As humanity emerged from the social isolation of the COVID-19 pandemic, people craved human connection. Instead, we got generative artificial-intelligence programs, complete with hallucinating chatbots and deepfake songs. Yet, music from diverse sources does find ways to unite people around the world, from the animated musical film KPop Demon Hunters to Spanish singer Rosalía singing in 14 distinct languages on her album Lux and Puerto Rican rapper Bad Bunny performing in Spanish at the half-time show of the 2026 American Football Super Bowl LX.
Could music be the “universal language of mankind”, as poet Henry Wadsworth Longfellow noted in 1835? Might it even be humanity’s antidote to the inhumanity of the AI era? The nature of music is a subject of hot debate.
Don’t talk science, play science: translate your data into music to improve its reach
Nineteenth-century scholars, including Longfellow, Herbert Spencer and Charles Darwin, were fascinated by music. Darwin described it as “among the most mysterious” faculties with which humans are endowed. Darwin and Spencer argued vigorously but inconclusively about whether music was unique to humans, and whether it evolved from language or vice versa1.
And heated discussions continue to this day, regarding whether music bonds people together more than language does2, and the extent to which non-human species share our musical capabilities3,4.
In my book Comparative Musicology1 (freely available at go.nature.com/4akfrne), I revisit several of these debates using the latest cross-cultural and cross-species data. Here, I outline some of those thoughts.
What can science say about music?
Since Longfellow and Darwin’s time, several controversial attempts have been made to scientifically test ideas about music’s universality. But it has proved difficult to pin down.
For example, during the 1960s, folk-song collector Alan Lomax invented a method called cantometrics for comparing songs across cultures. He rated songs according to 37 acoustic features — including rhythm, melody, vocal timbre and blend — and applied his technique to thousands of songs from hundreds of societies around the world5.
‘Shake it off’: Taylor Swift’s changing voice shows how our accents evolve
Lomax concluded that songs trace the evolution and migration of human societies, reflecting and reinforcing their social structures and cultural histories. By studying how the complexity of songs and societies developed around the world, he thought, human history might be mapped.
However, later reanalyses of Lomax’s work and genetic studies, which were impossible in the 1960s, did not support his conclusion. An analysis of 121 societies, for which matching genomic, musical and linguistic data were available, found instead that musical diversity is mostly independent of genetic and linguistic histories6. Songs do not universally follow the same paths as the people who sing them, or even the languages they are sung in. Instead, songs have their own stories to tell about the complexity of human cultural history.
Half a century after Lomax’s cantometrics, the Natural History of Song team at Harvard University in Cambridge, Massachusetts, came up with yet another answer7. The researchers analysed ethnographic texts written by anthropologists studying diverse societies, which, they showed, all had something the anthropologists labelled as song or music. They also found that listeners could identify dance songs, lullabies and healing songs more often than would be expected by chance. All of these similarities convinced the team that Longfellow’s idea of music as the universal language of humankind seemed to have been right.
Performers at Goroka’s annual sing-sing festival in Papua New Guinea.Credit: Norbert Eisele-Hein/imageBroker/Shutterstock
Other scholars — ethnomusicologists in particular — disagreed with that view on principle and challenged even the idea that scientific studies of music were meaningful1. Some described Western scientists studying the universality of music as ‘imperialist’ and ‘ethnocentric’ for imposing Western concepts on other societies without consulting them. Other critics contended that, even if music were universal, its meanings might not be.
How can scholars decide who is right? Like so many academic debates, it depends on definitions: what do we mean by ‘universal’, ‘language’ and ‘music’? And who gets to decide on those meanings?
Can we define music?
In the Garland Encyclopedia of World Music (1998), a 1997 audio recording of men from Madina village in Papua New Guinea performing is described as “sounds of large and small friction blocks, a swung slat, ribbon reeds, and human moaners” (see go.nature.com/4abgwjp). Is this music? The ethnomusicologists who recorded the clip thought so. But it doesn’t contain either of the hallmarks found in more than 99% of the world’s music: discrete pitches and regular beats8.
What did the Madina men consider their performance to be? It’s hard to know, because it is one of many secretive rituals found throughout Papua New Guinea, and they might have been forbidden from disclosing its meaning to outsiders.
What do dolphins talk about? One biologist is trying to listen in
However, societies of many cultures insist that what outsiders think is musical is not actually music. For example, many Muslims consider music to be divinely forbidden and that the Islamic sung recitation of the Koran is not music.
