Tuesday, February 7, 2012

Words Matter: Christopher Columbus, the Bergman Effect, and Google Translate

David Bellos is the director of the Program in Translation and Intercultural Communication at Princeton University. He is also an award-winning translator. He has written an interesting book on language and translation that is recommended to anyone with an abiding interest in those topics. It is titled: Is That a Fish in Your Ear? Translation and the Meaning of Everything. No attempt will be made to explain the title, nor will the subtitle be defended. You will have to read the book.

Bellos covers a lot of ground. Some topics are worthy of a longer discussion and may appear here at a later date. For now we will have to be satisfied with a few brief notes on topics that seemed both interesting and well-bounded.

Christopher Columbus

For the English speakers who are currently able to go almost anywhere in the world and survive without having to learn a native language, Bellos reminds us that learning to communicate in multiple tongues was, and will continue to be required of most people of the world..

Bellos uses Christopher Columbus to illustrate the language skills required for those of his time who would travel the world.

"The great explorer Christopher Columbus provides an unusually well-documented case of the intercomprehensibility and interchangeability of European tongues in the late Middle Ages. He wrote notes in the margins of his copy of Pliny in what we now recognize as an early form of Italian, but he used typically Portuguese place-names—such as Cuba—to label his discoveries in the New World. He wrote his official correspondence in Castilian Spanish but used Latin for the precious journal he kept of his voyages. He made a ‘secret’ copy of the journal in Greek, however, and he also must have known enough Hebrew to use the astronomical tables of Abraham Zacuto, which allowed him to predict a lunar eclipse and impress the indigenous people he encountered in the Caribbean."

Just how many languages did Columbus know?

"It’s unlikely Columbus even conceptualized Italian, Castilian, or Portuguese as distinct languages, for they did not yet have any grammar books. He was a learned man in being able to read and write the three ancient tongues. But beyond that, he was just a Mediterranean sailor, speaking whatever variety of language he needed to do his job."

Most people of the world live in multicultural environments where many different languages are encountered. Most manage to become polyglots.

"Five to ten languages seem to represent the effective limit in all cultures, however multilingual they may be."

Luckily, our children are not distracted by such irrelevancies. It allows them to spend more time honing their video-game skills.

The Bergman Effect

Bellos introduces us to the art of subtitling films.

"It has become conventional to regard average moviegoers as capable of reading only about fifteen characters per second; and in order to be legible on a screen as small as a television set, no more than thirty-two alphabetic characters can be displayed in a line. In addition, no more than two lines can be displayed at a time without obscuring significant parts of the image, so the subtitler has around sixty-four characters, including spaces, that can be displayed for a few seconds at most to express the key meanings of a shot or sequence in which characters may speak many more words than that....A further constraint on subtitling is the convention that a subtitle may not bleed across a cut."

It follows that the viewer is often seeing not only a translation of what is actually being said, but a significant amount of compression. Some directors have not been thrilled at the notion that some unknown person is going to, in effect, rewrite their dialogue.

"Filmmakers dependent on foreign-language markets are well aware of how little spoken language can actually be represented in on-screen writing. Sometimes they choose to limit the volubility of their characters to make it easier for foreign-language versions to fit all the dialogue on the screen. Ingmar Bergman made two quite different kinds of films—jolly comedies with lots of words for Swedish consumption, and tight-lipped, moody dramas for the rest of the world. Our standard vision of Swedes as verbally challenged depressives is in some degree a by-product of Bergman’s success in building subtitling constraints into the composition of his more ambitious international films. It’s called the ‘Bergman effect’...."

Consider a most skillful author who labors away writing wonderful novels in Albanian. Who will ever know, other than a few fellow Albanians? But if he can get his work translated into English, or one of the other major languages (Spanish, French, German, Russian, Arabic, for example) then there is a translation path to the readers of almost any country of the world. The need or desire to reach a worldwide audience using English, predominantly, as a pivot is beginning to affect the manner in which artists construct their works. A "Bergman effect" is beginning to be recognized in other areas.

"Steven Owen has argued that some contemporary poets from China, for example, write in a way that presupposes the translation of their work into English—and that all writing in foreign languages that now aspires to belong to ‘world literature’ is built on writers’ effective internationalization of translation constraints."

Google Translate

Anyone who has used Google Translate (GT) will have been both awed at its apparent power, and left dumbfounded by some of the silly mistakes it makes. Bellos explains what is going on.

 
"At present it offers two-way translation between fifty-eight languages, that is to say, 3,306 separate translation services, more than have ever existed in all human history to date. Most of these translation relations—Icelandic—>Farsi, Yiddish—>Vietnamese, and dozens more—are the newborn offspring of GT: there is no history of translation between them...."

So how does Google do it?

"It doesn’t deal with meaning at all. Instead of taking a linguistic expression as something that requires decoding, GT takes it as something that has probably been said before. It uses vast computing power to scour the internet in the blink of an eye looking for the expression in some text that exists alongside its paired translation....GT uses statistical methods to pick out the most probable acceptable version of what’s been submitted to it. Much of the time it works. It’s quite stunning."

There is a vast amount of material for GT to search through, and the more that exists, the more accurate it will be. GT also has the option of using intermediate languages as pivots in those cases such as Icelandic—>Farsi where there is likely little data. Icelandic—>English and English—>Farsi is likely to be a more productive means of connecting those two languages.

"When the only tool you have is a hammer, then every problem begins to look like a nail."

Hats off to Google for creating such a revolutionary tool! Although it may not be what it appears to be, it is powerful, and will only improve as the body of translations available to it continues to grow.

And thanks go to David Bellos for providing some interesting insights.

2 comments: