An experience-based sampling approach to examining prior experience in adaptive speech perception

Prior experience is a powerful mechanism for many types of learning. In speech, listeners draw on their prior experience to perceive sounds, predict words, and determine the likely sentence structure intended by the speaker. However, the impact of listeners’ long-term linguistic experiences understanding novel talkers’ speech and their accents remains unexplored. Researchers have attempted to quantify and assess prior experience, but their efforts are tainted by shortcomings related to recall and self-reporting biases. The goal of this thesis is to develop a means of assessing prior experience, as it relates to accents, by combining multiple different metrics to provide a holistic view on prior experience. Specifically, this paper proposes and validates a new mobile-based experience-based sampling survey of nonnative accent perception (SNAP). Combined with a perceptual experiment, the SNAP survey provides an individualized predictor of listeners’ accent-comprehension capabilities. The paper discusses how multiple research instruments including SNAP can provide sufficient evidence to be able to conclude that prior experience does play a role in facilitating accent adaptation.