PRINCIPAL COMPONENTS ANALYSIS OF EUROPEAN FOLKSONGS

• Like Cavalli-Sforza who used prevalence of genetic variations, we used prevalence of musical features for the basic information used in the components analysis.
  
  
• Began by measuring 60 musical features for about 1,000 songs (resolved to the city/village level). E.g. features included:
  
  
  • number of notes in the melody
  • average note density (notes/second)
  • mean amount of syncopation
  • percent scale degree redundancy
  • average metrical position of the last note
  • percentage of iambic/trochaic feet
  • etc.
  
  
• Sample results: For each degree moving north latitude, melodies increased in length by 0.5 notes. (!?!)
  
  
• As we will see, the sample of music proved problematic.
  
  
• But also we are inexperienced with these types of analysis.