Answers are grounded in this paper and cite their sources.
Evaluating AI-mediated simultaneous interpreting requires distinguishing information preservation from listener-perceived quality. This mixed-methods study introduces the Cultural-Cognitive Focus Flag (C-Flag), grounded in Cultural Linguistics, to annotate selected cultural requirements relative to a reference listener before output assessment. The analysis included 1,160 outputs for 232 assertions from four professional interpreters and one commercial system across two Chinese-to-English speeches, alongside six native English-speaking users’ transcript ratings and interviews. Human outputs received higher mean ratings, although comparative information loss varied across speeches. Adjusted human–machine rating differences in idiomaticity and usability were larger for flagged assertions, whereas flagging’s association with information loss did not differ significantly between output types. Case and interview analyses indicated that favorable listener evaluations did not necessarily correspond to information preservation. Through C-Flag, the study connects Cultural Linguistics with human–AI interpreting evaluation and informs design directions for selective access to culturally relevant explanations.
No takes yet. Share an insight, caveat, or question.
Tang et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: