research about musical genres

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

• View all journals
• Explore content
• About the journal
• Publish with us
• Sign up for alerts
• Open access
• Published: 10 May 2024

Global musical diversity is largely independent of linguistic and genetic histories

• Sam Passmore   ORCID: orcid.org/0000-0002-5302-356X 1 , 2 ,
• Anna L. C. Wood 3 ,
• Chiara Barbieri   ORCID: orcid.org/0000-0001-8827-5655 4 , 5 , 6 ,
• Dor Shilton 7 , 8 ,
• Hideo Daikoku 1 ,
• Quentin D. Atkinson   ORCID: orcid.org/0000-0002-8499-7535 9 &
• Patrick E. Savage   ORCID: orcid.org/0000-0001-6996-7496 9 , 10  

Nature Communications volume  15 , Article number:  3964 ( 2024 ) Cite this article

3772 Accesses

118 Altmetric

Metrics details

• Interdisciplinary studies

Music is a universal yet diverse cultural trait transmitted between generations. The extent to which global musical diversity traces cultural and demographic history, however, is unresolved. Using a global musical dataset of 5242 songs from 719 societies, we identify five axes of musical diversity and show that music contains geographical and historical structures analogous to linguistic and genetic diversity. After creating a matched dataset of musical, genetic, and linguistic data spanning 121 societies containing 981 songs, 1296 individual genetic profiles, and 121 languages, we show that global musical similarities are only weakly and inconsistently related to linguistic or genetic histories, with some regional exceptions such as within Southeast Asia and sub-Saharan Africa. Our results suggest that global musical traditions are largely distinct from some non-musical aspects of human history.

Similar content being viewed by others

Cultural macroevolution of musical instruments in South America

Commonality and variation in mental representations of music revealed by a cross-cultural comparison of rhythm priors in 15 countries

The pace of modern culture

Introduction.

Do people and their cultures move together or independently? Darwin proposed “curious parallels” 1 between biological and cultural evolution, such that “a perfect pedigree of mankind… would afford the best classification of the various languages now spoken throughout the world” 2 . Darwin’s proposal stimulated studies of cultural evolution that attempted to trace ancient population movements by combining linguistic, archaeological, and/or genetic histories 3 , 4 , 5 , 6 , 7 , 8 , 9 . Some have found support for a correspondence between the phylogenetic patterns in language and the movement of human populations 4 , 7 , 10 , 11 . For example, quantitative data comparing global genetic and linguistic diversity shows that genetic relationships between populations are generally tighter within language family groupings, but also that in around 20% of cases, populations are genetically closest to linguistically unrelated groups 12 . Critics of the phylogenetic approach argue for more complex relationships between people and their cultures, pointing out that basic vocabularies used to construct language phylogenies represent one limited dimension of cultural history that does not necessarily correspond to other markers of language or culture 13 , 14 , 15 , 16 , 17 .

Music, like language, is a universal cultural trait that varies within and between societies 18 , 19 , 20 , 21 , 22 , 23 , 24 . Could music play a significant part alongside language in research on human history? 70 years ago, Alan Lomax proposed that it could, arguing that musical style changes less than language or other cultural traits 25 .

There are three contrasting predictions regarding potential historical relationships between music, languages, and genes: (1) music correlates with genes due to parallel processes of migration and evolution 26 , 27 , 28 , 29 ; (2) music correlates with language due to shared process of cultural transmission via vocal and interactional domains (i.e., song and speech both primarily use words) 30 , 31 , 32 ; (3) musical patterns are unrelated to either genes or language, due to differences in the evolutionary shape, fabric, and/or tempo of musical, linguistic, and genetic evolution (e.g., rapid musical change independent of demographic or linguistic turnover) 33 , 34 , 35 .

Direct quantitative testing of these competing predictions with matched musical, genetic, and linguistic data has previously been restricted to regional studies and has produced mixed results. Studies have found evidence of significant correlations between musical and genetic diversity in Taiwan, Sub-Saharan Africa, and Eurasia 26 , 36 , 37 but not in Northeast Asia 38 . Global comparisons were not previously possible because detailed public data on cross-cultural musical diversity, genetic similarity, and language history were not available. Recently, however, globally representative datasets of musical, genetic, and linguistic diversity have been published, allowing us to test these hypotheses.

Quantifying global musical diversity has been the focus of several recent efforts 22 , 24 , 39 . In this project, we use The Global Jukebox, a dataset of almost 6000 songs from almost 1000 societies coded on 37 standardized “Cantometric” features of musical style 20 (Table  S1 ). The Global Jukebox is particularly well suited to comparing music, language, and genes across cultures. It contains over 15 times more coded songs than other datasets that use similar coding schemes 22 , 24 . While a slightly larger global dataset of 8200 audio recordings exists 18 , it is constructed around the unit of country, which does not allow for direct comparison with genetic or linguistic data at the unit of ethnolinguistic groups. The dataset also relies on automated signal analysis and machine learning using country labels, without validating against human perceptual data. A recent independent examination of the data 40 found no significant correlation between the automated similarity algorithm 39 and human similarity ratings for a global sample of songs (but did find significant correlations between naive human similarity ratings and similarity metrics based on Cantometric codings), raising questions about the interpretability of automatically identified dimensions.

In this paper, we leverage the publication of the Global Jukebox with recently published datasets of genetic diversity, and global linguistic evolution 12 , 41 , 42 , to directly compare these three domains on a global scale (Fig.  1 ). First, we use the musical data to extract five dimensions of musical style from the Cantometrics dataset and show that these dimensions contain between-group variability, making them useful markers for cultural history. Autocorrelational tests show that the between-group structure in our musical variables is organised geographically. We observe similar patterns in genetics and language, although the strength of the musical relationship is weaker than is found in genes or language. Finally, we show that the similarities among our five musical dimensions are only weakly related to the structure found in the linguistic and genetic data, such that musical traits capture largely independent information about human cultural history.

Each point is a society, sized by the number of songs recorded for that society. All points are used to estimate latent variables. 121 societies (represented by 923 songs) are matched to both genetic and linguistic data and are coloured red. Societies without matching genetic and linguistic data are in grey. See Fig. S 2 and S 3 for maps of the 10 or more-song sample, and the SCCS sample. Maps made with Natural Earth.

Musical, linguistic, and genetic samples

Musical data is drawn from the Cantometrics dataset from within the Global Jukebox. Using this dataset, we devise three sample sets: a set where societies must have two or more songs to be included (resulting in a sample of 5242 songs from 719 ethnolinguistic groups), where societies must have 10 or more songs (3063 songs and 222 societies; Fig. S 2 ), and a sample of societies matched to the Standard Cross-cultural sample (SCCS; 742 songs and 110 societies; Fig. S 3 ).

Genetic data is drawn from GeLaTo, a genomic database designed to investigate patterns between genetic and linguistic diversity 12 . The dataset collects published genomic data genotyped with the Human Origins SNP chip, a chip designed to maximise human genetic variability across continents and minimise the effects of ascertainment bias 43 , from 4000 individuals, representing 397 genetics populations and 295 languages. Most of the genetic data available, to our knowledge, was collected between the 1990’s and today.

Linguistic relationships are drawn from a recently produced Bayesian global language phylogeny to quantify linguistic affiliations between the societies in our sample 41 . The global language phylogeny was built from a taxonomy of extant languages 44 , together with previous Bayesian phylogenetic analyses of basic vocabulary data from major families, information on the timing of language diversification events, the geographic location of languages, and assumptions about the paths of human migration 41 .

A detailed description of the processing steps of the Cantometrics dataset, the GeLaTo dataset, and the pairing of datasets can be found in the Methods and Supplementary Note  1 . All statistical tests that follow are performed across the three Cantometric datasets to assess the robustness of effects, with the results found in the  Supplementary Information , with a general summary in Tables  S3 and S4 .

Five dimensions of Cantometric musical diversity

We first built a latent variable model containing five dimensions of musical style modelled after Lomax’s 45 factor analysis of a subset of Cantometrics. Our five dimensions are a subset of Lomax’s nine dimensions after excluding four dimensions due to coding interdependencies (See Table  S6 and S7 for details). By using dimension reduction to reduce Cantometric variation to 5 latent musical variables, we distil any repeating signal that occurs across several interdependent variables, while removing information that is variable-specific.

The five latent dimensions were designed to reflect: (1) Articulation (lyric repetition and enunciation); (2) Tension (vocal width, nasality, raspiness); (3) Ornamentation (the amount of decorative singing within a song), (4) Rhythm (meter and tempo); and (5) Dynamics (volume, register, and intensity; see Supplementary Note  2 for more detail on variable construction and examples for all variables). To examine whether the five-dimensional model (including additional covariances; see Table  S5 ) is a valid description of the data we use three tests of model validity: Root Mean Square Error of Approximation (RMSEA), Standardized Root Mean Square Residual (SRMR), and the Comparative Fit Index (CFI) 46 . Both RMSEA and SRMR values are considered appropriate model fits if they are <0.08. CFI is considered an appropriate fit if the score is above 0.9 (see Methods for descriptions of these measures).

The two or more song dataset (5242 songs, 719 groups) passes all model fit tests (RMSEA = 0.06 (90% CI: 0.056-0.068); SRMR = 0.05; CFI = 0.93). The model also fits a dataset where societies must have >10 songs to be included (RMSEA = 0.06 (90% CI: 0.059-0.061); SRMR = 0.06; CFI = 0.94) and a dataset that only contains societies within the Standard Cross-Cultural Sample (SCCS; RMSEA = 0.06 (90% CI: 0.059-0.067); SRMR = 0.05; CFI = 0.94). We performed additional sensitivity analyses on the latent variables by excluding all Cantometrics variables that show low inter-rater agreement (Cohen’s kappa agreements of <0.4), proposed as a minimum acceptable level of reliability [e.g., in clinical contexts 47 ]. All latent variables showed a correlation of 0.97 or higher between the two latent models, except Tension (Table  S10 ). Results involving Tension should be interpreted cautiously. For simpler comparisons in later analyses, we separately build an aggregate measure of musical similarity over all Cantometric variables, analogous to the ‘modal profiles’ or ‘musical distances’ used in previous Cantometric analyses 20 , 23 (cf. Supplementary Note  3 ).

Music varies between societies

Figure  2 shows that although musical diversity presents as a continuous phenomenon (Fig.  2a ), there is also an underlying structure aligning with cultural lineages (Fig.  2b ). Figure  2b shows overlapping yet different distributions of songs from two language families, Atlantic-Congo and Sino-Tibetan. Sino-Tibetan songs tend to be more ornamented, whereas Atlantic-Congo songs tend to contain more regular rhythms. Highlighting two societies within these language families (Fig.  2b Red squares: Ubangi; Blue squares: Burmese) shows the multi-level nature of musical diversity within and between societies.

A Each point is a song, and labels within the graph refer to exemplar songs described in S2.2: [A] Song with Xylophone—Burmese, [B] Djokobo—Mbendjele, [C] Caravan Song—Tibet, [D] Alima Song—Mbuti. B The same scatter plot as ( A ), with topographical gridlines showing the density of points in Fig. 2a, overlaid with points from the Atlantic-Congo language family (red) and the Sino-Tibetan language family (blue). Squares show songs from the Atlantic-Congo society Ubangi (red), and the Sino-Tibetan society Burmese (blue). The dispersal of squares reflects within-society diversity.

To formally test whether societies can be differentiated musically, we use an AMOVA (Analysis of Molecular Variance 23 , 48 , 49 test on a set of 636 societies with linked language families, totalling 5131 songs). AMOVA parses the variance of a trait to show the relative importance of within-society and between-society diversity (Supp. Data S 1 ; cf. Fig. S 5 for a visual comparison of results). In genetics, within-population variance accounts for 93–95% of variance and between-populations constitutes around 3% to 5% of the variance (Rosenberg et al.) 50 . Amongst our musical variables, within-society diversity explains between 54–72% of the total variation. Between-societies / within-macro group diversity (Macrogroupings can either be Language family or Macroarea) contains between 29–43%. Comparing musical and genetic fixation statistics (measures of similarity between populations) 51 ; showed that the differences in music were between 10% and 40% higher than genetic differences between populations, although between-society musical diversity was also more variable (Fig. S 6 ). In general, while musical diversity within societies is large, when compared to genetic populations there are substantial differences between societies. However, we caution that the relative between- vs. within-society variation is calculated slightly differently for music and genetics, and the greater diversity within music may partially reflect the diversity of different musical genres coexisting in each society’s repertoire. More information on AMOVA analyses is held in Supplementary Note  3 .

Music is more similar between geographically closer societies

Societies have identifiable musical differences, but do these differences reflect geographic patterns as they do in genes and languages? We estimated geographic autocorrelation (literally self - correlation) by comparing the musical, linguistic, or genetic similarity between pairs of societies whose geographic separation falls within a specified distance class (e.g. all societies within 500 km), against the similarity of those outside (e.g. beyond 500 km). Distance between societies is calculated using Haversine distance (distance between two points on a sphere). Haversine distance is a crude but common distance metric for large-scale comparative studies 49 , 52 since it is the lowest assumption distance metric. A high correlation statistic indicates that the similarity between societies within the distance class is higher than those outside it, implying geographic autocorrelation. We calculate geographic autocorrelation for musical, linguistic, and genetic processes at 500 km intervals up to 20,000 km (Fig.  3 ).

White shapes indicate significant autocorrelation and black shapes indicate non-significant autocorrelation. Error bars show the 95% confidence intervals for each distance. See Fig. S 7 for the same graph for the individual musical metrics. See Fig. S 8 and S 9 for the  10 or more-song sample and the SCCS sample. See Supplementary Data  2 – 4 for detailed statistical information.

Geographically closer societies are also more musically similar (Fig.  3 and Fig. S 7 –S 9 ). Within the averaged measure of musical similarity, spatial autocorrelation persists up to 4000 km on average ( p  < 0.01), slightly less than the level of autocorrelation seen between societies when using linguistic or genetic distance, which persists to around 5000 km and 5500 km, respectively (Fig.  3 ). The average level of autocorrelation seen in music within 4000 km is 0.17, which is comparatively lower than the average value seen in language ( r  = 0.24) and genes ( r  = 0.63) across the same distance. This result is robustness tested for each of our three data samples, with statistics for each available Supplementary Data  2 – 4 .

Musical style suggests tree-like structure

Cultural evolution research often tries to separate the contribution of vertical and horizontal transmission. “Delta-scores” have been identified as a useful tool for identifying the extent of this conflicting signal 53 , 54 , 55 , which quantifies how closely the distances between quartets of languages, music, etc., approximate the structure of a bifurcating tree (0 = a perfect branching tree with no reticulation, 1 = maximally non-tree-like). A perfectly branching tree represents a pattern of purely vertical transmission. Due to the computational expense of comparing all quartets, Delta scores are only calculated for 50 randomly chosen societies in each of Africa, Europe, and Oceania (the most sampled regions), for each latent variable. Delta scores range between 0.25 and 0.4 across all latent musical variables, with most results between 0.31–0.34 (Table  1 ). These values all fall between the range previously reported in the lexicon of 12 Indo-European languages (0.21) and 38 Polynesian languages (0.41) 53 . They are also more tree-like on average than those reported for lexical and structural data in 81 Austronesian languages (0.38 and 0.44, respectively) 54 .

Musical similarity contains independent structure compared to language and genetics

We introduced three possibilities for why we might observe geographical patterning in musical style: (1) a correlation with genes; (2) a correlation with language; or (3) music is unrelated to genes or language. To dissect which of these theories is most likely we test the correlation between our musical measures, against measures of spatial, linguistic, and genetic similarity. We test the relationships between musical, spatial, linguistic, and genetic data using partial redundancy analysis (RDA) 38 . We also report analogous analyses using partial Mantel tests (Mantel, 1967), but caution that Mantel tests are often considered unreliable 56 . Musical distances are represented by Phi ST distance matrices (one for each dimension, and one for aggregate similarity, six in total). Genetic distances are measured by F ST, and linguistic distances are measured through patristic distance.

Within our two or more song sample, music shows weak correlations with genes and language (Music—Genes (Controlling for Geography): Mantel’s r  = 0.15 ( p  < 0.001) RDA Adjusted R 2  = 0.09; Music—Genes (Language): r  = 0.1 ( p  < 0.05); RDA Adj. R 2  = 0.06; Music—Language (Geography): r  = 0.18 ( p  < 0.001); RDA Adj. R 2  = 0.1; Music—Language (Genes): r  = 0.11 ( p  < 0.05), RDA Adj. R 2  = 0.03; See Supplementary Data S 6 for more details). Results for the two or more-song sample show high agreement with the 10 or more-song sample (Supplementary Data S 7 ), but not the SCCS sample (Supplementary Data S 8 ; Table  S12 summarises the comparison). The SCCS sample does not show strong relationships with any process. Of all 36 tests performed 91% returned an Adjusted R2 value of <10%, and 83% returned an Adjusted R2 value of <5% (See Supplementary Data S 8 for specific test results). Since the SCCS sample is designed to maximise the independence of societies (and thus minimize autocorrelation), we should expect that autocorrelation between groups is low in this dataset. To test the sensitivity of our results we also perform this analysis in regional samples.

Correlations within the three regions with the largest samples, Africa ( n  = 20 societies), Europe ( n  = 27), and Southeast Asia ( n  = 12), show us how variable the evolution of music may be, with the caveat of smaller samples (Fig.  4 ). Within Africa, the strongest musical correlations are with linguistic distance, explaining up to 33% of the variance (for the Articulation latent dimension). Within Europe and especially Southeast Asia, music is most correlated with geographic distances (up to 23% and 50%, respectively). We advise caution when interpreting the regional analyses for two reasons. Firstly, the smaller sample sizes in each region mean there is likely substantial unaccounted variability in the estimates, and secondly is variability in cultural homogeneity. A proxy for cultural homogeneity is the number of language families found within a region. In Europe and Southeast Asia, there are five language families each. Most European languages are Indo-European, with only a few Uralic, Turkic, and two isolate languages, whereas Southeast Asia contains a slightly less biased spread across the smaller families of Austroasiatic, Hmong-Mien, Sino-Tibetan, and Tai-Kadai languages, with the majority of languages coming from Austronesian. Within Africa, there are 22 different language families, considerably more than the other two regions. The number of languages in each family is heavily skewed towards a few large families, with 73% of samples coming from three language families. More than half of the societies speak an Atlantic-Congo language, 15% are Afro-Asiatic, and 9% are Nilotic. In each regional case, there is a similar level of language family diversity, when accounting for their unbiased distribution. Nevertheless, the range and breadth of cultural diversity are likely to have an impact on regional calculations of similarity by acting as hurdles to the flow of genetic and cultural material.

RDA tests show the amount of variation explained by Genetic, Linguistic, or Spatial distances while controlling (ctrl) for a second process. From top-left, clockwise, Global distances, within Africa, within Europe, and within Southeast Asia. All correlations are rounded to two decimal places.

After over a century of debate and partial testing using indirect proxies or regional samples, our direct comparison of global musical, linguistic, and genetic data reveals that musical histories—as captured in stylistic features of traditional songs—are not consistently related to genetic and linguistic histories on a global scale. This does not imply that musical features do not preserve historical patterns, as our analysis reveals that these features may preserve a relatively tree-like structure suggesting largely vertical transmission across generations. Rather, musical histories capture partially independent features from linguistic or genetic ones.

