Gradient Effects of Reading Ability on Talker-Specific Perceptual Learning

Abstract The acoustic signal of speech is a complex signal that simultaneously cues the linguistic content of a talker’s message and the identity of the talker. Traditional models of speech perception proposed that the properties of the acoustic signal that cue linguistic meaning and talker identify were distinct. However, a growing body of evidence challenges this traditional view, indicating that linguistic and talker-specific phonetic information are tightly linked in language comprehension. One of these benefits is increased intelligibility of the spoken message when listening to a familiar talker compared to an unfamiliar talker. Just as experience with a talker’s voice influence linguistic processing, it has been shown that this link is bi-directional such that linguistic experience influence voice recognition ability, indicating the recruitment of phonological knowledge to perform talker identification tasks. Consistent with this account, recent evidence indicates that adults with dyslexia demonstrate reduced talker identification abilities when compared to a typically developing population. Other work suggests that there may be a gradient influence of phonological ability on talker recognition. Here we test the hypothesis that reading ability will demonstrate a gradient effect on the ability to not only learn talkers’ voices, but also the ability to incorporate talker-specific phonetic detail into lexical representations across the range abilities in the unimpaired population. Monolingual English readers were assigned to either the average (n = 15) or advanced (n = 15) reading group based on a median split of performance on a standardized assessment battery for reading sub-skills and comprehension. All readers participated in a five-session study, where at session 1 each completed a word transcription pre-test, sessions 2 – 4 consisted of talker identification training for six talkers, and session 5 consistent of a post-test identical to the transcription test used in session 1. The results indicated that compared to the average readers, the advanced readers (1) showed higher rate of learning for talker identification during training and (2) showed greater degree of improvement on speech perception at post-test. Correlational analysis showed that the degree of improvement at post-test was significantly correlated with performance on the standardized reading assessment battery. These results extend earlier findings to include a gradient effect of reading ability on talker-specific perceptual learning