Post-Editing

Post-editing (formally Post-Editing of Machine Translation, or PEMT) is the process of reviewing and improving machine-translated content to bring it to a specified quality level. A qualified human linguist reviews the machine translation output, correcting errors, improving fluency, and ensuring the final text meets the required quality standard.

Post-editing exists on a spectrum with two main service levels. Light post-editing (also called rapid post-editing) aims for comprehensible, accurate output without necessarily achieving publication quality. The editor corrects mistranslations, omissions, and factual errors but does not polish style or rewrite for fluency. This level is suitable for internal use, information gathering, and content where speed and cost outweigh stylistic perfection. Full post-editing aims to achieve the same quality as human translation. The editor addresses accuracy, fluency, style, terminology, and cultural appropriateness, producing output indistinguishable from direct human translation. This level is appropriate for any content that will be published or distributed externally.

The effectiveness of post-editing depends heavily on the quality of the machine translation baseline. Well-performing MT engines on suitable content types can produce output that requires relatively light editing, making the process significantly more efficient than translation from scratch. Poorly performing MT on unsuitable content types can produce output that is harder to edit than to retranslate.

LEXIGO offers post-editing services calibrated to the content type and quality requirements, advising clients on when PEMT offers genuine efficiency gains and when full human translation is the more effective approach.

Post-editing enables organisations to benefit from machine translation efficiency while maintaining quality through human oversight. For high-volume, repetitive content types where MT performs well, post-editing can reduce costs by 30-50% compared to full human translation while delivering comparable quality.

However, post-editing is not universally appropriate. Applying it to content where MT performs poorly wastes editor time and can produce inferior results. Understanding the right use cases for post-editing versus human translation is key to optimising both quality and cost.