NYT: A Web That Speaks Your Language
Sunday, May 17, 2009
- Organization: New York Times
IN the early years of the Web, nearly all of its content appeared in English. But that is changing quickly. Today, articles on Wikipedia are available in more than 200 languages, for example. And about 36 percent of the seven million blogs running on WordPress, a free software platform, are in languages other than English, according to the founder Matt Mullenweg.
Such changes create a challenge, says Ethan Zuckerman, a research fellow at the Berkman Center for Internet and Society at Harvard. “We are all experiencing a smaller Internet than we should be,” he said. “In the user-created Web, we’ve created a weird dynamic where there is more out there every day — some of it important — but each person can individually read less of it because it’s in multiple languages.”
A number of services, automated and human, are helping to translate what Mr. Zuckerman calls “the polyglot Internet.” Once-expensive machine translation technology is now available free at sites like Google Translate, which offers translations in 41 languages. At these sites, users can input a block of text, and a machine-generated translation pops up almost instantaneously.
Google Translate can also translate a search term into another language, then hunt for it on foreign-language sites. Results appear in two forms: in the target language and translated back into the original language.
Machine translations give workable renderings of basic texts, but complicated ideas or phrasings can trip up even the most sophisticated software, particularly in non-Romance languages. And when it comes to nuance, “machine translation just won’t get you there,” Mr. Zuckerman says.
People worldwide are stepping up to provide that nuance, free of charge.
Leonard Chien, a student and professional translator and interpreter living in Taiwan, charges $100 an hour as an interpreter. But two to three hours a day, he volunteers his translation skills to Global Voices, globalvoicesonline.org/, a citizen journalist site founded by Mr. Zuckerman and Rebecca MacKinnon. There, Mr. Chien translates posts from around the world into Chinese.
Mr. Chien is co-director of the Global Voices translation project, called Lingua, globalvoicesonline.org/, which uses volunteers to translate Global Voices posts into 15 languages. He receives a small monthly stipend for his work as a director, he says, but he is happy to donate his time as a translator.
“I am always excited to see new stories are up,” he says. “I want to tell my readers, but in different languages.”
Mr. Chien is among 104 people who volunteered as translators for the Lingua project last month. Volunteers around the world have also participated in the “Google in Your Language” program, helping the company translate its products into 120 languages. Last Wednesday, TED, www.ted.com/,an invitation-only conference featuring high-profile speakers like Al Gore and Bill Gates, posted translated subtitles and transcripts for many of the talks archived on its Web site. Of the 300 translations, 200 were done by volunteers.
Translators have various reasons for volunteering. “I enjoy the challenge of translating between two very linguistically and culturally different languages,” says Anas Qtiesh, an Arabic-English translator and editor living in Damascus who volunteers 15 to 20 hours a week with Lingua. The work also brings him exposure and experience, he says.
Alexander Klar, a graphic designer in Möhnesee, Germany, who estimates that he has spent 62 hours translating TED talks into German, is inspired by the content itself. “Sharing these ideas over the boundaries of language,” he says, “gives us a chance to forget about the walls and barriers that separate us.”
TED began the video-translation project expecting to use mostly professional translators, even though the site had received unsolicited translations from fans of particular talks. “We thought professional translation was the only way to ensure high-quality work,” explains June Cohen, executive producer of TED Media. The shift to volunteer translators came last fall, after Ms. Cohen and her colleagues — the roughly 20 full-time employees speak 14 languages among them, she says — read several volunteers’ translations and were impressed.
“The volunteers are deeply committed to making the best translation, and they don’t care how long it takes them,” she explains. “There is a passion there that you don’t get from hired guns.”
And then there are the cost savings. Ms. Cohen estimates that a professional translation service would charge $500,000 for the translations already completed by volunteers or in process.
The most obvious potential liability of crowd-sourced translation is quality control. “Google in Your Language” submissions are “reviewed by the company before they are launched,” said Nate Tyler, a company spokesman. Lingua and TED require a review by a second bilingual translator before publication and have translators sign their work; the signature discourages sloppy or deliberately malicious translations.
IT remains to be seen whether volunteer translation efforts can grow beyond isolated groups dedicated to specific causes. One solution may be a hybrid of machine and human translation. This is the approach of Meedan.net,beta.meedan.net/, a site for English and Arabic speakers to discuss the Middle East. Postings are automatically mirrored in the other language, using machine translation, then are refined by human translators.
Ed Bice, Meedan.net’s founder, calls this a “transitional model”; he says he believes that machine translation will continue to improve and may even be capable of human-quality translations within the next decade.
In the meantime, Mr. Zuckerman says, other solutions are needed. “The Internet has the potential to be a global conversation,” he notes. “But unless we solve this problem with languages, it cannot be, and it will not be.”