IBM have announced that they have been working on an internal project to create their own machine translation systemn.Fluent. It’s a great project from the sound of it. The system has essentially been built using crowd sourcing: all the IBM staff have been carrying out the translations during their normal work by using plug-ins to the existing software, which then takes the text off to a central repository.

It’s great that IBM have thought of doing this; I guess they’re trying to make inroads on the plethora of machine translation tools, and the way they’ve done it using crowd sourcing is great. n.Fluent isn’t available outside of Big Blue yet, but it’s only a matter of time.

Will this replace professional translation?

While machine translation is getting better and more accurate all the time, there are still too many issues even for the boffins at IBM to tackle. The very fact that they’ve crowd-sourced the translations means that there’s no central editor to ensure the consistency of the translations and of the language.

The best way to think about this is to imagine people who aren’t language experts translating documents, all with their individual slant on the text, and then bringing the whole thing together to try and make a cohesive translation. As you can imagine, the language of the resulting document would be hugely inconsistent, making it very confusing for a reader. How would a book co-written by 200 authors read, even if they were all working to the same story line?

While I salute IBM’s effort on this there’s still no substitute for professional translators who know their subject.