Whale Songs Reveal a Surprising Connection to Human Language

Two new studies have uncovered an eerily human-like sophistication in whale songs, challenging ideas about human exceptionalism and potentially illuminating how language evolved.

One study found that some whales can communicate as efficiently as humans, or even more so, with humpback whales (Megaptera novaeangliae) providing the closest competition. The other found that humpback songs follow a statistical structure regarded as a hallmark of human language.

“These findings challenge long-held assumptions about the uniqueness of human language, uncovering deep commonalities between evolutionarily distant species,” says Simon Kirby, professor of language evolution at the University of Edinburgh.

The whales’ abilities are both impressive and humbling, but they could also deepen our understanding of animal language and our own. A growing body of research suggests that many species have intricate communication systems, and that numerous animals possess qualities once thought to be exclusively human.

Whale songs and the efficiency of human language

In the study of communication efficiency, Stony Brook University ethologist and computational scientist Mason Youngblood applied two linguistic laws to 51 human languages and 65,511 whale-song sequences.

Youngblood notes that natural selection favours efficient communication because it allows individuals to exchange vital information quickly and simply. Complex signals can carry more information, he acknowledges, while redundancy helps ensure that information is transmitted accurately. Those advantages come at a cost, however: prolonged vocalising consumes valuable time and energy and may draw predators.

To measure efficiency in human and whale communication, he used Menzerath’s law and Zipf’s law of abbreviation. Menzerath’s law holds that efficiency rises when longer sequences, such as words, sentences or songs, are made up of shorter elements, such as words, phonemes or notes, Youngblood explains. Zipf’s law of abbreviation states that a communication system is more efficient when frequently used elements, including words, phonemes and notes, are shorter.

Youngblood applied both principles to vocalisation sequences from 16 cetacean species, encompassing baleen whales, dolphins and other toothed whales. He assessed 51 human languages for comparison.

Calls from 11 of the 16 species showed Menzerath’s law to the same extent as, or more than, human speech. The exceptions were orcas (Orcinus orca), Hector’s dolphins (Cephalorhynchus hectori), Commerson’s dolphins (Cephalorhynchus commersonii), Heaviside’s dolphins (Cephalorhynchus heavisidii) and North Pacific right whales (Eubalaena japonica).

Most species did not show Zipf’s law of abbreviation, Youngblood reports. It appeared only in humpback and blue whales (Balaenoptera musculus), and only humpbacks rivalled humans in their adherence to it.

Hidden language-like structure in humpback whale songs

The second study concentrated on humpback whale songs, using quantitative methods normally applied to speech in human infants.

Earlier research has identified a particular feature of human language that appears to support learning and help preserve language. Structurally coherent language units show a frequency distribution governed by a power law known as a Zipfian distribution.

The researchers behind the new study note that this seems to make language easier for babies to learn and “likely enhances the accurate preservation of language across generations”.

Humpback songs are similarly complex and often consist of nested hierarchical components. The whales use individual sound elements to create phrases, repeat those phrases to form themes, and combine themes into songs.

Like humans with language, humpbacks also transmit songs culturally. The study authors reasoned that, if the statistical properties of human languages evolved to make cultural transmission smoother, comparable signs should be present in whale songs.

To investigate, they analysed eight years of humpback recordings using speech-segmentation techniques developed for use with human infants. The analysis revealed hidden structure in the songs, including statistically coherent subsequences whose frequency followed a Zipfian distribution – traits present in all human languages.

The lengths of those subsequences also follow Zipf’s law of brevity, the researchers add. This law states that linguistic units used more frequently tend to be shorter.

“Using insights and methods from how babies learn language allowed us to discover previously undetected structure in whale song,” says first author Inbal Arnon, a developmental psycholinguist at the Hebrew University of Jerusalem.

Behavioural ecologist Ellen Garland of the University of St Andrews adds that discovering this hidden, “language-like” structure was surprising. Whale songs do not have the semantic meaning found in language, she explains, and may be more comparable to human music.

Nevertheless, Garland says the discovery “strongly suggests this cultural behaviour holds crucial insight into the evolution of complex communication across the animal kingdom”.

The studies were published in Science Advances and Science.