Abstract
This study assessed the perceptual assimilation model (PAM) predictions about assimilation types and discrimination performance through the lens of assimilation overlap in Quebec French listeners' perception of nine non-native Korean stop consonants. The consonants varied in voicing (fortis, lenis, and aspirated) and place of articulation (labials, coronals, and velars). In the identification experiment, the Korean three-way voicing contrasts were found to undergo an assimilation overlap to correspond to the French two-way equivalent contrasts across places of articulation. In the discrimination experiment, assimilation overlap tended to hinder detection of non-native speech distinctions, which is in line with the PAM-based suggestion.
Highlights
Discrimination of non-native speech contrasts should be more difficult when the two phonemes are perceived as highly similar
This study aimed to explore the applicability of the perceptual assimilation model (PAM) framework to three-way stop voicing contrast perception and examine the role of assimilation overlap in discrimination performance
Our results showed that discrimination generally tends to be modulated as a function of assimilation overlap
Summary
Discrimination of non-native speech contrasts should be more difficult when the two phonemes are perceived as highly similar. Nam (2018) found that Korean listeners identified English /tS/, /dZ/, and /Z/ as Korean /tS/, whereas they mapped English /S/ to Korean /s/. They were appreciably more accurate at discriminating the corresponding English affricate-fricative contrasts when the two non-native phones did not have identification overlap (/tSa-Sa/, 92% correct) than when they did (/dZa-Za/, 61% correct). This observation points to the relationship between non-native contrasts’ identification and their discrimination, the role of identification overlap. The PAM assigns six assimilation types to non-native contrasts: (1) For the two-category (TC) type, the two contrasting non-native phones are assimilated to different native phonemes; (2) for the single-category (SC) type, both non-native phones are assimilated to the same native phoneme with the same level of perceived similarity to that native phoneme; (3) for the category-goodness (CG) type, both phones are assimilated to the same native phoneme, but one phone is perceived as a better instance of the native category than the other; (4) for the uncategorized-categorized (UC) type, one non-native phone matches a native phoneme but the other is not assimilated to any single native phoneme above a categorization (or statistical) threshold; (5) for the uncategorized-uncategorized (UU) type, neither non-native phone is assimilated to any single native phoneme; and (6) for the non-assimilable (NA) type, neither non-native phone is perceived as speech
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have