You may have noticed that a few years ago Facebook created a translation app to crowd source translations. The basic concept works like this:
Step One – Translate the Glossary with all the core terms, people vote translations up or down to determine which ones stick.
Step Two – Translate all of Facebook into the other language, with people voting up or down the best translations.
Step Three: Review everything, with people voting to confirm it’s actually working and accurate.
Facebook provides all the tools that the crowd needs to help translate such as a style guide and a discussion board to talk about the nuances of the translation.
Thanks the Translation App, Facebook is now in all kinds of languages – even some relatively new ones…like Leet Speak.
Now, here’s the question. Why couldn’t other governments do the same thing? I’ve ranted before about some of the challenges facing government when it comes to languages other than English. While it’s nice to have a translator for in the office, trying to translate a whole website into another language can be a time-consuming and expensive project.
But there are real advantages to crowd sourcing. Can you speak Ethiopian? Neither can I, but there are hundreds of people in my service area that do. It would be a great community service project to try and get as many languages as we could get. It would be even better if we could share glossaries between different cities and agencies so that as one city progresses in it’s translations (say, Chicago in Polish) – other cities (like Dallas) could benefit as well.
There are some things that we would still need experts (teachers or lawyers) to take a look at – such as what the law says. However, couldn’t we crowd source some of our translations needs using an App like this? (Assuming somebody could replicate it – or knows somebody who knows Mark Zuckerburg)
If your agency were handed the technology would you feel comfortable crowd sourcing your translations?
Great post. I imagine you’d run into the issues of “how do you know its accurate”? But between Facebook and others I think the time/$ savings plus advantage of crowd makes sense
Well, for certain things you would have to have an expert. For instance, city ordinances should not have their contents crowd sourced to the masses. Those COULD be crowd sourced to just bi-lingual lawyers/teachers.
Yeah that’s the problem is that gov’t has the burden of being right where as facebook can roll something out and if it doesn’t work eh it doesn’t work on to the next thing. It is a very cool idea though. Does anyone know if that code is open source? Maybe whatever cities get CfA they could try tackling this.
What about using Google Translate to fix a page automatically? Wondering how accurate it is…
Google translate is pretty good (I think it works the best of all the ones that I have tried)… but until semantic technology is more advanced and/or applied we will always have the problem of context. I always tell people a story about how the word for ‘straw’ in Venezuelan Spanish means a completely different thing in Mexican Spanish… Hmmmm it puts one in a very awkward situation when employed by teenagers from both countries while having lunch! Even when sematics are more applied, languages evolve so it is a moving target.
Nothing beats a human translator
I think the most common error we run across is Workers Compensation. Our telephone system translates it as literally, Trabajadores de Compensación and it gets confused with a paycheck VS
Compensacion por accidentes del trabajo (Compensation for an accident as work) which is what we mean.
You can add Google Translate to your website with a cool widget, but you don’t have control over the weird errors the syntax or multiple-choice options result in (if I say “tea” in the UK am I talking about the drink or the meal? The tool can’t tell the difference and select the appropriate term in Spanish – te’ or cena). For government, I think the issue is primarily one of being blamed for bad translations because they have no control. So much Federal content has ties to legal content that even fairly benign text can become a big problem in the right hands. There is a Google Translate Toolkit which may be more what you are looking for, but I think that is more for assisting you in translations and not automated. There may very well be APIs you can use. Interesting thought at least!