The Russian language has a word for light blue and a word for dark or navy blue, but no word for a run-of-the-mill generic shade of blue. So when translators are tasked with converting "blue" from English to Russian, they're forced to choose a specific shade.
It's hard to imagine that this particular choice would have any serious implications, but interpreters are constantly translating concepts into other languages with words that have no exact match.
In his book, Is That a Fish in Your Ear?, David Bellos explores the history, the future and the complexity of translation — from the tangled web of simultaneous translation at the United Nations, to movie subtitles and the text on ATM screens.
NPR's John Donvan talks with Bellos, director of the program for translation and intercultural communication at Princeton University, about the art of translation, and what's lost — and gained — in the process.
On why translation is integral to relating to others
"We translate all the time. If we refuse to translate, refuse to listen to what other people have to say to us, whichever language it is in, we're not living as fully as human beings as we could be ...
"For translation to exist, you have to accept the fact that languages are all different and they don't describe the world in quite the same way. You also have to accept that languages are all the same in that anything you can say in one language can be said in any other. And it seems to me [that the] tension between the incommunicability of difference and ... the sharing of a common set of messages and meanings is ... human. I mean, we all live in that state, that I am not like you. My experience is not directly commensurable with yours, and yet, for us to get on and to be human and to be in a society, we have to also make the assumption that in another dimension, we're all the same. We have the same needs, the same fears, the same desires."
On why good translations can never be word-for-word
"People ... often have the idea that a translation ... has to be the same as the original that it's translating. And my big argument all the way through the book is no, no — a translation has to be like. And the ways in which it is like its original vary. They vary historically. They vary in the specific language patterns that you're dealing [with]. They vary depending on the kind of text or object that you're translating.
"Likeness is what translation seeks to provide. A good match is what you're after, but sameness ... well, that you just can't have, because even in the same language, no two utterances — even of the same sentence — are actually the same. You know, time has passed and the mere fact of saying it a second time makes it not like saying it the first time. So I think it's this ideology — not very explicit, not reformulated, but [a] quite powerful idea — that unless a translation is the same as the original, then it's no good.
"That's what I'm trying to get people to drop, to abandon, to realize it's much more subtle and much more interesting than that."
On the limitations of automated translation tools
"It's very silly to use Google Translate or any automatic translation service to produce text in a language you don't master completely. ... The output of any automatic translation device needs to be read and corrected by somebody who commands that language completely, because you can often see easily where the mistake is, or you can tell whether it's garbage or not. And if it's garbage, you disregard it.
"[In] the retranslation game, well, if you work for human translators ... the retranslation into English would not be the same as what you started with. It would be fluent English, because the translator's a human being, not a machine. But it would be different in some degree, in some detail, great or small, because language isn't a machine itself ...
"When you use it to take a letter from a Swedish girlfriend and check that you have understood what she meant, that's fine, if your Swedish is a bit ropey. ... Google Translate has many perfectly sensible and viable uses, and it's a most impressive intellectual and technical achievement. But ... Google itself wouldn't think of using Google Translate to produce its publicity literature in the languages in which it sells its services. It uses human translators to do that."
On the flexibility of languages
"Every human language can fulfill all the needs that its users want to make of it. And if it really needs a word to articulate the wrist and distinguish the hand from the arm, well, they'll jolly well invent one so as to do so. And if they haven't invented one, it's because actually their [are] sort of other ways around it, because life is a very flexible thing.
"I'm personally very skeptical of the idea that any language, any of the languages that human communities have, constrains them to talk about the world in any particular way. It may make it easier to talk about the world in some particular ways, but if you really need to make a distinction, well, you invent a word. You do something new. Language is forever changing in response to [its] users' need."