Western culture also includes sounds that challenge definitions of music. In 1952 in Woodstock, New York, composer John Cage staged the first performance of his composition entitled 4′33″, which consists of a musician sitting silently in front of their instrument for 4 minutes and 33 seconds (divided into three movements). The performer does not play a note, yet Cage calls it music.
Can scholars come up with a universal, objective definition of music? I argue that we cannot. Attempts to do so, such as ethnomusicologist John Blacking’s description of ‘humanly organized sound’, fail to distinguish between song and speech — both of which are found universally in all human societies. Researchers can, however, give a useful working definition of music as ‘sound organized into regular pitches or rhythms’, which characterizes most of the world’s music8 but does not generally occur in speech9.
Is music uniquely human?
Blacking’s description also fails to account for the possibility of music being produced by non-human animals or generative AI, both of which arguably can do so. Whereas spoken language is considered unique to humans, songs are not. The capacity to sing complex, culturally learnt (not genetically innate) songs has evolved independently several times in distantly related species, including whales and songbirds10. This has led some, including Darwin, to propose that human language evolved from a musical protolanguage, which might itself have first evolved to help us to bond or to attract mates1.
Although the ability to synchronize movements to a beat was once thought to be uniquely human, this capacity is now known to be shared with other animals. The first cases to be found were all in animals (such as the famous dancing cockatoo Snowball) capable of vocal learning — the capacity to mimic learned vocalizations. This capacity has evolved several times in songbirds, cetaceans and humans, but not in non-human primates. Over the past few years, however, scientists have published evidence of beat synchronization in non-vocal learning animals, including sea lions11 and macaques3. Although some of the results are debated4,12, musicality includes a range of capacities. We might share musicality with our non-human relatives more than previously suspected.
Machines, too, might be more musical than we thought. Developments in generative AI have seen the emergence of not just large language models that can mimic human language, but also large music models that can mimic human music13,14. Like their language counterparts, large music models provide a number of potential benefits — including cheap, convenient music on command and controlled stimuli for experiments — but they also have drawbacks, such as environmental and ethical concerns related to resource use, copyright violations and deepfake recordings that mimic artists without their consent1.
Is music a language?
But there does seem to be something special about music that speaks to us regardless of our culture or language. Several decades of cross-cultural psychology experiments show that listeners from distant cultures (such as those of Japan, India, the United Kingdom and Cameroon) can detect some of the intended emotional meanings in each other’s music1.
Generally speaking, fast and energetic music tends to be perceived as happier and is more likely to accompany dancing, whereas slow and less energetic music is often considered sadder and is more likely to be used in calming genres, including lullabies1.
However, this is not always the case. In one study7, listeners were able to identify dance songs and lullabies in unfamiliar languages around half the time. This was much better than their ability to identify love songs (26%) — effectively the same as choosing one of the four provided options randomly. But it’s nowhere near perfect.
And there are other counterexamples. Although many societies find ‘dissonant’ intervals (such as minor seconds or tritones) unpleasant, the Tsimané people (Indigenous Amazonians with limited exposure to Western music) generally do not15. Moreover, Western listeners usually find music in major keys happy and in minor keys sad. By contrast, Papua New Guinean listeners felt no difference in happiness between major and minor keys, and Pakistani listeners found minor keys happier than major ones1. In these cases, music does not seem to be a universal language.
What do you hear?
Arguably the most direct test of the universal language theory was led by Elizabeth Margulis, who studies music at Princeton University in New Jersey. Rather than having participants rate abstract emotions such as happiness and pleasantness, Margulis and her collaborators asked them to describe, in their own words, the kinds of story that they imagined when listening to music16.
The scientists curated a sample of 32 instrumental music recordings: 16 by Western composers and 16 by Chinese ones. They then asked some 600 participants in the United States and China to listen to those recordings and describe their imagined narratives.
For instance, when listening to an excerpt from Guan Pinghu’s Strains of Spring Morning, US participants imagined ‘cowboys’, whereas Chinese people conjured up ‘sorrow’. Likewise, for an excerpt from Ferde Grofé’s Sunrise, Americans imagined ‘birds’ and Chinese participants ‘man’. These experiments confirm that Longfellow’s ‘universal language’ metaphor is wrong in its most literal sense: the narrative meanings of musical sounds are not communicated in the same way across cultures. US and Chinese participants interpret the pieces’ meanings differently on the basis of their distinct exposures to similar musical styles.
