Back

The phonology of sperm whale coda vowels

Begus, G.; Dabkowski, M.; Sprouse, R.; Gruber, D.; Gero, S.

2025-06-10 animal behavior and cognition
10.1101/2025.06.09.658556 bioRxiv
Show abstract

In previous research, sperm whale codas (structured series of clicks used for communication) have been shown to resemble human vowels acoustically. Based on the number of formants, two different coda quality categories have been described: a-codas and i-codas. In the present paper, we demonstrate that sperm whale codas not only resemble human vowels acoustically, but also pattern like them across several dimensions. First, traditional count- and timing-based coda types interact with coda "vowel" quality (a vs. i). Second, a-codas are generally longer than i-codas. Third, the duration of i-codas has a bimodal distribution, showing a contrast between short i-codas and long[i] -codas. Fourth, the baseline coda length differs across whales. And fifth, edge clicks mismatching their coda often match an adjacent coda, a phenomenon that resembles human coarticulation. All five properties have close parallels in the phonetics and phonology of human languages. Sperm whale coda vocalizations thus represent one of the closest parallels to human phonology of any known animal communication system. SIGNIFICANCE STATEMENTSperm whales communicate using series of clicks known as codas. The codas acoustically resemble human vowels. In addition, they pattern in ways similar to human sound systems. For example, different coda types are correlated with particular click qualities, and their durations are intentionally controlled. This shows that sperm whale vocalizations are highly complex and likely constitute one of the most sophisticated communication systems in the animal kingdom. By studying it, we may be able to gain a broader understanding of animal intelligence and social behaviors, determine the impact of human activities on whale habitat, develop strategies to protect whales from threats such as noise pollution and ship traffic, and advance legislation which facilitates that protection.

Published in Proceedings of the Royal Society B: Biological Sciences (predicted rank #6) · training set

Matching journals

The top 4 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.