| name | localization-ready-dialogue |
| description | Prepare dialogue for localization by preserving meaning, context, speaker metadata, variables, gender or plurality constraints, line relationships, timing, and cultural clarity while avoiding source-language wordplay that cannot travel without annotation. |
Localization-Ready Dialogue
Use before dialogue is handed to translators or localization vendors.
Procedure
- Give every line a stable identifier plus speaker, addressee or scene context, emotional intent, and neighboring lines when meaning depends on them.
- Expand ambiguous pronouns, subjects, references, abbreviations, or standalone fragments in translator notes when the line itself must remain natural.
- Mark variables, placeholders, tags, markup, gender, plurality, grammatical choices, and runtime substitutions with clear constraints and examples.
- Identify wordplay, idioms, rhyme, cultural references, invented terms, names, and jokes that require creative localization rather than literal translation.
- Avoid concatenating sentence fragments or assuming English word order where runtime strings can be localized as complete units instead.
- Define character voice and relationship context so translators can preserve status, formality, humor, and personality.
- Record timing, subtitle length, UI width, lip-sync, audio, or branch constraints where the localized line must fit a technical boundary.
- Validate imported localized samples in context and feed recurring ambiguity back into the source dialogue or localization notes.
Decision rules
- Context is part of the translation source.
- Prefer full translatable units over runtime string concatenation.
- Do not force literal preservation of wordplay when equivalent effect is the real requirement.
- Stable line IDs matter more than file order in distributed production.
Quality gate
Dialogue is localization-ready when every line has enough context and metadata to translate meaning and voice, variables and technical constraints are explicit, language-specific traps are annotated, and localized samples can be imported without reconstructing intent from the original writer.
