While intelligence analysts are trying to prise open low-resource languages from the outside, native speakers of those languages are also taking matters into their own hands. They, too, want access to urgent information in other languages – not for espionage, but to improve their everyday lives.
"When this Covid-19 pandemic happened, there was a sudden need to translate basic health tips into many languages. And we couldn't do this with machine translation models, because of the quality," says David Ifeoluwa Adelani, a doctoral student in computer science at Saarland University in Saarbrücken, Germany. "I think this has really taught us that it's important that we have technology that works for low-resource languages, especially in time of need."
Adelani is originally from Nigeria and a native Yorùbá speaker, and has been building a Yorùbá-English database as part of a non-profit project called Cracking the Language Barrier for a Multilingual Africa. He and his team created a new dataset by gathering translated movie scripts, news, literature and public talks. They then used this dataset to fine-tune a model already trained on religious texts, such as Jehovah's Witnesses publications, improving its performance. Similar efforts are underway for other African languages like Ewe, Fongbe, Twi and Luganda, helped by grassroots communities such as Masakhane, a network of researchers from all over Africa.
One day, all of us may be using multilingual search engines in our everyday lives, unlocking the world's knowledge at the click of a button. Until then, the best way to really understand a low-resource language is probably to learn it – and join the multilingual, online human chatter that trains the world's translation robots.
--
Join one million Future fans by liking us on Facebook, or follow us on Twitter or Instagram.
If you liked this story, sign up for the weekly bbc.com features newsletter, called "The Essential List". A handpicked selection of stories from BBC Future, Culture, Worklife, and Travel, delivered to your inbox every Friday.
Keep it Clean. Please avoid obscene, vulgar, lewd, racist or sexually-oriented language. PLEASE TURN OFF YOUR CAPS LOCK. Don't Threaten. Threats of harming another person will not be tolerated. Be Truthful. Don't knowingly lie about anyone or anything. Be Nice. No racism, sexism or any sort of -ism that is degrading to another person. Be Proactive. Use the 'Report' link on each comment to let us know of abusive posts. Share with Us. We'd love to hear eyewitness accounts, the history behind an article.
For those who are acquainted with the sensation, accidentally squeezing citrus juice into your eye is irritating beyond words. However, with the push to fit “orbisculate” — a word coined in 1958 by late alum Neil Krieger ’62 — into the dictionary, the acidic sting of a grapefruit to the cornea won’t be nearly as indescribable.
Orbisculate was invented by Neil Krieger during one of his first-year writing seminars at the University more than 60 years ago. The intransitive verb — one that does not take a direct object — is defined on the Krieger family’s website as: “to accidentally squirt the inner content from fruits, vegetables and other foods onto one’s face, body, or clothing, or onto that of a person nearby.”
After Krieger’s unexpected passing in April 2020 due to COVID-19 at the age of 78, his wife Susan and children Jonathan and Hilary Krieger ’98 were forced to grieve without the closure of a proper eulogy or in-person shiva, a week-long mourning period for those of the Jewish faith.
Instead, Hilary and Jonathan have decided to honor their father by getting orbisculate into the English dictionary.
As such, they have begun a social media campaign to bring the word out of obscurity and into every-day use. To chronicle their success, the Orbisculate website lists 50 goals, 16 of which have already been completed, which they hope to achieve.
These obstacles range from getting the word in a grocery store to being included in a song by Hamilton creator Lin Manuel Miranda.
Leaderboard 2
Growing up, Hilary and Jonathan thought orbisculate was already in the dictionary. While most of Neil Kreiger’s classmates quickly forgot their linguistic inventions, he continued to use his regularly — and the opportunity arose frequently, considering Krieger made freshly squeezed orange juice for his family on weekends.
Hilary, a former editor-in-chief for The Sun, only found out orbisculate was invented by her father at the age of 23, when she had made a $5 bet with a Cornell friend and fellow Sun alum Alex Carey ’97, who doubted that the word was real.
Newsletter Signup
Hilary joked that, with this project, she is looking forward to reclaiming her $5.
To those close to Neil, orbisculate and the story behind it showcased his ability to turn a mundane, negative situation into something hilarious.
