The question of whether cultural traits evolve in parallel with biological and demographic histories has been a long-standing debate. Darwin noted parallels between biological and cultural evolution, suggesting a potential link between language phylogenies and human population movements. While some studies have found correlations between linguistic and genetic relationships, others highlight the complexity of these relationships, arguing that basic vocabulary, often used in linguistic phylogenies, provides only a limited perspective on cultural history. This paper explores the role of music, a universal yet diverse cultural trait, in understanding human history. Music, like language, is transmitted across generations, and its global diversity could potentially mirror patterns observed in genetics and linguistics. Three contrasting hypotheses are examined: (1) music correlates with genes due to parallel migration and evolution; (2) music correlates with language due to shared cultural transmission; and (3) musical patterns are unrelated to genes or language due to differing evolutionary dynamics. Previous regional studies have yielded mixed results, highlighting the need for a global-scale comparison using comprehensive datasets. The availability of the Global Jukebox dataset, a comprehensive collection of almost 6000 songs from nearly 1000 societies coded with Cantometric features, alongside recently published datasets of genetic diversity and global linguistic evolution, allows for a robust global-scale test of these hypotheses.

The existing literature on the relationship between cultural traits and demographic history is mixed. Some studies support the correspondence between linguistic and genetic phylogenies, indicating a shared history of human population movements. However, a significant portion of cases show discrepancies, with populations genetically closest to linguistically unrelated groups. Critics of purely phylogenetic approaches emphasize the limitations of relying on basic vocabulary to reconstruct cultural history, arguing that this approach may neglect other cultural dimensions. Alan Lomax's work, spanning seven decades, proposed that musical style might be a more stable marker of cultural history than language, offering a valuable perspective for this research. Prior studies examining the correlation between musical and genetic diversity have yielded mixed results, depending on the geographic region, with some showing significant correlations while others showed no relationship. The lack of globally representative datasets previously hindered a comprehensive global comparison.

This study utilizes three main datasets: The Global Jukebox (Cantometrics), containing nearly 6000 songs from approximately 1000 societies coded for 37 standardized musical features; GeLaTo, a genomic database examining genetic and linguistic diversity; and a recently produced Bayesian global language phylogeny. The researchers focus on three samples from the Global Jukebox: (1) societies with at least two songs (5242 songs, 719 ethnolinguistic groups), (2) societies with 10 or more songs, and (3) societies matched to the Standard Cross-Cultural Sample (SCCS). Genetic data is drawn from GeLaTo, with a focus on a subset of populations matched with the musical data. Linguistic relationships are derived from the global language phylogeny. Data pre-processing involved standardizing Cantometric features to a 0-1 scale and reversing some codes for consistency. A latent variable model was constructed, reducing the 37 Cantometric variables into five dimensions representing Articulation, Tension, Ornamentation, Rhythm, and Dynamics. The model's validity was assessed using RMSEA, SRMR, and CFI. Analysis of Molecular Variance (AMOVA) was used to assess the relative importance of within- and between-society musical diversity. Geographic autocorrelation was estimated using Haversine distance to assess if musical similarity increases with geographic proximity. Delta scores were calculated to quantify the tree-like structure of musical diversity, reflecting the relative contributions of vertical and horizontal transmission. Finally, partial redundancy analysis (RDA) and partial Mantel tests were used to examine the correlation between musical, linguistic, and genetic data, controlling for geographic effects. Regional analyses were conducted for Africa, Europe, and Southeast Asia to assess regional variations in these relationships.

The analysis revealed five distinct dimensions of Cantometric musical diversity. AMOVA results indicated substantial between-society variation in musical style, despite considerable within-society diversity. Geographic autocorrelation analyses showed that geographically closer societies exhibit greater musical similarity, though this relationship was weaker than observed for genetic and linguistic data. Delta score analysis suggested a tree-like structure of musical diversity, similar to previous findings in language evolution, indicating both vertical and horizontal transmission. Crucially, partial RDA and partial Mantel tests showed weak and inconsistent correlations between musical similarity and both genetic and linguistic relationships on a global scale. The adjusted R² values were generally low (<10% in 91% of tests). Regional analyses revealed stronger correlations between musical and linguistic diversity in Africa and between musical and geographic diversity in Europe and Southeast Asia. This variation across regions suggests that the overall lack of global correlation does not stem from methodological issues but rather reflects regional idiosyncrasies. The SCCS analysis showed no significant correlations, as expected due to the sampling methodology.

The findings challenge the notion of a strong, consistent relationship between musical diversity and linguistic or genetic histories on a global scale. The weak correlations observed do not negate the historical structure within musical traditions, as evidenced by the tree-like structure revealed by the delta-score analysis and the geographic autocorrelation. Instead, the results suggest that musical history captures information largely independent of linguistic and genetic histories. These results align with ethnomusicological observations of the independent movement of musical traditions across cultural and linguistic boundaries. The authors propose that mechanisms such as borrowing and cultural exchange could contribute to the independent evolution of music, suggesting that musical inheritance is not strictly confined to linguistic lineages. Future research should investigate the relative contributions of neutral drift (e.g., random variation) and functional coevolution (e.g., adaptation to social structures) in shaping musical diversity. While the current data does not definitively distinguish between these mechanisms, the existence of previously proposed neutral musical features suggests that drift is a potentially important factor.

This study provides compelling evidence that global musical traditions, as captured by the Cantometric features, demonstrate a largely independent evolutionary history compared to linguistic and genetic histories. While exhibiting geographical structure and a partially tree-like pattern suggestive of vertical transmission, musical similarities do not consistently reflect genetic or linguistic relatedness. Future research should focus on disentangling the influences of neutral drift and functional coevolution, exploring a broader range of musical features, and integrating diverse data sources to enhance our understanding of human cultural evolution.

The study's limitations include the relatively small sample size (121 societies) for the matched musical, linguistic, and genetic data, which represents a non-random subset of the larger musical dataset. This limited sample size might affect the generalizability of the findings. The Cantometric coding scheme, while validated, might not fully capture the nuances of human musical perception, potentially leading to an incomplete picture of musical diversity. Further, the use of basic vocabulary data in the linguistic phylogeny may not encompass the full complexity of linguistic evolution. Despite these limitations, the robustness checks conducted across different samples from the Global Jukebox provide confidence in the primary findings.