The birth of the Google Translate era

The rise of new technology is changing the way we think about language and the world. An expert explains how

Published October 9, 2011 7:00PM (EDT)

For most of human history, the notion of a "Star Trek"-style universal translator seemed as farfetched as a warp drive or American universal healthcare. Not anymore: In recent years, Google Translate has made automated translation as easy as copy-and-pasting text into a browser; you can now auto-translate entire news articles at the click of a button, and a host of mind-blowing translation apps have hit the iPhone. Word Lens, for example, allows you to point your camera at a piece of text and see it translated in real time on your phone. (Check out the app trailer here).

It's a change that raises a number of bigger questions: Will automation completely replace human translation? Are we about to see the end of multilingualism? According to David Bellos, a professor of French and comparative literature at Princeton and Booker Prize-winning translator, that's not likely to happen anytime soon. In his new book, "Is That a Fish in Your Ear?," about process and social meaning of translation, he persuasively argues that human translators are as crucial as ever. At a time when the world seems more globalized and small than ever, they play a central role helping us understand each other and bring art to a broader audience.

Salon spoke to Bellos over the phone about the future of English, the rise of the auto-translator and America's cultural aversion to multilingualism.

Why is Google Translate succeeding where other automated translators failed?

In terms of user experience I think it's a quantum leap in machine translation. In terms of its intellectual presuppositions it's not a quantum leap, it's just a very clever engineering solution. It uses a statistical method to find the most probably match for an expression -- it doesn't try to understand it or strip it down into vocabulary, syntax and meaning. It just looks for matches using a probabilistic computational device.

Obviously Google didn't invent all of that by itself -- statistical machine-based translation had sort of been invented in the 1980s, and there had been small systems using it in the 1990s -- but now Google has the Web to work with. There are now millions of pages in many languages that Google's huge service can scour in an instant. And it's also crowd based, in that when it gives you a result it invites you to present a better translation. And people do. They are pleased to show a machine where it went wrong, so the database gets more and more refined. But one thing people must remember is that Google Translate is not conceivable outside of a world in which millions of humans are producing and translating these texts.

Do you think the rise of Google Translate -- and astonishing iPhone translation apps like Word Lens, which allows you to automatically translate signs on your phone -- means people are going to stop learning other languages?

That could happen in the way the generalization of the pocket calculator means my grandchildren don't know how to do arithmetic. That would be a sad result. Obviously it removes the motivation for language learning for all sorts of minor, everyday trivial tasks. But I think that people in positions of responsibility in the educational system and in schools and public life should know full well that the existence of Google Translate in no way reduces the educational need and the utility of acquiring foreign languages. We should struggle ever harder so that some of the next generation have a proper understanding of a variety of other languages -- not just French, but Chinese, Japanese or Arabic.

There are about 8,000 languages in the world, but only a small number of them are spoken by large numbers of people across borders. You call these languages "vehicular." What are they?

A vehicular language is a language that is used as a vehicle of communication between people who are not native speakers of it. So when a Marathi-speaking citizen of Mumbai phones a Chinese business contact working in Hong Kong, they speak English to communicate. English is the vehicular language. There are maybe 80 such languages that play that role. If you know somewhere between 9 and 15 of the vehicular languages of the world you've got a means of communication with something like 95 percent of the world's population.

When a language does become vehicular, it escapes the control of the home country completely and starts to absorb and hybridize. English is the most obvious example. It's a very hybrid language -- it's got bits of Danish and Celtic and words from Bengali and obviously lots of words from French and Latin and Greek. It's like a vacuum sweeper, so the vehicular language acquires that central role and does in a sense become marginally more recognizable to speakers of any other language simply because it hybridizes bits of them.

So why did languages like English and German catch on while others didn't? Is it because they were imperial powers?

I think the imperial answer is too easy. Look, the Dutch had an empire at least as far flung and opulent as the British empire 300 years ago, but Dutch has not become a vehicular language of any importance in the world today. The Portuguese are similar.

In linguistic terms, there is absolutely no difference between a language that's become vehicular and one spoken by a small tribe in Papua New Guinea. All languages are equal and can be made to serve whatever purposes their speakers want to achieve with them. The old pre-1920s view that the hierarchy of language reflects something inherent in the languages themselves is really not true, and you can't map the complexity of the language onto the complexity of the society that uses it. There's nothing inherently better about English than about Chinese or Russian or whatever.

The role of English in the world at present is without historical precedent. Some people like Nicholas Ostler foresee the English language splitting up into a myriad of dialects that will bit by bit become mutually incomprehensible, but I think the conservative forces of print will slow that down very substantially. God knows how long that will take.

In the book you also point out that Eskimos don't actually have 30 words for snow, which I had actually bought into. And I didn't realize that myth had such racist connotations.