“[Orbisculate] is just a really great example of who he was and it encapsulated so much about him in one word,” Hilary said. “It was funny, it was creative, it was original, it took something that would otherwise be pretty negative, and turned it into something to have fun with. That was very much his personality.”
Krieger’s Cornell roommate, high school classmate and lifelong friend Paul Marantz ’62 also commented on how orbisculate illuminates Neil outside of his career.
“I think the word was very much in keeping with his very playful, humorous attitude he had towards life,” Marantz said. “Inventing the word orbisculate, keeping it alive and then not even telling his kids that the word was invented, very much sort of reflects his playful side.”
Attaining professorships in biochemistry at Harvard University and the University of Pennsylvania before founding a consulting firm for biochemical start-ups, Neil Krieger’s career strayed far from lexicography. However, Krieger’s held onto a love of libraries and an all-encompassing curiosity for words and the world around him, all of which manifested itself in orbisculate.
“He always wanted to understand and think through and be engaged with these different parts of life, so to him science, humor and creativity, the written creation of words, those were all connected for him,” Hilary said.
The best way to support the “Orbisculation Nation,” according to Hilary, is for people to fold the word into their daily vocabulary. Additionally, supporters can sign the petition on their website in support of the word or purchase an orbisculate t-shirt, of which all proceeds go to Carson’s Village, a nonprofit that helps with the logistical and bureaucratic challenges families face when they lose a loved one.
To Hilary and Jonathan Krieger, one of the most fulfilling aspects of the project has been watching their father’s story resonate with those who hear about it. In many ways the project functions in much the same way that Neil’s word does — it turns a negative into a positive.
“I hope what people take is a way to find something positive in the dark time where we see all this stuff with people passing away,” Jonathan said. “ I think one of the things special for us is to share the story of who our father was, but it’s also special to hear stories of other people and who they’ve lost that meant so much to them.”
The Rotary Club of Jackson donated dictionaries to JPS third-graders Monday.Cars lined up to receive the dictionaries at the Mississippi Agriculture and Forestry Museum for students within the JPS system.The Rotary Club of Jackson's President Leroy Walker said they plan to speak with teachers on how to properly use the dictionary."We're going to talk to the instructors about how to use it and what is it that they can get in terms of increasing their vocabulary and making decisions in terms of structuring their sentences and grammar," Walker said.According to Walker, administering such resources can provide a change from the technological norm. "What we normally would do under normal situations is go and see the kids ourselves and talk to them about the dictionaries and the process of looking up certain words. We're into the computer age, I understand, but at some point they'll may have to use a dictionary to be able to fashion themselves in the manner they should growing up," said Walker.Dictionaries were also donated to the seniors of Lanier High School.
The Rotary Club of Jackson donated dictionaries to JPS third-graders Monday.
Cars lined up to receive the dictionaries at the Mississippi Agriculture and Forestry Museum for students within the JPS system.
Advertisement
The Rotary Club of Jackson's President Leroy Walker said they plan to speak with teachers on how to properly use the dictionary.
"We're going to talk to the instructors about how to use it and what is it that they can get in terms of increasing their vocabulary and making decisions in terms of structuring their sentences and grammar," Walker said.
According to Walker, administering such resources can provide a change from the technological norm.
"What we normally would do under normal situations is go and see the kids ourselves and talk to them about the dictionaries and the process of looking up certain words. We're into the computer age, I understand, but at some point they'll may have to use a dictionary to be able to fashion themselves in the manner they should growing up," said Walker.
Dictionaries were also donated to the seniors of Lanier High School.
One of the more unexpected twists of an unprecedented year is that the little-known business of literary translation has become a source of public controversy.
It began in mid-January with an uncontroversial choice — the selection of Amanda Gorman, a then-22-year-old Black poet, to read at Joe Biden’s presidential inauguration. Gorman’s inaugural poem, “The Hill We Climb,” was a rousing success, a stirring call to the unfinished business of American democracy after an attack on the U.S. Capitol. Penguin Random House snatched up the poem for publication, and foreign publishers clambered to publish it abroad, which meant enlisting translators worldwide.
