Speaker Normalization in Speech Perception

Abstract

Talkers differ from each other in a great many ways. Some of the difference is in the choice of linguistic variants for particular words, as immortalized in the song by George and Ira Gershwin “Let’s call the whole thing off”.You say either [iðɚ] and I say either [aᴵðɚ],You say neither [niðɚ] and I say neither [naᴵðɚ]Either[ iðɚ], either [aᴵðɚ] Neither [niðɚ], neither [naᴵðɚ]Let's call the whole thing off.You like potato and I like potahtoYou like tomato and I like tomahtoPotato, potahto, Tomato, tomahto.Let's call the whole thing offListeners have experienced different pronunciations of words, and many of the variants that we know are tinged with social or personal nuance. This “multiple-listing” notion, that listeners store more than one variant of each word in memory is the dominant hypothesis, among sociolinguists regarding the cognitive representation of social phonetic variation (Thomas, 2011), and has been proposed as a way to account for the listeners’s ability to ‘normalize’ for talker differences in speech perception (Johnson, 1997).