Crowdsourcing translation
October 3rd, 2008
Finally I have discovered the correct (hip?) term for describing these community of users based translation efforts – crowdsourced translation.
This is the sort of thing we have seen recently used in the localisation of Facebook. An interesting development of this is the approach being taken by the social networking site Hi5. Hi5 was built to the OpenSocial specification which is i18n ready. In addition to crowdsourcing translations for Hi5 itself, they are extending it to include applications created by developers.
There is a nice blog entry and an earlier blog entry and podcast about this on mashable (thanks to Mike for the pointer). The Translation Service for OpenSocial Applications on hi5 and Developer Guidelines for Application Translation provide an interesting overview of the process.
Crowdsourcing translations would appear to have much to recommend it – it’s cheap and it might be a useful way of judging whether there is a market for your software in a given language. It may also be a good way of involving the language community and getting “buy-in”. On the down-side I guess they may be concerns about quality (depending of the size of your crowd), it may be slower than commercial translation and the ability to be localised needs to be built into the software from the start. Are there any drawbacks apart from these – if not it looks like a win-win solution to me.
Posted by djcunlif | Tagged with localisation, translation
October 3rd, 2008 at 09:59 AM On larger translation jobs you have to have a combination of both aspects. One of the key factors in the Facebook translation was that professional translators took part. A second key factor was that a few people took de facto lead roles in the translations, guiding style, grammar, and consistency. I would say why shouldn't the Welsh Language Board fund people to participate in the process of crowd or open sourced translation if the project is likely to be used by many. It would increase the speed of translation, ensure a level of quality control, whilst at the same time staying at arms length. You could call it the open source translation equivalent of match-funding!