Modeling rhythmic complexity in a corpus of polyrhythm examples from Europe and America, 1900-1950

Author: Poudrier, Ève & Shanahan, D.

Presentation details: Proceedings of the 15th International Conference on Music Perception and Cognition and the 10th Conference of the European Society for the Cognitive Sciences of Music Conference (Parncutt, R. & Sattmann, S., Eds.), Montreal (QC), Canada, 23-28 July 2018, pp. 355-360


Abstract: Rhythmic complexity, as represented by polyrhythm (the superposition of two or more contrasting rhythms, meters or speeds), is often identified as one of the central features of twentieth-century
Western art music. This project uses computational analysis to explore the development of polyrhythm in a corpus of 719 examples extracted from 450 works by composers from Europe and North America from 1900 to 1950 (Suter, 1980). The current study aims to develop complexity metrics to examine the use of polyrhythm in this period, test competing claims about its development, and explore the cognitive processing of complex rhythms. Corpus examples and associated metadata were processed to be analyzed using the Humdrum Toolkit (Huron, 1995). Exploratory analysis was conducted using a stratified sample dataset (N = 80) that includes four randomly selected examples for each composer (N = 20). Correlational analysis using global complexity measures (average nPVI, event density, entropy) showed that differences in entropy and variability between rhythmic groups within each excerpt could be predictive of genre, and that the entropy of the composite rhythms seems to decline in the 1930s and 1940s, before increasing again in the 1950s. These results are taken as starting points for future avenues of research on the interaction of rhythmic groups, variability, and complexity in the use of polyrhythm in Western art music of the twentieth century.