AIQA looks for language issues in text
AIQA can flag potential terminology, style, grammar, and other language issues using the source, translation, and project guidelines. Translators check each finding and decide whether and how to fix it. The number of AI flags is not the number of confirmed errors, and a text with no flags may still contain errors.
| Comparison | AIQA | In-game LQA |
|---|---|---|
| Environment | Source, translation and project rules | A playable test build |
| Focus | Terminology, style and language issues | Display, UI fit and scene context |
| Output | Human-reviewed language revisions | Traceable issues, fixes and regression checks |
In-game LQA checks what players experience
Correct text in a spreadsheet can still fail in the game. A button may clip, subtitles may appear in the wrong scene, or a form of address may conflict with the current story state. In-game LQA checks actual display and context in a test build, records traceable issues and supports fixes and regression checks. Agree on languages, modules, build versions and access conditions.
Connect text review to in-game verification
Identify potential language issues
AIQA flags potential issues; translators confirm which need fixing.
Update localization files
Correct actual errors and recheck affected strings.
Check how the text appears in the game
Check display, scene and interaction in the agreed modules.
Verify fixes
Retest corrected issues and affected locations.
Link findings to the files and scenes they affect
Text issues should identify the relevant localization file; in-game issues should point to the string or implementation involved. Record the string ID, source, translation, issue type, evidence, and status. In-game reports also need a build version, scene, screenshot, and reproduction steps. Recheck affected content after changes, and quote and schedule each type of review separately.