Methodlogical Research

Deconstructing the nPVI

I'm very excited that my first purely methodological paper has passed peer-review and been accepted for publication with the journal Music Perception! In this paper, I study, and critique the use of the normalized Pairwise Variability Index (nPVI) in symbolic music research. The nPVI is a mathematical measure designed to quantify the rhythmic quality of languages, especially regarding the stress-timing vs syllable-timing distinction. However, in 2003, Joe Danielle and Aniruddh Patel introduced the nPVI to music research, where it has seen much use since then. I argue that the nPVI is not particularly useful, meaningful, or transparent, as a descriptor of rhythm in symbolic music data, especially when compared to alternative measures that are more transparent. These issues with the nPVI are, of course, just examples of a broader problem in science, when we become reliant on fancy-sounding but opaque, "black-box" algorithms/calculations.