Demystifying Video Game Localization

By Siyang Gao

Photo by Glenn Carstens-Peters on Unsplash

The idea of “video game localization” can be intimidating. In fact, the number one misconception about it, as far as I have seen, is that it requires a lot of technological know-how. But does it? Now, relax and have a cup of tea. Let’s see what video game localization is really about.

Coming from the domain of science, I prefer to approach things from fundamental principles. What do we actually do when we translate? First, we decode the source text in one language to extract literal meanings, connotations, specific styles, plot devices, and terminology. Then, we encode everything in another language.

What is there to decode from a video game? In this day and age, with the exception of some vintage-style games, video games are overwhelmingly about storytelling. Games like Disco Elysium and Baldur’s Gate 3 dedicate millions of words to the art of storytelling. First-person shooters (FPS), a traditional shut-down-your-brain-and-shoot-everything-that-moves genre, are growing increasingly like interactive movies. Even the digital equivalent of dinosaurs, such as the Super Mario franchise, are visibly getting more story-heavy.

When translating a video game, the majority of your time will be spent on:

• Making sure plot devices in the source text works as they’re supposed to in the target text;
• Delivering dialogue lines that are credible and natural for specific characters;
• Fine-tuning the dramatic tension of each scene to “just right”;
• Ensuring that the final product could stand on its own, without requiring players to refer to the original edition. (Absolutely no one, except for localization translators, plays the same game twice in two languages.)

Essentially, video games are multimedia-based interactive novels, and video game translation is good old literary translation, albeit this time around it actually makes okay money. As a translator, if you’re ready to take on short stories or even novels, you’re already 80% ready to try your hands at video game localization.

The other 20% of readiness primarily comes from technical common sense and what I call “non-linear perspective.” The spread is about 5% and 15%, respectively. Mind that these figures are entirely subjective. It’s not rocket science, after all.

Working with game texts, every now and then you will see strings like this:

{charName} sizes you up suspiciously. “Can you really afford it?” {genderPronoun} asks. “It’s gonna cost you %s dollars.”

In this example, {charName}, {genderNoun} and “%s” are called “variables,” meaning display text segments that are to be filled out on the fly when the game is being played. As you can infer from the context, “charName” most likely means “character name,” “genderPronoun” is self-explanatory, and “%s” should probably be a number. In an actual game session, the player might be talking to a merchant called John Smith, where this line would appear like this:

John Smith sizes you up suspiciously. “Can you really afford it?” he asks. “It’s gonna cost you 50 dollars.”

The reason why there are variables mixed in otherwise plain texts is that depending on the player’s actions, these parts may appear differently. Say, if this player hasn’t been very nice to John Smith, he might jack the price up a bit.

John Smith sizes you up suspiciously. “Can you really afford it?” he asks. “It’s gonna cost you 100 dollars.”

Perhaps in another player’s session, John Smith got killed as a result of the player’s previous actions. His wife, Jane Smith, might take over the shop. In that case, the player will see something like this:

Jane Smith sizes you up suspiciously. “Can you really afford it?” she asks. “It’s gonna cost you 50 dollars.”

See? Instead of writing the mostly-same line with a bazillion small variations to cover every possible situation, the developer could just write it once to be reused countless times by introducing those three variables.

Of course, in real life, variables may not be named exactly like these. The one for character name doesn’t have to be “{charName}” or wrapped in curly brackets. It could be anything—“[mingzi]” or “$s21” or “%soihsfjdajhgfdslk”—because the computer that runs the game sees it strictly as a name. The bottom line is, there will always be some kind of symbol to make it stand out as a variable so that the computer that parses/compiles the code won’t get confused. Plus, its name will always make some sense in one way or another, because the programmer, a human juggling dozens of variables, must be able to tell what’s what. Naming variables like “a”, “aa”, “aaa” would certainly backfire.

As a translator, what you do with variables is…nothing. Keep them strictly as-is. Don’t translate variable names, no matter how much they look like natural language. Don’t tamper with their identifying symbols, or else the compiler won’t recognize them anymore. A special note: Chinese and English parentheses, although looking very much alike, are different symbols!

That’s about it, your 5% of readiness, and it’s really not that hard to master. All you need to do is pick a programming course and learn about the fundamentals of variables. I learned the C language, but these days I’d recommend something more relevant, like Python.

As to the other 15%, namely, “non-linear perspective,” it’s a bit more complicated.

Different from novels or movies, video games don’t just unfold in one direction. In the movie The Matrix, Neo took the red pill, independent of the will of movie-goers. In an interactive video game, however, players want the freedom to choose, even if it means game over, because endings, good or bad, are an integral part of the experience. So, the story would branch out when Morpheus presents two pills, with distinctively different dialogues and developments piled together. With one branching point leads to another, pretty soon the text becomes a scrambled mess. Being able to tell where one dialogue section starts and ends is critical. You’ve got to learn to be a Timelord in Doctor Who—perceive time differently.

How does one do that? By playing the game fully before diving into translation, as a matter of course. While having fun, get to know the lay of the land, mark story branching points, and chart nooks and crevices that you fail to reach in your playthroughs based on the overall context. If everything fails, ask the developer. Communication is a huge part of video game localization, as questions (a lot of them) are inevitable, given the nature of your source text.

Altogether, the technological bar of entry to game localization is in fact quite low. What it really takes is the love for the game, just like the translation of literary work. You’ve got to have enough love to wholly dig the game/novel and its nuances. In the overwhelming majority of failed game localization projects, the root cause, instead of “mistranslation,” is the failure to create a sufficient level of immersion. Sometimes 1+1=5, so to speak, if a justified situation calls for it. If you always wanted to do literary translation but got dissuaded by the grim outlook of compensation, consider giving game localization a try. Don’t worry about the know-how part, as you only need to work on 20% of readiness.

Siyang Gao is a programmer-repurposed-into-a-translator, starting off with video game localization (from English into Chinese), then branching into a wide spectrum of industries over the next 18 years, including but not limited to journalism, literature, legal, medical, IT, and fashion. Since 2022, he’s been returning to his root by picking up video game localization work again, this time from Chinese into English. He believes Chinese game developers are underrepresented in the global community and a big part of the reason is poor localization, from translation to project management. His current focus is to find a replicable methodology that could effectively facilitate teamwork in game localization.

Edited by Fang Sheng and Sijin Xian.