Presentation
Search Abstracts | Symposia | Slide Sessions | Poster Sessions | Lightning Talks
Does speech adaptation generalize across adverse listening conditions?
Poster D78 in Poster Session D, Wednesday, October 25, 4:45 - 6:30 pm CEST, Espace Vieux-Port
This poster is part of the Sandbox Series.
Yuting Gu1, Xin Xie1, Chigusa Kurumada2; 1University of California, Irvine, 2University of Rochester
[Rationale] Human speech perception exhibits remarkable adaptability in challenging conditions such as speech-in-noise or nonnative accents. Previous research has shed light on the neural basis of the flexibility and adaptive plasticity (Chandrasekaran et al., 2015; Guediche et al., 2014; Adank, 2020). Yet, a crucial question remains unanswered—Is there a shared mechanism that supports adaptive speech perception across different adverse conditions? Nonnative-accent adaptation has been considered exposure-specific: adaptation to speech-in-noise does not transfer to adaptation to systematic shifts of pronunciations common in nonnative accents (Xie et al., 2018). However, studies have also identified individual difference variables (e.g., receptive vocabulary size) that predict adaptation to multiple adverse listening conditions (McLaughlin et al., 2018). This may suggest that there is a common source of resilience to different types of challenging listening conditions. The present study provides a direct test of whether exposure to nonnative-accented speech improves subsequent recognition of speech-in-noise and vice versa. [Task Design and Predictions] Building upon previous work (Clarke Garrett, 2004; Xie et al., 2018), we employ a cross-modal matching task to assess speech recognition abilities. 120 participants (planned) judge whether a visual target (e.g., tents) matches the final word of an auditory sentence (e.g., “He pointed at the cents”). After task familiarization, participants are randomly assigned to one of three exposure conditions (Control, Noise, Accent, n = 40 each) presented over three blocks, followed by two test blocks (Noise and Accent, with order randomized across participants). Finally, a baseline block is administered to account for individual differences in response speed. Changes in baseline-corrected response times (RT) and error rates served as measures of adaptation. Data analyses use generalized mixed-effects regression models to distinguish between two competing hypotheses. If speech adaptation is exposure-specific, then listeners in the Noise and Accent conditions will exhibit better performance only for the corresponding type of speech during the test, compared to Control listeners. In contrast, if adaptation transfers across exposure conditions, listeners in either Noise or Accent conditions will outperform Controls in both test conditions. [Plans for Follow-up Studies] If results support exposure-specific adaptation, it would suggest that different mechanisms are engaged depending on the specific type of variability. In such a case, we will further delve into the source of such specificity by evaluating the effects of prior experiences (e.g., environmental exposure to Spanish-accented English) on accent adaptation. If the specific acoustic-phonetic properties of novel speech input are important for adaptation, then long-term experiences with a given condition may affect adaptation to that condition (e.g., Spanish-accent) without influencing another (e.g., Mandarin-accent). Alternatively, if transfer is observed, it would raise new questions about the cognitive, attentional, and neural mechanisms recruited for all adverse listening conditions. In such a scenario, we plan to conduct follow-up experiments involving populations with diverse attentional and cognitive abilities (e.g., young vs. older adults). In summary, the results of the current study will provide critical insights into individual differences in the ability to cope with challenging listening conditions.
Topic Areas: Speech Perception,