Date of Award

Spring 1-1-2025

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Music

First Advisor

Quinn, Ian

Abstract

This dissertation examines several ways in which listeners' expectations for musical structure are related to the usage of repetition across and within pieces of music in a range of different styles. In particular, it examines the relative role of long-term expectations (acquired from a lifetime of exposure to music) and short-term expectations (acquired in real time as a piece of music unfolds) in music listening. Throughout the dissertation, I assume that listeners acquire musical expectations via statistical learning, a cognitive mechanism that automatically extracts distributional properties of sensory input without explicit instruction. Part I of this project, comprising the first two chapters, problematises the usage of overall summary statistics from musical corpora as a proxy for the long-term expectations a listener might bring to the experience of hearing a piece for the first time. I present this issue by considering two types of musical patterns: conventional patterns (belonging to many examples in a corpus of music) and characteristic patterns (repeated extensively in just a handful of examples in that corpus). Assuming that listeners only ever hear a subset of pieces in a corpus, then conventional patterns are more likely to be shared between listeners due to their widespread coverage, whereas characteristic patterns are less likely to be shared even if their overall likelihood in a listener's mental model of music is comparable to that of a conventional pattern. Chapter 1 applies the concepts of conventional and characteristic patterns to build an understanding of harmonic practice in twentieth- and twenty-first-century popular music, showing that twentieth-century popular music relies more on conventional patterns and twenty-first-century popular music relies more on characteristic patterns. Meanwhile, Chapter 2 presents a behavioural study that explores the effect of manipulating a pattern's conventionality or characteristicness on listeners' ability to learn that pattern in the first place. Part II of this project, comprising the next two chapters, is more focused on investigating short-term expectations. In particular, it explores the structural properties of patterns of large-scale thematic repetition from the perspective of how listeners dynamically adapt their expectations as a piece of music unfolds in real time. Chapter 3 presents a number of different perceptual theories of large-scale repetition structure in music and uses an information-theoretic framework to implement them computationally. I subsequently show, through an analysis of a highly stylistically diverse corpus of music, that (according to this information-theoretic framework) many pieces exhibit patterns of repetition that feature a relatively moderate rate of musical surprise compared to alternative patterns. Chapter 4 presents a novel composition task that explores whether naive participants exhibit a bias towards generating musical structures that also exhibit a moderate rate of musical surprise. Overall, this dissertation presents new applications of concepts from computational music cognition to important problems in music theory. The two parts highlight the contribution of different modes of expectation formation to the emergence of musical structure, and each present methods that open up new possibilities for analysing musical style and theorising musical form.

Download

Share

COinS