postédition

Postédition (often rendered in English as post-editing) refers to the process of reviewing and correcting machine-translated text to produce an acceptable target-language output. It sits between fully human translation and fully automated machine translation, aiming to combine machine speed with human judgment to achieve clarity, accuracy, and appropriate style.

Post-editing is commonly classified into light (or gisting) post-editing, which focuses on comprehensibility and rapid turnaround, and full post-editing, which seeks publication-quality text meeting stylistic and terminological standards. The choice depends on purpose, budget, and domain requirements.

The workflow typically begins with machine translation output, followed by human review using computer-assisted translation (CAT) tools, terminology databases, and quality estimation software. Post-editors check for mistranslations, omissions, fluency problems, cultural appropriateness, and consistency with client-specific guidelines or glossaries.

Effective post-editing requires strong bilingual skills, familiarity with machine-translation error patterns, and competence with relevant tooling. Quality assessment may use human metrics (adequacy, fluency, error counts) and automated measures (BLEU, TER, or quality estimation scores), though human judgment remains central for final approval.

Challenges include variable MT quality across languages and domains, ambiguity in client expectations, and potential speed–quality trade-offs. Standards such as ISO 18587 provide guidance on post-editing processes and professional requirements. As neural machine translation continues to improve, the role of post-editing evolves toward specialized, higher-value tasks like domain adaptation, quality control, and creative revision.