A practical concern might be that the three datasets are not comparable, and therefore a consistent relationship is not expected. We believe this is not the case for two reasons: (1) previous studies have shown that it is possible to capture significant correlations between cultural and/or genetic data using the same types of data sources used here 12 , 26 , 42 ; and (2) our spatial autocorrelation analysis (Fig.  3 ) and regional analyses (Fig.  4 ) show the presence of stronger relationships in precisely the kind of local areas and scales that had previously been suggested in regional studies (i.e., sub-Saharan Africa and Southeast Asia 26 , 36 ). This suggests that our finding of substantial divergence on a global scale is not an artefact of our methodology but rather reflects the reality that music is largely independent of linguistic and genetic histories on a global scale. However, future analyses sampling music, language, and genes from the same population at the same time may be able to examine possible sampling biases more comprehensively (cf. 57 , 58 for examples of direct comparison of music and language recorded from the same individuals in diverse societies).

Our quantitative data seem to plausibly quantify what ethnomusicologists have long argued based on qualitative data: musical traditions often move independently from people or their languages. For example, the Cantometrics data has shown a large region of similar solo, heavily ornamented and richly accompanied singing styles stretching across the Eurasian “Silk Routes” from the Mediterranean to Japan, uniting groups speaking diverse languages from different families (e.g., Afro-Asiatic in North Africa, Indo-European, and Dravidian in India, Sino-Tibetan and Altaic in East Asia) and with partially independent genetic histories 42 . Instruments and musical systems have also been documented to have diffused and evolved across this trade route (e.g., the modern Japanese shamisen and European violin evolved from their shared ancestor with the Arabic oud 59 ). Importantly, our results show that the topographical pattern of history that music holds is relatively independent of genes or language. While it may be the case that the rate of change is different between music, language, and certainly genes, the largely independent information we obtain from these three sources cannot be explained as the result of different rates of change. If traits changed at different speeds but primarily co-evolved through the same phylogenetic process, we would expect any correlation between co-evolving traits to be at least as strong as their correlation with geography. This is not what we find, suggesting that the patterns we observe reflect a pattern of global musical ancestry that is largely independent of the phylogenetic history of genes and languages.

There are many reasons why topologies might differ while maintaining historical structure. We propose two possible reasons why the historical topography of music differs from the other two phenomena. First, is the influence of historical patterns of borrowing. While languages are often used to delineate cultural groups, the historical frequency of bilingualism means individuals likely drifted between groups 60 . It is unclear if music adheres to the same level of boundedness, but it seems unlikely. Regional evidence discusses the exchange of musical ideas across linguistic and cultural boundaries 61 . The ability to share music across cultural boundaries means the flow of musical inheritance is not restricted to linguistic lineages. This might be classed as historical borrowing but could also be conceived as an alternative path of inheritance.

A crucial question for future research is to characterize the specific mechanisms driving the independence between music and language or genetics. Two major types of contrasting mechanisms are (1) neutral drift and (2) functional coevolution. Genetic and linguistic studies have primarily focused on markers that are not subject to strong selection pressure, also known as neutral markers (e.g., genetic variants unaffected by strong selective pressure, basic vocabulary for language) precisely because these are less likely to be influenced by functional coevolution 62 . Drift in the biological sense is impossible to realise in the absence of a genotype-phenotype division 63 . Within the study of music, it is not clear whether this distinction applies. In cultural-evolutionary studies in general, we rely on metaphor where drift is linked to usage-dependent mechanisms, the dynamics for which there is no purpose choice or benefit 64 . We know that some musical features have been proposed to evolve in a relatively neutral (no purpose) manner (e.g., microevolution of ornamental notes in melodies 65 ), making drift a worthy hypothesis to test. Cantometric features were specifically designed to capture either functional coevolutionary relationships with social structure on a global scale or to track historical drift 20 . Modelling whether the hypotheses stand in the light of the collected data will be a revealing avenue for evolutionary theories of music.

The current data alone will not allow us to differentiate between neutral and functional mechanisms. For example, virtuosic accompanied solo singing may have spread along the Bronze Age trade networks and the Silk Routes in a neutral manner accompanying trade and other cultural exchange, and/or functionally signalling and validating powerful hierarchies and the complex division of labour required to sustain large-scale trade networks 20 , 66 as has been proposed for other aspects of culture such as religion 67 , 68 . A reanalysis of proposed correlations between musical style and social structure supports Lomax’s hypothesis of functional coevolution 42 . Future analyses directly comparing music, social structure, language, genes, and geography will be needed to explain causal mechanisms. Similarly, while our delta-score analysis suggests similar levels of horizontal and vertical transmission as found in previous analyses of language evolution (delta-scores ranging from roughly 0.2~0.4 53 , 54 ), precise specification of these mechanisms of horizontal and vertical transmission in musical evolution will require more explicit models of the evolutionary process 30 , 69 , 70 .

The global comparison of musical diversity to linguistic, and genetic diversity represents a substantial increase in size and geographic scope over previous regional analyses. Nevertheless, our data remain limited in important ways. In particular, the sample of 121 societies with matching genetic and linguistic data is only a small and non-random subset of the full musical sample of 719 societies (cf. Fig.  1 ), as available genetic data from Indigenous populations in the Americas, Africa, and Oceania is not well represented—often due to colonial legacies 71 . The full sample of Cantometrically coded musical data also only represents simplified reductions of the full complexity of cross-cultural musical diversity into 37 features, each of which has different levels of inter-rater reliability and cross-cultural universality (cf. 29 , 42 , 72 for critical discussion of the Cantometrics sample and methodology). While the reliability of the Cantometrics data has been validated using expert coders with substantial experience recording and analyzing music throughout the world 42 , they have yet to be validated against the subjective judgments of culture-bearers who may not necessarily perceive their music in the same way as outsiders 73 , 74 , 75 (also see ref. 58 for an example of how perceptions of musicians themselves can be incorporated in comparative analysis). At the same time, it is equally possible that culture members find outsiders’ observations interesting, useful, and validating when there is dialogue 76 . Similar limitations apply to the other datasets used for comparison: for example, linguistic phylogenies based on basic vocabulary data only capture certain aspects of language evolution 77 and biases in the design of genotype platforms can lead to skewed estimates of genetic diversity 78 . Our robustness analyses (Supplementary Data S 7 and S 8 ; Table  S12 ; Fig. S 11 –S 16 ) indicate that our current findings of broad independence between musical and linguistic/genetic histories are robust to specific sampling decisions regarding the populations or variables included in the current analyses. Future analyses comparing broader ranges of musical/linguistic features (e.g., grammatical features) 38 , 79 ; direct acoustic comparison of sung/spoken audio 58 , 80 may help to understand the mechanisms underlying the separation.

The relative independence of musical processes in our analysis highlights the possibilities for music to tell us more about the relationships between societies. Earlier work has noted that “human history is written in both our genes and languages” 12 (page 1), but our work has shown that traces of history can also be found in other parts of culture. By expanding what we consider can tell us about human cultural history, we can build richer and more complex stories about the human cultural past, as well as the breadth of evidence used for building holistic models of human cultural history 12 , 81 , 82 . Much research on cultural evolution has shown complex connections between cultural domains 83 , 84 . For example, that sex-biased movement creates distinct histories of language and material culture 85 , 86 , 87 , 88 , 89 . But it is equally possible for cultural domains to tell us about contrasting relationships in human history. Basketry traditions can transcend linguistic boundaries 90 , and folk stories show incredible conservation across large geographical and historical areas 91 , 92 . Creative arts including music, dance, and poetry may be subject to less functional constraints and so may offer even more avenues in which culture can evolve independently of other aspects of cultural and population history 93 , 94 , 95 . Integrating models of music and the arts alongside genetic, linguistic, and other cultural histories into a unified narrative may enhance our knowledge of the shape and fabric of cultural evolution, and allow us to tell richer tales of the human past 53 .

Ultimately, we show that relationships between musical styles are analogous to, yet largely distinct different from, linguistic and genetic relationships. Precisely how the interplay between neutral and functional mechanisms maintains musical similarity and drives musical change is still unclear. However, our publicly available data and code, combined with the recent release of complementary public datasets of global cultural and genetic data 12 , 42 , 79 , 96 provide an important foundation for future research into human cultural and biological evolution.

The Global Jukebox contains Cantometric codings for 5776 songs from 1026 societies on 37 different variables 42 (Table  S1 , Fig. S 1 ). The dataset used here was restricted societies with at minimum two songs, meaning we use a dataset with 5242 songs from 719 societies. We also restrict our analysis to 24 of the 37 variables, which are those without built-in redundancies (see Supplementary Note  2 , Table  S6 , Table  S7 , Fig. S 4 ). Songs can display multiple characteristics within a Cantometric Line throughout the performance, meaning some songs can have multiple codes for any particular variable. For analytical reasons, we require one value per song, per variable which we select at random. This affects 3% of the dataset. We built the latent variables 100 times with different randomly chosen values, finding that the average Pearson correlation between datasets was between 0.987 and 0.99.

All musical data is standardized to a 0–1 scale for comparability between features. We reverse the codes of several existing Cantometric variables so that all variables align high values with a more frequent occurrence of what the variable measures (Table  S2 ). See supplementary material for more information on data pre-processing.

The Cantometrics dataset was designed and curated by Alan Lomax and Victor Grauer as an alternative to Western staff notation that could capture broad dimensions of musical performance present in varying degrees in the world’s music, including not only rhythm and melody but also domains such as vocal timbre and social organization of the singers and accompanying instruments (Table  S1 ). The coding scheme and first batch of several thousands of coded songs were first debuted in Lomax’s landmark book Folk Song Style and Culture 20 and updated with thousands more songs over the subsequent decades. Most songs in the Global Jukebox database were coded by Lomax or Grauer themselves, and they have been recently validated for inter-rater reliability (mean κ  = 0.54) and accuracy (~0.4–1% rate of unambiguous coding/data entry errors; Wood et al., 2022). The 37 variables vary substantially in reliability, from chance levels (e.g., nasality [Line 37]) to near-perfect (e.g., musical organization of the vocal part [Line 4] κ  = 0.94/89% agreement 42 ). Our robustness analyses removing low-reliability variables (Table  S10 ) suggest that such variation does not affect our main results. Songs within the Global Jukebox were primarily collected from the 1940s to 1980s, with a maximum range of recording dates between 1904 and 1982.

In this paper, there are three sampling variants of the Cantometrics dataset, that are used to test the robustness of the results: all societies that have two songs or more coded in Cantometrics, societies with 10 songs or more, and societies that align with the Standard Cross-Cultural Sample (SCCS).

The GeLaTo data consists of a sample of 1729 individuals from 156 populations, with a median of 9 individuals per population 12 . The source, glottocode and sample size of each genetic population sample are described in Supplementary Data S 5 and can be found in a long format in the code repository. Genetic distances between populations are calculated with the Weir and Cockerham FST formula 97 implemented in the software PLINK v. 1.9 98 , using the following script ( https://github.com/epifaniarango/Fst_forLargeDatasets ). The genetic distances are elaborated and expanded from a collection already described within the dataset GeLaTo 12 . In GeLaTo, published genetic data is merged, filtered and curated for anthropological and linguistic contextualization, to be used for multidisciplinary studies on human history and diversity. The genetic platform utilized in all the publications considered is the Human Origins SNP chip, a platform designed to maximize human genetic variability across continents and minimize the effects of ascertainment bias 43 . The final dataset includes individuals with a minimum of 550,000 SNPs successfully typed and calculates F ST over autosomal chromosome SNPs. We subset this matrix to 121 societies that could be paired to Cantometerics, totalling 923 songs. We convert the F ST distances to a correlation matrix using a Matérn correlation function, and the parameters: kappa = 0.001, phi = 0.1.

These musical, linguistic, and genetic datasets, like other cross-cultural datasets 96 as standardized ethnolinguistic markers to label and match societies, genetic populations and languages. We matched 121 societies from our musical dataset of two or more songs to the genetic and linguistic datasets (Fig.  1 ). 65 societies are paired via a direct match of Glottocodes across all datasets. A further 56 societies are manually matched with proxies, using higher- or lower-level glottocodes. For example, The linguistic phylogeny and genetic database contains samples for Czech (czec1258), but Cantometrics contains data for the subordinate dialect Czech-Morovian (czec1259). In this instance, the Cantometrics data is linked to the higher-level Czech data. The 121 societal matches are represented by 923 songs, the genomic profiles of 1296 individuals, and 121 languages across 38 language families. Our 10 or more song robustness sample is paired with 44 societies with linguistic and genetic data, and the SCCS sample to 21 societies (Fig. S 2 and S 3 ).

Latent variable modelling

Latent variable modelling is performed using R v4.1 99 and the package lavaan v0.6-9 100 . In addition to the five latent variables, the model allows latent variables to correlate with each other and incorporates six correlations between Cantometric variables which were not explained by the latent variables. A written description of the latent variable model is given in Table  S5 . The coefficients for this model are standardized for both latent and observed variables, also known as a completely standardized solution. We use three common methods for the Latent Variable model Goodness of fit, RMSEA, SRMR, and CFI. RMSEA is the difference between the observed data variance (i.e. degrees of freedom) and the proposed model, penalizing for the number of parameters. A small value indicates the model explains close to the total variance in the data. A value of <0.08 is widely considered acceptable 46 . SRMR is a similar measure to RMSEA, but does not penalize for the number of parameters, again with values close to zero indicating better fit, and values <0.08 considered acceptable. These measures are called absolute measures of fit, and measure how far the model is from a perfect fit. CFI is a relative-fit measure, comparing the proposed model to a null model. A null model assumes all variables are independent. Values >0.9 are considered appropriate. The two or more song dataset (5242 songs, 719 groups) passes all model fit tests (RMSEA = 0.06 (90% CI: 0.056-0.068); SRMR = 0.05; CFI = 0.93). The model also fits a dataset where societies must have >10 songs to be included (RMSEA = 0.06 (90% CI: 0.059-0.061); SRMR = 0.06; CFI = 0.94) and a dataset that only contains societies within the Standard Cross-Cultural Sample (SCCS; RMSEA = 0.06 (90% CI: 0.059-0.067); SRMR = 0.05; CFI = 0.94). The five dimensions align with Lomax’s (1980) proposal of these five dimensions (a sixth dimension—organization—was not strongly supported and so was not included in analyses). Two of these dimensions—Ornamentation and Rhythm—share features with the two primary dimensions identified by Mehr et al. (e.g., their melodic complexity variable and our ornamentation variable both incorporate tremolo and melodic embellishment and their rhythmic complexity variable and our rhythm variable both incorporate tempo and metre). The other three dimensions—Articulation, Dynamics, and Tension—are not directly comparable because Mehr et al. did not include such features in their principal component analysis. Table  S8 compares the weightings of a principal component analysis to the weightings of the latent variable analysis.

We tested the robustness of our results by running the analyses on the two or more-song sample, 10 or more-song sample, and the SCCS sample. Correlations between societies that exist between these datasets showed strong and significant correlations. All correlations are >0.97, with significant two-sided p -values (Table  S9 ).

To ensure our results were not an artefact of coding bias, we performed additional sensitivity analyses on the creation of latent variables excluding all Cantometrics variables with Cohen’s kappa agreements of <0.4, the lower end of the threshold described as moderate agreement 47 . One exception was made for Line 31 because of its importance in model convergence. In two cases this meant there was only one variable remaining in the latent variable model (Rhythm and Tension). Since we cannot create a latent variable from a single dimension, we use the single variables for comparison. We compare the full model to the remaining variable in these two instances. The comparison of the full to restricted latent variable model showed that all variables, except Tension, had significant Pearson correlations >0.7. Tension showed a significant, but small correlation to the remaining Tension variable (Table  S10 ). Reported p -values are two-sided. Tension results then should be viewed with more caution than the other latent variables.

AMOVA analysis is performed using R v4.1 and ade4 v1.7-18 101 . Information on Language family and the geographic Region categorisation are taken from Cantometrics metadata. Euclidean distances are calculated between songs. Macroareas are a geographical categorisation within Cantometrics that broadly correspond to United Nations Regional Groupings and Cantometric Regions and Divisions ( https://unstats.un.org/sdgs/report/2019/regional-groups/ ). Results are available in Supplementary Data S 1 .

AMOVA analysis is additionally performed on societies with 10 or more songs, and the SCCS sample. The results show negligible differences between the two-song sample, and the 10 and SCCS samples (Supplementary Data S 1 , Fig. S 5 ).

Musical Phi ST and Genetic F ST

Musical Phi ST matrices are created using the pairPhiST function within the haplotypes R Package 102 . We build Phi ST matrices for each musical dimension, and for an aggregate musical similarity that uses all Cantometric variables, for a total of five Phi ST matrices. Each of these five matrices is created three times, once for each sample of musical data. See the recipe Phi ST in the MakeFile for more details.

Genetic distances between populations are calculated with the Weir and Cockerham F ST formula 97 implemented in the software PLINK v. 1.9 98 , using the following script ( https://github.com/epifaniarango/Fst_forLargeDatasets ). The genetic data comes from published sources that used the Human Origins SNP Chip, a panel that includes ~550,000 SNPs selected to be variable in populations from all continents 43 . F ST values are calculated from a sample of 121 populations, with a minimum of 5 individuals per population, a mean of 9, a maximum of 75, and a total of 1492 individuals.

Spatial autocorrelation

Spatial autocorrelation for musical, and genetic variables were calculated from the distance matrices produced through the process in the previous section. A linguistic distance matrix was additionally produced using patristic distance within the global phylogeny, and a geographic distance was produced using Haversine distance, and the longitude and latitude for each society in the Cantometrics metadata (as described in the main text).

Autocorrelation was calculated using the Excel add-in Genalex 103 . We used the procedure “Single Pop Spatial Structure” found in the Distance-based menu, under the Spatial subheading. We used 40 evenly distributed spaced distance classes, which equated to 500 km bands, with a maximum distance of 20,000 km. These results are performed for each latent variable, the aggregated variable for the two-song dataset, and the aggregated musical variable in the 10 songs, and SCCS samples (Fig. S 7 –S 9 ). We also produce a measure of autocorrelation for the genetic and linguistic data for all samples.

Delta scores

Delta scores are based on measuring distance from the four-point condition 55 . The four-point condition says that, given four taxa have come from a tree (x, y, u, and v), the distances between those points must satisfy the following formula:

That is, the summed distance between x and y (x–y) and between u-v, must be less than the summed distance between x-u and y-v, or the summed distance between x-v and y-u, whichever is larger. Delta scores measure the distance from a perfect tree, meaning small scores are more treelike. We use Euclidean distance between each society’s musical scores to calculate Delta scores.

We calculate Delta scores for samples of 50 societies in the two-song sample. The exponential increase in possible quartets with every increase in societies creates significant computational effort, and hence why we restrict the sample size. Societies are sampled at random from the two-song, 10-song, and SCCS datasets. The results for all these samples can be found in Table  S11 and Fig. S 10 .

Partial RDA and partial mantel

Musical Phi ST distance matrices were created using the function pairPhist in the haplotypes package (Aktas, 2020). The partial RDA analysis is a two-step process. First, we reduce all distance matrices to their primary dimensions using Principal Coordinate Analysis (PCoA). We extract all dimensions that explain >10% of the total variance. Then, we use RDA models to measure the correlation between the primary dimensions of the PCoA. Bi-variate RDA regresses a response variable set onto an explanatory variable set. Partial RDA allows us to assess the strength of correlation after controlling for the influence of a third confounder (e.g., regressing Articulation on genetic distance, controlling for geographic distance). We assess the strength of the relationship using adjusted R 2 . Partial Mantel tests, like Partial RDA analyses, aim to estimate the correlation between two distance matrices while parsing the influence of a third. Partial Mantel tests were performed using mantel.partial in the vegan package 104 . Results for partial Mantel tests are in Supplementary Data S 7 . Mantel tests calculate the correlation between the two matrices, and then permute the rows of the response matrix to determine if the correlation is significantly greater than chance in a way that accounts for the non-independent nature of the distance matrix.

