What is the relationship between translation and intellectual production? It is not obvious, you might think. Translation, by necessity, does not provide new insights but rather makes existing knowledge available in another language. If everyone spoke English it would not really be an issue.

Picture courtesy Hishaam Siddiqi on Flickr under CC license Attribution NonCommercial NoDerivativeWorks

At Meedan we do think there is a deep link. Let me try to explain why with some tidbits from the recent Arab Thought Foundation annual conference, FIKR9, that we were lucky to attend.

For many at FIKR9, the idea of making intellectual production available in Arabic was a matter of pride. Arabic is the fifth most widely spoken language in the world, containing within its breadth a vast resource of human scholarship and innovation through the ages. Arabic is a language so intimately connected with the intellectual history of Islam and the modern sense of shared cultural affinity within what Prince Khalid Al Faisal, President of the Arab Thought Foundation, refers to as Al Umma Al Arabiya (the Arab Nation).

There was real concern from many quarters about the dearth of intellectual production on the web in Arabic, not least Jordanian social entrepreneur Maher Kaddoura. Google estimates that less than one percent of all web content is in the language – a widely quoted figure at the conference. ICANN’s Middle East officer
recently predicted that the next twenty million web users in the region will speak little or no English – a warning perhaps that Arabic speaking users risk being trapped in a tiny corner of the web, unable to contribute to global intellectual production. When you factor in the wide disparity in cheap access to good internet, the web will have contributed to global knowledge gaps, rather than helping to alleviate them.

So this is the first point – translation is a pressing duty of those interested in increasing access to knowledge around the world. Societies are not going to learn English en masse at a rate capable of checking these growing knowledge imbalances. Instead, we need methods for scaling translation.

The other issue is to do with the value of translation itself. It would be a reasonable premise to suggest that in periods of high intellectual activity, societies invest in translation. Why? There is evidence to suggest that access to diverse perspectives enables better intellectual outputs. To quote Clay Shirky, it’s not how many people you know but how many kinds of people you know. He cites Ronald Burt’s research into the way in which a major US electronics firm was able to generate new ideas. When the company piloted a scheme to harvest ideas from across the company, the managers in charge were more impressed by ideas generated by people whose social networks included employees from outside their department. In other words, access to diverse networks and forms of knowledge improves our ability to innovate.

Ed Bice suggests this is the critical question of the digital age: the extent to which the internet will ‘increase the network diversity of information exchange or whether, given free choice to create our own channels and refine our information networks, we will evolve distribution structures that narrow our networks, and subsequently, narrow our thinking.’

Translation increases network diversity and it reduces knowledge divides. All of this also increases the likelihood of Arabic users contributing more content in Arabic, for three reasons: first, there is more intellectual output to build upon; second, Arabic users know that their Arabic content can be translated into other languages; and third, Arabic-speaking users who see other Arabic-speaking users writing in Arabic are more likely to contribute in Arabic themselves. And the beautiful thing is we now have a model for doing it at scale without too much cost.

Translation today can be conducted cheaply and to high quality by a combination of machines and humans. Automated translation can provide a first draft which can be edited by one or many individual translators working together on small chunks of text – much like a Wikipedia page entry. Translation revisions show the lineage of the translation, and help alert moderators to problems or vandalism. And the fun part is that each translation contributes to the improvement of the automated translation – so you can continually translate more, and better. The humans focus increasingly on the really hard bits.

The list of organizations working on this model is wide, and Meedan is in the thick of it. If we want to make the web more polyglot, and increase the amount of Arabic content in the next ten years, we need to put energy into the tools to make scaled social translation a natural and intuitive publishing gesture on the web. This is not to suggest that professional translators are going to be out of a job any time soon – quite the reverse, the more we translate content the more demand there will be for the services of translators with real expertise for the most difficult translation problems. But respect for the translation profession should not be a straitjacket into which we put the vision of a polyglot web.

At FIKR9, Tom Trewinnard and I took part in a social media roundtable with some of the best thinkers in the region’s emerging social media landscape. We put these arguments to the group and suggested that to be effective, our number one priority was to enable intellectual production to circulate more freely. The nub of the challenge then is to enable scaled social translation; to make translation of small pieces of text a normative publishing gesture on the web for those who have the skills.

This really depends on our being able to make minimum interventions into existing social practices on the web, to build upon existing social behaviours for sharing content, and to create solutions to the problem that harness the incentives that have infused social services such as Twitter with global community activity – our collective desire for social recognition and our eagerness for human contact premised on cooperation.

This was originally posted on the Meedan blog.

The BBC experiment connected web users across languages

The BBC appears to be thinking seriously about using translation to connect its global audience online.

On Thursday the World Service hosted a cross-language discussion between English, Arabic, Chinese, Portuguese, Persian, Indonesian and Spanish speakers with Google’s Machine Translation service providing translations.

What ensued was a bizarre disjointed discussion about nothing much in particular, resembling a collection of spam attacks.

‘I would l like to use this opportunity to introduce you about Somaliland’ … ‘May we be healthy, live calm and help those who need us most’ … ‘Someone knows Bournemouth, UK? Is an interesting city to visit?’ … ‘I want to tell to the world that Jesus is coming. read Jhon 3:16 please look for in on the BIBLE’.

Full marks for the effort and the vision to think about using translation. But perhaps the BBC should be talking to some of the players in the field who could help them build a more useful cross-language debate with better translation.

They could have used the translation community from Global Voices Lingua for example, with the translation tools developed by Worldwide Lexicon which enable humans to edit and improve MT translations right there on the page, supplying data that is housed on an open source translation memory (WWL has just released a brilliant translation plugin for Wordpress I’ll be blogging about shortly).

It’s also worth comparing what the BBC did with Meedan. Our cross-language Arabic-English interface – which clearly shows the status of a comment translation, the original language it was posted in, the direction of translation, who is author of the comment and who translated it, and whether it is a human or automatic translation – is leaps and bounds ahead of this early BBC experiment.

Still, it’s great to see a major publisher on the internet thinking creatively about crossing languages, which does benefit to everyone in the space working to build the polyglot web.

George discussing Meedan's approach to collaborative translation at the Centre for Translation Studies

There was a great turn out, particularly from Arabic students, and I was absolutely amazed be the quality of the feedback.

We discussed Meedan’s approach and how to get started using the tools, and I tried to demonstrate how getting involved would increase translators’ opportunities by boosting their profile, increasing their technology awareness and honing their translation skills with a live audience.

I used a delicious tag to organize pages for discussion and then used the new beta Delicious browsebar to scroll through them (with one or two minor slip-ups). You can see my notes on this google doc.

The students were really interested in ensuring quality and resolving disputes.

Students get to see Meedan's cross-language news sharing site for the first time

One suggested we should have a translation discussion page alongside the translation history of any comment – a great idea. Another suggested a feed of human translated content – a feature we’ve had in the pipeline for some time at Meedan.

There was also a suggestion that you should be able to receive updates for any piece of content you create on Meedan. Perhaps you’ve written a comment or posted a link and want to know when someone translates it. Or maybe you’re a translator and you’d like to be informed if someone modifies your edits.

These seem like such brilliant ideas – and it is always refreshing to hear them live from an audience.

So I hope we can take this presentation to other universities and translation studies centres in the coming months because it has proved to be a great way for us to learn, and get the Meedan message out at the same time.

We’ve so far spoken in the UK at London Metropolitan, Westminster, and SOAS in addition to Leeds – let us know if you’d like to hear from us at your university.

We are @meedan on Twitter or I am gweyman [at] meedan.net.