how it does not work: translation-as-decoding In 1949 one of the inventors of the theory of information and communication, Warren Weaver, wrote an essay entitled "Translation." Weaver explored the idea that computers might be used to translate texts from one language to another. He proposed that translating texts was like decoding secret, encrypted messages and wrote the following:

When I look at an article in Russian, I say, 'This is really written in English, but it has been coded in some strange symbols. I will now proceed to decode'.

Weaver's "Translation" essay was enormously influential and, arguable, still informs computer scientists' approaches to translation. But, after half a century of sustained work on Weaver's translation-as-decoding problem, how much progress has been made? Not much! Anyone with a web browser has the means to check. If you haven't yet, play with some of the automatic translation software now available on the web.

how it does work: translation-as-collaboration Perhaps, fifty years later, it's finally time to admit Weaver's folly: translation is not a task of decoding. Instead we propose an alternative that is simple common sense to translators: translation is a form of collaborative writing between people, specifically between authors and translators. Instead of trying to build a computer program that can translate authomatically, we are attempting to build a computer program that can help connect people together over the Internet facilitating a collaborative re-writing process. Like any translation, the result will be good if the contributing translators are good writers.

The Translation Map incorporates

  • a collaborative editing system allowing several people to work on the translation of any message; included in this system are also tools for printing and viewing the message as it appears partially or fully translated;
  • a set of databases that record where approximately 6000 languages are spoken; and,a set of routing rules that uses these databases to help you choose a route for your message; for instance, a message written in English in the United States that is suppose to arrive in France in French might get routed through Canada where many English-French bilinguals might be found to help you with the translation.

When you send a message with the Translation Map system you will be asked to help pick the route of the message. To navigate the chosen route, the message is posted into online, geographically-specific, public discussion forums (e.g., Usenet newsgroups, Yahoo groups, etc.). Once you have sent your message it will both be uploaded to the Translation Map website and mailed to the public discussion forums that you choose. Your message will be publically available and most probably posted to several areas of the Internet. You will be given a URL to your message's webpage and you can periodically check the webpage to see how the translation of your message is proceeding.

No message is guaranteed to be translated. If your message finds its way to people who want to help you out with its translation, it will be translated. The Translation Map does not do any automatic translation. It simply connects together hundreds of public, online discussion sites and provides a place for collaborative authoring/translating to occur. Our proposal is part of a long genealogy of collaborative/experimental art and writing techniques and technologies including, for example, the Surrealists' exquisite corps writing games.