Pretty rough: plurals and singulars are mixed up, the proper nouns are a mess, and the verb is not recognized. Now I edit this as:

“In 1898, the area around the Morago cave was very different from what it is today: the river banks were terraced and were used for irrigation with water from the Ullals River.”

Why would anyone want to translate this way? Well, it’s actually a bit quicker than translating from scratch; I get a lot of good terminology on-screen instantly; and I don’t have to type so much. Most important, no one will ever have to translate this sentence again: my edited version is not just in my translation memory but in Google’s online Catalan-English database. And since this is statistical machine translation, the more we use the system, the better the output will get, for this sentence and for all sentences using these phrases. Sure, Google gets the fruit of my work, but so do all users, hopefully producing more and better translations to and from Catalan.

This means that free online machine translation is going to improve very rapidly—soon anyone will be able to translate in this way. Translating will be like singing: we can all do it. It will be a participatory, social activity, as it is now for the fans who have translated Facebook, Twitter, and the like.

Of course, just as some people sing better than others, some translate better, and are worth paying for. They will do more than tidy up bad machine translations.

At the time Picasso was living in that Morago cave, at age sixteen, he would go down to the oil mill in the village of Horta, in northeast Spain, where I am now, and eat the following:

"a large slice of toast put into a dish of olive oil so that it became saturated, then with a little salt and garlic—it 'tasted like heaven.'”

That's what the Catalan text says. My problem, though, is that when the people around me eat toast this way, they put the garlic on first—you cut the garlic in two and just rub it on the bread. That doesn't work if the toast is already saturated with olive oil. So should I correct the text and improve the recipe? Or condemn a generation of English-speaking would-be Picassos to years of messy oil on their fingers?

As technology allows everyone to translate, professional translation problems will move to questions of what to choose, what to trust, and thus of ethics. It's not a question of reproducing something, or of seeking equivalent experience—it’s about the quality of new cross-cultural experiences.

During World War II, Weaver became aware of a cryptographer who intercepted a message written in Turkish. Without knowing in what language the message was written and without knowing a word of Turkish, the cryptographer nevertheless decoded the message correctly. This gave Weaver the idea that cryptography might provide the basis for universal translation. Let us assume, he wrote in a memorandum in 1949, that Chinese is “really” English encoded using cryptographic methods (in other versions, he used Russian, another language fraught with Cold War tensions). Therefore, using the same cryptanalytic techniques that succeeded in breaking the ultra-secret German Enigma codes (and later the Japanese codes) in World War II, could not translation into English from any other language be handled through mechanical cryptanalysis?

Although his proposal was never implemented, one part of it is still intriguing enough to provoke continuing commentary. Weaver sketches a scene in which various people, all housed in towers distant from one another, try to communicate by yelling to each other, with little success. But then he imagines the towers are all connected by a common basement. By going down into the basement, the people find they can communicate without difficulty. Many commentators have noted that Weaver’s towers recall not only the Tower of Babel but also the “Tower of Languages,” as digital media scholar Rita Raley calls it, that is used by networked and programmable machines and includes scripting languages and programming languages such as Java and C++ on down to assembler and, below that, binary code.

Although not a language itself, binary code is probably as close to a universal translator as the world will ever see. Not only does it translate all text written, stored, and transmitted in networked and programmable machines; it also encodes everything from sounds and images to graphics and animation. Through extensive sets of standards set by the IEEE (Institute of Electrical and Electronics Engineers), computers with different specifications are able to process and produce binary code, of which every other kind of programming code is a construction. So in a sense, Warren Weaver’s proposal has been realized, not only in its general thrust but also with the vertical imagery he used.

The irony is that while programmable machines everywhere can use this code, very few humans can understand it without translation into higher programming languages. So we are left yelling from our towers and hoping that the computers down in the basement can offer translations good enough for us to understand one another.

One might think that in diplomacy close translations are a priority. And sure enough, a lot of time is spent in international organizations on securing agreed definitions of key terms and on ensuring that these terms are translated “correctly,” or at least with as little ambiguity as possible. One of the main jobs of the World Trade Organization’s Dispute Settlement Mechanism, for example, is to resolve ambiguities that have provoked dispute. But even though clarity and precision may be central to diplomacy, they are not always essential, or even desirable. Ambiguity has many advantages: it can allow two divergent positions to be reconciled in a single “piece of language,” and as a result it can buy time, prevent loss of face, and prepare the ground for the next round of talks.

An example of translation-related “constructive ambiguity” concerns the Hainan Island incident of April 2001, when a U.S. spy plane collided with a piloted Chinese interceptor. The Chinese demanded an apology from the U.S. government that included the word “regret.” But because regret can be interpreted as an admission of responsibility, the U.S. insisted on the word “sorry,” which may or may not entail guilt. In what came to be known as “the letter of the two sorries,” the U.S. stuck to being “very sorry,” both about the loss of the Chinese pilot’s life and about entering Chinese airspace. The Chinese, however, translated the first of these as “sincere regret” (shen biao qian yi), thus capitalizing on the wiggle room provided by translation. The second of these sorries (translated as zhen cheng yi han) is, if not an “illuminating pitfall,” a dark snake pit of impersonal passives and agentless verbs: “We are very sorry the entering of China's airspace and the landing did not have verbal clearance.”

One of the most controversial examples of ambiguity in diplomatic language concerns United Nations Security Council Resolution 242, which asks for “the immediate withdrawal of Israeli armed forces from territories occupied in the recent conflict”—that is, the 1967 Six-Day War. Does the lack of a definite article “the” and the quantifier “all” before the word “territories” imply, as the Israelis claim, that only some territories are being referred to? The French version, calling for “retrait des forces armées israéliennes des territories occupés,” does have a definite article (“des”), but then the French language requires an article in that position: des territoires occupés. Although this is often cited as an example of constructive ambiguity, it might better qualify as an instance of bad-faith interpretation. Where one party imposes its own unilateral interpretation, bad faith may turn into bad blood.

If success depends on achieving one’s aims, then translations that are linguistically faulty may nevertheless be successful in a diplomatic context where the aim is constructive ambiguity. I concluded my last piece by suggesting that the incongruities and serendipities found in translation are essential to art. They are, I would like to argue here, equally essential to the art of diplomacy.