UCLA

This dissertation investigates the prominence relationship between lexical stress and tonal rhythm across multiple languages and tests whether cross-linguistic differences in tonal rhythm are perceptible to listeners. Specifically, the experiments in this dissertation test Jun’s (2014:537) hypothesized inverse correlation between the strength of lexical stress and the strength of tonal rhythm or macro-rhythm (MacR). Lexical stress strength is the difference in duration and intensity between stressed and unstressed syllables, and MacR strength is the presence, regularity, and frequency of word-sized F0 alternations within an utterance. This hypothesis was tested by comparing lexical stress strength with MacR strength in English, Uyghur, and Bengali. The predicted ranking of lexical stress strength is English > Uyghur > Bengali, while the predicted ranking of MacR strength is Bengali > Uyghur > English. The lexical stress production experiment compared the duration ratios of stress and unstressed vowels of disyllabic nonce words produced by speakers of each language. The results found that English had the strongest realization of lexical stress, but Uyghur had the weakest stress instead of Bengali (English > Bengali > Uyghur), possibly due to the use of nonce words and Bengali speakers’ experience with English. However, the results of the MacR Frequency Index (Jun, 2014:538), which calculates peak-per-Prosodic Word ratio, confirm the predicted strength ranking (Bengali > Uyghur > English). In the MacR perception experiment, participants rated the melodicity of utterances that were phonetically manipulated in two conditions: low-pass filtered (Filtered) and hummed (F0-only) stimuli. The results found that Bengali utterances were rated significantly more melodic than Uyghur in both conditions, and more melodic than English in the Filtered condition, but Uyghur and English were not rated significantly different in either condition. Overall, the results support the predicted inverse relationship between lexical stress strength and MacR strength, and they also demonstrate that listeners can perceive differences in MacR strength in the predicted direction. This study is the first of its kind to directly test the hypothesized inverse relationship and the perceived MacR strength across languages. The results contribute to our understanding of MacR, its effect on speech rhythm perception, and prosodic typology.