global voices, one world

Ethan Zuckerman has a wonderfully provocative post on how he sees translation as the biggest challenge facing the future of the internet. If the internet is truly to deliver the promise of connecting people worldwide, one of the main barriers, if not the biggest one right now, is that we as people don’t really have the equivalent of TCP/IP for interfacing with each other. For the less geek-oriented, that means we really don’t have a way of having a conversation with each other, with all of us who are connected online (technically, but not linguistically), unless huge innovations in translation will bring about a polyglot internet. English so far is doing the job (poorly) as the lingua franca of the internet.

Ethan warns us that machine translation will never be up to the task completely by itself. It will take a combination of tools and communities to achieve better conversation through translation. Current examples I was thinking of include the Lingua project by Global Voices. Lingua seeks to translate the blog posts by Global Voices from English to many other languages, including but not limited to German, Spanish, Malagasy, Farsi, Bangla, Hindi, Chinese, and some others, creating an infrastructure that allows people who can only read Farsi to know what is happening in the Chinese blogosphere and vice versa. Another project that came to mind is Yeeyan. Yeeyan is a community consisting of people who translate content from around the web into Chinese – people can submit posts they want translated, express which posts you want to see translated, give ratings and comments. Are these models we can extrapolate? 

One way forward to think about this is whether we can learn from the most succesful peer production (as coined by Benkler) and apply them to translation. Yeeyan and Lingua are two budding flowers that can hopefully grow into a collection of well-maintained gardens. A few questions arise: Who will maintain the quality of translations, especially once work scales up? It is easy to maintain quality when there are only a few articles to be translated, but how will we oversee translations once they hit thousands or even millions in number? Can we also peer produce the kind of editing that is needed to maintain quality? Or peer produce the kind of filtering that is needed to be able to filter out the quality?

And can we rely on pure peer production based on volunteerism to scale this up? Do we need to monetize translation in order to scale up to a polyglot internet? With monetize, I don’t mean having a system in place that can pay professional translators who work for profit, although that certainly wouldn’t hurt. Instead, I mean being able to compensate volunteers for their time. Viviana Zelizer talks about the crowding-in and crowding-out effect – sometimes once you start paying money, people actually get offended and start leaving the community. Imagine going to a friend who invited you over for dinner and offering him/her to pay for the meal after you finished eating. Also, by paying some but not others, you might have people leaving, although sometimes it is okay to pay some if they do the kind of work nobody wants to do but which is necessary to keep the community and project going. Some people volunteer to translate because they love translating, others do it because they see it as contributing to a greater good. People have different reasons to translate; we will need to understand what motivates people to contribute, and learn how we can encourage these people, with and without money. 

Finally, even once we have translation in place, there still remains a lot of work to be done to overcome cultural distance. Context, recognition and responsiveness are only a few things we would also need beyond (linguistical) access. My Masters degree is, of course, in China Studies – essentially a bridge discipline that seeks to teach and educate students how to serve as the connector between China and the rest of the world. Greater funding for language and area studies, particularly in the United States, is another key component that would go a long way towards the realization of a polyglot internet.