There is a fine
line between news and entertainment on a new Web site that
translates hundreds of the world's newspapers into English. Consider
this headline translated from The Beijing Daily: "Sells which roast
the dried bean curd girl." From The Okinawa Times comes this
gripping lead: "The light where the body of the strong so stainless
steel material is sluggish is shot."
Those are samples of "gist level" translation by Newstran.com
(www.newstran.com). The three- month-old site, created by Rick
Perez, links to AltaVista's free translation service, named Babel
Fish. Babel Fish, which uses dictionary software from Systran, was
the first such online translation service when it appeared in late
1997; today it supports 19 language pairs.
As a result, Newstran.com can translate newspapers into English
from German, French, Italian, Spanish, Portuguese, Japanese, Chinese
and Korean. (It also can translate English-language newspapers into
any of these languages.) Whether or not readers will understand the
result is another matter. At the very least it takes some patience
to muddle through the babble.
"Sure, it stumbles on slang and idioms, but you can get the gist
of most things," said Mr. Perez, who runs an advertising and public
relations company in New Orleans. "I wouldn't base any major
business or life decisions on what you read here."
To be fair, Asian languages are the most difficult for computers
to translate into English, given the extremely different character
sets. French and Spanish newspapers tend to be the most readable; a
few attempts at German-into-English were laughable and resulted in
Spiegel Online headlines like "Single dumps from the telex." The
site also translates magazines, radio broadcasts and online news
sources, including BBC News and Air Force Radio.
While Newstran.com may not be perfect, it can still be useful.
The ability to translate Web sites (news or otherwise) is
increasingly important in a world where an estimated 400 million
people are communicating and conducting business online. According
to the Internet marketing company Global Reach, 53 percent of Web
users speak something other than English as their native language.
Mr. Perez said Newstran.com was becoming popular with travelers
who wanted to learn about the countries they were visiting, but his
main goal is to open the world's media to all readers — especially
those without access to a free press. To protect user anonymity he
integrated the security system SilentSurf.com into the site.
"It's more of a crusade than a business project," said Mr. Perez,
who finances the site through his nonprofit organization Humanitas
International. "I'm going to put every dime I have into this to make
it successful."
Mr. Perez may be headed for some legal trouble, however, because
of the way his site works. When a user calls up a newspaper for
translating, the newspaper site is contained within a frame and is
surrounded by any Newstran advertisements.
The New York Times Company,
for one, has sent a "cease and desist" letter to Mr. Perez. The
Times Company's legal department "will follow up if the illegal
framing and unauthorized use of The New York Times's content is not
stopped," a company spokeswoman said.
Mr. Perez said he was not trying to steal any traffic from
newspapers. He doesn't want to take The New York Times on the Web
off the site, but isn't sure what he will do. "According to my
attorneys I'm not violating any laws," he said. "We're a nonprofit
organization providing a service."
Aside from any legal problems, professional translators say that
Newstran may be spreading itself too thin by trying to translate so
many newspapers between so many languages.
"The problem with a general-purpose site is that there's no such
thing as a general-purpose language," said Rose Lockwood, director
of research at Berlitz GlobalNET, a subsidiary of Berlitz
International.
Systran, the company that provides the dictionary software for
Babel Fish and, by extension, Newstran.com, is a pioneer in the
realm of machine translation, but its system works best when
customized. The European Commission, for example, has used Systran
since the 1970's to translate thousands of documents in multiple
languages. Because new words are always being added to the system,
over time the commission has built extremely robust databases of
vocabulary.
As ragged as the newspaper translations are, they will only
improve as Babel Fish expands its vocabulary. Unknown words are
added to the dictionaries, so the longer it is used, the better it
gets.
"We're only in our infancy," Mr. Perez said. "As computers get
faster and smarter, the translations will get more
precise."