Machine Translation Post-editing
Machine translation post-editing is human revision of machine-translated text against the source, context, terminology, and agreed quality requirements.
Machine translation post-editing is human revision of machine output against its source and intended use. ISO 18587:2017's public abstract covers full human post-editing and post-editor competences. ISO public abstract.
Meaning and structure need different checks
Fluency editing asks whether target prose reads well. Source-target review also asks whether the meaning survived: conditions, negation, quantities, terminology, and claims. A mechanical placeholder check examines a narrower structural requirement. These checks can contribute to one release without answering the same question.
Illustrative fictional fixture: an English help message permits export after admin approval and says not to delete an archive. A hypothetical target preserves every placeholder but drops the approval condition. Another hypothetical target preserves the condition while changing {{workspace_name}} to {{workspace}}. Neither example is a reviewed foreign-language translation.
The first defect needs qualified meaning review; the second can be found by exact placeholder comparison. A fluent rewrite or another model's back-translation does not establish that a human reviewer compared the target with the approved source.
The approval record should identify both source and target versions. A later source edit can reopen the affected segment even when the earlier target was reviewed carefully.
Preserve the reviewed version
The localization-review workflow records segment IDs, context, glossary version, findings, and reviewer approval. DeepL is one drafting option to evaluate within that process.
The localization glossary builder prepares terminology from supplied material. The localization copy reviewer proposes source-target questions, while check-translation-tokens checks supported named placeholders. The language reviewer resolves meaning and locale judgments and approves an identifiable target version. Citing the public ISO abstract here defines scope; it does not claim that this workflow meets the full standard.
Filed under
machine-translation-post-editing · knowledge-work
Related
- Check Translation TokensCompare recognized named placeholders in a source-target translation CSV with Python, preserving token multiplicity and flagging blank targets. Supports exact {name}, {{name}}, and %(name)s/%(name)d forms; does not validate full ICU, HTML, positional placeholders, or translation meaning.
- DeeplDeepL provides AI translation for text and documents, with preferred-term glossaries and paid options for team translation and customization workflows.
- Localization Copy ReviewerReview paired source and translated marketing or help-document text for a specified locale against supplied terminology and style rules: flag lost meaning, negation, conditions, numbers, protected terms and ambiguous segments, with evidence-linked proposals for a human language reviewer.
- Localization Glossary BuilderBuild a draft bilingual terminology glossary from supplied source-target samples and product references: retain term IDs, approved equivalents, protected strings, context, disallowed variants and unresolved translations. Use before reviewing localized marketing or help copy; do not translate a whole document or invent approved terms.
- Review AI Localization with a Glossary and Source-Target ChecksReview AI-translated copy with a source-target register, approved terminology, protected placeholders, contextual checks, and qualified language review.