Last month, two of those translators ceased work on the project. — first in the Netherlands, after criticism that a white author had been chosen to translate the work of a Black woman. Dutch translator Marieke Lucas Rijneveld, who last year became the youngest writer to win the International Booker Prize, handed back the assignment. Most recently, Catalan translator Victor Obiols was removed from the job.
“They have told me that I am not suitable to translate it,” Obiols told the Agence France-Presse news agency March 10. “They did not question my abilities, but they were looking for a different profile, which had to be a woman, young, activist and preferably Black.”
Advertisement
Some see the debate over “The Hill We Climb,” out this month in the U.S. and expected to sell millions of copies worldwide, as an opportunity to interrogate literary diversity everywhere. Others say world literature wouldn’t have spread without white translators and ask that we judge the translation, not the translator. Still others worry about the ethics of pressing U.S. notions of race on foreign readers.
Online, the usual battle lines are being drawn. Thomas Chatterton Williams, who tweets frequently about what he considers overreaction on the left, called the change in translators “an international moral panic.” Obiols recently told Spain’s ABC newspaper that he was “banned.” He suggested that to get another contract, “I will have to look for bitumen,” a material used for blackface.
(Representatives for Obiols and Gorman did not respond to interview requests; Rijneveld declined to comment.)
Advertisement
Such debates are uncommon in this specialized field. “The translation world doesn’t tend to have many large controversies like this,” noted Aaron Robertson, a writer, translator and editor at Spiegel & Grau. “It’s always extremely surprising for us when it’s thrust into the middle of a larger spotlight.”
Yet it makes sense at a time when so many institutions are being scrutinized. Conversations about representation and inequities in many industries, including book publishing, gathered momentum after last summer’s police brutality protests. And now the field of translation — which remains overwhelmingly white — is having its own reckoning.
The course of that reckoning is familiar: social media backlash leading to institutional reversal. Last month, after the Dutch publishing house Meulenhoff announced Rijneveld as the translator, journalist and activist Janice Deul led social media critics with an opinion piece in de Volkskrant newspaper calling the choice “incomprehensible” and a missed opportunity to hire someone like Gorman — “a spoken word artist, young, a woman and unapologetically Black.”
Advertisement
It was a Dutch author, young-adult writer Corinne Duyvis, who in 2015 coined #OwnVoices, the hashtag advocating that stories about marginalized people be written by authors who share the same identity and experiences. The idea was much debated in young-adult circles before reaching adult trade publishing with the controversy over Jeanine Cummins’ bestseller “American Dirt” — a thriller about Mexican refugees written by a white American. Now it has reached international publishing.
“As far as I know, American publishers have not historically used the background of the author and translator as part of their calculus when deciding who is going to work on the book,” said Chad Post, publisher of Open Letter Books, a nonprofit translation press. “That’s not to say these conversations aren’t valid or shouldn’t be happening. And it will make more sense with certain projects than with others.”
The #OwnVoices movement brings special challenges to translation — a job that is inherently about making work accessible to audiences different from the author. Is the act of translation also an extension of a particular identity? Is the experience of a person of color in Holland analogous to that of an African American?
Advertisement
There is also the relatively limited pool of translators; finding someone to do a job for modest pay that requires expert fluency in at least two languages can be hard enough. Adding identity further complicates the search.
Nonetheless, debates on translating works across race or identity are not new. “When people have to translate books from the 1920s that use very loaded terms, and you have characters that are racist or using derogatory terms, that gets to be a tricky issue,” said Post. “How do you deal with that?”
Identity is one factor. Lawrence Schimel, an author and translator, brings up a Spanish translation of his own work by a straight man who “made biased, wild assumptions” based on ignorance of gay subculture. “He didn’t know the word ‘leather queen’ and instead of finding out what it meant, he used ‘paganos’ (pagans) because the characters were dressed in leather and engaging in ‘ritual’ activities.”
Schimel then offered a counterexample: Another straight man, while translating a story of Schimel’s into German, called a gay sex chat line to ensure that there wasn’t a gay slang term for sex he didn’t know about. “The publisher reimbursed him” for the chat, Schimel said, “since he had put in the legwork to do his job right.”
