5-03 - A Corpus-based Analysis of Syncopated Patterns in Ragtime
Phillip B Kirlin
Keywords: MIR tasks, Pattern matching and detection, Applications, Music retrieval systems, Automatic classification, Indexing and querying, Music summarization, Similarity metrics
Abstract:
In this paper, we build on and extend a number of previous studies of rhythmic patterns that occur in ragtime music. All of these studies have used the RAG-C dataset of approximately 11,000 symbolically-encoded ragtime pieces to identify salient rhythmic patterns in the corpus and qualify how they are used. Ragtime music is distinguished from other musical genres by frequent use of syncopation, and previous computational studies have confirmed a number of musicological hypotheses regarding the use of syncopated patterns in ragtime compositions. In this work, we extend these studies to investigate further questions involving the use of syncopation. Specifically, we introduce a new methodological framework for processing the RAG-C dataset and confirm that experiments from previous studies obtain similar results using the new methodology. We investigate the use of the common ``short-long-short'' syncopated pattern in different time periods and present new results detailing its use by three well-known ragtime composers. We describe how the use of other syncopated patterns has evolved over time and the different distributions of patterns that result from those changes. Lastly, we present novel results identifying statistically significant patterns in the way composers varied the amount of syncopation in consecutive measures in compositions.