Is music universal? Scientists who spent years trying to find out now have an answer

American poet Henry Wadsworth Longfellow famously noted: "Music is the universal language of mankind."

That might seem obvious, but no one's actually been able to show how that assumption is in fact true — until now.

Samuel Mehr at Harvard University led a team that has just completed the most comprehensive scientific study into music as a universal cultural phenomenon — one that took five years and spanned 300 distinct societies worldwide.

The findings were published on Friday in the journal Science. 

Mehr spoke with As it Happens host Carol Off about what they found. Here is part of their conversation.

The answer to the question, "Is music universal?" seems like a given. We make this assumption. It's in poetry, it's in theatre, it's in Shakespeare. But you wanted to find out scientifically if that's the case. What were you looking to find out?

So we wondered first, in the most trivial sense, is music universal? Does it appear wherever people appear in the world? And we show — very straightforwardly — that yes, wherever we study people, wherever there's anthropology of human behaviour, we see music.

But the rest of the work that we're doing is, I'd say, a bit more interesting than that. It's less trivial, in that we're asking how music is universal. What is it about musical behaviours, the things that we do while we make music, that make it similar or different across societies? And what is it about musical forms — the guts of the music, the way it sounds across different societies — that makes it work the way it does?

And what did you find that the societies had in common when it came to music?

We find that when we take annotations of objective facts about the behaviours — things like how many people are there, or what time of day it is — we find that there is some structure to these behaviours: that everywhere in the world, music seems to vary in terms of how formal or informal it is, in terms of how exciting or calming it is, and in terms of how religious or secular that music is.

Those three sets of features characterize different types of songs very well. So they make it clear that a dance song has different behaviours associated with it than a lullaby.

But what they don't do is distinguish very well between societies. So every society seems to have some formal music and some informal music, or some exciting music and some calming music. But there aren't any societies that are only exciting music or only calming music.

So what you found is that no matter where we, no matter where human beings have been, they have lullabies, they have love songs, they have ballads, drinking songs, dance music, hymns. Not just that music itself is universal, but even the kinds of music, the ways we use music is universal, right?

That's right. So the behavioural contexts in which music appears seem to be quite consistent worldwide.

How did you test it? You must have had to collect an awful lot of music in different places.

First we built a database of anthropological texts — ethnographies from 60 different societies around the world. These are reports of anthropologists writing down the behaviours that they see in the world.

From this huge database, we found lots and lots of examples of singing and the behaviours associated with it. 

And then the other database was even harder to build, because we had to go out to archives and libraries, and even contact at the musicologists and anthropologists to provide recordings from their personal collections to build a database of audio recordings where we can actually measure things about the music itself, and not just the behaviour surrounding the music.

One of the big questions that we had in this work was: is there something about the musical forms — the things that the singer is actually producing with his or her voice — that informs the listener about the behaviours that are present? Is there something about what the singer is doing that tells the listener: "This is a song for dancing?" 

And what we find is that, yes, there is. Just the pitches and rhythms that the singer is producing are enough to inform listeners and machine-learning algorithms that this is a dance song.

Did you find anything in your research that wasn't universal, that wasn't unique?

What we're finding are essentially universal building blocks that appear in general worldwide, but then that cultures can elaborate in their own ways. 

So one way of thinking about this is if we play the songs to people who've never heard them before — which we do in this massive online experiment — we had 30,000 people who come to my lab's website, themusiclab.org, and they listen to these songs and guess what are they listening to — and we find that they're pretty good at guessing them.

But they don't get them 100 per cent of the time. In fact, they get them wrong quite a bit. They just get them wrong less than you would if they were answering at chance. 

So this is to say that there's still quite a bit of variability in music worldwide. And our paper doesn't make the claim that that doesn't exist.

Our paper makes the claim that there's there are some universal building blocks, and that those get elaborated, that the diversity in human music turns up as a function of these universal building blocks.

Interview produced by Kevin Ball. Q&A edited for length and clarity. The original post was corrected to reflect that the study spanned 300 societies worldwide, not 60, and involved researchers from multiple universities.