Advertisement
Post believes there are certain titles for which identity makes a difference — for instance, memoirs about motherhood or sexual assault. “Having those translated by a generic white guy is viscerally irritating,” he said, “and isn’t dissimilar from the emotions people are having in the Gorman situation.”
Fostering real diversity
The ability to find the right translator for any book depends on the depth — and diversity — of the field. According to a recent survey by the American Literary Translators Assn., 73% of the translator community is white, 11% is Asian, 10% is Hispanic/Latino, 4% is Middle Eastern/North African and 2% is Black.
The association has taken steps in recent years to broaden the diversity of its conference attendees by conducting outreach to historically Black colleges and other organizations. It’s also ensuring that more people of color are included in its programming.
Advertisement
“A lot of people are very hopeful for the future of diversity efforts like these; I’m still in the wait-and-see-box,” said Miki Turner, a photojournalism professor at USC and founder of the Annenberg Cross-Cultural Student Assn.
“We had this big wave of diversity efforts in the ‘80s and the early ‘90s that worked for a while, and then they went away,” she said. “It’s just like the entertainment industry. It’s all very cyclical. We have all these mandates of ‘More minorities in movies and shows,’ and they create an ensemble cast with the Black person and the Asian person, and they essentially say nothing but they’re on-screen, and they call it diversity.”
Real diversity in the field of translation would have to run deeper than in other industries to foster true representation — to ensure that, say, a Black woman in her 20s would be available to translate poetry from English to Catalan.
Advertisement
Regina Brooks believes there is a lot of talent out there, if publishers know where to find it. The founder and president of Serendipity Literary Agency, Brooks calls the idea that there aren’t enough diverse potential translators “absolutely ridiculous … There are all sorts of things that can be done. The translation community just has to identify what’s going to work for them.”
Getting your foot in the door often comes down to luck and privilege, as Robertson, the Spiegel & Grau editor, can attest. He received scholarships to an elite high school, followed by Princeton and Oxford universities. “I was learning how to navigate these exclusionary spaces. … I knew what questions to ask, and I knew where to look,” he said. “But to even be here, as a Black translator and as someone who works at a well-respected publishing house, I’m sort of the exception to the rule.”
The debate comes down to opportunity, he added: “The people who are asking for Black women to translate Amanda Gorman’s work, they are looking not at the unifying task of translation as pure art, but they are asking, ‘Why do we turn to certain translators and not others?’”
Advertisement
It’s a valid question, but Robertson worries that it distracts from the larger task at hand — building pipelines among the publishing world and communities that don’t have access to it. That could mean opening up access to universities with robust language departments and expanding opportunities for student translators, many of whom can’t afford to accept unpaid internships in publishing.
Is translation appropriation?
It’s notable that such systemic questions should arise from a situation as unusual as Gorman’s. Typically, foreign rights to a book’s publication are “subrights,” usually owned and sold (as in Gorman’s case) not by the American publisher but the literary agent. The international publisher then decides on the translator in consultation with the author and agent. Sometimes the writer has suggestions. Other times (as in Gorman’s case, according to Dutch publisher Meulenhoff), agents require foreign publishers to hire sensitivity readers tasked with finding biases, racism, stereotypes and misrepresentations in translations.
Major authors have some discretion, but many writers are fortunate to be translated at all. Between 550 and 650 works of fiction and poetry in translation are published in the U.S. every year, according to Post, who tracks the numbers in the Translation Database. Only roughly 80 of those books are poetry.
“Authors have a limited amount of power in their careers,” said one publishing industry journalist, who asked to remain anonymous while discussing sensitive issues. “Gorman’s fame has given her more say than most. So insisting on a translator who is young, female, an activist and if possible Black, she’s giving herself leverage which she might never have again. … She’s using that platform to foster or foment the values she believes in.”
Advertisement
The question of whether those values can or should apply to all translations is likely to be debated for a long time. Even some Black writers are wary of a translation system governed by an #OwnVoices ethos.
Alain Mabanckou, a widely translated French Congolese writer and professor at UCLA, believes that vetting a translator’s national or ethnic origin is a form of “discrimination” and “racism.”
“One simply cannot fight against exclusion by reinventing new ways of marginalizing people,” Mabanckou said in an email, “for this would ultimately lead to a situation whereby one could only understand or speak for people who are assumed to be like us.”