It was first debunked by Geoffrey Pullum, and it's really significant that this factoid became lodged in the popular consciousness without people realizing the baggage it brings with it. It suggests that "Eskimo" is a primitive language, because, unlike languages like English, it has all these concrete terms for snow but not the abstract or general term [i.e. "snow"]. There's this idea that languages with a general term are able to cope with the world in a more abstract or sophisticated way. It's a very 19th-century colonialist view, which was a time when people were trying to figure out why English or French were better than Swahili or Eskimo. And on top of that, Eskimo is not a language. There are many languages of the Inuit peoples. It's like saying there's a language called "European."

The idea of speaking multiple languages is much more celebrated in Europe than it is in the U.S. Why do you think that Americans are so much less interested in being multilingual?

I think one needs to be a bit more subtle about it than that. One out of every five Americans is functionally bilingual -- either because they are recent immigrants or they inherited it from their family or because they learned it at school or from residence abroad. That's quite a high percentage for an English-speaking country. America isn't as monolingual as people say it is -- but it is culturally monolingual.

The motivation of most immigrant groups when they get here is to naturalize and assimilate as soon as possible. If they do carry on speaking Italian or Hungarian at home, they don't especially want it to be known because they want to be new Americans. That's a special emotional and political ideology that's particular to the U.S. In Europe, where people move across borders, that kind of aspiration -- to pass yourself off as French, for example  is first of all unachievable because the French won't let you (and the British even less). Multilingualism is by no means absent from the U.S., but it configures differently in American cultural values.

Canada, where I'm from, has an official bilingual policy -- so the packaging on all products, from potato chips to diapers, has to be in both French and English. We obviously have our own historical and political reasons for that, but what's interesting is that many parts of America are now also becoming bilingual, albeit in a very chaotic, unlegislated way. In New York subways, for example, you'll see lots of ads in Spanish with no English translation, and it's completely unremarkable at this point.

I think it says a lot about the past of America. If you look outside of the big cities of the Northeast, the real areas of creeping bilingualism are precisely the areas that used to belong to the Spanish empire: California, New Mexico, Arizona, Louisiana, Florida. So it's kind of a reversion. I'm not a real academic sociolinguist, but I understand from colleagues that some people are skeptical about the next generation maintaining their Spanish. The generations growing up in the U.S. and getting an elementary U.S. education may, like the grandchildren of Italian immigrants, just drop that language in the next 20 to 30 years. Or maybe for political reasons they will assert their identities as Spanish-speaking American and we'll move towards something more like the Canadian model. All these things are open to negotiation.

In some parts of the world, foreign movies are dubbed over instead of subtitled. I remember watching dubbed TV shows and movies when I was a kid, and even then I thought it was very strange. In the U.S., we clearly prefer subtitles.

Even among the countries of the European Union, there are very different traditions. When sound came into movies in the 1930s and the industry de-internationalized, different countries established traditions and expectations that remained quite firmly fixed. France, for example, dubs its foreign films -- except for films considered as art -- while in England and America that's unheard of. The way America treats subtitles says a lot. In Germany and France and Italy, the people who produce the local versions of foreign movies, whether dubbing or subtitling, are respected professionals with a public profile. Here they are a completely obscure network of guys who do it for a few dollars an hour in a garret at night. And I'm hardly exaggerating actually. And these are some of the most difficult operations you can do with language.

Does that suggest that Americans just don't care about translation as an art form?

Twenty years ago, if you were an academic and also a translator you never put your translations on your C.V. because they would count against you when it came to promotion time. Translation was seen as second-rate. But I think that is changing. I think there is much more respect now for translators in the academy and elsewhere.

Far more titles are translated from English than into English every year. What does that say about the relative status of English-speaking culture in the world?

There is a huge asymmetry. Somewhere around 3 or 4 percent of books appearing in the U.S. each year are of foreign origin, whereas in other developed book cultures, like France or Germany, it ranges from 15 to 20 percent. The English language book industry is vastly bigger than any other, so 4 percent of all the books published in English is actually quite a large number of books.

It's partly because English is written by so many different cultures around the world -- South Africans, Australians, and so forth -- so the English book world is already diverse. But the fundamental reason is that publishers remain convinced that translated books are a hard sell and that they don't have much of a public. They don't really believe in translated books as way of making money. It's true that very few New York Times bestsellers were originally written in another language, but these are to some degree self-fulfilling prophecies. It's something I really want to carry on struggling against. Publishers should be more courageous and less enclosed in the received idea that translation is expensive and needs a subsidy and is never going to appeal to anybody.

And there's the idea that a translation of a book is just a degraded copy of the original.

That argument is self-contradictory. There are good translations and less good translations, but the general argument that a translation is less good than the original is impossible to support with any kind of intellectual self-respect. To put it in a nutshell: How do you know?


By Thomas Rogers

Thomas Rogers is Salon's former Arts Editor. He has written for the Globe & Mail, the Village Voice and other publications. He can be reached at @thomasmaxrogers.

MORE FROM Thomas Rogers


Related Topics ------------------------------------------

Education Google Nonfiction