Understanding and improving machine translations for emergency communications

This study investigates the use of machine translation (MT) in urgent "behavior-change communication," aiming to improve emergency messaging for multicultural communities. It stands out as the first research to evaluate not just raw machine translation output, but also two distinct editing processes: post-editing (where a human corrects the machine-translated result) and pre-editing (where a human corrects the source text before it is translated automatically). Through this comparative approach, the researchers generate concrete data on the kinds of problems that can arise when relying on machine translation for emergency communications, along with insights into what is needed to address these issues.

Based on these findings, the study offers practical, actionable suggestions for reducing or avoiding machine translation problems and risks. The ultimate goal is to help emergency communicators develop messaging strategies that are more efficient, effective, and culturally appropriate — ensuring that critical safety information reaches multicultural and multilingual communities accurately during urgent situations.