Cumulative input sensitivity predicts both attenuation and stability of lexically guided perceptual learning.

Listeners can use lexical information to accommodate ambiguity in speech input. Some evidence suggests that lexically guided perceptual learning persists over time. However, other evidence suggests that lexically guided perceptual learning attenuates throughout the test session, consistent with learning that occurs given exposure to the test stimuli. The current study aimed to determine whether this apparent discrepancy could be resolved when viewed through the lens of the belief-updating theory of speech adaptation, which posits continuous sensitivity to statistical cues in speech input. During exposure, listeners (n = 160) heard speech sounds ambiguous between /ʃ/ and /s/ in a lexically-biasing context. At test, listeners categorized tokens from an ashi-asi continuum. Critically, the duration of the initial test phase was manipulated between subjects to be either brief or long. Approximately 24 hours later, all listeners completed a second test phase. Though evidence of lexically guided perceptual learning was present at all timepoints for both biasing groups, attenuation of learning was observed given continued testing. Strikingly, the position of the 24-hour time delay relative to duration of the initial test had no significant effect on learning. These results and a re-analysis of previous work align a seemingly mixed literature under a unifying account characterizing lexically guided perceptual learning as an iterative, continuous sensitivity to dynamic changes in speech input that is insensitive to time alone in the initial 24-hour learning window.