Retained or lost in transmission? Analyzing and predicting stability in Dutch folk songs

The goal of the dissertation is to investigate the transmission of Dutch folk songs: which parts of melodies change, and which remain stable? To this end, melodies are compared computationally, using similarity measures established in Music Information Retrieval. The computational comparison quantifies the stability, i.e., resistance to change, of a given melodic segment within a group of related melodies. This quantified stability of melodic segments is then related to hypotheses which might predict stability. The hypotheses are based on previous evidence from music cognition and computational musicology. They concern the length, the number of repetitions and position of melodic segments, as well as the melodic expectations created by the notes composing a segment. The results show that indeed stability can be predicted to some extent through the proposed hypotheses. This contributes to understanding the retention and transmission of cultural heritage. There are also other factors of cultural transmission which cannot be investigated with the proposed methods, providing perspectives for future research.