Advertisement
As an example, Mabanckou cites two major influences on his work, Toni Morrison and Maya Angelou — translated by white women into his native French.
“When the only way of looking at the world is through the lens of identity politics, then we have moved into a space that is contrary to what literature is about,” he said. “Literature is destined to liberate us, to change the way in which we see the world and to transport readers on previously uncharted adventures, to delineate the contours of a world in which fear gradually recedes into the background and the ‘other’ is invited into our hearts.”
Some writers have addressed the question of whether it’s OK to write about another group by answering that it requires careful work to do well. Schimel said the same applies to translation.
Advertisement
“I made a personal commitment years ago to make sure I translate at least one writer of color per year, as a way of using my privilege to try and effect change,” Schimel said. “Can people translate writers with vastly different experiences than their own? Obviously, but also with obvious variations in how good a job they’ll do, based on their arrogance and assumptions in treating the work and the amount of effort they expend to unlearn their ingrained prejudices.”
Brooks, the literary agent, is more concerned about finding translators who aren’t privileged in the first place.
“No one is saying at this point that white translators are not going to continue to get jobs or opportunities, because that’s always going to be the case,” she said. “But the question is: Is there room at the table for more voices? For opportunities for other voices? My point on that is that yes. Yes, there are.”
For those who are acquainted with the sensation, accidentally squeezing citrus juice into your eye is irritating beyond words. However, with the push to fit “orbisculate” — a word coined in 1958 by late alum Neil Krieger ’62 — into the dictionary, the acidic sting of a grapefruit to the cornea won’t be nearly as indescribable.
Orbisculate was invented by Neil Krieger during one of his first-year writing seminars at the University more than 60 years ago. The intransitive verb — one that does not take a direct object — is defined on the Krieger family’s website as: “to accidentally squirt the inner content from fruits, vegetables and other foods onto one’s face, body, or clothing, or onto that of a person nearby.”
After Krieger’s unexpected passing in April 2020 due to COVID-19 at the age of 78, his wife Susan and children Jonathan and Hilary Krieger ’98 were forced to grieve without the closure of a proper eulogy or in-person shiva, a week-long mourning period for those of the Jewish faith.
Instead, Hilary and Jonathan have decided to honor their father by getting orbisculate into the English dictionary.
As such, they have begun a social media campaign to bring the word out of obscurity and into every-day use. To chronicle their success, the Orbisculate website lists 50 goals, 16 of which have already been completed, which they hope to achieve.
These obstacles range from getting the word in a grocery store to being included in a song by Hamilton creator Lin Manuel Miranda.
Leaderboard 2
Growing up, Hilary and Jonathan acted as if orbisculate was already in the dictionary. While most of Neil Kreiger’s classmates quickly forgot their linguistic inventions, he continued to use his regularly — and the opportunity arose frequently, considering Krieger made freshly squeezed orange juice for his family on weekends.
Hilary, a former editor-in-chief for the Daily Sun, only found out orbisculate was invented by her father at the age of 23, when she had made a $5 bet with a Cornell friend and fellow Sun alum Alex Carey ’97, who doubted that the word was real.
Newsletter Signup
Hilary joked that, with this project, she is looking forward to reclaiming her $5.
To those close to Neil, orbisculate and the story behind it showcased his ability to turn a mundane, negative situation into something hilarious.
“[Orbisculate] is just a really great example of who he was and it encapsulated so much about him in one word,” Hilary said. “It was funny, it was creative, it was original, it took something that would otherwise be pretty negative, and turned it into something to have fun with. That was very much his personality.”
Krieger’s Cornell roommate, high school classmate and lifelong friend Paul Marantz ’62 also commented on how orbisculate illuminates Neil outside of his career.
“I think the word was very much in keeping with his very playful, humorous attitude he had towards life,” Marantz said. “Inventing the word orbisculate, keeping it alive and then not even telling his kids that the word was invented, very much sort of reflects his playful side.”
Attaining professorships in biochemistry at Harvard University and the University of Pennsylvania before founding a consulting firm for biochemical start-ups, Neil Krieger’s career strayed far from lexicography. However, Krieger’s love of libraries and an all-encompassing curiosity for words and the world around him maintained his interest in orbisculate.
