The role of gradient speech categorization in accent perception

A key question in speech perception research is how listeners categorize highly varied signals into discrete units. The long-held assumption was that listeners discard variation and focus on the category itself (categorical perception, Liberman et al., 1957). However, listeners often encounter highly varied speech signals (e.g., processing unfamiliar accents). In such cases, discarding variation can be detrimental for listeners and can lead to more processing difficulty. Here, in an ongoing online study (n = 80), we measure English-speaking adult listeners’ speech categorization patterns with a continuous measure (the Visual Analogue Scaling Task: Kutlu et al., 2022; Apfelbaum et al., 2022) to see whether they are more susceptible to phonetic variation in their language environments. We index adults’ language exposure with an extensive social network survey which quantifies the extent to which they are exposed to varied accents on a regular basis and what those accents are. We then asked participants to recognize spoken sentences from a diverse set of unfamiliar accents. We predict that adults who are more gradient in their speech categorization are more accurate in transcribing sentences with a diverse set of accents.