just as likely.
I don't buy that all science coming from 1/6th of the world (by population) will somehow surpass all the science coming from the other 5/6ths. The only way for Mandarin to rival English would be convincing non Chinese to use Mandarin. But I am glad to hear that China plans to begin developing its own science and tech.
Most spoken language English 18.8%, Mandarin 13.8%
Most spoken _first_ language Mandarin 12.3% vs English 5.1%
The short/medium term competition is 14% of native mandarin speakers vs 5% native english speakers + 13% non native speakers of different fluency. Long / very long term (2050-2100), Mandarin speakers will likely decline to 1/12 of the world, while English speakers will likely increase, maybe 1/4 of the world. But Ultimately only a fraction of talent is properly captured and exploited, so it really depends on state policies, i.e. PRC graduates STEM at 2x the rate of the US or 8x more graduates per year, more than west taking in talent via brain drain. Hence IMO enough to rival English, which is the best PRC can hope for because displacing English with Chinese is too culturally difficult.
It is useful for there to be some lingua franca. There are arguments for why English is a good candidate: - already widely spoken - economically important - simple word morphology - no grammatical gender - an alphabetic writing system that is relatively easy to learn
There are reasons why it is a bad candidate: - terrible spelling for an alphabetic writing system - definite articles, which confuse speakers of languages which lack them - arguable unnecessary tense and aspect distinctions - obligatory gender in animate pronouns
In any case, it's what we've got.
But it is also true that communicating professionally in a language you learned for that purpose imposes a burden that native speakers don't face. If you find it boring to read attempts to quantify this, maybe this article isn't for you.
The only way to fix this is to make the lingua franca a constructed language. I'm not exactly against the idea of Esperanto being the lowest common denominator. However, there's this implication in your post and the article that English being the wrong language is some thinly veiled imperialism, racism, or something. Science is not that way. If there was an easier language to communicate to an audience with they'd use that. The "base" language of English is far, far simpler than other languages. Gendered pronouns are available in almost every language and English is one of the only languages where you can replace them with non-gendered pronouns if you so choose. In fact, I wrote this entire post without using any gendered language. This would be impossible in nearly any other widely spoken language.
It would be easier if the frothing at the mouth pro-DEI-at-all-costs people would just say what they mean. Then the argument could be addressed correctly - English replaced Latin because it was easier and evolving. Chinese is the only language that would replace it based on paper-origin/language ratio and Chinese is an order of magnitude or more complicated. Even just speaking about the alphabet it's something like 100x larger.
If we wanted a more neutral language ideally suited to scientific discourse then Lojban would be the "logical" choice. But of course in the real world it's totally impractical.
> obligatory gender in animate pronouns
While it's just one in a long list of things, I'll point out that singular "they" is available (and has a long history) for people wishing to use non-gendered, animate pronouns. (I have also heard "it", but, while everyone's mileage will differ, my brain has a much harder time getting over the difficulty with reading "it" as animate than it does the difficulty with reading "they" as singular.)
It's why Queer Theory is so insane - very very low stakes. (If you don't believe me, read some Gayle Rubin or the less mainstream Foucault stuff).