Found In Translation

While planning a trip to Morocco this summer, Alexander Bevilacqua turned to Google to find things to do and places to see. But his search for accommodations and sightseeing destinations in the North African kingdom left a lot to be desired, returning mostly sites for backpackers and budget travelers.

Lucky for Mr. Bevilacqua, he speaks French, a language closely associated with Morocco, a former French protectorate. So the 22-year-old student in Cambridge, Mass., took a stab at a Google search in French, yielding a much richer set of results, including historical information about particular tourist attractions that he wouldn’t have found in U.S. guidebooks. Even so, Mr. Bevilacqua says, “I wish there were some neutral term that would give me both sets of results.”

Linguistically speaking, the World Wide Web has a way to go to live up to its name. As the number of Internet users expands globally, most consumers can still only access content in their native language, unless, like Mr. Bevilacqua, they are multilingual. But now, well-known search companies are trying to help bridge those language gaps with more sophisticated language-translation tools aiming to provide the most comprehensive search results no matter what language the user is searching in.

The push comes as U.S. search giants are focused on gaining market share overseas, where the search and search-advertising industries are generally less mature than in the U.S. At the same time, the relative share of the searchable Web that is in English is falling fast, roughly estimated at around 30% to 40%, according to Roger Bohn, a professor at the University of California in San Diego, requiring search engines to find a way to translate more pages faster or risk falling behind.

Many initiatives are designed to help the growing population of non-English-speaking Internet users who are shut out from the English-language Web. They are also likely to benefit English-speakers as well, unlocking Web content that only exists in other languages, like local tourism information or Web sites on less widely known celebrities and public figures.

That means that English-speaking users seeking information about music, foods or fads popular in a specific country could find information about those issues written or compiled by people with more authentic knowledge about them — a biography of a popular Italian singer written by an Italian, for instance. Professionally, the new services could be useful for doing research on foreign markets, locating primary research written by natives.

Google Inc., of Mountain View, Calif., yesterday began to roll out a translation service that allows users to search Web pages in a dozen different languages but still enter their query — and view their results — in the language with which they are most familiar.

An English-speaking user, for example, could type “Bordeaux tasting” and a particular town in France and select from a pull-down menu — an option that lets him search French Web pages. Google then automatically translates the query into French, finds the most relevant results among French Web pages, and presents those pages back to the user in English.

Similarly, a Spanish-speaking searcher traveling to the U.S. could enter the name of particular activity or event in Spanish and direct Google to search for its query among English Web pages before translating the results back to Spanish.

“If you’re not an English speaker, the most valuable corpus of information you have to improve your search is the English Web,” says Sergey Brin, co-founder of Google, who acknowledges that while the translation software might be “irritating” if you were trying to read a novel, it suffices for digging up basic information like locations, people and facts.

Rival Yahoo Inc. is taking a more human-intensive approach based on its Yahoo Answers service, which is available in nearly a dozen languages, including Chinese, French and Portuguese and soon, Vietnamese and Thai. The service allows users to ask other Yahoo Answers users any question and have it answered by a native speaker. Yahoo then indexes the responses into its Web search results for others to find in the future. “Outside the U.S., there is information that is still in people’s heads that doesn’t appear on the Web,” says Tim Mayer, vice president of product management for Yahoo Search. “This is a very large opportunity.”

Meanwhile, Microsoft Corp. is working on improving its natural-language processing software that translates documents by extracting implied meanings behind a string of phrases or words as opposed to translating each word literally. The company says it is considering ways to implement technology into its Live search service, which it currently offers in 46 markets from Germany to Hong Kong.

As the number of international Internet users has skyrocketed in recent years, U.S.-based search engines have been working to keep their services relevant by improving their ability to search in other languages. Their existing algorithms alone, which determine the popularity of a particular Web page based on the content of the page and what other pages link to it, do a poor job for languages that have a relatively few number of Web pages and fewer links among them.

As a result, a search in a foreign language often pulls up a combination of results in the language that the user was searching in, and occasionally results in other languages if the result, like the home page of the organization being searched for, scores highly in its algorithm.

Google, Yahoo, Microsoft and Ask, owned by IAC/InterActiveCorp, have also launched sites for particular geographic regions tailored to particular markets. The sites draw assumptions about the mindset and interests of searchers from a particular country and rank queries accordingly. A search for “government” on Google.de will bring up information about the German government, for instance, versus a search for the same word on Google.cn which will bring up results about the Chinese government.

The companies have implemented language-translation software that allows users to translate search results back to their native language by hitting a “translate this” option next to some search results or by cutting and pasting the text or the Web address into an online translator tool.

But competition from search engines launched in and customized for a specific country like Baidu.com in China — along with demand from consumers seeking more comprehensive results — has prompted them to do more behind the scenes to make the translation more seamless and to improve its accuracy.

Yahoo, for instance, takes into account variations in grammar and syntax across languages to better guess what users are searching for. Since German words are often composed of a series of words, Yahoo will break up German queries into the parts of a compound word and treat them as separate words.

Some feel that the efforts are already paying off. Henri de Saint Mars says the quality of his Google searches about French topics, from pop music to politics, which he prefers to do in English, is leaps ahead of where it was a few years ago. “The improvement is fantastic,” says Mr. de Saint Mars, who is 27 years old and works for an investment bank in New York City. He says he now finds many more specific results, like song lyrics, audio files and relevant user comments by searching in English than he did before.

But many believe there is room for improvement. Miriam Asnes, who works for a nongovernmental organization in New York, frequently needs to search for locations, people and news particular to Israel and the Middle East. But she finds that searching for the name of a public figure or a mosque in Google in English, even when she specifies that she wants to get results in Hebrew or Arabic, often returns a bunch of Web sites where her search terms appear somewhat randomly and without much context. “Google in English is just better,” says Ms. Asnes, 26, who instead finds more detailed information like official home pages by switching over her keyboard and searching in Hebrew or Arabic instead.

Udi Manber, a vice president of engineering at Google, says Google’s translation service will improve as it learns to better account for the fact that foreign words often have different meanings depending on their context. Nuances in language are particularly difficult to discern, he says, given that searchers are only accustomed to entering a few words at a time into a search field. “There is a lot of work involved besides translating the individual words.”

– By Jessica E. Vascellaro

Popularity: 1% [?]