Search Abstracts | Symposia | Slide Sessions | Poster Sessions
How does perceptual learning of speech shape neural encoding? An EEG investigation
Poster Session C, Thursday, October 1, 10:45 am - 12:45 pm, Wangari Maathai
This poster is part of the Sandbox Series.
Sahil Luthra1, Drew McLaughlin2, Barbara G. Shinn-Cunningham3, Joseph Toscano2; 1Stony Brook University, 2Villanova University, 3Carnegie Mellon University
To comprehend speech, a listener’s perceptual system must accommodate each talker’s individual accent. Research has robustly established that listeners can adapt to phonetic variation, but it is unclear whether such phonetic recalibration induces perceptual changes or whether adaptation relies on decision-level mechanisms. At least some past results hint that different mechanisms may underlie recalibration of different types of speech sounds. Adaptation to fricative sounds (e.g., /f/ and /s/) has been shown to be talker-specific, suggesting a perceptual locus. In contrast, adaptation to plosive consonants (e.g., /d/ and /t/) can generalize across talkers, suggesting a post-perceptual, decision-level adjustment. This difference may arise because fricatives’ acoustics tend to vary across talkers, whereas acoustics of plosives are relatively consistent within a language. Here, we leverage two event-related potential (ERP) components, the auditory N100 and P300, to assess how phonetic recalibration influences speech perception. Past research indicates that the N100 reflects fine-grained acoustic detail in the speech signal, whereas the P300 indexes phonetic categorization of the sounds (Toscano et al., 2010). Participants are assigned to one of four biasing groups (/d/, /t/, /s/, /f/; target n = 20 per group). During an initial exposure phase, listeners hear ambiguous speech sounds (#) in lexical frames that bias their interpretation of the sound (e.g., the /d/-bias group hears /d/-/t/ blends in frames like croco#ile); they also hear clear productions of the contrastive speech sound, such as cemetery with a clear /t/. During subsequent test blocks, listeners hear stimuli from two continua (dime-time, sign-fine) and monitor for an oddball target word (e.g., dime). Past work indicates that listeners should accommodate the talker’s idiosyncratic speaking style, categorizing ambiguous stimuli (e.g., #ime) in line with previous exposure. Of interest is whether the N100 and/or P300 to ambiguous test stimuli will be modulated by listener knowledge of the talker’s typical speaking style (e.g., whether #ime will elicit a /d/-like ERP). We predict that results will differ across speech sounds: For fricative sounds that vary across talkers, adaptation should drive changes in the N100, reflecting perceptual adjustment, but for plosive sounds that are relatively stable across talkers, only the P300 should be affected, reflecting decision-level changes. By capitalizing on established electrophysiological indices of perceptual encoding, this work tests whether accent adaptation involves perceptual or decision-level changes, as well as whether such changes depend on the specific speech sounds being accommodated and how likely, a priori, those sounds are to vary across talkers.
Topic Areas: Speech Perception,