Morpheme-like prosodic functions: evidence from acoustic analysis and computational modeling

In this paper we address the long-standing issue of how prosodic patterns are linked to meanings. We explore the idea that prosodically realized communicative functions, such as focus and sentence modality, are analogous to lexical morphemes, the smallest sound units that carry meaning. We considered evidence of a four-way similarity between lexical morphemes and prosodic functions. First, similar to lexical morphemes, each prosodic function consists of multiple phonetic components. Second, like segmental phonemes, individual prosodic components are meaningless themselves, but act jointly to mark both intra- and inter-functional contrasts. Third, like lexical morphemes, prosodic functions have allomorph-like variants whose occurrences are conditioned by factors like location in sentence and interaction with other prosodic functions. Finally, similar to lexical morphemes, prosodic functions are language-specific and the specificity has likely historical sources. We examined the evidence by a) reviewing existing literature on speech prosody, b) conducting two new experiments on the production of focus and sentence modality in General American English and Mandarin Chinese, and c) training an articulatory-functional model on focus, modality, tone and stress in English and Mandarin, and synthesizing fully-detailed F0 contours with the learned functional targets. Overall, all the evidence examined is in support of the hypothesis. In particular, the consistency between the target parameters obtained from acoustic analysis and computational modeling, and the close match between functionally synthesized and naturally produced F0 contours demonstrate the plausibility of establishing a clear link between function-specific categorical representations and fine-detailed surface prosody