Localization Issues: Why Do Game Translators Make Mistakes?

Life can be unfair for game translators. You may come up with a fantastic translation for a 100k+ words adventure game, but a single typo or mistranslation will get you trashed on game forums worldwide. One may however wonder why even the best translators can make absurd mistakes and why everybody down the localization chain fails to notice them.

I decided to write this article after reading Sony’s presentation about the future of localization at GDC 2014. In particular, I liked the following series of witty Q&As.

● Why do we need to test localized assets?
– Give us the assets earlier and in one batch and it’ll
need less testing

● Why don’t the translators get it right first time?
– Give us the assets earlier and in one batch and it’ll
need less re-work

● Why is localization testing so expensive?
– Give us the assets earlier and in one batch and it’ll
be cheaper

Further into the presentation, we are given a series of pie charts about localization issue distribution: we first learn that for the games covered (a number of SCE titles, pre-testing), 56% of localization issues were actually implementation problems rather than something wrong on the translator’s side.

As for translation quality issues themselves, less than half of them were translator mistakes. The majority of localization quality issues were caused by lack of context, errors in source strings and file issues.

Of course, this is a small study on a limited number of games by a single editor. But it gives you an idea of what can go wrong during localization.

Lack of context

Game translators will usually send query sheets to their clients for any context issues. But sometimes it just isn’t enough. Sometimes we don’t receive answers by delivery time and have to go with our best guesses. At other times the text’s context can be misleading and trick us into thinking we got it when we haven’t. That’s why game debugging is not simply a matter of functional testing. We do need to see the localized strings in the game itself to be sure they make sense.

Source text and file issues

That type of issue is actually similar to the previous one in some ways. Sometimes source text errors are pretty obvious (typos, missing words, etc.) and we can report and fix them as we go on with translation. Sometimes they are more subtle (wrong skill/object name displayed for example) and only become apparent once you see the game in action, if you are very attentive to what’s going on. That is if you manage to maintain an absolutely perfect concentration after 50 hours testing the same game.

Yes, sometimes we screw up

Breaking news: We, translators, are human beings! As such, we have our failures and make the odd mistake, may that be a mistranslation or an awful typo. It doesn’t matter how many times you proofread your own text, it just happens every now and then.

Of course game testers are here as a backup, and while they do an excellent job overall (Sony indicates they are able to fix 98% of localization issues in the above-mentioned presentation), they will occasionally miss a bug or two. Remember, they’re human too. And it’s hard to blame them when they spend whole days scrutinizing texts, looking for errors. I mean, anybody would lose their concentration at some point if they had to talk 3 times to every single NPC in a RPG.