Representation and processing of pitch variation in tonal languages
This project examines how speakers store and process regular pitch variation.
In connected speech, spoken words may vary considerably from their forms in isolation. Much of the variation is regular. There have been heated debates on whether and how speakers store regular variants in their long-term memory, and how they process them in speech production and comprehension. Most psycholinguistic work, however, focuses on how listeners cope with segmental variation.
The current project instead proposes to examine how speakers store and process regular pitch variation. Our empirical focus is on Chinese dialects, where pitch changes (tones) are used to distinguish word meanings in isolation, but tonal realization of words in phrases is determined by domain formation and pitch changes within the domain, together known as tone sandhi. Crucially, both how a tonal domain is formed and how sandhi variants are realized within the domain are influenced by a complex mixture of pragmatic, syntactic, and phonological factors. The pitch variation in these dialects thus manifests at all levels of linguistic organization.
We hypothesize that the way pitch variation is stored in the mental lexicon and processed during speech production varies according to the types of tone sandhi changes and the constituent structures of sandhi domains. We will take an interdisciplinary approach to test this hypothesis. Cross-dialect speech corpora will be compiled to obtain a comprehensive understanding of how pitch variation is conditioned in linguistic contexts. Acoustic and perceptual experiments will be conducted to understand the categorical vs. gradient nature of pitch variation. Both voice onset latencies and speakers’ eye movements during speech will be monitored to tap into the representation as well as the processing of pitch variation.