Partial Mantel and RDA tests are performed with the 10 or more-song sample and the SCCS sample. Results show a strong correlation between the 10 or more-song sample and the two or more-song sample, but neither sample showed a strong correlation to the SCCS sample (Table  S12 ). The absence of a correlation with the SCCS sample is expected given that the SCCS sampling strategy intended to reduce autocorrelation in the data. All RDA p -values (Supplementary Data S 6 –S 8 ) are two-sided tests.

Changes to pre-registration

We registered a preliminary pre-registration of secondary data analysis, also available within the OSF archive. In the process of carrying out the analysis, our methods have changed substantially such that the current analyses should not be considered strictly pre-registered. The analyses here focus on RQ1 and RQ2 of the pre-registration, modelling the autocorrelational structure of the musical data. Our hypotheses of the major axis of musical style were expanded from three to five dimensions after realizing that our three originally proposed latent variables (social context, song structure, and singing style) were not the best variables to capture the primary dimensions of musical diversity (this change was done before analyzing correlations between these musical dimensions with genes, languages, or geography). The autocorrelational models proposed in RQ2 did not converge, therefore we shifted from a Bayesian to a frequentist modelling approach for the same hypotheses. We did not have time or space to sufficiently explore RQ3, which we will explore in future projects. The original pre-registration was registered on May 4, 2021, and can be found at https://doi.org/10.17605/OSF.IO/VE2DC .

Inclusion and ethics statement

This study uses publicly available data 12 , 41 and so did not require additional ethical approval. For information about ethics and inclusion in primary data collection, please see refs. 12 and 41 —particularly the “Ethics, Rights, and Consent” and “Inclusivity in Global Research” sections. As noted there: “Repatriation of Lomax’s recordings to their communities of origin, in partnership with those communities, is ongoing and has reached over 50 communities, descendants of artists, and national libraries. North American and Australian Indigenous audio samples will be streamed on the Jukebox only with the agreement of each community. To improve ethical practices, ACE [the Association for Cultural Equity] convenes with cultural advocates from diverse communities…. To further improve access to Lomax’s recordings and research, ACE engages with community arts leaders, artists and other culture bearers to connect their constituencies to the Global Jukebox and our online archive in meaningful ways. They are invited to contribute Journeys and Exhibits, correct metadata, interpret the songs, suggest new songs and codings, and add their documentation to the songs.” No identifiable information within the publicly available genetic dataset is available since all data is aggregated to the population level. Genetic data is used conforming to the associated informed consents and ethical permits, which allow the use of the data for studies of population history.

Reporting summary

Further information on research design is available in the  Nature Portfolio Reporting Summary linked to this article.

Data availability

All data processed and used in this study are accessible at https://doi.org/10.5281/zenodo.10817212 . All data is freely accessible. The processed data are available at the same address within the folder ‘processed_data’. Some results and data processing take significant computing time, so we keep pre-computed results in the same repository and folder. This project only utilises existing datasets. The sources of the data are as follows: The Global Jukebox 42 , with data accessible from https://github.com/theglobaljukebox/cantometrics ; GeLaTo 12 , with data accessible from https://github.com/gelato-org/gelato-data , additionally, the source of the population samples used are also listed in Supplementary Data  S5 ; The global language phylogeny 41 , with data accessible at https://osf.io/yzxv9/ . To listen to the audio, and read more detail on the Cantometric coding scheme visit http://theglobaljukebox.org . Please cite 42 if using Cantometrics, or other Global Jukebox data. Global Jukebox datasets are archived with ZENODO, and the DOI provided by ZENODO should be used when citing releases of Global Jukebox datasets, which are available within the GitHub organization.

Code availability

All necessary code for replicating our analyses has been deposited into the ZENODO repository: https://doi.org/10.5281/zenodo.10817212 . Each step of the analysis is detailed in chronological order within the Makefile.

Atkinson, Q. D. & Gray, R. D. Curious parallels and curious connections—phylogenetic thinking in biology and historical linguistics. Syst. Biol. 54 , 513–526 (2005).

Article   PubMed   Google Scholar  

Darwin, C. The Descent of Man, and Selection in Relation to Sex Reprint edn, 864, Vol 1 (Princeton University Press, 1871).

Atkinson, Q. D. Phonemic diversity supports a serial founder effect model of language expansion from Africa. Science 332 , 346–349 (2011).

Article   ADS   CAS   PubMed   Google Scholar  

Bouckaert, R., Bowern, C. & Atkinson, Q. D. The origin and expansion of pama–nyungan languages across Australia. Nat. Ecol. Evol. https://doi.org/10.1038/s41559-018-0489-3 (2018).

Cavalli-Sforza, L. L., Piazza, A., Menozzi, P. & Mountain, J. Reconstruction of human evolution: bringing together genetic, archaeological, and linguistic data. Proc. Natl Acad. Sci. USA 85 , 6002–6006 (1988).

Article   ADS   CAS   PubMed   PubMed Central   Google Scholar  

Diamond, J. & Bellwood, P. Farmers and their languages: the first expansions. Science 300 , 597–603 (2003).

Gray, R. D., Drummond, A. J. & Greenhill, S. J. Language phylogenies reveal expansion pulses and pauses in pacific settlement. Science 323 , 479–483 (2009).

Henn, B. M., Cavalli-Sforza, L. L. & Feldman, M. W. The great human expansion. Proc. Natl AcaS 109 , 17758–17764 (2012).

Article   ADS   CAS   Google Scholar  

Robbeets, M. et al. Triangulation supports agricultural spread of the transeurasian languages. Nature 599 , 616–621 (2021).

Grollemund, R. et al. Bantu expansion shows that habitat alters the route and pace of human dispersals. PNAS 112 , 13296–13301 (2015).

Koile, E., Greenhill, S. J., Blasi, D. E., Bouckaert, R. & Gray, R. D. Phylogeographic analysis of the Bantu language expansion supports a rainforest route. Proc. Natl Acad. Sci. USA 119 , e2112853119 (2022).

Article   CAS   PubMed   PubMed Central   Google Scholar  

Barbieri, C. et al. A global analysis of matches and mismatches between human genetic and linguistic histories. Proc. Natl Acad. Sci. USA 119 , e2122084119 (2022).

Boas, F. Race, Language and Culture New edn, 668, Vol 1 (Macmillan, 1940).

Donohue, M. & Denham, T. Farming and language in island Southeast Asia: reframing austronesian history. Curr. Anthropol. 51 , 223–256 (2010).

Article   Google Scholar  

Graeber, D. & Wengrow, D. The Dawn of Everything: A New History of Humanity 704, Vol 1 (Allen Lane / Penguin UK, 2021).

Kroeber, A. L. Cultural and Natural Areas of Native North America, 38, Vol 1 (Univ of California Press, 1939).

Terrell, J. History as a family tree, history as an entangled bank: constructing images and interpretations of prehistory in the South Pacific. Antiquity 62 , 642–657 (1988).

Brown, S. & Jordania, J. Universals in the world’s musics. Psychol. Music 41 , 229–248 (2013).

Jacoby, N. & McDermott, J. H. Integer ratio priors on musical rhythm revealed cross-culturally by iterated reproduction. Curr. Biol. 27 , 359–370 (2017).

Article   CAS   PubMed   Google Scholar  

Lomax, A. Folk Song Style and Culture 1st edn, 386, Vol 1 (American Association for the Advancement of Science, Washington, DC, 1968).

McDermott, J. H., Schultz, A. F., Undurraga, E. A. & Godoy, R. A. Indifference to dissonance in native Amazonians reveals cultural variation in music perception. Nature 535 , 547–550 (2016).

Mehr, S. A. et al. Universality and diversity in human song. Science 366 , aax0868 (2019).

Rzeszutek, T., Savage, P. E. & Brown, S. The structure of cross-cultural musical diversity. Proc. R. Soc. B: Biol. Sci. 279 , 1606–1612 (2012).

Savage, P. E., Brown, S., Sakai, E. & Currie, T. E. Statistical universals reveal the structures and functions of human music. Proc. Natl Acad. Sci. USA 112 , 8987–8992 (2015).

Szwed, J. F. Alan Lomax: The Man Who Recorded the World Reprint edn, 448, Vol 1 (Random House, 2011).

Brown, S. et al. Correlations in the population structure of music, genes and language. Proc. R. Soc. B: Biol. Sci. 281 , 20132072 (2014).

Grauer, V. Looking Backward: Music and Language in Deep history. Music in Human Experience: Perspectives on a Musical Species (Friedman, J. L. (eds)) 73–90, Vol 1 (Cambridge Scholars Publishing, 2022).

Leroi, A. M. & Swire, J. The recovery of the past. World Music 48 , 43–54 (2006).

Google Scholar  

Savage, P. E. Alan Lomax’s Cantometrics project: a comprehensive review. Music Sci. 1 , 2059204318786084 (2018).

Ozaki, Y., Kloots, M. de H., Ravignani, A. & Savage, P. E. Cultural evolution of music and language. PsyArXiv https://doi.org/10.31234/osf.io/s7apx (2023).

Patel, A. D. Music, Language, and the Brain 520, Vol 1. (Oxford University Press, Incorporated, NY, 2008).

Savage, P. E., Tierney, A. T. & Patel, A. D. Global music recordings support the motor constraint hypothesis for human and avian song contour. Music Percept. 34 , 327–334 (2017).

Feld, S. Sound structure as social structure. Ethnomusicology 28 , 383–409 (1984).

Nettl, B. Response to victor grauer: on the concept of evolution in the history of ethnomusicology. World Music 48 , 59–72 (2006).

Stock, J. P. J. Clues from our present peers?: a response to victor grauer. World Music 48 , 73–91 (2006).

Callaway, E. Music is in our genes. Nature https://doi.org/10.1038/news.2007.359 (2007).

Pamjav, H., Juhász, Z., Zalán, A., Németh, E. & Damdin, B. A comparative phylogenetic study of genetics and folk music. Mol. Genet Genom. 287 , 337–349 (2012).

Article   CAS   Google Scholar  

Matsumae, H. et al. Exploring correlations in genetic and cultural variation across language families in northeast Asia. Sci. Adv. https://doi.org/10.1126/sciadv.abd9223 (2021).

Article   PubMed   PubMed Central   Google Scholar  

Panteli, M., Benetos, E. & Dixon, S. A computational study on outliers in world music. PLoS One 12 , e0189399 (2017).

Daikoku, H. et al. Agreement among human and automated estimates of similarity in a global music sample . PsyArXiv https://doi.org/10.31234/osf.io/76fmq (2022).

Bouckaert, R. et al. Global language diversification is linked to socio-ecology and threat status. https://doi.org/10.31235/osf.io/f8tr6 (2022).

Wood, A. L. C. et al. The global jukebox: a public database of performing arts and culture. PLoS One 17 , e0275469 (2022).

Patterson, N. et al. Ancient admixture in human history. Genetics 192 , 1065–1093 (2012).

Harald Hammarström, Robert Forkel, Martin Haspelmath, & Sebastian Bank. (2024). glottolog/glottolog: Glottolog database 5.0 (v5.0) [Data set]. Zenodo. https://doi.org/10.5281/zenodo.10804357 .

Lomax, A. Factors of Musical Style. in In Theory and Practice: Essays presented to Gene Weltfish . 29–58 (ed. Diamond, S.) (Mouton, The Hague, 1980).

Kline, R. B. Principles and Practice of Structural Equation Modeling . Vol 1 (Guilford Publications, NY, 2015).

Sim, J. & Wright, C. C. The kappa statistic in reliability studies: use, interpretation, and sample size requirements. Phys. Ther. 85 , 257–268 (2005).

Excoffier, L., Smouse, P. E. & Quattro, J. M. Analysis of molecular variance inferred from metric distances among DNA haplotypes: application to human mitochondrial DNA restriction data. Genetics 131 , 479–491 (1992).

Ross, R. M., Greenhill, S. J. & Atkinson, Q. D. Population structure and cultural geography of a folktale in Europe. Proc. R. Soc. B: Biol. Sci. 280 , 20123065 (2013).

Rosenberg, N. A. et al. Genetic Structure of Human Populations. Science 298 , 2381–2385 (2002).

Handley, C. & Mathew, S. Human large-scale cooperation as a product of competition between cultural groups. Nat. Commun. 11 , 702 (2020).

Bromham, L., Hua, X., Cardillo, M., Schneemann, H. & Greenhill, S. J. Parasites and politics: why cross-cultural studies must control for relatedness, proximity and covariation. Royal society open . Science 5 , 181100 (2018).

Gray, R. D., Bryant, D. & Greenhill, S. J. On the shape and fabric of human history. Philos. Trans. R. Soc. B: Biol. Sci. 365 , 3923–3933 (2010).

Greenhill, S. J. et al. Evolutionary dynamics of language systems. PNAS 114 , E8822–E8829 (2017).

Holland, B. R., Huber, K. T., Dress, A. & Moulton, V. δ Plots: a tool for analyzing phylogenetic distance data. Mol. Biol. Evol. 19 , 2051–2059 (2002).

Harmon, L. J. & Glor, R. E. Poor statistical performance of the mantel test in phylogenetic comparative analyses. Evolution 64 , 2173–2178 (2010).

PubMed   Google Scholar  

Hilton, C. B. et al. Acoustic regularities in infant-directed speech and song across cultures. Nat. Hum. Behav. 6 , 1545–1556 (2022).

Ozaki, Y. et al. Globally, songs are slower and higher and use more stable pitches than speech: A registered report. Sci. Adv. 10 , adm9797 (2024).

Sachs, C. The History of Musical Instruments. Vol 1 (W W Norton & Co Inc, 1940).

Evans, N. Did language evolve in multilingual settings? Biol. Philos. 32 , 905–933 (2017).

Daikoku, H., Wood, A. L. & Savage, P. E. Musical diversity in India: a preliminary computational study using Cantometrics. Keio SFC J. 20 , 34–61 (2020).

Levinson, S. C. & Gray, R. D. Tools from evolutionary biology shed new light on the diversification of languages. Trends Cogn. Sci. 16 , 167–173 (2012).

Bickel, B., Giraud, A.-L., Zuberbühler, K. & van Schaik, C. P. Language follows a distinct mode of extra-genomic evolution. https://doi.org/10.31234/osf.io/nxf8c (2023).

Derex, M. & Mesoudi, A. Cumulative cultural evolution within evolving population structures. Trends Cogn. Sci. 24 , 654–667 (2020).

Savage, P. E. et al. Sequence alignment of folk song melodies reveals cross-cultural regularities of musical evolution. Curr. Biol. https://doi.org/10.1016/j.cub.2022.01.039 (2022).

Savage, P. E. et al. Music as a coevolved system for social bonding. Behav. Brain Sci. 44 , 1–22 (2021).

Norenzayan, A. et al. The cultural evolution of prosocial religions. Behav. Brain Sci. 39 , e1 (2016).

Turchin, P. et al. Disentangling the evolutionary drivers of social complexity: a comprehensive test of hypotheses. Sci. Adv. 8 , eabn3517 (2022).

Youngblood, M., Baraghith, K. & Savage, P. E. Phylogenetic reconstruction of the cultural evolution of electronic music via dynamic community detection (1975–1999). Evolution Hum. Behav. 42 , 573–582 (2021).

Youngblood, M., Ozaki, Y. & Savage, P. E. Cultural Evolution and Music. in The Oxford Handbook of Cultural Evolution (eds. Tehrani, J. J., Kendal, J. & Kendal, R.) C42S1–C42N14 https://doi.org/10.1093/oxfordhb/9780198869252.013.42 (Oxford University Press, 2023).

Hudson, M. et al. Rights, interests and expectations: indigenous perspectives on unrestricted access to genomic data. Nat. Rev. Genet 21 , 377–384 (2020).

Wood, A. L. C. Like a cry from the heart”: an insider’s view of the genesis of alan lomax’s ideas and the legacy of his research: part I. Ethnomusicology 62 , 230–264 (2018).

Jacoby, N. et al. Cross-cultural work in music cognition challenges, insights, and recommendations. Music Percept. 37 , 185–195 (2020).

Lomax, A. Mister Jelly Roll: The Fortunes of Jelly Roll Morton, New Orleans Creole and ‘Inventor of Jazz’ Vol. 368 (University of California Press, 1950).

Tilley, L. A. Making It Up Together: The Art of Collective Improvisation in Balinese Music and Beyond Vol. 336 (University of Chicago Press, 2019).

Wood, A. L. C. Musical practice and memory on the edge of two worlds: the kalymnian tsambóuna and song repertoire in the family of nikitas tsimouris. in Greek Music in America. Vol 1 96–153 (University Press of Mississippi, 2011).

Evans, C. L. et al. The uses and abuses of tree thinking in cultural evolution. Philos. Trans. R. Soc. B: Biol. Sci. 376 , 20200056 (2021).

Clark, A. G., Hubisz, M. J., Bustamante, C. D., Williamson, S. H. & Nielsen, R. Ascertainment bias in studies of human genome-wide polymorphism. Genome. Res. 15 , 1496–1502 (2005).

Skirgård, H. et al. Grambank reveals the importance of genealogical constraints on linguistic diversity and highlights the impact of language loss. Sci. Adv. 9 , eadg6175 (2023).

Albouy, P., Mehr, S. A., Hoyer, R. S., Ginzburg, J. & Zatorre, R. J. Spectro-temporal acoustical markers differentiate speech from song across cultures. Nat. Commun. (in the press).

Aguirre-Fernández, G. et al. Cultural macroevolution of musical instruments in South America. Humanit Soc. Sci. Commun. 8 , 1–12 (2021).

Tambets, K. et al. Genes reveal traces of common recent demographic history for most of the Uralic-speaking populations. Genome Biol. 19 , 139 (2018).

Turchin, P. et al. Big Gods and big science: further reflections on theory, data, and analysis. Relig., Brain Behav. 13 , 218–231 (2023).

Henrich, J. The WEIRDest People in the World: How the West Became Psychologically Peculiar and Particularly Prosperous (Penguin UK, 2020).

Buckley, C. D. & Boudot, E. The evolution of an ancient technology. Open Sci. 4 , 170208 (2017).

Lansing, J. S. et al. Kinship structures create persistent channels for language transmission. Proc Natl Acad. Sci. USA 114 , 1291–12915 (2017).

Barbieri, C., Butthof, A., Bostoen, K. & Pakendorf, B. Genetic perspectives on the origin of clicks in Bantu languages from southwestern Zambia. Eur. J. Hum. Genet. 21 , 430–436 (2013).

Arias, L., Barbieri, C., Barreto, G., Stoneking, M. & Pakendorf, B. High-resolution mitochondrial DNA analysis sheds light on human diversity, cultural interactions, and population mobility in Northwestern Amazonia. Am. J. Phys. Anthropol. 165 , 238–255 (2018).

Zhang, M., Zheng, H.-X., Yan, S. & Jin, L. Reconciling the father tongue and mother tongue hypotheses in Indo-European populations. Natl Sci. Rev. 6 , 293–300 (2019).

Jordan, P. & Shennan, S. Cultural transmission, language, and basketry traditions amongst the California Indians. J. Anthrop Archaeol. 22 , 42–74 (2003).

Tehrani, J. J. The phylogeny of little red riding hood. PLoS One 8 , e78871 (2013).

Article   ADS   PubMed   PubMed Central   Google Scholar  

Bortolini, E. et al. Inferring patterns of folktale diffusion using genomic data. Proc. Natl Acad. Sci. USA 114 , 9140–9145 (2017).

Hoeschele, M. & Fitch, W. T. Cultural evolution: conserved patterns of melodic evolution across musical cultures. Curr. Biol. 32 , R265–R267 (2022).

Brown, S. The Unification of the Arts: A Framework for Understanding What the Arts Share and Why , Vol 1 (Oxford University Press, 2021).

Lomax, A. in International Encyclopedia of Communications 230–233 (Oxford University Press, 1989).

