Babel: of languages and some tech issues with the UNDP gig

(Intended for [Matthias])

This contract is a sort of Mission:Baltic times four. We seed and animate online conversation – aimed at some sort of foresight exercise – in four countries. We also have a global layer where we bring these conversations together for perspective. Countries might be something like: Georgia, Bangladesh, South Africa, Ukraine – at any rate, no Western European country where everybody speaks English. The time frame is also brutal: finish the work by end of March.

So we have opposite constraints. On the one hand, the need for interoperable conversation, compounded by the time issue. These press for a tech solution of hosting everything on the existing ER platform and using English as the connecting language. On the other hand, we have countries in which English literacy might be low, and languages are a big political deal with the UN.

The solution might be the original ER1: some briefing and explanatory material translated in several languages. Conversation in English. Google Translate integration allowing anyone not comfortable with English to write in whatever language. I ran this through Millie and it seems an acceptable compromise. This is also good for us, as we might be able to pull into the core ER community some of the participants to this exercise


Can you, Matt, think of an implementation that fits into our information architecture? There are choices to make that could be very important. For example, is the whole exercise one group on Edgeryders? Is each country a separate group? What are the advantages and disadvantages of each solution? These are never just (or even mainly) technical choices, but it is important to get a feel for tech constraints. Also, I would like to get an idea of the experience of the Internet in those countries. Are people used to looking at menus in English, for example?

i18n, l10n and TMS

Just looked into how multilingual sites are meant to be done with Drupal Commons. It says it can work with Drupal’s internationalization and localization modules, and it was recommended to also add the Lingotek translation management system. Together, these Drupal modules would enable to have multilingual versions of any Drupal content, based on fields. Translations can be contributed by machine translation, users with a special “translator” role, and any commercial translation agency working with the Lingotek API. The TMS would show and manage all the translation tasks.

I know you once said, translation is a boring task that nobody wants to do. Ture, but now there would be some money in this project to pay bilingual content contributors from the countries of the exercise to also translate the content of those who are not bilingual. This could even be an added selling point, since that added cognitive processing of others’ content could lead to deeper / improved discussion quality on the platform.

This proposal of course depends on how great the Lingotek solution is … but it’s seemingly the best thing available for Drupal, and has tools such as syncing tanslations after the original was updates, a translation memory etc… Machine translation can be a starting point for editing it afterwards … seems to be faster than pure human translation …

One group or many: a group is largely a notification / cohesion mechanism, so it might make sense to set up groups per topic, not per country, also enabling cross-country collaborations. As also done in the original Edgeryders 1.0.