Users Translating for a Difference: An African Experience
By: Dwayne Bailey (Translate)
06 June 2012
The crowd is the buzzword. But what if the crowd is smaller than a football stadium, maybe an unskilled crowd of one? We're helping those teams by pretending it's easy.
Letting go of the translation process is very scary, especially for traditional localizers.1 But letting go can also be exciting. If you assume that all of the things that we traditionally think of as important to localization - translator qualifications, style guides, terminology lists - are all optional, then you will find a wealth of new talent, new languages and new markets.
Letting users translate your software allows them to make it more useful for themselves and allows the software to touch markets that weren't even on your radar. Now that sounds much more exciting than trying to control localization, doesn’t it?
There is so much useful and powerful software out there, software that could literally change lives. In many cases the only reason it isn’t changing lives is because it’s not available in a local language. While we could excuse this in the days before phones, apps and tablets, companies can’t hide anymore from the fact that the world really can and should be their market. Their products can go anywhere, if they just speak the language of the user. So how do we achieve that? Hide overly complicated translation processes from passionate translators and let them do what they’re good at. In other words, get out of the way.
Complicated localization processes
When we've engaged with companies who localize a lot, I've always been surprised at the complexity of their localization processes. I wouldn’t be worried if I felt it resulted in better translation quality, but I'm not that sure it does. In spite of their desire for high quality translations, it seems that most of the time is spent navigating the process, not improving translations.
Which I guess is my way of saying that I don't think the holy grail of very strict formal localization processes results in higher quality localization. And this is important when considering, in our case certainly, that people are volunteering their time to translate. They aren't volunteering to follow a strict process.
Why do we see so much localization – even with bad processes?
We see a lot of translation even with bad processes. We've all seen the brag numbers and we'd all love to get to 100+ languages. People will put up with a lot because first, they don't know any better and second, they want your software in their language. But if you aren't the latest internet hot property, you might want to consider how to make the engagement with your community feel comfortable.
We find many localizers in Africa that are passionate about their languages. Like anyone else in the world, language is a large part of our African identity. We're ready to roll up our sleeves and make a difference for our language and our people. What holds us back? Many things. The lack of a computer. No tools, access to tools or understanding of the localization process. No dictionaries. A poor education in their mother tongue.
Now you add a complicated programmer-driven localization process and you have people who are completely unqualified. But do they really need those skills to translate? It is obvious the answer is no, but our processes and tools scream yes.
So in Africa we have two groups. The first group contains the people most qualified to manage localization; those that can do the tech work, who don't care about the language. (Would you if local languages weren't what you saw as your future financial success?) Then we have another group: those who do care and have the translation skills but can't manage the process.
We have complicated localization processes and we have some seriously under-skilled translators. So what do you do?
Pretending it’s not that hard
Hide the process from translators. We set ourselves in the middle doing localization engineering and building tools that make it seem really easy to localize the product. Behind the scenes we're doing some amazing feats of localization engineering, working to eliminate problem areas in the localization process and translating “techie speak” into human.
The programmers are happy and none the wiser because we mostly don't touch their process. (Although we try to educate them as we go along.) And the translators are happy because they’re making a difference. They see a simple process of translation. They either don't know or don't care about what happens behind the scenes.
To achieve this we built tools to do the engineering. The Translate Toolkit is an open source toolset to manage large parts of localization engineering. We also built Pootle and Virtaal, which allow simple web-based and desktop-based translation. The aim with all of these endeavors has been to create tools that are simple yet powerful. We are trying to bring the best of localization that might be missing but presenting it in a way that volunteers can use.
What happens when you do this?
Let’s look at the Firefox example. Firefox is the web browser that really heralded the new age of the Open Web. Remember when IE6 ruled supreme and the Internet was rather dull? Along came Firefox and changed that.2 As open source software developed by an open community, Firefox is a little easier to translate than other browsers. What’s more, with our tools it can be done by a small crowd, or even a crowd of one. It’s one of our shining examples of what can happen when we let translators focus on what they do best, rather than on complicated processes.
Keep in mind that we're not the whole Firefox localization process. There are many other people and other languages (and continents) involved. We've set our focus on helping African teams. A few years ago Afrikaans was the only African language in Firefox. Now volunteers are translating another eighteen. Of those, about nine are active and are regularly updated by the volunteers. We've also added a number of other languages that were translated by a team of one or by teams who just needed a little help. So Urdu, Sakha (Russian Federation), Welsh and Scottish Gaelic are also being assisted.
Hide the truth
What we've found is that getting more done is not about education, manual writing and up-skilling our translators. It is also not about changing the fundamentals of the localization process. Rather, what we've found that works is to hide the truth from the translators and work hard behind the scenes to build tools that make that possible – to change processes so that they match the reality of the translators that they are serving.
And we have the results to prove it.
Dwayne Bailey is a localiser at large for Translate, a South African based localization company. His primary focus now is community localization. Helping people change the world, helping companies harness the skills of their most passionate users and building community translation tools that make it all happen. He is assisting Mozilla to support and bring new African languages into Firefox. He also runs a team of developers extending translations community localization tools for clients around the world.
Translating in a simple web tool allows the translators to focus on translation. It is a continual fight to keep the interface simple, yet powerful. Here we see Translation Memory, Alternate Source languaage and Terminology.
 For tech companies who know nothing about translation and localization it seems like the hip, Internet way to do things, but sometimes I think it's a good thing most of the staff in those companies don't speak any of the languages that the crowd is translating into.
 I have a personal love affair with Mozilla, the non-profit that develops Firefox, and with the browser itself. While the new kids are on the browser block and the old have pulled up their browser socks, there aren't any whose sole purpose is to build a great tool, protect me, the user, and take the internet to a wonderful new place. I trust Mozilla's motives, I'm not sure I trust the motives of other organizations.