Return to search

Speech perception of Canadian English sibilants: Processing of acoustic information or underlying (articulatory) vocal tract configurations?

Acoustic and articulatory speech perception theories are proposed to explain how listeners map acoustic signals to phonetic categories. Different from acoustic theories, articulatory theories hypothesize that listeners would recover articulatory information during the mapping. To test this hypothesis, we altered the acoustic information of the Canadian English sibilants /s/ and /ʃ/ while keeping their articulatory information to signal places of articulation of the original sibilants. The manipulated sibilants were articulatorily cued as the original sibilants, but acoustically cued as the alternative sibilants (/s/ as /ʃ/ and /ʃ/ as /s/). We conducted an identification task to examine whether altering acoustic information would switch our Canadian English listeners’ identification. The listeners identified acoustically /s/-like /ʃ/ completely as the alternative sibilant /s/, but the acoustically /ʃ/-like /s/ as 60% the alternative sibilant /ʃ/ and 40% the original sibilant /s/. There was a categorical switch in the /ʃ/ stimuli but not the /s/ stimuli. This asymmetry of identification between two sibilants can be explained by two accounts: an acoustic plus articulatory account would be that the listeners relied more articulatory information only when identifying /s/ but not /ʃ/; and a purely acoustic account would be that the asymmetry was only a result of still existing small acoustic differences. While the acoustic plus articulatory account cannot explain why articulatory information only influenced the /s/ identification of the /s/ stimuli even after adding a set of assumptions, the purely acoustic account allows us to explain our results consistently without additional assumptions. Although our results cannot be used as evidence to reject the possibility that listeners will recover articulatory information, the results do suggest that even if we assume that articulatory information is recovered, acoustic information is more dominant than articulatory information in the identification process, at least for Canadian English /s/ and /ʃ/. / Thesis / Master of Science (MSc)

Identiferoai:union.ndltd.org:mcmaster.ca/oai:macsphere.mcmaster.ca:11375/24112
Date January 2018
CreatorsLuk, San-Hei Kenny
ContributorsPape, Daniel, Service, Elisabet, Cognitive Science of Language
Source SetsMcMaster University
LanguageEnglish
Detected LanguageEnglish
TypeThesis

Page generated in 0.0018 seconds