Cross-Cultural Corpus Creation and Statistical Tendencies in Music

Abstract

The notion that some musical features can be found in a majority of cultures across the globe has garnered scholarly attention within the past decade [6, 29]. However, the lack of both large and diverse musical corpora has made furthering work on this topic difficult. This paper addresses this issue by testing four purported statistical tendencies in melodic organization across four substantial samples of culturally diverse music. Three corpora include musical samples of European, Native American, and Chinese folk music. The fourth, a new corpus devised specifically for the purposes of this study, is a corpus of cross-cultural folk music including material from nearly 700 sampled audio recordings collected from 44 distinct cultures. The creation of this corpus involved establishing clear guidelines on how to parse audio materials, how to define a musical “phrase”, and how to transcribe musical information (e.g. pitch and duration). While the primary purpose of this paper is to contribute to the dialogue on corpus creation techniques involving culturally diverse music, results for the testing of all four hypotheses are included. Results are consistent with a number of broad statistical tendencies in musical phrases that are evident above chance levels, and are common across the repertoires sampled. These include the tendency for small over large pitch movements, for large leaps to ascend, for musical phrases to fall in pitch, and for phrases to begin with an initial pitch rise. Limitations of our corpus creation methods and empirical tests are highlighted.