Postediting

Post-editing is the process of revising machine-translated or draft text to meet defined quality standards. It is commonly applied in translation, transcription, subtitling, video and audio captioning, and broader localization workflows. The term distinguishes this work from editing of text written by humans, since the starting point is machine output or an imperfect draft rather than a finished source.

Two main types are commonly described. Light post-editing aims to render the content understandable and reasonably accurate with minimal changes, prioritizing speed. Full post-editing seeks near-human quality: faithful meaning, fluent target-language style, and thorough localization, with comprehensive corrections to terminology, tone, and readability.

Workflow and tasks typically involve assessing the MT or draft output, correcting inaccuracies in meaning, fluency, and grammar, and ensuring terminology consistency with glossaries and style guides. Editors adjust formatting, numbers, dates, and other metadata, and may restructure sentences or rewrite passages to fit the target audience. They verify cultural appropriateness and localization of references, and perform quality checks using built-in CAT tools, terminology management, and QA software.

Tools and metrics: Post-editing commonly uses computer-assisted translation tools, MT engines, term bases, and QA utilities. Metrics for evaluation include editing time, post-editing distance or effort, and acceptance rates. Standards such as ISO 18587 define requirements for post-editing of machine translation output and guidance for professional practice.

Contexts and considerations: In localization, post-editing balances speed and quality according to use case—consumer content versus regulated or specialized texts. Ethical considerations include data confidentiality, client guidelines, and adherence to applicable laws and industry norms.