Effects of contextual and noncontextual nasalization on perceived vowel height

Abstract

Listeners' ability to separate the spectral effects of nasal coupling from those of tongue position was examined by comparing labeling responses to three continua: oral /bɛd‐baed/; contextually‐nasalized /bɛnd‐baend/; and noncontextually nasalized /bɛd‐baed/. The continua were generated by articulatory synthesis and were identical to one another in tongue position. Results showed the same percent /ɛ/ responses to the oral and contextually nasalized vowels, but significantly fewer /ɛ/ responses to the noncontextually nasalized vowels. That is, the noncontextually nasalized vowels were perceived as lower than the oral and contextually nasalized vowels. Perceptual lowering of the nasalized vowels relative to the oral vowels is consistent with upward shifts in F1 due to nasal coupling. However, the absence of a lowering effect in the contextual nasal condition needs to be explained. We propose that subjects' vowel judgments are influenced by spectral effects of nasal coupling when no conditioning environment is present. When a nasal consonant is present, these spectral effects are perceptually analyzed as due to the consonant and do not influence vowel height judgments. [Work supported by NIH grants HD‐01994 and F32‐NS‐07196, NSF grant BNS‐81‐11470.]