Linguistic Context and the Social Meaning of Voice Quality Variation

Abstract

This dissertation investigates the linguistic and social constraints on the occurrence of creaky voice quality (creak) in Beijing Mandarin (BM), as well as the effect of linguistic and prosodic context on creak's social meanings for Mandarin listeners. It is a two-phase study, composed of 1) a production study of the distribution of creak in the naturalistic speech of speakers of BM and 2) an experimental study exploring listeners' evaluations of creak in different linguistic environments. Not only does this dissertation expand on our knowledge of the patterning of suprasegmental variation, as well as of sociolinguistic variation in Chinese more broadly, it also opens up inquiry into the interaction between linguistic factors and the social meanings of linguistic variability.

I collected and examined a corpus of sociolinguistic interviews with 15 Beijing area residents, 18-20 years old. Using acoustic and auditory methods, I quantified the distribution of creak according to a number of factors, both social (sex, region) and linguistic (tone, position in intonational phrase). Creak is more common on low tones and toneless syllables (so-called 'neutral tone'), as well as phrase-finally. Meanwhile, acoustic measures also indicate an interaction between tone and prosodic environment: toneless syllables in phrase-final environments are less creaky than other tones in the same prosodic positions.

I also performed a "matched guise" experiment, soliciting native Mandarin listeners' reactions to short stretches of speech with and without creak, which were otherwise identical. These audio stimuli were designed to investigate reactions to creak in different tonal (low tone vs. high falling tone) and prosodic environments (phrase-medial vs. phrase-final). The results indicate that creak in IP-final environments indexes increased enthusiasm and interest compared to baseline. IP-medial creak, meanwhile, indexes decreased enthusiasm and interest compared to baseline. This result provides intriguing support for the hypothesis that linguistic environment plays a role in determining the social and affective meanings of linguistic variation.