Games Localisation: Why Translation Is Not Enough

Translation is the start, not the result

Translating a game means moving words from one language into another. Localising a game means adapting it so that, in the target language, it feels as if it had been created there in the first place. Those are two different things.

Localisation is not just about the text. It is about cultural references, humour, units of measurement, date formats, naming conventions, legal requirements and the question of whether a pun that works in English also works in German. Most of the time, it does not.

What localisation actually covers

Anyone who has never worked with game text underestimates the scope. A mid-sized game can easily run to 50,000 or 200,000 words of text. And that text behaves differently from text in a book or on a website.

Variable strings

Game text contains placeholders: {player_name} has received {item_name}. In English this works. In German the sentence has to change depending on the grammatical gender of the item: "has received den Stahlhelm" (the steel helmet), "has received die Rüstung" (the armour), "has received das Schwert" (the sword). A single string, three variants. Machine translation fails here reliably.

Character limits

UI elements have a fixed width. "Save" fits on a button. "Speichern" needs twice the space. "Zwischengespeichert" blows it apart completely. Localisation also means working with the space available, without losing meaning.

Cultural adaptation

Humour is culture-specific. References to American pop culture that every child in the US understands mean nothing to anyone in Germany. Good localisation replaces them with equivalents that trigger the same emotion – not with a literal translation that falls flat.

Consistency across the whole game

A sword that is called "Blade of the North" in the description, "Northern Blade" in the inventory and "Nordklinge" in dialogue destroys immersion. Localisation requires terminology lists that apply across every piece of text – from the tutorial to the endgame content.

The five most common mistakes

1. Machine translation without review. DeepL and Google Translate produce usable output for emails. For game text with variable strings, gender adaptation and context that is only visible in the game, they produce errors that players spot at once.
2. Translators without gaming experience. Anyone who translates "HP" as "horsepower" rather than "hit points" has not understood the subject. Gaming has its own vocabulary, and that is not a matter of taste.
3. No context for the translators. Strings without context are a guessing game. "Fire" can mean "ignite", "shoot" or "dismiss". Supply no screenshot and no description, and you get random hits.
4. Localisation only at the end. When localisation starts after the game is finished, the text does not fit the UI, the terminology is inconsistent and the time pressure produces errors. Professional studios start localisation in parallel with development.
5. No Language Quality Assurance (LQA). Translated text has to be tested in the game. Does the text fit the box? Is the context right? Does the form of address work? LQA is not a luxury but the part that makes the difference between "translated" and "localised".

Why AI reaches its limits here

AI-assisted translation is getting better. For games localisation it still falls short. The reasons:

• Variable strings: AI cannot reliably adjust gender, case and number in dynamic sentences.
• Missing context: A string in a table has no on-screen context. AI translates what is there, not what is meant.
• Cultural nuance: Puns, allusions and humour cannot be transferred by rule. That takes creativity and a feel for the target audience.
• Space limits: AI does not optimise for character length. It translates meaning, not layout.

AI can speed up the process, for instance in the pre-translation of large volumes of text. But the result needs a human review. In practice that means AI quality assurance as a fixed part of the workflow, not as a substitute for localisation expertise.

My background

I come from the gaming industry. At Goodgame Studios and InnoGames I tested game releases and ran the quality assurance for localisation projects. At Jung von Matt I checked campaigns for BMW and Sparkasse against quality criteria. Whether software or language, the quality process is the same.

Today I offer specialist translation and localisation for the German-speaking market – with the technical understanding and industry knowledge that only come from working directly on the product.

Annemarie Andersen

Freelance language services provider

Specialised in Easy Language to BITV 2.0, editing and accessible communication. Before going freelance: quality management at Jung von Matt, InnoGames and Goodgame Studios. References: German Future Prize 2024 + 2025, Seniorpartner in School e.V. More about me.