Kirby, K. R. et al. D-PLACE: A global database of cultural, linguistic and environmental diversity. PLoS One 11 , e0158391 (2016).

Weir, B. S. & Cockerham, C. C. Estimating F-statistics for the analysis of population structure. Evolution 38 , 1358–1370 (1984).

Chang, C. C. et al. Second-generation PLINK: rising to the challenge of larger and richer datasets. GigaScience https://doi.org/10.1186/s13742-015-0047-8 (2015).

R Core Team. R: A Language and Environment for Statistical Computing. https://www.R-project.org/ (2021).

Rosseel, Y. lavaan: An R package for structural equation modeling. J. Stat. Softw. 48 , 1–36 (2012).

Dray, S. & Dufour, A.-B. The ade4 package: implementing the duality diagram for ecologists. J. Stat. Softw. 22 , 1–20 (2007).

Aktas, C. Haplotypes: Manipulating DNA Sequences and Estimating Unambiguous Haplotype Network with Statistical Parsimony . https://CRAN.R-project.org/package=haplotypes (2020).

Peakall, R. & Smouse, P. E. genalex 6: genetic analysis in excel. population genetic software for teaching and research. Mol. Ecol. Notes 6 , 288–295 (2006).

Oksanen, J. et al. Vegan: Community Ecology Package. R-Package Version 2.0-10. https://github.com/vegandevs/vegan (2013).

Download references

Acknowledgements

We thank all the individuals and researchers who previously contributed and curated the genetic, musical, and linguistic data. We thank Alan Lomax, Luca Cavalli-Sforza, Victor Grauer, Steven Brown, Sarah Tishkoff, Floyd Reed, and Armand Leroi for inspiration and discussion about comparing global patterns of musical and genetic diversity. We thank Russell Gray, Shinya Fujii, and members of the CompMusic Lab, NeuroMusic Lab, and Language, Culture, and Cognition Lab for feedback on earlier versions of the manuscript. The Global Jukebox has been developed with support from the National Endowment for the Arts, the National Endowment for the Humanities, the Concordia Foundation, the Rock Foundation, and Odyssey Productions. SP, HD, and PES are supported by funding from the Japan Society for the Promotion of Science (Grant-in-Aid #19KK0064); the Yamaha corporation; and grants from Keio University (Keio Global Research Institute and Keio Academic Development Fund). SP is also supported by the Evolution of Cultural Diversity Initiative at the Australian National University. PES is also supported by the Royal Society Te Apārangi (Rutherford Discovery Fellowship RDF-UOA2202 and Marsden Fast-Start Grant MFP-UOA2236). CB was supported by the University Research Priority Programme of Evolution in Action of the University of Zurich, the NCCR Evolving Language, the Swiss National Science Foundation Agreement (#51NF40_180888), and the SNSF Sinergia project ‘Out of Asia’ (183578). Article Processing Charges are supported by the University of Auckland Faculty of Science Open Access High Impact Publication Fund. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Author information

Authors and affiliations.

Graduate School of Media and Governance, Keio University, Fujisawa, Japan

Sam Passmore & Hideo Daikoku

Evolution of Cultural Diversity Initiative (ECDI), Australian National University, Canberra, Australia

Sam Passmore

Association for Cultural Equity, New York, USA

Anna L. C. Wood

Department of Evolutionary Biology and Environmental Studies, University of Zurich, Zurich, 8057, Switzerland

Chiara Barbieri

Centre for the Interdisciplinary Study of Language Evolution (ISLE), University of Zurich, Zurich, 8050, Switzerland

Department of Life and Environmental Sciences, University of Cagliari, 09126, Cagliari, Italy

Cohn Institute for the History and Philosophy of Science and Ideas, Tel Aviv University, Tel Aviv, Israel

Dor Shilton

Edelstein Centre for the History and Philosophy of Science, Technology, and Medicine, Hebrew University of Jerusalem, Jerusalem, Israel

School of Psychology, University of Auckland, Auckland, New Zealand

Quentin D. Atkinson & Patrick E. Savage

Faculty of Environment and Information Studies, Keio University, Fujisawa, Japan

Patrick E. Savage

You can also search for this author in PubMed   Google Scholar

Contributions

PES, SP, and ALCW conceived the project. SP conducted the analyses with methodology recommended by PES, CB, QDA, ALCW, and DS. CB provided and curated the genetic data. SP, CB, DS, and HD curated, matched, and cleaned the data. HD conducted a code review. SP wrote the initial draft, with substantial revisions by PES and ALCW. CB, DS, and QDA contributed to the final draft.

Corresponding authors

Correspondence to Sam Passmore , Anna L. C. Wood or Patrick E. Savage .

Ethics declarations

Competing interests.

The authors declare no competing interests.

Peer review

Peer review information.

Nature Communications thanks Tudor Popescu and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Supplementary information, peer review file, description of additional supplementary files, supplementary data 1–8, reporting summary, rights and permissions.

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ .

Reprints and permissions

About this article

Cite this article.

Passmore, S., Wood, A.L.C., Barbieri, C. et al. Global musical diversity is largely independent of linguistic and genetic histories. Nat Commun 15 , 3964 (2024). https://doi.org/10.1038/s41467-024-48113-7

Download citation

Received : 28 June 2023

Accepted : 19 April 2024

Published : 10 May 2024

DOI : https://doi.org/10.1038/s41467-024-48113-7

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

By submitting a comment you agree to abide by our Terms and Community Guidelines . If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.

Quick links

• Explore articles by subject
• Guide to authors
• Editorial policies

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Music Genre Classification Revisited: An In-Depth Examination Guided by Music Experts

• Conference paper
• First Online: 24 November 2018
• Cite this conference paper

• Haukur Pálmason 17 ,
• Björn Þór Jónsson 17 , 18 ,
• Markus Schedl 19 &
• Peter Knees 20  

Part of the book series: Lecture Notes in Computer Science ((LNISA,volume 11265))

Included in the following conference series:

• International Symposium on Computer Music Multidisciplinary Research

1169 Accesses

5 Citations

Despite their many identified shortcomings, music genres are still often used as ground truth and as a proxy for music similarity. In this work we therefore take another in-depth look at genre classification, this time with the help of music experts. In comparison to existing work, we aim at including the viewpoint of different stakeholders to investigate whether musicians and end-user music taxonomies agree on genre ground truth , through a user study among 20 professional and semi-professional music protagonists. We then compare the results of their genre judgments with different commercial taxonomies and with that of computational genre classification experiments, and discuss individual cases in detail. Our findings coincide with existing work and provide further evidence that a simple classification taxonomy is insufficient.

This is a preview of subscription content, log in via an institution to check access.

Access this chapter

Subscribe and save.

• Get 10 units per month
• Download Article/Chapter or eBook
• 1 Unit = 1 Article or 1 Chapter
• Cancel anytime
• Available as PDF
• Read on any device
• Instant download
• Own it forever
• Available as EPUB and PDF
• Compact, lightweight edition
• Dispatched in 3 to 5 business days
• Free shipping worldwide - see info

Tax calculation will be finalised at checkout

Purchases are for personal use only

Institutional subscriptions

Similar content being viewed by others

Music Genre Classification: A Semi-supervised Approach

Music Overflow: A Music Genre Classification Web Application

Machine learning for music genre: multifaceted review and experimentation with audioset

George Tzanetakis, the author of the dataset, has repeatedly confirmed being aware of these issues, but has chosen not to correct them since the dataset has been used so many times and changing the files would render comparisons of results infeasible.

The genre “classical” was included even though no track was classified as such.

Aucouturier, J.J., Pachet, F.: Representing musical genre: a state of the art. J. New Music. Res. 32 (1), 83–93 (2003)

Article   Google Scholar  

Aucouturier, J.J., Pachet, F., Roy, P., Beurivé, A.: Signal + context = better classification. In: Proceedings of the 8th International Conference on Music Information Retrieval (ISMIR), Vienna, Austria (2007)

Google Scholar  

Craft, A., Wiggins, G., Crawford, T.: How many beans make five? The consensus problem in music-genre classification and a new evaluation method for single-genre categorisation systems. In: Proceedings of the 8th International Symposium on Music Information Retrieval (ISMIR), Vienna, Austria (2007)

Drott, E.: The end(s) of genre. J. Music. Theory 57 (1), 1–45 (2013)

Fabbri, F.: A theory of musical genres: two applications. Popul. Music. Perspect. 1 , 52–81 (1981)

Gjerdingen, R.O., Perrott, D.: Scanning the dial: the rapid recognition of music genres. J. New Music. Res. 37 (2), 93–100 (2008)

Guðmundsson, G.Þ., Jónsson, B.Þ., Amsaleg, L.: A large-scale performance study of cluster-based high-dimensional indexing. In: ACM Multimedia Workshop on Very-Large-Scale Multimedia Corpus, Mining and Retrieval, Florence, Italy (2010)

Krumhansl, C.L.: Plink: “Thin Slices” of music. Music. Percept. Interdiscip. J. 27 (5), 337–354 (2010)

Lamere, P.: Social tagging and music information retrieval. J. New Music. Res. 37 (2), 101–114 (2008). Special Issue: From Genres to Tags - Music Information Retrieval in the Age of Social Tagging

Lippens, S., Martens, J.P., De Mulder, T., Tzanetakis, G.: A comparison of human and automatic musical genre classification. In: Proceedings of the IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) (2004)

Lo, H.Y., Wang, J.C., Wang, H.M., Lin, S.D.: Cost-sensitive multi-label learning for audio tag annotation and retrieval. IEEE Trans. Multimed. 13 (3), 518–529 (2011)

Mandel, M.I., et al.: Contextual tag inference. ACM Trans. Multimed. Comput. Commun. Appl. 7S (1), 32:1–32:18 (2011)

Marques, G., Domingues, M.A., Langlois, T., Gouyon, F.: Three current issues in music autotagging. In: Proceedings of the 12th International Society for Music Information Retrieval Conference (ISMIR), pp. 795–800 (2011)

McKay, C., Fujinaga, I.: Musical genre classification: is it worth pursuing and how can it be improved? In: Proceedings of the 7th International Conference on Music Information Retrieval (ISMIR), Victoria, BC, Canada (2006)

Novello, A., McKinney, M.F., Kohlrausch, A.: Perceptual evaluation of music similarity. In: Proceedings of the 7th International Conference on Music Information Retrieval (ISMIR), Victoria, BC, Canada (2006)

Pachet, F., Cazaly, D.: A taxonomy of musical genre. In: Proceedings of Content-Based Multimedia Information Access (RIAO) Conference, Paris, France (2000)

Scaringella, N., Zoia, G., Mlynek, D.: Automatic genre classification of music content: a survey. IEEE Signal Process. Mag. 23 (2), 133–141 (2006)

Schedl, M., Flexer, A., Urbano, J.: The neglected user in music information retrieval research. J. Intell. Inf. Syst. 41 , 523–539 (2013)

Seyerlehner, K., Widmer, G., Knees, P.: A comparison of human, automatic and collaborative music genre classification and user centric evaluation of genre classification systems. In: Detyniecki, M., Knees, P., Nürnberger, A., Schedl, M., Stober, S. (eds.) AMR 2010. LNCS, vol. 6817, pp. 118–131. Springer, Heidelberg (2012). https://doi.org/10.1007/978-3-642-27169-4_9

Chapter   Google Scholar  

Sturm, B.L.: An analysis of the GTZAN music genre dataset. In: Proceedings of the 2nd International ACM Workshop on Music Information Retrieval with User-centered and Multimodal Strategies (MIRUM), Nara, Japan (2012)

Sturm, B.L.: The state of the art ten years after a state of the art: future research in music information retrieval. J. New Music. Res. 43 (2), 147–172 (2014)

Turnbull, D., Barrington, L., Torres, D., Lanckriet, G.: Towards musical query-by-semantic-description using the CAL500 data set. In: Proceedings of the 30th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR), Amsterdam, the Netherlands (2007)

Turnbull, D., Barrington, L., Torres, D., Lanckriet, G.: Semantic annotation and retrieval of music and sound effects. IEEE Trans. Audio Speech Lang. Process. 16 (2), 467–476 (2008)

Tzanetakis, G., Cook, P.: Musical genre classification of audio signals. IEEE Trans. Audio, Speech, Lang. Process. 10 (5), 293–302 (2002)

Wang, S.Y., Wang, J.C., Yang, Y.H., Wang, H.M.: Towards time-varying music auto-tagging based on cal500 expansion. In: Proceedings of the 2014 IEEE International Conference on Multimedia and Expo (ICME) (2014)

Download references

Acknowledgements

Supported by the Austrian Science Fund (FWF): P25655 and the Austrian FFG: BRIDGE 1 project SmarterJam (858514).

Author information

Authors and affiliations.

Reykjavik University, Reykjavík, Iceland

Haukur Pálmason & Björn Þór Jónsson

IT University of Copenhagen, Copenhagen, Denmark

Björn Þór Jónsson

Johannes Kepler University Linz, Linz, Austria

Markus Schedl

TU Wien, Vienna, Austria

Peter Knees

You can also search for this author in PubMed   Google Scholar

Corresponding author

Correspondence to Peter Knees .

Editor information

Editors and affiliations.

Laboratoire PRISM, AMU-CNRS, Marseille, France

Mitsuko Aramaki

INESC TEC, Porto, Portugal

Matthew E. P. Davies

Richard Kronland-Martinet

Sølvi Ystad

Rights and permissions

Reprints and permissions

Copyright information

© 2018 Springer Nature Switzerland AG

About this paper

Cite this paper.

Pálmason, H., Jónsson, B.Þ., Schedl, M., Knees, P. (2018). Music Genre Classification Revisited: An In-Depth Examination Guided by Music Experts. In: Aramaki, M., Davies , M., Kronland-Martinet, R., Ystad, S. (eds) Music Technology with Swing. CMMR 2017. Lecture Notes in Computer Science(), vol 11265. Springer, Cham. https://doi.org/10.1007/978-3-030-01692-0_4

Download citation

DOI : https://doi.org/10.1007/978-3-030-01692-0_4

Published : 24 November 2018

Publisher Name : Springer, Cham

Print ISBN : 978-3-030-01691-3

Online ISBN : 978-3-030-01692-0

eBook Packages : Computer Science Computer Science (R0)

Share this paper

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Publish with us

Policies and ethics

• Find a journal
• Track your research

Music encyclopedias

Articles, dissertations, conference proceedings, the four r-projects of international music scholarship, music research portals.

• Streaming audio and video
• Find music scores
• Online scores
• Musical images, manuscript sources, and musical instruments
• Music theory study aids
• Citations and writing guides

Performing Arts Librarian

Ask Laura Stokes

• Library of Congress: Performing Arts Encyclopedia The Performing Arts Encyclopedia is a collection of performing arts materials located at the Library of Congress.
• Rock's Backpages 1960-present; Rock's Backpages is a database of rock music writing, sourced from the pages of the music and mainstream press. Thousands of articles (reviews, interviews, features and more) for over 2,000 artists are available.
• MusicID This link opens in a new window Provides access to perpetually-updated charts information, including Billboard charts, iTunes downloads, Spotify and Apple Music streams, and Shazam searches. Also includes a revenue calculation tool. more... less... Current; Music ID is an academic research tool that utilizes powerful analytics to compile global music industry data. Incorporating 5,452 different charts spanning 74 countries, Music ID Data provides access to perpetually-updated, week-to-week information on iTunes downloads, Spotify and Apple Music streams, and Shazam searches, as well as contemporary and historical data from Billboard, the Official Charts Company, GfK, and more.
• RISM (Répertoire internationale des sources musicales) RISM's catalog is a finding aid for music manuscripts and prints; the online catalog includes Series A/1 and A/II (Music Manuscripts after 1600) and parts of B/1 as well as a list of library sigla. Other components can be found in print in the Orwig Music Library. more... less... http://www.rism.info/en/publications.html
• RIdIM: Répertoire International d'Iconograpie Musicale (International Repertoire of Musical Iconography) The RId'IM database contains descriptions of visual objects that depict music-making, musical instruments, and similar related objects. Where possible, there are links to online images of the artwork.
• Online Resources for Music Scholars (Harvard University) The electronic research guide from Harvard's Loeb Music Library, with many free resources covering musicology, music theory, composition, ethnomusicology, and more.
• MusRef (Brigham Young University) MusRef, a portal from the music library at Brigham Young University, includes an online bibliography of reference sources for music, both print and online, as well as guides for getting started with music reference sources. This site includes a huge collection of links to the vast world of freely available information about music on the web.
• Digital Resources for Musicology (DRM) A portal to open-access projects that may be of use to musicians and musicologists.
• << Previous: Home
• Next: Streaming audio and video >>
• Last Updated: May 2, 2024 12:42 PM
• URL: https://libguides.brown.edu/music

Brown University Library  |  Providence, RI 02912  |  (401) 863-2165  |  Contact  |  Comments  |  Library Feedback  |  Site Map

Library Intranet

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here .

Loading metrics

Open Access

Peer-reviewed

Research Article

Instrumentational Complexity of Music Genres and Why Simplicity Sells

Affiliation Section for Science of Complex Systems, CEMSIIS, Medical University of Vienna, Austria

* E-mail: [email protected]

Affiliations Section for Science of Complex Systems, CEMSIIS, Medical University of Vienna, Austria, Santa Fe Institute, Santa Fe, New Mexico, United States of America, IIASA, Laxenburg, Austria

• Gamaliel Percino, 
• Peter Klimek, 
• Stefan Thurner

• Published: December 31, 2014
• https://doi.org/10.1371/journal.pone.0115255
• Reader Comments

Listening habits are strongly influenced by two opposing aspects, the desire for variety and the demand for uniformity in music. In this work we quantify these two notions in terms of instrumentation and production technologies that are typically involved in crafting popular music. We assign an ‘instrumentational complexity value’ to each music style. Styles of low instrumentational complexity tend to have generic instrumentations that can also be found in many other styles. Styles of high complexity, on the other hand, are characterized by a large variety of instruments that can only be found in a small number of other styles. To model these results we propose a simple stochastic model that explicitly takes the capabilities of artists into account. We find empirical evidence that individual styles show dramatic changes in their instrumentational complexity over the last fifty years. ‘New wave’ or ‘disco’ quickly climbed towards higher complexity in the 70s and fell back to low complexity levels shortly afterwards, whereas styles like ‘folk rock’ remained at constant high instrumentational complexity levels. We show that changes in the instrumentational complexity of a style are related to its number of sales and to the number of artists contributing to that style. As a style attracts a growing number of artists, its instrumentational variety usually increases. At the same time the instrumentational uniformity of a style decreases, i.e. a unique stylistic and increasingly complex expression pattern emerges. In contrast, album sales of a given style typically increase with decreasing instrumentational complexity. This can be interpreted as music becoming increasingly formulaic in terms of instrumentation once commercial or mainstream success sets in.

Citation: Percino G, Klimek P, Thurner S (2014) Instrumentational Complexity of Music Genres and Why Simplicity Sells. PLoS ONE 9(12): e115255. https://doi.org/10.1371/journal.pone.0115255

Editor: Dante R. Chialvo, National Scientific and Technical Research Council (CONICET), Argentina

Received: August 18, 2014; Accepted: November 20, 2014; Published: December 31, 2014

Copyright: © 2014 Percino et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License , which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Data Availability: The authors confirm that all data underlying the findings are fully available without restriction. All relevant data are within the paper and its Supporting Information files.

Funding: P.K. was supported by EU FP7 project MULTIPLEX, No. 317532, and G.P. by the National Council for Science and Technology of Mexico with the scholarship number 202117. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Competing interests: The authors have declared that no competing interests exist.

