Acceptable Noise Levels and Speech Recognition Measures in Bilingual Spanish-English Listeners

Purpose.;The current study had two main purposes. The first purpose was to investigate how acceptable noise levels (ANLs) might compare between Spanish-English bilinguals and English monolinguals. The second purpose was to investigate how ANL might be related to speech recognition in these two groups of listeners.;Methods.;Forty normal-hearing adult listeners participated in this study, 10 monolingual English listeners (Group E) and 30 Spanish-English bilingual listeners (Group S). ANL is obtained using an English passage in English 12-talker babble and a Spanish passage in Spanish 12-talker babble. For speech recognition, listeners were presented with high-predictability sentences from the Speech-Perception-in-Noise (SPIN) test at four signal-to-noise ratios: -3, 0, +3, and +6 dB. A psychometric function was then computed to establish the 50%-correct performance level, theta.;Results.;Comparable ANL was found across listener groups and test languages, suggesting that language does not play a significant role in determining one's ANL. By contrast, Group S performed significantly more poorly on the SPIN test across all SNRs than Group E, as indicated by a significantly higher theta in the former group of listeners. ANL was not significantly correlated to theta. Almost all Group E listeners yielded an ANL close to the high end of their psychometric function (near-perfect performance), whereas Group S listeners' ANL corresponded to a wide range of SNRs on their psychometric function. Last, Group S listeners' language background was not correlated to their ANL, but significantly correlated to their theta, suggesting that ANL was not affected by one's language experience or language proficiency.;Conclusions.;The current study revealed that ANL and conventional speech recognition tests differed in more than one way. First, ANL was comparable regardless of test language or listeners' language background, but compromised speech recognition was found in bilinguals. Second, ANL was not equivalent or correlated to the 50%-correct level on the SPIN. Third, language background variables accounted for the theta on the SPIN, but not for ANL. Based on these findings, the current study supports the notion that ANL is a language-independent measure and assesses a unique aspect of speech perception.;Keywords: Bilingual, Acceptable Noise Level, Speech Recognition, Background Noise, Psychometric Function, Language Background.