Battlelore, take two...

I mentioned last week how the board game Battlelore started me thinking about readability and my dissertation. I said at the end of my post how Battlelore had got me thinking about the problem, but how it also got me thinking about the solution as well.

Battlelore comes with 8 or 9 battle scenarios. You set the board up according to the rule book, and then play the game. But the first 4-5 scenarios are ‘limited’ scenarios. They are more basic in order to teach the game in a step-by-step method. So in all actuality, you only have a handful of scenarios to play.

So the game company set up a web site where players could submit their own scenarios. Since the game is produced in Germany and France, some of the user scenarios are in a different language. I came across one of these scenarios in French, a language I don’t speak. At all.

But being a fan of Douglas Adams, I’m familiar with babelfish, software that translates text from one language into another.

The software isn’t perfect. Not anywhere nearly so. In fact, sometimes it’s fun to take a piece of text, translate it into another language, and then back into English. You’re left with all sorts of crazy stuff. For example, here is the Gettysburg Address after being translated into Korean and back.

But the software does basically what you want it to. You end up with text that is at least understandable. So I copied the instructions for the scenario into the field, hit submit, and was left with English that wasn’t even close to grammatically correct, but made sense (with a little bit of thought). I copied the translation, fixed all the errors, and resubmitted the game as an English translation.

I had just translated text from French to English, without knowing French.

To take an article in wikipedia written at a post-graduate level, and try to simplify it is a daunting task. Even more daunting is the thought of getting a machine to do it automatically. With current technology it’s not possible. But what if, like the translation software, we could get the machine to do most of the work. Then a human can come in and ‘clean things up’, just as I did with my French translation.

I think it’s our best bet to get the vast amounts of information currently in wikipedia to a larger audience. A group I work with is looking into the matter, and we’re working on software that will help facilitate the work. I’ll keep you posted.