Introduction

The composer Arnold Schönberg held that joy or excitement in listening to music originates from the struggle between two opposing impulses, ‘the demand for repetition of pleasant stimuli, and the opposing desire for variety, for change, for a new stimulus.’ [1] . These two driving forces – the demand for repetition or uniformity and the desire for variety – influence not only how we perceive popular music, but also how it is produced. This can be seen e.g. in one of last year's critically most acclaimed albums, Daft Punk's Random Access Memories . At the beginning of the production process of the album the duo behind Daft Punk felt that the electronic music genre was in its ‘comfort zone and not moving one inch’ [2] . They attributed this ‘identity crisis’ to the fact that artists in this genre mostly miss the tools to create original sounds and rely too heavily on computers with the same libraries of sounds and preset banks [3] . Random Access Memories was finally produced with the help of 27 other featured artists or exceptional session musicians, who were asked to play riffs and individual patterns to give the duo a vast library to select from [4] . The percussionist stated that he used ‘every drum he owns’ on the album; there is also a track composed of over 250 different elements. The record was awarded the ‘Album of the Year 2013’ Grammy and received a Metacritic review of ‘universal acclaim’ for, e.g. ‘breath[ing] life into the safe music that dominates today's charts’ [5] . However, the best-selling album of 2013 in the US was not from Daft Punk, but The 20/20 Experience by Justin Timberlake. The producer of this album, Timothy Mosley, contributed 25 Billboard Top 40 singles between 2005–2010, more than any other producer [6] . All these records featured a unique production style consisting of ‘vocal sounds imitating turntable scratching, quick keyboard arabesques, grunts as percussion’ [7] . Asked about his target audience, Mosley said ‘I know where my bread and butter is at. […] I did this research. It's the women who watch Sex and the City’ [8] . These two anecdotes illustrate how Schönberg's two opposing forces, the demand for both uniformity and variety, influence the crafting of popular music. The Daft Punk example suggests that innovation and increased variety is closely linked to the involved musicians' skills and thereby to novel production tools and technologies. The example of Mosley shows how uniformity in stylistic expressions can satisfy listener demands and produce large sales numbers over an extended period of time. There is indeed substantial evidence now that it is the delicate balance between repetitiveness and surprise that shapes our emotional responses to music [9] , [10] .

The complexity of music is a multi-faceted concept [11] . Aspects of this complexity that are amenable to a quantitative evaluation include acoustics (the dynamic range and the rate of change in dynamic levels of audio tracks), timbre (the source of the sound and the way that this source is excited), as well as complexity measures for the melodic, harmonic, and rhythmic content of music (that are often based on time-frequency analyses) [12] . The so-called ‘optimal complexity hypothesis’ suggests that audiences prefer music of intermediate perceived complexity [13] , as has recently been experimentally confirmed for modern jazz piano improvisations [14] . It is worth to note that commercial success or popularity of music (as measured by the numbers of sales or listeners, respectively) is not determined by quality or complexity of music alone [15] . The number of record sales of a given artist is in general also not correlated with the record sales of similar artists [16] . In an ‘artificial music market’ it has been shown that success is determined by social influence, i.e. people showed the tendency to prefer music that they perceived was also preferred by many other listeners [15] . Music preferences are also shaped by nationality, language, and geographic location [17] . Interestingly, a geographic flow of music has been detected between cities, where some of them consistently act as early adopters of new music [17] . Over the last fifty years popular music experienced growing homogenization over time with respect to timbre [18] , which is the fingerprint of musical instruments and was found to exhibit similar statistical properties as speech [19] . Another important application of a quantitative evaluation of trends in the music industry is the development of music recommendation systems that are based on the similarity of artists [20] , or on collective listening habits of users of online music databases [21] – [24] .

Here we assume that instrumentational complexity of a style is related to the set of specialized skills that are typically required of musicians to play that style. Instrumentational complexity of a style increases with (i) the number of skills required for the style and (ii) the degree of specialization of these skills. A highly complex music style, in terms of instrumentation, requires a diverse set of skills that are only relevant for a small number of other styles. A style of low instrumentational complexity requires only a small set of generic and ubiquitous skills, that can be found in a large number of other styles. If a music style requires a highly diverse set of skills, this will to some degree also be reflected in a higher number of different instruments and production technologies. In general, demand for variety translates into a larger number of instruments used in the production process. Desire for uniformity favors a limited variability in instrumentation in a production. Music styles with high instrumentational complexity therefore have large instrumentational variety and at the same time low instrumentational uniformity. It follows that the desire for variety and uniformity are not only relevant for the perception of musical patterns. The notions of variety and uniformity also apply to the instrumentations that musicians use for their pieces. Note that instrumentational complexity can be regarded as a timbral complexity measure and is not informative about, for example, rhythmic, tonal, melodic, or acoustic complexity [12] .

In this work we quantify the variety and uniformity of music styles in terms of instrumentation that is typically used for their production. We employ a user-generated music taxonomy where albums are classified as belonging to one of fifteen different music genres that contain 374 different music styles as subcategories. Styles belonging to the same music genre are characterized by similar instrumentation, a fact that has already been exploited in the context of automatic genre detection [25] . We construct a similarity network of styles, whose branches are identified as music genres. We characterize the instrumentational complexity of each music style by its instrumentational variety and uniformity and show (i) that there is a remarkable relationship between instrumentational varieties and uniformities of music styles, (ii) that the instrumentational complexity of individual styles may exhibit dramatic changes across the past fifty years, and (iii) that these changes in instrumentational complexity are related to the typical sales numbers of the music style.

Music styles and genres are characterized by their use of instruments

We introduce a time-dependent bipartite network connecting music styles to the instruments that are typically used in that style. The dataset is extracted from the online music database Discogs and contains music albums and information on which artists are featured in the album, which instruments these artists play, the release date of the album, and the classifications of music genres and styles of the album. For more information see the methods section and S1 Table in S1 File .

• PPT PowerPoint slide
• PNG larger image
• TIFF original image

https://doi.org/10.1371/journal.pone.0115255.g001

https://doi.org/10.1371/journal.pone.0115255.g002

From instrumentational variety and uniformity to complexity

https://doi.org/10.1371/journal.pone.0115255.g003

Complexity-lifecycles of music styles

However, the position of individual styles can change dramatically over time, as it is shown for ‘indie rock’, ‘new wave’, ‘disco’ and ‘synth-pop’. Some styles, such as ‘folk’, show almost no change in their position.

https://doi.org/10.1371/journal.pone.0115255.g004

https://doi.org/10.1371/journal.pone.0115255.g005

A simple model

Materials and Methods

To measure the average sales numbers of music styles we use a dataset that contains information on the Amazon SalesRank of music albums as of 2006 [27] . The Amazon SalesRank can be thought of as a ranking of all records by the time-span since an item last sold [28] . Albums in the Discogs dataset are assigned their Amazon SalesRank by matching album titles between the two datasets. As the Amazon SalesRank dataset only contains information on album titles, it was matched to entries in the Discogs dataset by choosing only albums whose title appears only once in both datasets.

Style similarity network

Goodness-of-fit between data and model

Supporting Information

Supporting information.

https://doi.org/10.1371/journal.pone.0115255.s001

Acknowledgments

We thank Álvaro Corral for valuable feedback to an early draft of this manuscript and Joan Serrà for many helpful comments.

Author Contributions

Conceived and designed the experiments: GP PK ST. Performed the experiments: GP. Analyzed the data: GP PK. Contributed reagents/materials/analysis tools: GP PK ST. Wrote the paper: GP PK ST.

• 1. Schönberg A (1978) Theory of harmony. London: Faber & Faber.
• 2. Weiner J (13 April 2013) Daft Punk reveal secrets of new album - exclusive. Rolling Stone.
• 3. Mason K (06 May 2013) Daft Punk on EDM producers: ‘they're missing the tools’. Billboard Magazine.
• 4. Daft Punk (2013) Random Access Memories (liner notes). Columbia Records: Sony Music Entertainment.
• 5. MetaCritic (22 Nov 2014) Daft Punk - Random Access Memories. Available: http://www.metacritic.com/music/random-access-memories/daft-punk . Accessed: 2014 Nove 22.
• 6. Phillips R (08 Sep 2010) Timbaland logs most Top 40 Billboard Hot 100 Singles since '05. HiphopDX. Available: http://www.hiphopdx.com/index/news/id.12165/title.timbaland-logs-most-top-40-billboard-hot-100-singles-since-05 . Accessed: 2014 Nov 22.
• 7. Parales J (13 Dec 2009) Critic's choice: Timbaland presents Shock Value II. The New York Times.
• 8. Reid S (11 Dec 2009) Timbaland says he's done with hip-hop. MTV News. Available: http://www.mtv.com/news/1628135/timbaland-says-hes-done-with-hip-hop/ . Accessed: 2014 Nov 22.
• 9. Huron D (2006) Sweet anticipation: music and the psychology of expectation. Cambridge: MIT Press.
• 10. Ball P (2010) The music instinct: how music works and why we can't do without it. London: Bodley Head.
• View Article
• Google Scholar
• 12. Streich S (2006) Music complexity: a multi-faceted description of audio content. PhD Thesis, Universitat Pompeu Fabra.
• 17. Lee C, Cunningham P (2012) The geographic flow of music. ASONAM'12 Proceedings of the 2012 International Conference on Advances in Social Networks Analysis and Mining, pages 691–695.
• 22. Celma O, Cano P (2008) From hits to niches? Or how popular artists can bias music recommendation and discovery. 2nd Workshop on Large-Scale Recommender Systems and the Netflix Prize Competition (ACM KDD).
• 28. Rosenthal M (2013) Amazon sales ranking and author rank. Available: http://www.fonerbooks.com/surfing.htm . Accessed: 2014 Nov 22.

How to Explore and Experiment with Different Music Genres

October 3, 2023.

As a musician, exploring and experimenting with various music genres can be a creative and career-enhancing endeavor. Here are five steps to help you break boundaries and dive into new musical territories:

• Research and Listen: Start by researching different music genres. Listen to a wide range of artists and tracks within those genres. Make notes about the characteristics, instruments, and vocal styles that define each one.
• Collaborate with Diverse Musicians: Collaborating with musicians from various backgrounds can be eye-opening. Join forces with artists who specialize in genres different from your own. Their insights and skills can broaden your musical horizons.
• Music Theory and Workshops: Invest in learning the theory behind different genres. Attend workshops, online courses, or work with a music teacher who specializes in the genre you want to explore. Understanding the fundamentals is key to successful experimentation.
• Genre Fusion: Experiment with genre fusion. Combine elements from different genres to create a unique sound. This innovative approach can lead to groundbreaking compositions and attract a diverse fan base.
• Live Performances: Take your experimentation to the stage. Perform live and gauge the audience's response. Their feedback will guide your exploration. Be open to adjusting your approach based on their reactions.

Exploring different genres can ignite your creativity and introduce you to exciting musical possibilities. Dive into research, collaborate with diverse musicians, study music theory, experiment with fusion, and bring your explorations to life on stage. Your musical journey will become richer and more captivating.

Rivet Electric Works, Inc. © 2023

An official website of the United States government

The .gov means it’s official. Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

The site is secure. The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

• Publications
• Account settings

Preview improvements coming to the PMC website in October 2024. Learn More or Try it out now .

• Advanced Search
• Journal List

Genre Complexes in Popular Music

Daniel silver.

1 Department of Sociology, University of Toronto Scarborough, Toronto, Ontario, Canada

2 Department of Sociology, University of Chicago, Chicago, Illinois, United States of America

C. Clayton Childress

Analyzed the data: ML DS. Wrote the paper: DS ML CC. Developed conceptual framework: DS ML.

Associated Data

All data files are available from the dataverse database (Url: http://hdl.handle.net/10864/11402 ; Study Global ID: hdl:10864/11402).

Recent work in the sociology of music suggests a declining importance of genre categories. Yet other work in this research stream and in the sociology of classification argues for the continued prevalence of genres as a meaningful tool through which creators, critics and consumers focus their attention in the topology of available works. Building from work in the study of categories and categorization we examine how boundary strength and internal differentiation structure the genre pairings of some 3 million musicians and groups. Using a range of network-based and statistical techniques, we uncover three musical “complexes,” which are collectively constituted by 16 smaller genre communities. Our analysis shows that the musical universe is not monolithically organized but rather composed of multiple worlds that are differently structured—i.e., uncentered, single-centered, and multi-centered.

Introduction

How do popular artists form their public identities by mobilizing existing stylistic forms? Strong evidence suggests that “categorical imperatives” [ 1 ] impose penalties on producers for illegitimate role performance, especially when performance is evaluated by critics and discriminating audiences, as it is in the music industry. Much research, moreover, argues that musical genre expectations in particular profoundly organize the music industry [ 2 – 4 ], shaping how band members meet [ 5 – 6 ], producers choose and venues book bands [ 7 – 8 ], radio stations choose what to play [ 9 – 10 ], record label divisions are organized, music news is reported, as well as how fans find music to enjoy and people to enjoy it with [ 4 ]. As such, genre designations and expectations provide crucial reference points that inform the way musicians construct their public presentation of self.

At the same time, other research indicates that (some) genre expectations are weakening [ 11 – 12 ], being more flexibly reimagined [ 13 ], being redefined as search algorithms (such as Pandora’s Music Genome Project) that create new ways to sort music independent of style, and even fading away in some contexts to the point that major digital musical stores like iTunes barely mention genre [ 10 ]. These transformations may in turn reconfigure the traditional genre frameworks through which musicians present themselves to their various audiences; rather than being a fixed and static system, genres emerge, evolve, and change over time [ 5 ].

Building on this research, we examine the structure of genre self-classification by popular musicians. We propose that big data sources such as MySpace.com make it possible to empirically and comprehensively evaluate debates about the strength and types of genre classification at work in popular music. Specifically, we ask:

• Which, if any, genre conventions structure popular music?
• How does the strength and structure of genre conventions differ across musical styles?

We pursue these questions in the context of two related literatures, one on music specifically and the other on the sociology of classification more generally.

We start in the sociology of music, where a pressing question concerns the extent to which traditional genre categories continue to structure the social production of music in the face of various pressures toward more flexible modes of categorization. We join this literature with ideas from the sociology of classifications. This literature helps us to move beyond the binary question of whether categories are strengthening or weakening to more fruitful questions about how various boundary characteristics, such as their clarity or scope, interrelate.

Following DiMaggio [ 14 ], we suggest ritual strength and differentiation as key dimensions for cultural classification, and treat these as different but complementary aspects of genre classification. Crossing these two dimensions produces a four-fold typology of what we call “genre complexes”: multi-centered , uncentered , single-centered , and free interchangeability . Our analysis of some 3 million musician profiles on MySpace.com indicates that Rock musicians categorize themselves in a multi-centered way, Hip Hop musicians in a single-centered way, and musicians in non-commercial or “niche” genres in an uncentered way; free-interchangeability, although theoretically possible, was not present in our data.

The primary goals of this paper are thus three-fold. A first contribution is theoretical-synthetic : we review existing literature on both genre and classificatory structure more broadly and sort key positions in the field into a more generalizable typology of theoretically possible intersections of two simple principles, differentiation and boundary strength . A second contribution is methodological : we demonstrate how to use network-analytical techniques to discern aggregate patterns of individual musicians’ genre choices from big data sources. A third contribution is empirical : we report on the structure of MySpace.com users’ genre self-presentation as of 2007 and show that the resultant patterns closely approximate fields within our theoretically derived typology. While much sociological work examines genre boundaries as they intersect with industry imperatives and audience demand, we investigate musical genres as a self-contained relational system and aim to illuminate its structural principles and aggregate patterns. We conclude with suggestions for extending this research further, in particular by examining how characteristics of musicians’ genre categorization relate to their popularity and their surrounding social contexts.

Genres in the System of Music Production

Musicological vs. sociological conceptions of music genre.

Literary and music scholars often treat genres as common stylistic elements that differentiate classes of artworks from each other. Thus imaginative writings in the romance novel genre exhibit certain key character-types, plot developments, and story structures [ 15 ]. Similarly, musics exhibit generic forms of melody, lyric, mood, rhythm, harmony, and instrumentation. Music genre in this conception is a musicological category discerned by analyzing the musical and lyrical contents and structures internal to musical works [ 16 , 3 ].

Sociologists by contrast have stressed genre as a system of social classification [ 14 ]. Sorting people and things into categories is a crucial social process [ 17 – 20 ], defining reference groups and appropriate behaviors while channeling gatekeeper and consumer attention in specific directions that may favor or hurt the chances of success for products and performers [ 1 , 21 – 25 ]. Musical genres in particular provide a set of shared expectations (about music and sometimes life more generally), which deeply structure musical production and consumption [ 3 , 10 , 26 ]: collaboration among musicians, music media writing, radio airplay, concert listings, record label marketing and talent acquisition, and more [ 3 , 4 , 9 , 10 , 27 ].

On this sociological view, genres are not so much common musicological elements as typical forms of interactions based on normative expectations. Hence Lena [ 3 ] discerns four major genre forms among American popular musical styles: avant-garde, scene-based, industry-based, and traditionalist. The difference between these four lies in the social dimensions that differentiate musical styles, such as organizational form, organizational scale, or the function of typical dress and argot. Musics classified within a given genre form are subject to different normative expectations and conflicts: musicians working within scene-based genres are expected to sustain local communities organized around their music, and face sanctions for producing work for the mass market; musicians working within industry-based genres are expected to sell records, and face sanctions for reducing their marketability.

Two major conclusions flow from this sociological intervention in the analysis of music genre. First, music genres are not wholly defined by their sonic qualities. Second, the non-sonic qualities of music genres often emerge relationally and categorically: relationally, because different musicological genres can share structural similarities and differences with sonically “unrelated” genres; categorically, because any given genre acquires some part of its meaning in virtue of its position within a broad and diverse topology of higher-order genre forms.

Genre Boundaries Might Be Weakening

A stream of research in the sociological literature proposes several mechanisms that may be fundamentally altering the way traditional boundary categories operate. The central process is a version of Durkheim’s division of labor thesis, which we could re-formulate in cultural sociological terms as DiMaggio’s [ 14 ] “proposition C-6”: “the more differentiated the system of genre classification, the less universal.” Or, in a statement by Lester Bangs quoted by Bruce Springsteen at his South by Southwest keynote speech: “Elvis was probably the last thing we were all going to agree on” [ 28 ].

We can spell out this logic in more detail. Less differentiated fields tend to be marked by more total oppositions that demand more wholehearted commitment. As specialization advances, new combinations become possible. In the realm of musical genres an example would be rock music differentiating into “hard” subgenres such as hard-core and metal while at the same time “rap” spins out its own hard-core variants, thereby creating the possibility of metal-rap crossover through the hardcore sub-genres of each genre. Such a hybrid would have seemed “impure” if only larger, homogenous classifications such as rap and rock were available. But with finer-grained classifications, new categories that need not respect old boundaries become possible. Combinations between seemingly unrelated sub-genres such as rap-folk or Nintendo Core should become more likely to occur, along with other novel combinations such as pop-punk or Avant-garde metal. By extension, as genre categories approach infinity, their capacity to constrain behavior would approach zero and in effect genre classification itself would become socially meaningless.

Differentiation, as Durkheim observed, is often driven by increasing “dynamic density” in a field. This occurs when producers are brought into closer communication, stimulating competition, enabling collaboration, and expanding their reference points for evaluating themselves. In the latter-half of the twentieth century, key drivers of genre differentiation include domestic and international migration, new communication technologies, legal frameworks, inter-organizational competition [ 26 , 29 – 31 ], and technologies that permitted increased competition among radio stations and the targeting narrower market niches [ 32 – 33 ].

