atranslation

atranslation is a neologism used in some discussions of translation practice to denote a hybrid approach that blends machine-generated text with human curation. The term is not widely standardized and its precise meaning can vary by context. In general, atranslation aims to combine the speed and scale of automation with the accuracy and nuance of human editing.

Origins and usage: The term has appeared in online forums, practitioner blogs, and some academic writings in the 2010s and 2020s. It is not part of formal canon and lacks a single canonical definition. Common interpretations include: (a) post-edited machine translation with emphasis on maintaining style and voice; (b) an autonomous AI draft that is subsequently refined by human translators for terminology, register, and cultural adaptation.

Practice and workflow: Typical atran workflows begin with machine translation to generate a draft, followed by human post-editing, glossing of cultural references, and possible authorial or brand-level adaptation. Some proponents emphasize iteration, domain-specific glossaries, and annotated translations where editor notes capture decisions for future reuse.

Criticism and status: Because there is no agreed standard, the term can blur lines between human translation, post-editing, and fully automated output. Critics warn of inconsistent quality, terminology drift, and over-reliance on automation.

See also: Translation studies; machine translation; post-editing; localization; adaptation.