Translation by Computer (Sort Of)
Every now and then I return to the topic of why we don’t use Google translate or something similar to translate the Bible from English into all of the minority languages of the world (you can find some of my musings on the subject here).
Please don’t get me wrong, I think Google translate is a wonderful system. It does a fantastic job of translating between distant languages such as English and Korean. This means dealing with changes in word order, mismatches between tenses, grammatical categories which exist in one language and not the other and widely differing figures of speech. It’s frankly stunning that it copes with all of that.
The way that Google translate (other translators are available) is by comparing large amounts of translated texts in both languages and learning all of the things I mentioned above and much more, too. This raises a problem for Bible translators, as I wrote in 2009:
In other words, the model for doing translation is built on translations which have previously been done by humans. What is the text that is most often the first to be translated into minority languages? Yes, you’ve got it, the Bible (or sometimes the UN declaration of human rights). So, in order to do an instantaneous translation of the Bible, we will need to load up the system with a translation of the Bible. I don’t actually see much of a short-cut here.
However, there is a possible shortcut…
When two languages are closely related, there are far fewer grammatical and semantic differences than between two distant languages such as English and Korean. In these cases, a software tool called AdaptIt can come in useful. Using AdaptIt, translators can produce a first draft of a translation from a translation which has already been completed in a related language. This is a great step forward, but it doesn’t do away with the need for people to be involved.
Take the case of the Néyo language in Ivory Coast, which is closely related to Kouya. Twenty years ago, we were wondering how we would ever see a translation completed in Néyo. However, using AdaptIt, it has been possible to produce a draft translation in Néyo using another related language, Godié as a starting point.
It’s at this point that the role of the humans in the translation process becomes important. The first thing that has to be done is to tidy up the Néyo text so that it sounds natural, rather than like a translation. Then the various key terms in the text (prophet, justification, Holy Spirit and such like) have to be worked through. In the case of the Néyo translation, the various churches in the area have agreed on which terms should be used and the translation team has to check through the whole manuscript to make sure they are all ok (computers help with this).
It is then important to make sure that Néyo people can actually understand what the translated text is trying to say, so the team visit lots of people and read the text to them, asking questions to see how well the translation communicates.
The final stage is to make sure that the Néyo text accurately reflects the original Greek. This is where Sue, and another technology come in. This week, Sue is working with the Néyo translator, checking the translation of Matthew’s Gospel. Pasteur Mattias is in Abidjan and Sue is in our spare room; they communicate by Skype. For Sue, it’s great fun to work in a language which is so closely related to Kouya; but it’s hard work too. They communicate in French and Sue has the text of Matthew’s Gospel on her screen in Greek, French, English, Kouya and Néyo.
The dream of an instant system to translate the Bible into minority languages is still a long way off; but there is no doubt, that technology, wisely used, is a huge help to translation work.