“He always wanted to understand and think through and be engaged with these different parts of life, so to him science, humor and creativity, the written creation of words, those were all connected for him,” Hilary said.
The best way to support the “Orbisculation Nation,” according to Hilary, is for people to fold the word into their daily vocabulary. Additionally, supporters can sign the petition on their website in support of the word or purchase an orbisculate t-shirt, of which all proceeds go to Carson’s Village, a nonprofit that helps with the logistical and bureaucratic challenges families face when they lose a loved one.
To Hilary and Jonathan Krieger, one of the most fulfilling aspects of the project has been watching their father’s story resonate with those who hear about it. In many ways the project functions in much the same way that Neil’s word does — it turns a negative into a positive.
“I hope what people take is a way to find something positive in the dark time where we see all this stuff with people passing away,” Jonathan said. “ I think one of the things special for us is to share the story of who our father was, but it’s also special to hear stories of other people and who they’ve lost that meant so much to them.”
PyTorch, the open source framework used to build machine learning models, including those used for machine translation, released its newest version, 1.8, on March 4, 2021.
According to PyTorch’s official announcement, “highlights include updates for compiler, code optimization, frontend APIs for scientific computing, large scale training for pipeline and model parallelism, and Mobile tutorials.” Prior to this release, PyTorch released version 1.7 in October 2020.
Facebook’s AI Research Lab released PyTorch in September 2016. Since then, PyTorch has attracted attention among developers for its flexibility, speed, and ease of debugging.
Advertisement
In an informal poll on LinkedIn, Tyler Folkman, Head of Artificial Intelligence at Branded Entertainment Network, asked colleagues to name their “go-to deep learning framework.”
“I personally prefer PyTorch because I think it has a good balance of usability and extendability,” Folkman wrote. “By that, I mean that I find the common stuff is pretty easy to do without too much code and the more complex architectures are also relatively straightforward to implement.”
Pro Guide: Translation Pricing and Procurement
Data and Research, Slator reports
45 pages on translation and localization pricing and procurement, human-in-the-loop models, and linguist compensation.
Though respondents alternately voiced support for PyTorch and TensorFlow, Dennis Sawyers, a Senior Cloud Solutions Architect at Microsoft, wrote, “TensorFlow is passé at this point.”
On Twitter, Josh Tobin, an instructor at University of California, Berkley and past research scientist at OpenAI, suggested a generational divide: “Why do people always ask what ML framework to use? It’s easy: Jax is for researchers; PyTorch is for engineers; TensorFlow is for boomers.”
Microprocessor engineering legend Jim Keller, himself a boomer, told Lex Fridman in a February 2021 podcast that he knows many people who have switched from TensorFlow to PyTorch.
Slator 2021 Data-for-AI Market Report
Data and Research, Slator reports
44-pages on how LSPs enter and scale in AI Data-as-a-service. Market overview, AI use cases, platforms, case studies, sales insights.
“The native language of people who write AI network programs is PyTorch now,” he said, noting how it is built to scale naturally. “If you write a piece of PyTorch code that looks pretty reasonable, you should be able to compile it and run it on hardware without having to tweak it and do all kinds of crazy things to get performance.”
Very Much an Applied Endeavor
What does this mean for machine translation (MT)? Experts hesitate to call PyTorch, which has yet to be widely deployed in production, the new standard. Other platforms, such as TensorFlow, are more established, with better visualizations, more pretrained models, and more support and tutorials.
Adam Bittlingmayer, CEO and co-founder of MT risk prediction API ModelFront, told Slator that the factors boosting PyTorch’s popularity for other machine learning tasks may not immediately apply to MT.
In the meantime, Bittlingmayer said, “It’s hard to know who is using an open-source library, but it looks like MarianNMT has the most traction right now.” (MarianNMT is notable for being written in pure C++, whereas most others are written in Python.)
“Machine translation is very much an applied endeavor. There’s so much that goes into making a good production system, so the tools that provide the most out of the box will win,” Bittlingmayer added. “The core algorithms themselves are easy to transfer across libraries and frameworks.”