After each excerpt had been played, volunteers were asked what they thought the song’s function was, and how sure they were of that on a scale of one to six. The possibilities offered were: “for dancing”; “for soothing a baby”; “for healing illness”; “for expressing love for another”; “for mourning the dead”; and “for telling a story”. The first four of these were real functions, as stated by the people from whom the song in question had been collected. The last two were made up, and were included as foils.
Dr Mehr and Mr Singh found that volunteers’ perceptions of a song’s function were generally in good agreement with its actual function—with one exception. Dance songs were particularly easy to identify. They rated 2.18 points higher on the certainty scale as being used “for dancing” than lullabies did; 1.38 points higher than love songs; and 1.09 points higher than healing songs. Similarly, lullabies were rated 1.53 points higher than dance songs as being “to soothe a baby”, 1.42 points higher than healing songs and 1.19 points higher than love songs.
Healing songs proved a bit more troublesome. They scored only 0.47 and 0.31 points higher than dance and love songs respectively for “to heal illness”, and were statistically indistinguishable from lullabies. The outlier, though, was love songs. Listeners could distinguish them from healing songs, but not from lullabies or dance songs.
Why love songs were hard to identify is unclear. Because such songs involve showing off to the object of one’s affections, they may require more creativity, and thus generate more variety than lullabies or dance songs. Perhaps the fact that both dancing and cooing are involved in romance confused listeners. This genre aside, however, Andersen was clearly onto something.
This article appeared in the Science and technology section of the print edition under the headline “Beyond Babel”