More recently digital technologies have promised to rapidly and deeply transform –and ultimately weaken –systems of musical genre classification [ 34 – 37 ]. Several mechanisms may be at work. Music scenes are no longer restricted to a specific physical locality. A band’s social media profile can be viewed anywhere in the world, making it possible for musicians working anywhere to know about, influence, and remix each other’s work, regardless of genre or sub-genre [ 35 ]. Online music stores are essentially unlimited in size and their products can be categorized in infinite ways. This makes it both harder for consumers to sift through the various offerings and possible to generate more flexible classifications sensitive to consumer preferences and behavior rather than pre-defined genre categories. Algorithms like the ones used by iTunes, Amazon, and Pandora strive to predict consumers’ musical preferences based not on genres but on past choices of similar consumers or analogies between a purchased song and other songs. Genres thus fade in consumer salience. Similarly, social networking websites (like Facebook or Twitter) may increase the salience of network ties in music diffusion while decreasing the salience of genre labels and other traditional forms of classification [ 10 ]. Instead of looking to genre classifications to provide information about how to program or advertise a new song, industry professionals may instead look to social network information. One might accordingly imagine a world in which music sales are sorted not into charts based on generic categories but instead based on user-specific contexts (top songs for people in my Facebook friend, friend-of-friend, etc. network).

This line of research implies at least two general hypotheses about the overall trajectory of the popular music system: 1) an increasing complexity of genre categorizations over time, and 2) as genre-based boundaries of classification systems break down, genres are replaced as signaling mechanisms by alternative social indicators of taste and preference. Or, put another way, some recent work in sociology—not to mention Lester Bangs and Bruce Springsteen—(implicitly) point toward a hypothesis about the contemporary structure of popular music genres: as genre boundaries become more fully porous, and genre as a signaling device to audiences and industries grows increasingly unimportant, there should be large and growing domains of free interchangeability in musicians’ selection of genre combinations in defining their own work.

Genre Boundaries Should Nevertheless Persist

At the same time, the sociological literature gives us strong reason to believe that genre categories should persist in structuring musical production and consumption. Consider just a few possible reasons for this persistence.

Producers are likely to rely on generic categories, even if consumers and critics have less need to do so. Record company executives and radio programmers have to allocate scarce resources. Even Internet radio has to allocate time and attention. Such gate-keeping decisions are likely to rely on genre categories [ 38 ], even while online consumers presented with customized streams of information may have less need for genre categories in their decisions [ 10 ].

Musicians are also likely to rely on genre categories for finding collaborators. Again unlike iTunes shoppers, musicians do not necessarily select collaborators based on flowing information streams. They look for others (musicians, producers, agents, venues) with whom they can work and from whom they can learn. Genre classifications send strong signals in this regard, helping musicians to sort one another into those with whom they might collaborate or not. For instance, musicians in a scene-based genre may be hesitant to work with representatives from an industry-based genre [ 3 ].

Musicians moreover may rely on genre categories for self-advertising. To the extent that music industry gate-keepers and scene members continue to use genre categories to make decisions, musicians aspiring to commercial success or scene-acceptance will feel pressure to do so as well. For instance, if country music airplay and commercial marketing strongly depends on conformity to genre conventions (of dress, speech, lyrical content, and even political orientation), then self-identifying as a “country” musician will be an important and expected professional statement by a musician aspiring to commercial success in that field [ 5 , 10 ].

Even beyond showing allegiance to genre-specific codes, genres, as a form of categories more generally, may also serve as strong incentives in connecting with audiences. Thus the sociology of classification has long been preoccupied with the fuzziness or clarity of the boundaries between categories. Fuzziness or clarity, according to this literature, is a function of how strongly those boundaries are demarcated. Much of this work examines how producers fare in a market where genres/categories are more or less clear [ 21 , 24 , 39 ], finding that attention and interest impose powerful incentives on maintaining clear categorization. Other work, by contrast, highlights categorical diversity, noting that a market is made up of multiple genres or categories, and producers face the challenge of situating themselves within or across this multiplicity [ 24 , 40 ].

While for creators the maintenance and clarity of genre categories may be a device to attract attention and remuneration from audiences, for their part, music fans may continue to rely on genre classifications for identity formation and inter-personal relationships. Music consumption goes beyond buying songs on iTunes. Music “fandom” often involves ethical convictions, political attitudes, styles of sociability, manners, race privilege and protest, and more [ 6 , 41 – 45 ]. To the extent that these are encoded in genre categories, such labels should be relatively sticky forms of social classification.

This line of research thus implies the following general proposition: despite the increasing complexity of genre designation and increased reliance on extra-genre sorting mechanisms, genres should nevertheless persist as significant methods of sorting and sense-making in musical spheres.

Multiplicity and Complexity

The above discussion highlights research focused on the question of whether traditional genre classifications are strengthening or weakening, and at the extremes whether there is reason to believe they will continue to exist at all. Yet there need not be one monolithic pattern describing the current state of genre boundaries in the overall system of popular music. Some regions within that system may have weak boundaries, others strong; some may be highly differentiated, others relatively undifferentiated. Any given genre is part of a larger complex or world, whether understood as a stream [ 46 ], social form [ 3 ], or otherwise. These worlds exert distinctive types of normative pressures, which have implications for how musicians are expected to relate to genre boundaries. Thus, genres likely exhibit different structural tendencies, depending on how they coalesce into larger complexes. For instance, some complexes may promote sub-cultural identity and novel mixing while others may encourage integration, expansion, and commonality [ 11 , 45 ]. The more general literature on the sociology of classifications helps to articulate this multi-dimensional possibility.

Although much work on classification systems focus on single dimensions of classification (such as fuzziness-clarity or diversity-homogeny), this is not true of all work within this stream. Clarity and multiplicity, for instance, are two separate dimensions that together define four possible types of categorical schemas. There may be a small number of categories with clear boundaries; there may be a large number of categories with fuzzy boundaries, and so on. For instance, Kovacs and Hannan’s [ 47 ] examination of audience ratings of restaurants in San Francisco considers the effects of combining different levels of boundary clarity and categorical diversity. Restaurants may remain within one category, they may span multiple fuzzy categories, or they may span multiple distinct categories. Kovacs and Hannah [ 47 ] find that when categories are low contrast, there are relatively few penalties for spanning them; when categories are high contrast, spanning them incurs harsher penalties. Likewise, Negro et. al’s [ 48 ] examination of winemakers finds that specialists have a big advantage in high contrast conditions but less advantage in low contrast conditions.

These works incorporate research on both boundary strength and categorical diversity, a trend that is in large part a rediscovery within a different research domain of DiMaggio’s study of artistic classification systems [ 14 ]. DiMaggio proposed that artistic classifications vary according to four dimensions—differentiation, hierarchy, universality and ritual strength. Differentiation—the number of genres into which a classification system is divided—and ritual strength—the intensity with which genre boundaries are defended in artistic production and consumption—closely resemble the concepts of categorical diversity and boundary strength that are central to the literature cultural categorization. Classification systems may be examined empirically according to any or all of these dimensions.

A Typology of Musical Worlds

This line of inquiry suggests treating genre systems not one-dimensionally but as complexes of different but compatible subsystems that cohere in what might be termed “musical worlds.” Given that boundary strength and differentiation have been key themes both in the recent literature and in DiMaggio’s [ 14 ] classic formulation, we take their intersections as a theoretical starting point for our empirical analysis of how musicians in fact tend to combine genres. Some genre complexes may have very strong boundaries, rarely intermingling with other styles; some less so. Some are internally differentiated, breaking down into sub-genres, while others do not break down in such an organized manner. As a result, arguments for and against the prevailing importance of genre differentiation may not apply to the entire musical landscape, but may instead reflect some “musical worlds” and not others within the overall topography of music. Accordingly, we construct a 2x2 table showing how genre complexes may differ according to their boundary strength or internal differentiation.

Table 1 shows four ways genre complexes may be organized in terms of the intersection of boundary strength and internal differentiations. High strength and high differentiation describes a multi-centered system in which distinct subcultures interpenetrate. High strength and low differentiation describes a single-centered complex, where, within a strong external boundary, genres fluidly mix in an unpatterned way. Low strength with high differentiation describes unbounded and uncentered space of subcultural mixing; multiple constellations that do not stay within a given galaxy.

Our analysis as it unfolds below gives more substance to this typology of genre systems. To preview our main finding: MySpace musicians classify themselves according to three major complexes of genres—one composed of Rock music genres, one of Hip Hop genres, and one of Niche or non-commercial genres. These complexes, as we will see, define three major “musical worlds,” and exhibit divergent structural patterns.

MySpace.com: A Window onto Popular Music

To discern the structure of the genre complexes into which musicians sort themselves, we employ a powerful dataset: musician profiles on MySpace.com, downloaded in January, 2007. MySpace.com is an internationally known website popular for its social networking capabilities, and used heavily by musicians seeking to promote their work [ 49 ]. Since 2007 other services like facebook.com have also become popular among musicians, but as of 2007 MySpace.com was the most widely-used and lacked serious competition. At that time, it received even more traffic than google.com. MySpace.com thus provides an extremely powerful window into the activities of popular musicians.

Although the use of data derived from social networking sites is a relatively new phenomenon, there is considerable literature that utilizes the MySpace.com website as a primary data source [ 50 – 59 ]. All have included smaller sample sizes than our study: across the nine articles just cited, N’s ranged between 1 MySpace user profile and 1.9 million user profiles, with most (6 of 9) including less than 30,000 MySpace user profiles. Our database includes nearly 3 million profiles: all musician profiles that were available for public viewing.

As have other researchers, we use data acquired through a software tool that scans the MySpace site and extracts user profile information for further analysis. Our data were initially gathered by the University of Chicago Cultural Policy Center with a custom algorithm, and was first used in Rothfield et al. [ 60 ] (see p. 50, where details about the script and data gathering methods can be found). All data were collected and used according to the company’s Terms and Conditions. Prior research using this dataset has examined descriptive features such as city-by-city genre distributions [ 61 ].

A key feature of MySpace profiles makes them especially useful for our purposes. Musicians self-selected genres for their profiles; their profiles thus record choices about how musicians deploy genres in their self-presentations. They did so from a list of 122 categories: a large but manageable number, especially for self-identified musicians who likely are more keenly aware of subtle connotations of genre labels than the general populace. Moreover, musicians could choose up to three genres. This means that if they wished to do so, musicians could combine genres in conventional ways (e.g. Rap, Hip-Hop, R&B) or unconventional ways (e.g. Showtunes-Americana-Death Metal). Such genre combinations constitute the core of our analysis, below, as they allow us to observe the clusters into to which musicians tend to consistently combine genres.

( Fig 1 ) shows basic descriptive statistics for the genres included in the MySpace dataset. It ranks the 122 MySpace genres by how frequently bands select them. Rap, Hip Hop, R&B and Rock, are the most commonly chosen; Samba, Tango, Italian Pop, and Swing, are the least common.

To be sure, this data set has its limitations. Online identities are malleable [ 62 – 64 ]. People may brag, file incorrect information as a joke, lie to increase their status, and the like. Many MySpace profiles may not meet normal social standards for being a “band” or “musician” (e.g. they may not be public or popular enough). Even though we have a large enough N to claim some degree of generalizability for this study’s results, certainly not all musicians have a MySpace page. Moreover, the available genres on MySpace may influence musicians’ choice patterns.

Despite these limitations, there are considerable benefits to the Myspace data over previously available datasets. First, it is one of the largest available datasets of currently active (at the time of 2007) popular musicians. That MySpace users do not include all musicians likely means that the dataset is somewhat biased away from older and more traditional musicians and toward younger, more digitally-savvy musicians. This means that all of the processes associated with the Internet reviewed above should be especially salient in this group, making it all the more meaningful that we find (below) strong genre boundaries even here. Second, whereas these data are broad in their definition of “band” or “musician” through the use of self-selection into the category, they are freed from the limitations of extreme censoring of “small”, “marginal”, “upstart”, or generally “unknown” bands through other collection strategies. In addition, these data provide a powerful and relatively unique opportunity to examine bands who have self-defined their music along genre indices for public appraisal and consumption. As such, due to both our large N and the richness of the data we are able to treat atypical genre combinations as useful information rather than “error” or outliers; they may indicate that (some) musicians do not feel bound to respect standard or traditional genre combinations and conventions, and perhaps may even seek unconventionality. To examine the exceptions, however, we must first discern the rule, and this is the major goal of the present paper. Finally, while different data sources collected at different times may reveal different patterns from those we find in MySpace, in order to develop and test coherent hypotheses about the nature and direction of such differences, we need revenant theoretical frameworks, technical methods, and empirical reference points, which this paper seeks to provide.

Analytical Strategy

Do musicians choose genres in regular patterns that form discrete complexes or are their choices relatively unbounded? If they are bounded, what holds them together? How do they vary in terms of their boundary strength and their internal differentiation? These questions guide our analysis of the structure of genres in the MySpace musical universe.

We follow a three-stage strategy in analyzing the MySpace data. Methods are discussed in the course of the analysis. A first step is to reject the null hypothesis: that there is a completely random relationship among a band’s genre choices. Whatever a band chooses for genre 1 would be arbitrarily related to its choice for genre 2 and genre 3, and vice versa. This is admittedly an unlikely scenario, but it does provide a useful baseline while simultaneously doubling as a test of whether the free interchangeability hypothesis holds as a system-wide phenomenon, in which genre assignments are combined seemingly at random.

Using a community detection (modularity optimization) algorithm, we find that genre choices are far from random. Certain genres are paired with one another with great consistency. To demonstrate this, we catalogue all musician-supplied genre combinations as a network defined by the frequency with which bands co-select them. For instance, if one band chooses Rap and Metal, there would be one edge between the “Rap” and “Metal” nodes, and so on with numerical frequency for “Rap” and “Metal” and all other activated genre-by-genre ties.

Given that bands’ genre choices do evince latent structural patterns that are not captured by modularity, the next ( second ) question is to examine the nature of these patterns. By applying modularity clustering to the MySpace genre hierarchically—that is, repeatedly subdividing genre communities until it is impossible to do so again with statistical significance—we are able to characterize in greater detail the MySpace universe’s organization of musical genres. We find a fundamental first division among Rock, Hip-Hop, and Niche/non-commercial musical worlds that break down further into 16 distinct genre communities.

Given that we do find that MySpace musicians group genres into consistent complexes (3 worlds, 16 genre communities), we turn to our third question, about the structural dimensions of these complexes. To do so, we examine the extent to which the permeability of genre communities’ boundaries vary. Here we return to the 2x2 table from above, and show how the major genre complexes in the MySpace universe fit within it.

Musicians Combine Genres in Strongly Patterned Ways

MySpace users face no shortage of possibilities for representing their music’s genre. Indeed, the fact that they could choose up to three genres from 122 different options means that they had 302,743 different ways to describe their distinctive style. This sort of freedom to represent oneself in so many different ways is certainly resonant of a potentially highly fluid system. However, MySpace musicians stick to a relatively small and recurrent subset of these possibilities, which they combine in highly regular and patterned ways.

This social fact becomes evident through analyzing co-selection of genres as a network, and then examining modularity in that network. This reveals the extent to which genres are likely to be paired with some other genres rather than with others. If such pairings consistently draw from a common pool of MySpace genres, they would not be isolated units combined haphazardly but rather anchored in higher order musical groupings. We shall refer to these complexes as “worlds” and “communities”—islands of musical inbreeding—each of which is structurally distinct from the next. The existence of such higher order groupings would suggest that musicians evidently respect the boundaries of their genres’ communities and worlds, mixing within them rather than between; collective musical norms would evidently structure musicians’ genre choices.

We create a large and complex genre network by mapping the band-provided (self-identified) co-listings of (up to three) genres. Genres are considered “related” once when a band lists them together. In the event that a musician chooses only one genre—thereby providing no information about how genres are associated with one another—the musician’s choices are not included in the analysis. To make sure this did not bias our results, we compared the distribution of genre nominations for genres that are listed alone vs. genres that are listed in some sort of combination. We found that the distributions do not differ substantially. Thus, eliminating single genre selections from our analysis does not bias the genre clusters we write about in any significant way, nor would providing some kind of unique score for single genres change the combinatory patterns that we currently see. In further analyses of the available data it seems that single genre selection has as much to do with incomplete or less active profiles as it does with a full-throated allegiance to a single genre.

For the full population of bands in the data which did list more than one genre designation, Greedy Modularity Optimization is employed to identify genre communities. Greedy Modularity Optimization was developed by Clauset, Newman, and Moore ([ 65 ]; see also [ 66 ]). This algorithm partitions a network by maximizing its modularity , a measure that quantifies a network’s community structure by providing a value for every clustering within a given graph. The general idea is to employ a random graph on the same vertex set that does not have any community structure, and compare the edge density of the clusters in the original graph with the edge density of the clusters in the random graph. The greater the difference between the two edge densities, the more community structure the given clustering describes. We use the version operationalized in R’s IGraph package, which outputs the best community structure (structure with the highest modularity score) possible.

But modularity algorithms, like most clustering algorithms, have no universally accepted significance tests. In other words, there is no consensus as to whether a modularity score of .1, .3, or any value indicates a “real” vs. an arbitrary community structure. In certain situations, however, it can be relatively easy to apply statistical techniques that approximate a significance test. While it is unclear how we would define, let alone test, the significance of the entire community structure discovered in this study, it is relatively straightforward to test whether a single identified community is significantly structurally “separate” from the rest of the large network. This can be done with a Wilcoxon rank-sum test, which, applied to this context, assesses whether there is a statistically significant difference between the number of in-edges and out-edges adjacent to members of a given community. If a community has significantly more in-edges than out-edges, the community is considered statistically significant—a relatively unified group of genres with relatively strong boundaries. And if all its constituent communities are significant, it is reasonable to consider an entire community structure statistically significant.

In this study, running the modularity optimization algorithm and significance testing was procedurally united. In order to identify the most specific genre complexes possible, we do not simply run the modularity optimization once. Instead we run the modularity optimization and the rank-sum test in direct succession and progressively until further dividing a community into smaller, more specific groupings no longer yields statistically significant communities. Our results therefore present a community structure in which all identified communities are indivisible into smaller significant communities and are themselves significant at the p <.01 level.

We used modularity-based approach to community detection because our primary interest is to identify areas of density in a graph composed of weighted and undirected edges. Our data are simple: musicians select genres, and genres are considered related when they are co-selected by many musicians. Areas of density therefore represent in a clear and straightforward way groups of genres that are commonly associated with one another across the millions of musicians in our sample. With only 122 genres and millions of musicians, structure comes from edge weights, so the community detection algorithm chosen must be able to work with edge weights. Since our edges are undirected, the community detection algorithm must be chosen accordingly. Modularity-based approaches are a primary example of internal density approaches that operate on weighted, undirected edges [ 67 ]; here we use Igraph's implementation of greedy modularity optimization.

While some modularity-based approaches facilitate the detection of overlapping clusters, we have chosen here to identify non-overlapping genre clusters. This is because we are primarily focused on genre classification. We seek to understand how genres may be categorized and the extent to which those categorizations are clear. Future research may fruitfully pursue potential overlaps between genre clusters—not only measuring boundaries' fuzziness/clarity, but also closely investigating those areas of fuzziness. For this purpose, detecting overlapping communities would be useful.

Many researchers have pointed out that modularity optimization presents a resolution problem, where smaller communities and communities varying in size are difficult to detect [ 68 ]. As a solution, we have recursively optimized modularity for discovered clusters (as is also done in [ 69 – 70 ]). While we were able to find reasonable criteria for deciding when to stop partitioning a cluster (Wilcoxon rank-sum test), we were unable to solve the problem of recursive partitions being inconsistent with each other (the "Rock" world is partitioned according to a different mathematical standard than the “Niche” world). However, we do not regard this as a significant problem for our analysis. Our clusters represent different musical worlds that operate according to different logics. The criteria for division and boundary creation should therefore be localized to each sub-cluster instead of generalized to the entire graph.

