The Computational Processing of Intonational Prominence: A Functional Prosody Perspective

Abstract

Intonational prominence, or accent, is a fundamental prosodic feature that is said to contribute to discourse meaning. This thesis outlines a new, computational theory of the discourse interpretation of prominence, from a FUNCTIONAL PROSODY perspective. Functional prosody makes the following two important assumptions: first, there is an aspect of prominence interpretation that centrally concerns discourse processes, namely the discourse focusing nature of prominence; and second, the role of prominence in language processing in general, and discourse processing in particular, is not essentially separate from the processing of other grammatical, nonprosodic information. This thesis develops a computational theory of prominence interpretation by explaining how prominence serves as an inference cue in discourse processing. Prominence signals changes in the attentional status of entities in a discourse model, while nonprominence signals that the realized entities are already in discourse focus. Evidence for the new theory is provided by distributional analysis of a spontaneous narrative monologue. New discourse processing algorithms that integrate form of expression, grammatical function and intonational prominence information for reference resolution and attentional state modeling show how the principles of the theory may be applied in SPEECH UNDERSTANDING systems. Finally, aspects of the new theory are explored in accent prediction experiments on a corpus of spontaneous and read direction-giving monologues. Machine learning is used to investigate the extent to which the analyzed higher-level linguistic features associated with prominence may combine with lower-level features that are known to influence accent assignment. Original constituent-based accent prediction experiments attempt to bootstrap off of established knowledge about citation-form accenting, and begin to develop an understanding of how the examined features of discourse context may be integrated into accent assignment systems for text-to-speech synthesis.