A Search for Structural Similarities of Oral Musical Traditions in Eurasia and America Using the Self Organizing Cloud Algorithm

Abstract

The main ethnomusicological contribution of this work is that a significant part of the similarities revealed by a comparative study of 36 folk music cultures in Eurasia and America can be traced back to certain well identifiable subsets of melody types jointly appearing in well-identifiable groups of musical cultures. We attribute this phenomenon to the effect of certain musical ‘primal languages’. The research was based on a newly developed algorithm for clustering and visual representation of large symbolic folksong corpora. We applied the system to find subsets of melody contour types jointly appearing in different musical cultures, as well as to find groups of cultures having many common melody types.

Keywords:

• data mining
• probabilistic clustering
• unsupervised self-learning
• multidimensional scaling
• comparative ethnomusicology
• contour analysis

Additional information

Funding

This work was supported by the Hungarian National Research Found (grant no. K33312).