Open-set identification of non-native talkers' language backgrounds

Listeners are skilled at detecting native talkers of a language, but can they identify specific non-native language backgrounds? Open-set identification was used to explore this question. Eighty monolingual American English-speaking listeners labeled the language backgrounds of 30 talkers with 5 different native languages (L1s) on the basis of syllableand word-length samples of English. As expected, listeners often identified L1 American English talkers correctly, despite hearing extremely short auditory stimuli. While listeners were sometimes unwilling to assign labels to L1 Korean, L1 Spanish, and to some extent L1 Mandarin talkers, L1 Hindi talkers were labeled frequently, and often the labels were correct. Responses revealed that listeners perceived many more language backgrounds than were actually represented by the talkers, a result which cannot be conveyed by the closed-set identification tasks commonly used. Impacts of listeners’ perceptions of talker language background on cross-cultural communication are discussed.