Manual

Localisation testing

Localisation testing checks translated builds for linguistic accuracy, cultural appropriateness and the technical problems that appear when text changes length or script. German overflows buttons, Japanese needs different line breaking, and Arabic reverses the entire interface.

What it catches

• Text that overflows or truncates in longer languages
• Untranslated strings and leftover placeholder keys
• Concatenated sentences that produce ungrammatical results
• Date, currency and number formats wrong for the locale
• Imagery or symbols that will not pass regional review

How the work actually goes

The most common technical localisation bug is length. English "Save" becomes German "Speichern" and Russian "Сохранить". A button sized to the English string clips both. Testers check every screen in every language, which is why localisation passes are scheduled late but planned early.

The second most common is concatenation. Code that builds "You found " + count + " " + itemName works in English and fails everywhere with grammatical gender, cases or different word order. Testers who can identify this pattern save their teams repeated rework.

Cultural review sits alongside the linguistic pass. Colours, gestures, flags, religious imagery and historical references all carry meaning that varies by market, and getting one wrong can block a release in a territory. You do not need to speak every language to be useful here, but you do need to know what to flag.

Tools you'll meet

• Pseudo-localisation builds
• String management (Crowdin, Lokalise)
• Font fallback inspectors
• Native-speaker review workflows