Working title: The Distribution of Disfluencies in Spontaneous Speech: Empirical Observations and Theoretical Implications
Supervisor: Mark Liberman
Proposal Committee: Jianjing Kuang (Chair), Kathryn Schuler, Gareth Roberts
Time & Location: 10-11:30 am May 9th, Linguistics Department library (Room 301C, 3401-C Walnut St.
Abstract: Speech disfluencies, or hesitation phenomena such as silent or filled pauses, repetitions, restarts and repairs, are prevalent in spontaneous speech. The regularities behind the distribution of disfluency phenomena have been extensively studied in the fields of psycholinguistics (e.g., Eilsner-Goldman 1958, Tannenbaum et al 1965, Levelt 1983, 1989, Swert 1998, Clark and Fox Tree 2002), sociolinguistics (e.g., Acton 2011, Tottie 2011, 2014, Freuhwald 2016, Wieling et al 2016) and speech technology (e.g., Oviatt 1995, Liu et al 2006, Skantze and Hjalmarsson 2010). However, there is still the need to further elaborate on how the multivariate feature space jointly explains the distribution of disfluencies. In the proposed dissertation, I will attempt to explore further into this joint feature space by addressing specific outstanding questions from the following three perspectives: the unexplored covariates underlying the observed disfluency variations, the under studied forms of disfluencies, and the overall of cross-linguistic understanding of disfluent phenomena. A large amount of spontaneous speech data selected from Fisher, SCOTUS and Czech Spontaneous Speech Corpus will be extensively analyzed. This corpus-based approach makes it relatively easy to bring evidence to bear on relevant questions. It is hoped that this work will make contributions by offering an account on disfluencies simultaneously and comprehensively from multiple perspectives, using the information gathered from large amount of representative spontaneous speech. Implications of the distributitional properties of speech disfluencies on sociolinguistic and psycholinguistic research, as well as on some practical applications will also be discussed.
The proposal document can be found here <https://hozh3497.github.io/dissertation_proposal.pdf?raw=true>