Quality in multilingual crisis communication: a reception-oriented perspective

This position paper examines how "quality" should be understood in multilingual crisis translation, arguing that textual accuracy alone is an insufficient measure of a good translation in high-stakes emergency contexts. Reviewing decades of translation theory, the paper traces a conceptual shift away from source-text fidelity and toward evaluating translations by their real-world impact on readers. It contrasts two broad approaches to assessing quality: an output-focused paradigm that relies on error analysis and metrics like BLEU or MQM, and a reception-oriented paradigm that centers the lived experience of message recipients. 

The author proposes four interconnected dimensions for judging translation quality in crisis settings: timeliness (can people access the message when they need it), understandability (can they make sense of it), trustworthiness (do they find it credible), and actionability (does it prompt the right behavior). Furthermore, the paper draws on a series of studies conducted by a Melbourne-based research team between 2020 and 2025, spanning the COVID-19 pandemic through recent work on generative AI. Early pandemic research with CALD community leaders revealed that official translations were often delayed, linguistically inconsistent, or culturally misaligned, prompting some communities to produce their own materials via bilingual leaders and social media. A later project on machine-assisted emergency messaging found that raw machine translation, while fast, produced dangerously misleading instructions, and that pre-editing or post-editing by humans significantly improved comprehension and perceived trustworthiness.