• Version
• Download 3
• File Size 695 kb
• File Count 1
• Create Date 15 September 2026
• Last Updated 15 September 2026

Masked speech recognition among multinational talker and listener cohorts

By Rebecca E Bieber, Matthew J Makashay, Benjamin M Sheffield, & Douglas S Brungart

Large multinational operations such as those conducted by the North Atlantic Treaty Organization (NATO) require participants who come from a variety of language backgrounds to communicate with team members in a common language that they speak with varying levels of proficiency. The present study used operators at the NATO Trident Jupiter Exercise to evaluate the intelligibility of English sentences in noise as a function of the NATO Standard Agreement 6001 Language Proficiency Levels of the talkers and listeners. Design. Stimuli included American English Matrix Test sentences produced by 20 NATO personnel, including 6 native and 14 non-native speakers of English, which were mixed in either multitalker babble or speech-shaped noise. Recognition accuracy and response times were recorded and were analyzed using mixed-effects regression modeling. Study Sample. One hundred and seventeen personnel in attendance at a NATO joint exercise completed the study, including 28 native speakers of English and 89 native speakers of 11 other languages. The limitedly available demographic data were included in the statistical analyses to control for these variables to the extent possible. Results. Talker and listener proficiency interacted with masker to predict recognition accuracy, with more prominent effects observed in babble than in noise. Sentences presented in babble also elicited longer response times than those in noise, but only when recognition accuracy was low. Conclusion. Non-native talker and listener proficiency impacts communication dynamically with changes to masker type. Results are discussed in the context of the NATO standard language profile descriptors.