Inferring causes during speech perception

Abstract

One of the central challenges in speech perception is the lack of invariance: talkers differ in how they map words onto the speech signal. Previous work has shown that one mechanism by which listeners overcome this variability is adaptation. However, talkers differ in how they pronounce words for a number of reasons, ranging from more permanent, characteristic factors such as having a foreign accent, to more temporary, incidental factors, such as speaking with a pen in the mouth. One challenge for listeners is that the true cause underlying atypical pronunciations is never directly known, and instead must be inferred from (often causally ambiguous) evidence. In three experiments, we investigate whether these inferences underlie speech perception, and how the speech perception system deals with uncertainty about competing causes for atypical pronunciations. We find that adaptation to atypical pronunciations is affected by whether the atypical pronunciations are seen as characteristic or incidental. Furthermore, we find that listeners are able to maintain information about previous causally ambiguous pronunciations that they experience, and use this previously experienced evidence to drive their adaptation after additional evidence has disambiguated the cause. Our findings revise previous proposals that causally ambiguous evidence is ignored during speech adaptation.

Now You Hear Me, Later You Don’t: The Immediacy of Linguistic Computation and the Representation of Speech

“…The adaptation additionally began to fade over time: The magnitude of adaptation (in conditions in which it was present) was larger in the first half of testing than in the second half of testing (see the Supplemental Material for additional visualizations). This reduction in the adaptation effect over time is in line with previous findings (Liu & Jaeger, 2018) and is perhaps not surprising given the remarkably limited sample during exposure (only 22 edited tokens out of 142 total) and the comparatively long testing phase.…”

“…The independent variables were experimental condition: shifted phone (shifted /t/ vs. shifted /d/, sum-coded), timing (text before and text after, sum-coded), and their interaction. VOT (continuous variable, scaled and centered) and test half (first vs. second, sum-coded) were included as main effects and interaction terms with experimental conditions to test whether the effects of interest changed over the course of the test phase; this followed previous observations (e.g., Liu & Jaeger, 2018) that perceptual adaptations may be unlearned, to some degree, throughout testing. We attempted to include block number (1 to 9, centered) as a factor, but no models with this factor converged, so we used test half (first four blocks vs. last five blocks) instead.…”

Maintaining information about speech input during accent adaptation

“…Previous studies have demonstrated that despite this variability, web-based experiments on speech perception are feasible [ 45 – 47 ]. This includes studies on accent adaptation [ 48 – 50 ]. To reduce the between-speaker variability, we required participants to use either in-ear or over-ear headphones when taking the experiment.…”

Online pragmatic interpretations of scalar adjectives are affected by perceived speaker reliability

“…When the idiosyncrasy is likely attributable to the speaker, listeners would develop a speaker-dependent expectation for the subsequent input. Alternatively, when an unexpected input is likely caused by an incidental factor, it would not be expected to recur once the cause is removed [ 79 , 80 ]. Contextually modulated interpretations of disfluencies (e.g., ‘Click on [pause] thee uh…”) have also been discussed under this model [ 25 , 26 , 81 ].…”