Event
Xin Gao will be defending their dissertation, "Compositional and Conversational Pressures on Silent Pauses in Spontaneous and Read Speech" on Monday, June 20th at 9:30 am EST.
The defense will take place in the Linguistics department library, and on Zoom.
----------------------------------------------------------------
Title: Compositional and Conversational Pressures on Silent Pauses in Spontaneous and Read Speech
Supervisor: Mark Liberman
Committee: Sunghye Cho, Meredith Tamminga
This dissertation investigates the interaction between prosodic realizations and the message structure of phrasing in natural speech, comparing patterns between spontaneous and read speech. Employing a corpus-based approach to address the limitations of lab-based speech studies and analyze real-world language, the study uses silent pause as its test case. As a frequent prosodic cue, silent pause is selected because it serves multiple functions in speech, is minimally affected by morphophonological processes, and can be precisely measured using current techniques.
In this dissertation, I address two questions: (1) How does the message structure of phrasing shape silent pause realization? Specifically, syntactic-structural and non-syntactic influences are examined. Through systematic analyses of audiobook readings and telephone conversations, I extend prior research by examining the interaction between prosody and the message structure of phrasing. (2) What underlies differences in silent pause realization across speech styles? By analyzing the same speakers under varying conditions, I test two explanatory mechanisms: compositional pressure and conversational pressure. I examine how these dimensions of pressure drive observed variations in silent pause realizations.
Chapter 2 addresses the first research question, demonstrating that both syntactic and non-syntactic structural factors shape silent pause realization. I establish that although syntactic structure substantially influences prosodic realization, prosody functions as an independent dimension where non-syntactic factors play a critical role, not merely mirroring syntax. Chapter 3 examines stylistic differences in the interaction between silent pause and message structure. Key findings show different structural effects across speech styles and weaker syntax-prosody alignment in spontaneous vs. read speech. I argue this misalignment arises from cognitive-prosodic trade-offs under heightened conversational/cognitive loads of spontaneous speech. Chapter 4 further investigates potential explanations for observed stylistic differences. By isolating compositional and conversational pressures while examining their individual effects on the same speakers’ speech, I systematically test these dual mechanisms' influence on syntax-prosody alignment. Empirical results demonstrate strong effects of compositional pressure, while revealing inconsistent evidence for conversational pressure.
Chapter 5 examines genre and individual differences in silent pause realization and its structural interactions. Analyses reveal divergent mediation patterns between political and non-political speech: stylistic and syntactic influences on silent pause realization are significantly stronger in non-political speech, alongside substantial individual variation. Finally, Chapter 6 synthesizes implications for understanding how compositional and conversational mechanisms mediate prosody-syntax interactions through distinct speech styles.
Collectively, these findings establish that prosodic realization dynamically interacts with the message structure of phrasing. Stylistic variations in prosody arise from divergent compositional and conversational pressures inherent to different speech styles. By framing these styles as causal sources of pressure differentials, this work proposes a unified account of structural-prosodic alignment. It reveals fundamental divergences between spontaneous and read speech, advocating a paradigm shift from laboratory-controlled approaches to naturalistic speech studies grounded in real-world complexity.