(Inside Science) -- Whether songs of love or dance, sung by Beyoncé or the Guarani people in Paraguay, nearly every society makes music of some kind. Music, many might say, is a universal language.
To see if that's indeed the case, a team of researchers has performed what they say is the most systematic and comprehensive analysis of the world's music to date.
"We're starting to know what is essentially the building blocks of what music is and how music works," said Sam Mehr, a cognitive scientist at Harvard University, who's leading the project, which has been dubbed the Natural History of Song. These building blocks would imply a universal musical grammar -- a basic structure on which all the world's music stands.
"It's easy to say music shows cross-cultural similarities," said Manvir Singh, a graduate student in anthropology at Harvard who co-leads the project. "The much harder thing is to figure out what are those building blocks."
The project's new paper, published today in the journal Science, is controversial, with criticism from ethnomusicologists, anthropologists and other social scientists. Citing music's wide diversity, many scholars doubt that universal features exist, said Patrick Savage, an ethnomusicologist at Keio University in Japan, who isn't an author on the new paper but participated in the study by listening to and classifying songs. The new study is "trying to show conclusively to the skeptics that, in fact, yes, there's diversity," he said. "But if you quantitatively analyze the diversity, you see that there's a lot of things that are consistent."
One such consistency, the researchers found, is the behaviors associated with songs. Almost every society the researchers studied sang songs in the context of 20 types of behavior, including dance, infant care, healing, religious activity, entertainment and mourning. To show the universality of these behaviors, the researchers statistically analyzed the written descriptions of song performances ethnomusicologists compiled while studying 60 primarily small-scale societies around the world.
Very few people have tried to analyze these ethnographies in a cross-cultural fashion, Savage said. "And no one has done it in a rigorous scientific way until now."
That's what makes the new analysis unique, said Savage, whose own search for music universals involved analyzing a database of global music recordings. He and his team found several acoustic features common to most cultures, such as simple scales and rhythms.
The new study also analyzed a database containing 118 recordings of songs from primarily small-scale societies in 30 regions around the world. The analysis confirmed one of the group's earlier published findings: Listeners who were unfamiliar with a song's culture could still identify a love, healing or dance song, or a lullaby, suggesting these types of songs -- which serve particular functions -- are universal across societies. While the first study involved 750 listeners, the new one recruited nearly 30,000 online volunteers to classify the songs. The researchers also found acoustic features like musical accents and tempo that are predictive of whether a song is one of the four types.
One shocking discovery, Mehr said, was that expert listeners recruited for the study found almost all of the songs belong to a musical key. "Even I was surprised to hear that," he said.
But the study is not without caveats. For instance, the volunteer and expert listeners were not from the societies being studied, so the way they interpret songs may not be how someone in that particular culture would interpret them. Whether a listener -- expert or not -- perceives a song to be tonal, or is one of love or healing, might depend on what kind of music they've listened to before.
The fact that the study comes from an exclusively western perspective is just one of many fundamental flaws with the study, said David Novak, an ethnomusicologist at the University of California, Santa Barbara.
Reducing a song to a category based on its function and analyzing its meaning based on a few acoustic features ignores the nuance and complexity through which listeners understand and interpret music, he said. Simply extracting terms and data from an ethnography disregards the holistic nature of the text, which contains specific information that's needed to accurately capture the context of the song.
While the question of whether music is universal is a valid one, he said, the problem with the NHS project is that it fails to take into account decades of scholarship and research in ethnomusicology, which shows that the way people understand and interpret music depends on their background and experiences.
The myriad ways and contexts in which you can sing a song can alter its meaning and purpose, he said. A love song sung with irony or sarcasm might not function as a love song anymore. So such a song can only be properly classified in terms of social experience and practice.
Research in ethnomusicology has also found that types of music don't correspond to particular behaviors, as the researchers claim. "That is not an idea that most ethnomusicologists share," he said. These issues don't mean it's impossible to identify universal features in music, but it does put the burden on the NHS project to address them.
Indeed, the researchers should have engaged more with ethnomusicologists, Savage said, noting that of the 19 authors on the paper, none are ethnomusicologists. While he agrees with the study's general conclusions, the resulting conflict may threaten progress in the field, he said. "I'm worried that this project is driving these two sides apart rather than bringing them together."
The researchers say they have improved their methods based on some of ethnomusicologists' criticisms. "Our paper builds on the insights they've had in the past," Singh said. And, Mehr added, the study relies on ethnomusicologists and acknowledges them in the paper.
Despite the study's limitations, it's still important for pushing the research forward, said Sandra Trehub, a psychologist at the University of Toronto, who wasn't part of the study. "It's likely to stimulate research," she said. "And even if it stimulates some criticism, well that's fine too if it leads other people to tackle a similar question in other ways."