Though modularity-based community detection makes clear analytical sense for our purposes, as a robustness check we tested several other algorithms that can analyze weighted, undirected edges and clusters genres according to network density or closeness (which we see reasonably similar to density). We found that the clustering produced by Greedy Modularity Optimization was robust across algorithms. If we compare how Greedy Modularity Optimization and alternative algorithms classified genres, we find clusterings that ranged from a low of 93% (Walktrap Community Detection) to a high of 96% (Multi-level Modularity Optimization) similar to Greedy Optimization. Alternative algorithms tended to produce clusters that were less balanced in terms of size, where, for instance, a few genres that Greedy Optimization placed into the “Hip-Hop” world would be placed into the larger and more diverse “Niche” cluster. Although a number of algorithms would produce similar findings, we chose Greedy Modularity optimization because it is the most computationally efficient algorithm and its logic reflects in a straightforward way our intention to find densely interconnected areas of the genre network.

Three Worlds, 16 Communities

( Fig 2 ) displays the network of genres in the MySpace universe highlighting how they cluster into genre communities. As modularity in a complex network is difficult to display in a simple visual, ( Fig 2 ) is a modified version of the network. It was created by plotting each of the clusters and its three strongest out-edges. Genre communities are defined by node color, where “warm” colors (red, orange, yellow, pink) represent Rock ‘n’ Roll genres [henceforth referred to as the broad “Rock” world], “cold” colors (blues, greens, purples, grays) represent Niche or Underground genres [henceforth referred to as the broad “Niche” world], and brown represents traditionally African-American and Latino Hip-Hop and Latin-based genres [henceforth referred to as the broad “Hip-Hop” world]. Edge width represents the frequency with which genres are co-chosen. ( Fig 3 ) is a dendrogram that details how a first-order clustering into Rock, Hip-Hop and Niche music worlds is broken down into 16 genre communities and the modularity coefficients at each division. For every division, the community’s in-edges outnumber its out-edges at a statistically significant level (p <.01). Table 2 places each of the 122 genres available on MySpace into 1 of 16 genre communities (see also Figs ​ Figs2 2 and ​ and3 3 ).

The graph in ( Fig 2 ) indicates how some communities overlap more than others (e.g Pop/Rock and Homegrown American vs. Punk Rock and Rave), and how a few genres—Rock, Hip Hop, Acoustic, Experimental—do much of the work of binding the disparate elements of the musical universe. It is also visible that “Rock” is the touchstone binding the Rock world, an extremely prevalent triad of Hip-Hop, R&B, and Rap is the touchstone for the Hip-Hop world, and Experimental and Electronica are the closest thing to touchstones for the Niche world, which otherwise has no clear center.

Given the strong and statistically significant modularity illustrated in ( Fig 2 ) and detailed in ( Fig 3 ), the analysis suggests that genre distinctions strongly structure the self-presentation of contemporary popular musicians. This dataset does not allow us to compare the strength of these conventions to the past (where much of the extant literature claims that genre conventions are weakening over time), but it does indicate that, at least as of 2007, the boundaries between genre worlds were far from extinction. As much as traditional musical genres may have been rhetorically subdivided time and again, those subdivisions still cohere together, operating within distinct boundaries rather than by way of free mixing of musical styles.

Thus, we can satisfyingly answer our first question about whether there is significant clustering of genres at all and can move on to our second: genre selection on MySpace exhibits substantial structural patterning, and those patterns cohere around three musical worlds, two of which may be decomposed into sub-complexes, or genre communities. We performed a placebo test to ensure that the modularity observed in the MySpace network is not due simply to network density, randomly rewiring the network 1000 times. While the modularity coefficient for the actual MySpace network is 0.31, the average modularity for our 1000 random simulations is only .04. This very dense network exhibits almost no modularity at all when its edges are randomly allocated. Thus, the clustering patterns we observe in the MySpace network are highly unlikely to be due to random chance.

Patterning around three musical worlds is most visible in the dendrogram in ( Fig 2 ). The final “end” communities are in the dark green cells. Progressive modularity clustering moves from the left to right of the dendrogram. Furthest on the left, the first-order breakdown separates out three main categories of popular music—Rock, Hip-Hop, and Niche genres. As we move to the right, the Rock world breaks down to its “Subcultural” varieties, which we call “Countercultural,” “Mainstream,” and “Punk Offshoots,” and then finer categories therein. The Hip-Hop world, dominated by Rap, Hip-Hop, and R&B, breaks down no further into statistically significant communities. They comprise both a major musical world and an “end” community—a fact of great significance, as we will see. The “Niche Genres” world represents a variety of less popular (as defined by the frequency with which they are selected by musicians) genres and communities. It is essentially a category encompassing musics not strongly tied to the two dominant poles of Rock and Hip-Hop. These include most notably Electronic music genres, Dark or “Extreme” Metal, and various underground and World Music genres, which emerge and further subdivide as we move to the right of the chart. The Genre-Community membership table demonstrates the considerable face validity of this clustering technique, though the variability in how sub-divided the different worlds are is an intriguing fact that we examine in more detail, below.

Strength and Differentiation

Turning to our third question, we now seek to uncover the structural properties that distinguish musical worlds from one another. To do so, we are guided by Table 1 . Specifically, we ask: What genre complexes have the most and least permeable boundaries? An effective way to measure permeability is by (1) comparing out-edges that reach across the three primary musical worlds that emerge from our first-order clustering above: Rock, Hip-Hop, and Niche. This indicates the proportion of genre communities’ out-edges that bridge great distance in the MySpace universe of musical style. In addition, we measure (2) the proportion of each genre’s adjacent edges that are shared with other genres in its community vs. with genres outside its community: in-edges vs. out-edges, in other words. This measure helps us to unpack what the modularity analysis already indicated, that some musical worlds are more internally variegated than others. Combining (1) and (2) allows us to empirically measure musical worlds in terms of their boundary strength and internal differentiation , and so to situate them in the typology articulated in Table 1 .

Permeability

( Fig 4 ) displays the proportion of each genre community’s edges that are external to its broader musical world. The percentage of extra-world edges for each musical world is listed in Table 3 .

Niche Genre communities are relatively unbound by their world. Quite often, they reach out to genres outside their world. In fact, ~53% of their edges reach not only outside of a given genre’s home community but also outside of its world to genres in the Rock or Hip-Hop worlds.

By contrast, only ~32% of Hip-Hop bands’ adjacent edges bridge to a different music world. The Rock world is even more impermeable. Only rarely (~28% of the time) do Rock musicians bridge across musical worlds. The worlds of Rock and Hip-Hop thus exhibit strong boundaries; the Niche world has relatively weak boundaries.

Differentiation

( Fig 5 ) turns our attention to the internal variegation of these worlds. The modularity analysis already indicated the key point: Rock and Niche worlds are internally differentiated; Hip-Hop is not. ( Fig 5 ) helps to articulate in more detail how finer-grained genre communities are associated with other communities in their own world and to communities in other musical worlds—the extent to which these differentiated genre communities mix and intermingle. ( Fig 5 ) shows the proportion of each genre community’s edges that are external to that community. The percentage of extra-community edges associated with each broad musical world is listed below it, in Table 4 .

On the whole, Niche communities have the most porous boundaries. On average, ~84.5% of edges adjacent to Niche genres bridge across genre communities. Rock music also exhibits substantial inter-penetration, with over 77% of its edges bridging communities. Since the Hip-Hop community is also a world, its community-bridging rate is not comparable to that of the Rock and Niche worlds. Hip-hop’s community-bridging rate is identical to its world-bridging rate of ~31.7%.

Structural Properties of Musical Worlds

Table 5 synthesizes all of these results. It returns to the typology articulated in Table 1 , but now situates the musical worlds within that typology on the basis of our analysis.

The Rock world is a complex of multiple interpenetrating sub-communities, surrounded by a strong external boundary. Genre mixing across these sub-communities is common; genre mixing beyond the limit of the Rock world is rare. Punk Rock musicians, for example, are more likely to describe themselves through Punk Rock and some non-Punk Rock genre (e.g. Pop Punk and Indie) instead of through Punk-based genres alone (e.g. Pop Punk and Grunge). But they rarely use non-Rock genres to describe themselves. Hip-Hop, however, largely resides in its own musical world, where Hip-Hop bands are extremely likely to describe their music with only other Hip-Hop genres; they are, like Rock musicians, rather unlikely to transcend their musical world. The line between Hip-Hop and not-Hip Hop is strong and rarely crossed. Within its bounds, it is a relatively boundless world, as genres mix fluidly, with little discernible internal sub-cultural differentiation.

The Niche world illustrates low boundary strength and high internal differentiation. Niche genre communities (with notable exceptions of Rave and Extreme Metal music) are the least strictly maintained. Musicians cross them with relative impunity. Obscure Underground Club musicians, for example, are much more likely to define themselves vis-à-vis genres in various other communities (e.g. Happy Hardcore and Electronica) than through a selection of Underground Club genres alone (e.g. Happy Hardcore and Jungle). Niche genre communities are a set of musical sub-cultures only very loosely bound together in a common world, like free-floating solar systems without a strong galactic center. This lack of a center is also visible in ( Fig 3 ), where there is no apparent touchstone binding the disparate Niche communities. They are essentially defined by what they are not, Rock or Hip Hop.

Conclusion: The Multiple Structural Logics of Popular Music

Drawing from some 3 million bands who have self-classified within a matrix of 122 available genres, we provide the first empirical analysis of this size we are aware of that investigates the structure of popular music by genre. We find that the world of MySpace musicians is not a single world that is made up of “tight” or “loose” genre associations, but instead, is made up of three meso-level genre complexes. A primary result of this paper is to have uncovered the basic properties of these worlds. Rock is a world of sub-cultural differentiation and sub-cultural mixing, operating within a broader common culture of Rock N’ Roll. Sub-cultural formation and transformation produce prevalent interpenetration of sub-cultural identities. The experience of a musician located within such a world is thus one in which musical boundaries are noticeably present, but are flexible and open to constant redefinition and transgression. A heuristic for the generative structural principle at work appears to be: “connect frequently to a common center, but combine some sub-sets more than others.”

Hip-Hop by contrast is a world with a single center but no sub-centers. Surrounded by a strong boundary, Hip-Hop musicians freely combine genres that fall within that boundary. It is a world of structureless fluidity, maintained by a shared connection to a central musical core (composed of Hip-Hop, Rap, and R&B)—a common reference point that over-rides any potential sub-cultural divisions. The experience of a musician located in such a world is of being part of an ever-expanding circle, in which multiple genres may be creatively joined in an egalitarian way as so many faces of a common tradition. Here the generative principle seems to be: “connect anything with anything else, as long as all connect equally to the same center.”

Hip-Hop and Rock ‘n’ Roll are the two major worlds of musical system revealed by the activity of the over 3 million musicians and bands on MySpace.com. Niche genres are the “other” to both: porous communities fluidly combining, lacking any larger touchstone to bind them together.

The worlds of Rock and Hip-Hop do not, however, merely co-exist side-by-side; they are part of the same broader musical system. The boundaries between worlds have a qualitatively different character from the boundaries within a world, for here structural principles collide. To cross over from Rock to Hip-Hop is not simply to engage in more of the same prevalent inter-penetration that characterizes the world of Rock ‘n’ Roll. It is to instead face a basic challenge to the principle of sub-cultural differentiation as a form of musical identity. To cross over from Hip-Hop to Rock is equally challenging. To do so is to potentially abandon the basic principle of egalitarian mixing about a common center in favor of sub-cultural dis-integration. What is differentiation for the one is fragmentation for the other; what is unity for the one is homogeny for the other. The result is that in the popular music system as a whole, the boundary between Rock and Hip-Hop is especially fraught with questions of authenticity and identity.

The complex of multiple creative principles at work among MySpace musicians holds a lesson for the sociology of culture more generally, which in the past decades has been pre-occupied with the rise of a culture of boundary crossing, under the heading of “omnivorousness” [ 71 – 73 ]. However, our analysis indicates that this conceptualization of culture has selective application. It fits the world of Rock ‘n’ Roll well, but it fails to capture the pattern of genre identification and mixing that characterizes the Hip-Hop world, which lacks the sub-cultural structure necessary for bridging to occur but nevertheless exhibits fluid, creative mixing of genres around a common center. The implication, then, is that sociologists of culture would do well to contextualize omnivorousness as one orientation to cultural creativity among others, and to understand how multiple orientations emerge, grow, change, interact, and conflict with one another, in different ways in different places.

This same basic point also suggests that sociologists of culture and scholars in related fields should proceed with caution when making generalizable claims about the ostensible decline or continued importance of genre as a boundary making device. Rather, genre boundaries may simultaneously be highly porous in some complexes and more impenetrable in others. For single or small N case-studies of musical genres, the properties of genre boundaries unearthed may not signal wider trends, but rather the meso-level complexes in which the genre or genres under question operate.

We also suggest that the unearthing of these three meso-level worlds opens up the possibility for more robust longitudinal analysis in future studies. Does the strong core of Rap/hip-hop provide the opportunity space for more unconventional mixing over time, or do the firm boundaries restrict mixing and shed off impure experimentation within the complex? As a niche genre gains in popularity are its genre boundaries more strongly policed, or does it open up space for other popular genres to engage in mixing? Pursuing these questions through alternative data sources—ideally from multiple time periods –would additionally allow us to examine the robustness of the patterns we observe in MySpace as of 2007 to different data collection techniques.

Lastly, rather than treating musical genres as a self-contained relational system, most sociological work examines genre boundaries as they intersect with industry imperatives and audience demand. Based upon our work, more robust analysis of genre complexes and their relationship to both industries and audience is possible. How does musicians’ popularity affect their patterns of genre choices (and vice versa) and do these relationships vary by musical style, place, and available industry resources? We hope to take up these questions in future work.

The major ambition of this paper, however, has not been to examine the antecedents or consequents of popular music’s various genre complexes. It has been to illuminate the structure of those complexes themselves. The discovery of form amidst apparent chaos and parsimonious explanation of complexity are our primary goals. From some 3 million musicians, 122 available genres, and over 300,000 possible combinations, we distill 16 communities nested within 3 worlds at the intersection of 2 dimensions. We lay bare simple heuristics concatenating to form major structural patterns which, for better or worse, continue to govern the classification system that binds and separates popular musicians to and from one another.

Supporting Information

Acknowledgments.

The authors would like to thank Noah Askin, Andrei Boutyline, Kim de Laat, Omar Lizardo, John Levi Martin, Ben Merriman, James Murphy, and Fabio Rojas for feedback on earlier drafts of this work. All mistakes are our own.

Funding Statement

Facebook Inc. provided support in the form of salary for Lee, but did not have any additional role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript. The specific roles of this author are articulated in the ‘author contributions’ section.

Data Availability

Music genre

Hip hop music, rhythm and blues, middle eastern music, classical music, electronic music, music of latin america, music for children, new-age music, vocal music, music of africa, christian music, music of asia, traditional music, independent music.

25 Most Popular Genres of Music

Musical genres are an inexplicable phenomenon. With each one that is discovered, there are a seemingly endless number of others waiting to be revealed and enjoyed by music fans around the world.

Most genres are born out of the fusion of two or more existing styles, with musicians taking certain elements and combining them. The rapid advances in music technology are also responsible for the creation of many genres.

If we were to go through all of the different music genres in existence, this article would take days to read!

Therefore, we’ve condensed them down to the 25 most popular genres of music. In this guide, we’ll discuss the origins of each musical style, their place in the world of modern music, and their main stylistic attributes.

Pop music is perhaps the most obvious addition to this list. It could be said that pop is a compilation of other genres, which are considered to be mainstream in a certain time period.

For example, in the 1960s, rock bands like The Rolling Stones, The Beatles, and The Beach Boys were labeled as “pop music”. Four decades later, the term pop was being used to describe hip-hop artists like Eminem or Kanye West.

In the modern era, pop music is heavily influenced by electronic dance music, trap – which is a sub-genre of hip-hop and uses other elements taken from alternative styles. These songs commonly top the charts on streaming services and dominate popular radio stations’ playlists.

One of the top musical genres in existence, rock music originated in the 1940s and 50s in the form of rock & roll. However, its roots can be traced back to the rhythm and blues of the African-American culture in the 1920s, merged with country music.

When recording techniques and musical equipment became more advanced in the 1960s, rock music exploded in the Western world. The genre is synonymous with high-gain electric guitar tones , accompanied by bass guitar, drum kits, and powerful vocals.

Rock’s influence has stretched far and wide, with a plethora of sub-genres being created from its key aspects. Although it perhaps doesn’t enjoy the mainstream popularity it did in the ‘60s and ’70s, rock bands still sell out huge stadiums today.

3. Hip-Hop & Rap

In its relatively short history, hip-hop has emerged as one of the most popular and innovative genres of music. Hip-hop originated in the Bronx, a borough of New York City, in the late 1970s when DJs would use samples and breakbeats to create backing tracks for MCs to rap over.

Like rock music before it, hip-hop took the world by storm in the 1990s. Artists like Dr. Dre, Tupac, Biggie Smalls, and many others rose to worldwide fame, paving the way for a new generation of rappers and hip-hop producers in the new millennium.

Digital equipment has made hip-hop much easier to make these days, with producers and rappers able to compose albums with a microphone and a laptop. This has led to huge exposure for hip-hop and its many sub-genres.

Country music can be traced back to the beginning of the twentieth century. It was created mainly in the south of the USA, by working-class people. These people would use country as a means to tell stories through music, about the realities of everyday life and their perspectives.

Some of the most influential country artists include Woodie Guthrie, Johnny Cash, and Hank Williams. Today, the genre continues to be incredibly popular in America but is often combined with other popular styles.

R&B, or rhythm and blues, is rooted in African-American culture from the 1940s. In the decades that followed, record labels used the term to describe recordings that were targeted towards that community in the US, and the style eventually inspired many rock artists of the 1960s.

Today’s R&B music commonly uses a blend of acoustic and electronic instruments, with bluesy melodies and soulful vocals.

Folk music has existed in many different parts of the world for centuries. Traditionally, this genre is essentially built upon people gathering to sing and play songs with others in their community.

In the 1960s, artists like Bob Dylan and Joan Baez popularized folk music to the masses. The genre is relevant today and is commonly associated with archaic melodies, acoustic instruments, and insightful lyrics with strong messages.

In the early 20th century, musicians in the city of New Orleans experimented by blending musical elements from European and African genres. This resulted in the origination of jazz, which would go on to become one of the most popular and unique musical styles in existence.

Jazz is defined by its swung rhythms, a diverse range of instruments, and smooth melodies. While it rarely tops the charts today, it still has a huge following around the world.

8. Heavy Metal

Heavy metal music is a sub-genre of rock and is characterized by loud volumes, crashing cymbals, pounding rhythms, and distorted guitars which often use drop tunings . Black Sabbath and Motorhead are two prime examples of classic heavy metal bands.

The live metal scene is as strong as ever today, with bands constantly coming up with new ways to innovate within the genre by using effects and complex playing techniques.

EDM is short for “electronic dance music”, which is a very broad category. In a popular music context, this genre describes songs that feature classic elements from dance music, such as four-to-the-floor drum beats, synthesizers, and repeated loops.

There are many sub-genres of EDM, some of which I’ll discuss later in this guide. The genre also has a huge live scene, with DJs playing sold-out shows all over the globe.

Soul music is a blend of traditional R&B, gospel, and jazz. It originated in the African American communities in the late 50s and early 60s, with popular artists like Ray Charles, Sam Cooke, and later the likes of Aretha Franklin and Marvin Gaye.

Soul and R&B are often categorized together in the modern era because both genres have changed significantly over the years which has resulted in many similarities.

Like the soul, funk was also the result of African Americans blending jazz and R&B. Funk has a strong rhythmic pulse, prominent bass lines, and syncopated rhythm guitar playing.

Although James Brown was known as “The Godfather of Soul”, many of his songs are considered the blueprint for funk music.

