Abstract
Much research on machine translation post-editing (MTPE) has focused on productivity, cognitive effort and error reduction. Comparatively less attention has been paid to the culturally and pragmatically consequential textual interventions that remain once a machine-generated draft is already in place, particularly in Arabic-English narrative translation. This article examines 50 Arabic-English post-edited narrative texts produced by undergraduate translators and identifies 372 culturally salient interventions. In the available task instructions, cultural adaptation was not stated as an explicit requirement, so the analysis focuses on interventions that emerged in the final products rather than on instructed cultural rewriting. Using an exploratory qualitative corpus design with descriptive quantitative support, the study codes these interventions into five recurring strategy types: lexical substitution (41%), register and pragmatic adjustment (24%), explicitation and glossing (18%), stylistic reframing (10%), and cultural preservation or hybridization (7%). The Cultural Post-Editing Strategy Framework (CPESF) is proposed as a provisional descriptive-analytical framework for naming these visible response types, not as a validated model of translator cognition or a prescriptive hierarchy. The findings show that culturally consequential post-editing in this corpus extends beyond literal error repair, while remaining uneven across intervention types. The study therefore treats “cultural agency” cautiously as an analytical description of visible textual departures from machine output, not as direct evidence of translators’ intentions or cognitive processes.