CONTINUITY AND CHANGE IN KAKHETIAN INTONATION: A COMPUTATIONAL STUDY OF ARCHIVAL RECORDINGS (1948–1971)

Abstract

This study presents the first corpus-based computational analysis of harmonic interval distributions in traditional Kakhetian polyphony, examining 331 field recordings recorded in Kakheti from the Tbilisi State Conservatoire archive spanning 1952–1967. Using automatic multiple F0 estimation and Gaussian Mixture Model fitting, we extract harmonic intervals from vertical pitch slices and characterize their distributions across temporal, performer-based, geographic, and genre-based dimensions. The Wasserstein distance quantifies distributional shifts between recording periods and analytical strata.

      Results reveal a stable core of structural intervals—the perfect fifth (~700 cents) and octave (~1200 cents)—persisting across all recording periods and contexts. Mid-range intervals (thirds and fourths) show greater variability, with later recordings (1962–1967) exhibiting more complex distributions absent in earlier material. Distinct intervallic profiles emerge across performer groups: women's ensembles display more fragmented distributions compared to men's ensembles. Geographic sub-regions within Kakheti show measurable variation, with Telavi recordings standing apart from other areas. Genre-based analysis reveals differences between feast/table songs (“supruli”) and work songs ("shromis"), the latter showing higher temporal variability.

      These findings establish a quantitative "harmonic fingerprint" for Kakhetian polyphony, providing empirical evidence for both continuity and change in intervallic practice across mid-20th century field recordings. The stratification approach with respect to chronology, performer group gender, geography, and genre helps disentangle the contributions of each component to observed variation, while highlighting methodological considerations for corpus-based analysis of archival ethnomusicological collections.

 

Keywords. Computational ethnomusicology, music information retrieval, historical musicology, audio corpus analysis, harmonic fingerprinting, traditional Kakhetian singing.

Published
2026-06-17
Section
SCIENTIFIC ARTICLES -SECTION OF CULTURAL STUDIES

Most read articles by the same author(s)