Reggae was invented in Jamaica in the late 60s, and quickly become the country’s favorite music genre. In the decades that followed, it reached the UK, USA, and Africa, where it amassed huge audiences.

Bob Marley is regarded as one of the pioneers of reggae music. The genre commonly uses a distinctive rhythm, where the emphasis is placed on the off-beat.

Disco music rose to prominence in the late 60s and early 70s, making its way into the upmarket nightclubs of major US cities.

In terms of musical characteristics, disco typically has a strong four-on-the-floor backbeat, syncopated basslines, funky guitar riffs, and reverb-draped vocals. The genre’s popularity began to dwindle after the 1970s, but a new wave has emerged in the past decade.

14. Punk Rock

Perhaps the ultimate rebellion against technical musicianship, punk is a hugely influential sub-genre of rock.

Punk rock has an aggressive sound, with fast-paced tempos and simple guitar riffs often played using only downstrokes. It was seen as a departure from the technical styles of the main music genres that had dominated the 1970s, with bands like The Ramones, The Sex Pistols, and The Clash bringing it into the public eye.

15. Classical

Classical music is amongst the oldest genres features in this list, but it continues to be popular around the globe.

This style features a range of orchestral instruments, and composers use pre-established forms to create emotive symphonies with soaring melodies and ear-friendly chord progressions.

Technically a sub-genre of EDM, house music has a huge global fanbase. Musically, it is often characterized by a tempo of between 120 to 130 BPM, with the kick drum being played on every beat.

House is especially popular in Europe and America, in raves, nightclubs, and at music festivals. It is one of, if not the most prominent subgenres of electronic dance music in the past thirty years.

Techno shares some similarities with the house music but tends to feature electronic sounds more heavily. It is incredibly popular in the rave scene, with its powerful, thudding drum beats making the genre perfect for long dancing sessions.

Electronic instruments like samplers, synths, and retro drum machines are commonly used to compose techno music. Its tempo usually sits somewhere between 120 and 150 BPM.

18. Indie Rock

In the past thirty years or so, indie has developed into one of the most popular sub-genres of rock music. With its D.I.Y ethos inspired largely by punk, indie reached its peak in popularity in the 2000s, with bands like The Strokes and Arctic Monkeys paving the way.

Like punk before it, indie rock doesn’t require technical proficiency to be played. It is a genre more focused on songwriting, with catchy melodies, jangly guitar tones, and introspective lyrics often used by composers in the genre.

In the 1980s in Seattle, a collective of aspiring musicians had become disillusioned with the mainstream rock music that dominated the radio stations. Out of their frustration, a new genre was born – grunge.

As the genre amassed a growing fan base, bands like Nirvana and Pearl Jam were propelled to stardom. Highly distorted guitar riffs, pounding drums and gritty vocals are three common characteristics of this genre. It continues to inspire the new generation of rock musicians today.

20. Ambient

With dreamy atmospheric layers that constantly change and evolve, ambient music is a unique instrumental genre. It includes a blend of acoustic and electronic instruments and samples, placing more emphasis on tonal qualities rather than rhythm.

The ethereal nature of ambient music makes it a popular choice for people who want to relax or meditate.

Whatever your religious or spiritual beliefs, you can enjoy the beautiful sound of gospel music. This Christian genre is often performed in churches and takes influence from traditional blues, R&B, and country.

Typically, gospel often includes rich layers of vocal harmonies sung by choirs. Pianos, guitars, and other acoustic instruments are used to create the backing music for the soulful vocalists to sing over.

22. Latin Music

Latin music is a genre that includes several styles originating in Spain, Portugal, Latin America, and parts of the USA.

This genre often heavily features syncopation and is incredibly catchy. Rhythmically, it takes inspiration from African beats and blends acoustic instruments like guitars and bass with electronic instruments like keyboards and synthesizers.

Grime originated in London, England in the early 2000s. For many years, it was seen as an underground sub-genre of hip-hop, but in recent years it has enjoyed prolonged mainstream success and become one of the most popular music genres in the UK.

Compared to traditional hip-hop and rap music, grime has a considerably faster tempo, which is usually around 140 BPM. Grime MCs use repeated phrases known as “bars”, and rap in a quick, technical way.

Trap music has its roots in the south of the United States and features a minimalistic production style with syncopated hi-hats and prominent snare drums.

Over the past decade, trap has emerged as one of the top music genres in the world. Producer Lex Luger is credited with popularizing the distinctive modern style of trap beats which dominates the US music charts consistently.

25. Psychedelic Rock

To conclude our list of the 25 most popular genres of music, we have psychedelic rock. As guitar effects experimented within the 1960s, artists like Jimi Hendrix and The Doors created mind-bending compositions that were inspired by psychedelic experiences.

Psych-rock doesn’t often feature in mainstream music charts these days, but artists like Tame Impala and King Gizzard and the Lizard Wizard have revived the genre and in recent years.

This list provides you with an overview of the most popular music genres in existence today. While these words can provide you with a brief description, the best way to experience the genres is to dive straight in and listen for yourself.

Gideon Waxman

Gideon Waxman is a songwriter and multi-instrumentalist from England. His solo project Guiding Force blends fantasy-infused soundtracks with powerful riffs and soaring atmospheres. 

Having previously toured in bands and studied music at university, Gideon has extensive knowledge of drums, music production, recording equipment, and the business side of the music industry too.

Related Posts

How to Stop Feedback On a Guitar Amp

17 Most Popular Canadian Female Singers

15 Famous Raspy Voice Singers

Are Gold Plated Guitar Cables Worth It?

What Are The Best Woods for Electric Guitar Bodies?

Why Do Guitar Necks Warp? (Plus Solutions & Fixes)

35 Types of Music Genres and Styles (With Examples)

Music is like a melting pot of genres. There are countless genres and sub-genres, encompassing rock, jazz, rap, and so much more. Let’s explore the world of sounds and styles, as we delve into the top 35 types of music.

Let’s get right into it!

9. Classical

15. dubstep, 16. drum and bass, 17. r&b, 20. instrumental, additional genres.

Pop music is an eclectic genre that features catchy melodies, and simple chord progressions, and usually deals with themes of love and relationships. This genre is defined by its accessibility and broad appeal.

Its production often involves a glossy, high-fidelity sound that is both polished and meticulously crafted. Pop music is known for its ability to evolve, absorbing elements from various other genres like hip-hop and rock to stay fresh and relevant.

Sub-genres: Electro pop, teen pop, power pop, pop-rock, and many more.

Popular artists:

• Taylor Swift
• Ariana Grande
• Billie Eilish

Rock music originated from rock and roll and quickly became a force of its own. It’s known for strong, driving rhythms, amplified instruments, and emotive lyrics. Rock music has always been associated with rebellion and social commentary, and it’s often lauded for its raw, unfiltered sound.

Derived from several popular genres at the time like blues, country, folk, jazz, and other styles, rock culminated in a widely accepted and hugely popular genre that took the world by storm. Rock incorporates electric instruments like the electric guitar, bass, keyboard, and acoustic drums. 

Sub-genres: alternative rock, indie rock, hard rock, and punk rock, among others.

• The Rolling Stones
• Foo Fighters

Rap is a genre centered around rhythm and rhymes. Often combined with hip-hop , it’s known for its powerful lyrical content that often discusses societal issues, personal experiences, and storytelling.

It’s unique in that the musical focus is on the rhythm and flow of the spoken words, making the vocals the most important aspect of the genre. This genre has taken the world by storm with a massive following behind it and names like Eminem, Kendrick Lamar, and Tupac, who have transcended the genre and brought it to the world.

Sub-genres: gangsta rap, trap, mumble rap, and conscious rap, among others.

• Kendrick Lamar

Jazz is often characterized by its swing rhythms, improvisation, and complex harmonies. It’s a highly expressive genre, allowing musicians to inject their own style and emotion into their performances. This creativity and spontaneity make every jazz performance a unique experience.

While not universally embraced, jazz stands as a groundbreaking genre, showcasing virtuosic artists who have pushed the boundaries of their instruments. These trailblazers have not merely made music, but masterfully pushed the envelope of what is possible with a musical instrument, rewriting the lexicon of sound and performance, and thus, catapulting jazz into an unparalleled realm of artistic expression.

Sub-genres: bebop, cool jazz, smooth jazz, and free jazz, among others.

• Miles Davis
• Ella Fitzgerald
• Herbie Hancock

Blues music is known for its emotive expression of struggle and triumph. This genre is characterized by its use of the “blues scale”, repetitive chord progressions, and soulful, often melancholic lyrics.

Its roots in African-American communities have helped shape other genres of music such as rock and jazz. Despite that, blues remains a genre that can stand on its own with numerous popular tones and widely-regarded artists.

Sub-genres: delta blues, Chicago blues, and blues rock, among others.

• Muddy Waters
• Eric Clapton

Folk music is often rooted in tradition, telling the stories and experiences of a people or culture. Characterized by acoustic instruments and simple melodies, it places a strong emphasis on lyrics and storytelling.

Folk music’s authenticity and connection to cultural heritage give it a timeless appeal. Deeply rooted and associated with people’s folklore, folk music is battle-tested and has evolved into a number of subgenres that are widely accepted today.

Sub-genres: contemporary folk, folk rock, and indie folk, among others.

• Mumford & Sons

Metal music is known for its aggressive and intense sound, often featuring distorted guitars, heavy drumming, and powerful vocals. Despite its sometimes extreme nature, it has a large, dedicated fan base due to its raw energy and technical complexity.

The genre derived from rock music and its various subgenres like blues rock, psychedelic rock, and acid rock, among other styles, to form its own culture and stand on its feet. Highly prominent today, metal music has a substantial following and is one of the more popular music genres out there.

Sub-genres: heavy metal , death metal, and black metal, among others.

• Iron Maiden

Country music is rooted in rural American folk and Western music. Known for its emotive storytelling, country music often features themes of love, heartbreak, and everyday life. It’s characterized by its distinctive twang, making use of instruments like the guitar, piano, banjo, and fiddle.

Country music is still highly regarded today with young artists like Taylor Swift keeping the genre at the top with captivating and interesting songs. A genre known for its ballads, Country music is still at the top of the music tree.

Sub-genres: country pop, outlaw country, and bluegrass, among others.

• Johnny Cash
• Dolly Parton
• Blake Shelton

Classical music is a vast genre that encompasses music from the Middle Ages to the present day. Known for its complexity and rich harmonies, classical music requires a high level of skill and understanding to perform.

It’s often seen as the foundation of many other music genres. Classical music is characterized by its intricate compositions, emotional depth, and rich orchestration and despite being around for a very long time, some classical compositions are still widely regarded today.

Sub-genres: baroque , romantic , and modern classical, among others.

Popular artists/composers:

• Ludwig van Beethoven
• Wolfgang Amadeus Mozart

Reggae music is most associated with Jamaica and is characterized by its offbeat rhythms and relaxed tempo. Often featuring themes of peace, love, and social commentary, it has a distinct sound that’s both danceable and thought-provoking.

Originating in the late 60s, Reggae was highly influenced by American jazz and rhythm and blues and it traveled worldwide because of artists like Bob Marley, Peter Tosh, and others.

Sub-genres: roots reggae, dancehall, and reggae fusion, among others.

Punk music is characterized by its raw sound, often featuring short songs with fast, hard-edged melodies and stripped-down instrumentation. It’s also known for its rebellious spirit and countercultural lyrics, making it a powerful medium for social and political commentary.

Punk supports the DIY mentality; many bands self-produce recordings and distribute them through independent record companies as they embody the anti-establishment mindset.

Sub-genres: punk rock, pop punk, and hardcore punk, among others.

• The Ramones
• The Sex Pistols

Techno is a genre of electronic dance music that’s characterized by its repetitive beats and synthesized sounds. Originating from Detroit in the 1980s, it’s known for its futuristic themes and is often associated with rave culture.

Techno often uses a tempo between 120 and 150 BPM (beats per minute) which is a general and unspoken rule of thumb within the genre. Also, it incorporates electronic instruments such as drum machines, sequencers, and synthesizers.

Sub-genres: minimal techno, dub techno, and acid techno, among others.

• Richie Hawtin

Trance is another form of electronic dance music. It’s characterized by its fast tempo, repetitive melodic phrases, and build-ups and breakdowns that create a hypnotic, euphoric feel.

Its emphasis on melody and atmosphere sets it apart from other electronic genres. Much like techno music, Trance usually varies between 135 and 150 beats per minute with repeating melodic phrases.

Sub-genres: progressive trance, psytrance, and uplifting trance, among others.

• Armin van Buuren
• Paul van Dyk
• Above & Beyond

EDM , or Electronic Dance Music, is a broad term that encompasses various sub-genres like house, techno, and dubstep. It’s known for its catchy melodies, synthesized sounds, and heavy use of samples and loops.

EDM has become synonymous with party culture due to its energetic and danceable tunes. It often includes a catchy chorus performed by guest artists, hard-hitting drum patterns, and a big moment of tension followed by a powerful drop.

Sub-genres: big room EDM, tropical house, and future bass, among others.

• Calvin Harris
• David Guetta
• Martin Garrix

Dubstep is a genre of electronic music that originated in South London. It’s characterized by its heavy basslines , sparse rhythms, and emphasis on sub-bass frequencies. The distinctive “wobble” sound and syncopated rhythms give dubstep its unique, energetic feel.

The genre was highly popularized by artists like Skrillex and Zomboy. It often incorporates elements of broken beat, grime, and drum and bass with syncopated rhythmic patterns that give it a unique feel.

Sub-genres: brostep, post-dubstep, and melodic dubstep, among others.

Drum and bass is a fast-paced genre of electronic music characterized by breakbeats, heavy bass, and sub-bass lines. Its high-tempo rhythms and intricate drum patterns make it a favorite among dance music enthusiasts.

Drum and bass is often within the 165 to 185 beats per minute range which is pretty fast. Originally inspired by Jamaican dub and reggae, Drum and bass or DnB now features complex syncopation and has spawned a whole dance culture that is highly addictive and fun.

Sub-genres: liquid drum and bass, Neurofunk, and jump-up, among others.

R&B, or Rhythm and Blues, is a genre known for its soulful vocals and steady, groovy rhythms. It often features themes of love, relationships, and personal experiences.

R&B’s smooth and emotive style has a broad appeal and has greatly influenced different types of music genres like pop and hip-hop. Popular artists like Beyonce and Alicia Keys have kept the genre at the top of the music industry and belted out songs that are considered timeless.

Sub-genres: contemporary R&B, soul, and neo-soul, among others.

• Alicia Keys

Indie music , short for “independent,” is characterized by its deviation from mainstream music norms. It often features unconventional song structures, experimental sounds, and introspective lyrics.

The indie ethos of valuing creativity over commercial success gives this genre a unique and diverse sound. Much like punk, Indie music also favors the DIY approach to recording and publishing.

Sub-genres: indie rock, indie pop, and lo-fi, among others.

• The Strokes
• Arctic Monkeys
• Tame Impala

Trap music is a subgenre of hip-hop music, characterized by its dark, ominous beats and aggressive lyrical content. It’s known for its use of 808 kick drums, fast hi-hats, and synthesized melodies. Trap’s gritty sound and streetwise lyrics have gained it a large, dedicated fanbase.

Sub-genres: drill, emo trap, and Latin trap, among others.

Instrumental music is a genre that emphasizes musical instruments over vocals. It spans a wide range of styles, from classical symphonies to jazz improvisations to electronic compositions. Instrumental music’s focus on melody and harmony makes it a favorite for background music or studying.

Sub-genres: instrumental rock, instrumental hip hop, and ambient, among others.

• Explosions in the Sky
• Chet Atkins

21. Bossa Nova

Known for its smooth, relaxed sound, Bossa Nova is a Brazilian music style that fuses samba and jazz. Artists: Antônio Carlos Jobim, João Gilberto.

Known for its catchy beats and flamboyant style, Disco dominated the music scene in the 70s. Artists: Bee Gees, Donna Summer.

Ska is a genre that combines elements of Caribbean mento and calypso with American jazz and rhythm and blues. Artists: The Specials, Madness.

World music incorporates styles and sounds from around the globe, creating a rich tapestry of musical diversity. Artists: Angelique Kidjo, Manu Chao.

Known for its upbeat rhythms and passionate vocals, Latin music includes styles like salsa, reggaeton, and bachata. Artists: Shakira, Marc Anthony.

Opera is a genre that combines music with theatrical performance. Artists: Luciano Pavarotti, Maria Callas.

27. Shoegaze

Shoegaze is known for its heavy use of effects pedals, creating a wall-of-sound backdrop for the vocals. Artists: My Bloody Valentine, Slowdive.

Funk is known for its groovy basslines and rhythmic, danceable beats. Artists: James Brown, Parliament-Funkadelic.

A form of jazz music known for its upbeat tempo and rhythmic “swing”. Artists: Duke Ellington, Benny Goodman.

A subgenre of rock that emerged in the mid-1980s, Grunge is characterized by its distorted guitars and introspective lyrics. Artists: Nirvana, Pearl Jam.

A genre deeply rooted in the Christian faith, Gospel music is known for its fervent vocals and emotive performances. Artists: Mahalia Jackson, Kirk Franklin.

32. Bluegrass

A form of American roots music, Bluegrass is characterized by its acoustic stringed instruments and complex harmonies. Artists: Bill Monroe, Alison Krauss.

Soul music is known for its deep emotional resonance and incorporation of gospel, rhythm and blues. Artists: Aretha Franklin, Sam Cooke.

34. New Wave

A genre that emerged in the late 70s, New Wave combined the energy of punk rock with electronic music and pop sensibilities. Artists: The Police, Blondie.

35. Industrial

Industrial music is known for its abrasive and often harsh soundscapes, utilizing non-traditional instruments like scrap metal and noise generators. Artists: Nine Inch Nails, Ministry.

From the lush melodies of Classical music to the gritty beats of Trap, the world of music is a vast, vibrant landscape of sound. It’s a testament to human creativity and expression, encapsulating our emotions, stories, and experiences in a language that transcends barriers. No matter your taste, there’s a genre out there that resonates with you, and hopefully, this guide has helped you explore the rich diversity of music.

Brian Clark

I’ve been a writer with Musician Wave for six years, turning my 17-year journey as a multi-instrumentalist and music producer into insightful news, tutorials, reviews, and features.

Music Festival Preparation Checklist – 80 Essential Items

What is Bandcamp Friday (And When is Bandcamp Friday?)

The 20 Best Music Producers of All Time

Sped-up Versions of Songs – Why Are There So Many of Them?

All Music is the same consisting of scales cords approach,etc its no need to label music its either good or bad either the person can play or they can not either they can sing or they can not. ( as the saying goes if you cant run with the big dogs stay on the porch). I listen to all kinds I personally like musicians who are constantly trying new things because with music you never stop learning some musicians grow old before there time because their music sounds the same.( this is just my opinion its not worth much have a good day! )

Genre is a very subjective thing but the problem with the ‘only 2 types good and bad’ is that it is even more subjective than genre. In truth most types of popular music are closely related, containing similar chords beats and often samples. Pop will often have a section of rap or a rock guitar. Many hip hop and electronic songs use latin beats, horn riffs (sampled or real) etc etc. Genre is often no more than a marketing tool so that commercial interests can determine a target audience.

Leave a reply Cancel reply

Save my name, email, and website in this browser for the next time I comment.

Musician Wave is owned and operated by Indigo Media Ltd.

Address: Indigo Media Ltd, 8 The Green, Suite #14648 Dover, DE 19901 United States

Phone: +1(302) 487-1715 Email: [email protected]

Get in touch via our contact page .

This page may contain affiliate links. As an Amazon Associate we earn from qualifying purchases.

• Privacy Policy
• Terms & Conditions

Subscribe to our  YouTube Channel