Off the coast of Dominica, a small Caribbean island nation, researchers have spent years lowering hydrophones into deep water and tagging sperm whales with acoustic sensors, capturing thousands of the rapid-fire clicks these animals exchange as they socialize near the surface. Now, a peer-reviewed study published in April 2026 in Proceedings of the Royal Society B reports that those clicks are organized into patterns that function remarkably like vowels and consonants in human speech.
A multidisciplinary team of more than a dozen researchers from UC Berkeley and the Cetacean Translation Initiative (Project CETI), drawing on a data-collection effort that has spanned several years of fieldwork in the Caribbean, identified two distinct vowel-quality categories within sperm whale codas, the short, patterned click sequences these animals use to communicate. The discovery adds a phonological dimension to earlier findings that whale codas operate as a combinatorial system, and it raises pointed questions about how complex communication can arise in a species whose evolutionary lineage split from ours roughly 90 million years ago.
Two “vowels” hidden in whale clicks
Sperm whale codas may all sound like staccato clicks to the human ear, but spectrographic analysis tells a different story. The study found that codas contain two types of spectral signatures, labeled “a-codas” and “i-codas,” that differ in their formant-like spectral peaks. The distinction mirrors the way the vowel sounds in “father” and “see” differ in human speech: each occupies a different region of acoustic space.
A-codas tend to be systematically longer in duration, while i-codas appear in both short and long variants. Perhaps more striking, the researchers documented coarticulation-like effects: adjacent clicks within a single coda influence each other’s spectral shape. In human language, coarticulation is what allows listeners to parse a continuous stream of sound into discrete units. Finding an analogous process in whale clicks suggests the animals are not simply producing isolated signals but are shaping sequences with internal acoustic structure.
Pratyusha Sharma, a machine-learning researcher at Project CETI and one of the study’s lead authors, has described the moment the spectral clustering first emerged from the data as unexpectedly clear. “We were not looking for vowels,” Sharma noted in a UC Berkeley summary of the work. “The structure showed up on its own once we let the analysis run without forcing human-language categories onto it.” That element of surprise, shared across the team, underscored how little scientists had previously understood about the fine acoustic architecture of whale clicks.
Building on earlier discoveries
The new paper extends a line of research that has been building momentum since 2024. An earlier study published in the MIT Press journal Open Mind first proposed that coda spectral patterns include vowel-like and diphthong-like features. That work analyzed spectral glides within individual clicks and argued that whales exercise structured control over these acoustic features. Diphthong-like glides, where the spectral peak shifts during a single click, parallel the way English speakers produce the “oi” in “coin.”
A separate 2024 study in Nature Communications provided the structural foundation. Working from an annotated dataset of 8,719 codas recorded near Dominica, that team showed whale vocalizations combine tempo, rhythm, and two context-sensitive features called rubato and ornamentation into a large inventory of distinct coda types. It was this combinatorial structure that led researchers and journalists to describe the system as “alphabet-like”: individual features combine and recombine, much the way letters form words.
Together, the three papers establish that sperm whale communication operates on at least two phonological axes: the quality of the spectral peak and its trajectory over time. The datasets behind all of them draw on recordings collected by Project CETI and the Dominica Sperm Whale Project, processed through an automated detection and annotation pipeline described in Scientific Reports. That infrastructure allows scientists to track which whale produced which coda and in what social context, a level of resolution that older manual methods could never achieve at scale.
What the research does not claim
The strongest findings here are structural: sperm whale codas contain measurable spectral contrasts and follow combinatorial rules. Whether those contrasts carry semantic meaning, the way human vowels help distinguish “bat” from “bit,” is a separate question, and one the published studies do not claim to have answered.
The “alphabet” framing is an analogy for the system’s building-block logic, not evidence that whales encode propositional content the way humans do. When press releases from UC Berkeley describe whale communication as “closely paralleling” human language, that language is useful shorthand but compresses real scientific uncertainty. The parallel holds at the level of acoustic structure. At the level of meaning, intention, and grammar, the evidence is not yet there.
Geographic scope is another open question. The annotated coda datasets come from a single population of eastern Caribbean sperm whales. Clans in the Pacific, Indian Ocean, and elsewhere use different coda repertoires, sometimes described as dialects. Whether the a-coda and i-coda distinction holds across those populations has not been tested. Without cross-population data, the vowel-like categories could reflect a local acoustic convention rather than a species-wide phonological principle.
Behavioral interpretation is similarly unresolved. Researchers have not yet linked specific coda types to specific actions, whether coordinated dives, prey sharing, or social bonding. The combinatorial richness of the system suggests it could support complex signaling, but that suggestion rests on analogy with human phonology rather than direct observation of how whales respond to particular sequences.
What ocean noise could disrupt in whale phonology
For anyone tracking cetacean cognition, the practical significance is that sperm whale social bonds depend on acoustic systems far more structured than scientists previously recognized. That realization carries weight for conservation.
As ocean noise from shipping, military sonar, and industrial activity increases, disrupting these acoustic systems could degrade the social coordination whales rely on for foraging and raising calves. The new phonological findings give conservation scientists a more precise framework for measuring what noise pollution might actually be interfering with. Rather than treating all whale clicks as equivalent, future impact assessments can now ask whether specific spectral categories are being masked or distorted. That is a finer-grained question, and it could lead to different policy answers than the blunt metrics used today.
The research also reframes a broader scientific debate. For decades, complex combinatorial communication was treated as a hallmark of human cognition. Sperm whales, with brains six times the size of ours and social structures built around matrilineal clans, are now offering the clearest evidence yet that the building blocks of structured communication can evolve independently, in an environment and a body plan utterly unlike our own.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
