LocJam Post-Mortem: On Grandpa's Localization and its Challenges

Today I will share another of my blog posts on video game localization. This time, I am sharing my impressions on the LocJam 2, a video game localization contest that took place just last week. This localization project offered various challenges to both experienced and aspiring translators. Many thanks to the organization team - they've done an amazing job so far!

After missing last year's edition, I took a shot at the LocJam 2. Contests of this kind are rare, and it was a fantastic chance to brush up my skills and measure them against other professionals of the game localization industry.

This year, game translators were offered a nice interactive story from the GameJam, Grandpa. A creepy atmosphere and a disturbing twist (although somewhat predictable?) gave us translators good material to test both our technical knowledge and literary talent.

The Process of Localizing Grandpa

Translators were in charge of the entire localization process for this game, up to final checks. The idea was to deliver a localized file that would be ready for commercialization.

Familiarization

Some games don't necessarily need a lot of familiarization. But as Grandpa is an illustrated interactive story, it was absolutely vital to understand exactly what was going on and how things looked like in-game. With less than 20 minutes needed to complete the adventure, it was a small but crucial time investment.

First draft

Game over. Time to take a break and digest the story. Now let's get down to the real business. The organizers recommended the use of a classic text editor, which I found to be a rather surprising choice. Instead, I used my personal computer-assisted translation tool, and anybody could have downloaded some of the free ones available out there. Considering the characteristics of the source file, it was a no-brainer for me, but I'll justify my choice in detail.

Drafting a first version based on what I remembered from the game took me about half a day. A couple of sentence got me scratching my head to produce a fluent French text, but overall it all went smooth. The excitation of the competition seems to have had a positive effect on my creative power!

Proofreading and basic QA

Before plugging the translated text into the game, I gave it a round of proofreading and QA. I had a good read at my version, spell/grammar-checked it and used automated QA tools to detect potential critical mistakes. In a regular game translation project, I would have had a last look at my text to improve what can be (especially style), but I decided to do it while testing the localized game.

In-context QA

My plan was to go through the game just once slowly, checking out every possible situation and improving my translation while fixing any possible bugs. In the end, I gave it two good rounds and another half a day of intense testing. Source text inconsistencies and minor bugs added to my own small mistakes, and it just turned out to be too much for a single run.

I could have gone for 10 extra runs and kept changing things every single time, but would have it benefited the game? Probably not. One never feels ready to let the loved ones go, but sometimes you just have to (right Emi?). Holding my breath, I uploaded my translation, feeling both relieved and sad as a different kind of interactive story was ending.

Localizing Grandpa: A Trying Challenge

In just around 1,500 words, Grandpa offered a complete challenge for even for the most experienced game translators. You want to get started in the localization industry quickly? Localize this game and you will know 80% of what you need to understand in this industry.

Code and text tightly mixed: It's unusual at all to see variables and other code bits in a source text. But in the case of Grandpa, it was sometimes really hard to tell what would happen to strings elements. Don't translate tags: pretty common, I'll take that. Now what do I do with [img[Poor Torn Hat|badhat 26]]? Knowledge of HTML helped me understand what was going on here, but I still proceeded very carefully (for the record, you needed to translate "Poor Torn Hat", which is the tooltip that appears when you stop your mouse on the object's image).

The dangers of copy and paste: The second part of translation was composed of game's inventory repeated for each screen with the heavy presence of tags and HTML elements similar to the ones mentioned above. How many occurrences of this inventory? 48 times. I take my hat off to all the translators who went on with a text editor and copied and pasted this without a mistake 47 times. By the way, have you noticed that ONE of these occurrences was slightly different than the others? :) My translation tool did, and it also took care of the copying and pasting tasks for me. It saved me a lot of time and helped me avoid painful mistakes.

Translators who didn't take notice shouldn't panic, though, that one slightly different occurrence actually works the same even if you copied one of the 47 other inventory placeholders, it will look the same to jurors. In other circumstances, it could have been a fatal mistake!

Stylistic choices: If you forget about the technical aspects of game localization, the project felt more like translating a book. Thus, style was maybe the most important element of the translation. Finding the right words and expressions to render the original tone gave me a few headaches.

Besides that, two things really bothered me: On occasions, I was never able to figure with a perfect certitude if certain actions were instructions from Grandpa or generic action buttons. "Look in oven" sounds very mechanical for example, but "Let's go somewhere else" are definitely the words of a human living (well, not quite for this game...). My other interrogation was: "How old is Emi (the game's main character) supposed to be?". Sometimes she will sound very childish, but then on the next screen she would use a sewing machine with great dexterity. "supposed to be" is really important here, since the whole problem is made even more complicated by the fact that...

The game ends on a twist: While you play the game, you can tell that something is going to happen by the ending screen. Looking back at the game, there are elements I interpret as hints, but were they really? Was it the intention of the developers, or just a bit of a stretch from my imagination? How much should I give away in the French text? A lack of subtlety would spoil the game. But give too few hints, and the game loses some of its atmosphere. Here again, there were heart-breaking decisions to take.

Translation AND QA: For this contest, we were asked to wear two hats as we acted as both translators and testers. Have you ever let a typo slip despite proofreading your text three times? This is why we have independent testers in the first place: our brains generously fix mistakes in the things we write. Even the best of the best make the odd mistake.

Testing your own work requires an incredible level of concentration, and it was one of the toughest challenges of this project.

The game had its small bugs: We're talking about really small stuff here, and you have to remember the context in which the game was developed. If I had to code for two days straight with no or so little sleep, I don't think I could put out a functional Pong-clone. Still, as testers, it's hard to close our eyes on the occasional bug. There a few really small inconsistencies in the punctuation, which is easy to fix when you translate.

The hard part came with transition screens, which all showed "Office" even if you were switching between two totally different places. One of these screens also had an image of the office in question. To get this one fixed, we had to look at the different screen IDs to figure out which string corresponded to which place. That was a hard one for those who already had trouble figuring out what was code and what wasn't.

The layout was a bit messy at times as well, as line returns were occasionally used for conditional elements even if they were not display. Sometimes you would have as many as 5 line returns before the next text, which looked a bit weird. Or on occasions there would be 2 line returns before an image instead of 1. I doubt it will have any influence on the final results, but I fixed these small issues and made sure the layout was homogeneous throughout the game.

Conclusion

Localizing video games requires linguistic AND technical skills. As a translator, Grandpa was very challenging from both points of view, and as such, it was a perfect game for the needs of the LocJam. It is also a very interesting projects for those who are new in the industry or interested in becoming a video game translator in the future. As a professional, it served as an excellent practice exercise. The contest is over, but you can still give it a try!