Perception of phonetic details in multiple cues: An individual approach
Listed in
This article is not in any list yet, why not save it to one of your lists.Abstract
There is an unresolved debate as to whether speech perception is essentially categorical or gradient. How phonetic details are processed is a crucial argument in this debate. Studies have long revealed individual differences in the perception of phonetic details. While the traditional position considers that a heightened sensitivity to phonetic details is detrimental to the formation and processing of phonological categories, more recent views suggest that individuals who are sensitive to phonetic details are able to pick up all available cues in phonological categorisation. To investigate the latter issue, the present study examines individual differences in processing primary and secondary cues to the French voicing contrast, by further breaking down listeners' behavioural performances into pre-categorisation and categorisation levels and comparing the strengths of correlations among all performances. Results show strong positive correlations between sensitivities to VOT and F0 when the cue is redundant or when the task concerns within-category discrimination. Correlations between sensitivities to cues become weaker when the task involves category mapping, and then become absent when a categorisation decision needs to be made. It suggests that the causal link between low-level and high-level sensitivities to cues is probably weak or blurred by other mechanisms. In line with recent neurophysiological evidence, these findings suggest that individuals not only differ in low-level perception in that some listeners detect fine-grained details more than others, but also in their retention of phonetic details from low-level to high-level processing. The paper will be concluded with a discussion on the role of individual variations in the perception of multiple cues on multiple levels and their implications for the definition of categoricalness in speech perception models.