Sperm whales may use vowel-like sounds — UC Berkeley and Project CETI study reframes whale communication
A UC Berkeley and Project CETI study finds sperm whales produce slow, continuous, vowel-like sounds that mirror core features of human speech, suggesting more structured, language-like communication and prompting AI-assisted efforts to decode these vocalizations.
Key takeaways
- UC Berkeley and Project CETI study finds sperm whale calls show acoustic features analogous to human vowels, a core element of spoken language.
- The work challenges models that treated whale codas as discrete signals — suggesting instead a more continuous, structured system.
- Project CETI combines AI, robotics and large datasets to classify clicks, dialects and behavior; funder data on Project CETI’s funder page cites high technical accuracy for early models.
What the study found
Led by Berkeley linguistics professor Gašper Beguš, researchers analyzed recorded sperm whale vocalizations and compared their acoustic patterns to features common in human speech. Rather than short, fixed clicks, the sounds showed slow, continuous modulations that the team describes as “very, very slow vowels.”
“Very, very slow vowels.”
Earlier models treated sperm whale codas — the patterned clicks whales make — as discrete units carrying fixed messages. This analysis suggests codas may contain continuous acoustic information, potentially allowing whales to encode more nuanced content much like vowels and syllables shape human words and grammar. Read the UC Berkeley news release for full details.
Why this matters for understanding language
Vowels are a defining feature of many human languages: they let speakers string sounds together, form words, and convey subtle meaning differences. Finding vowel-like properties in whale calls narrows the perceived gap between human language and some animal systems, raising the possibility of a deeper, structured communication system in sperm whales.
If validated, this shifts how scientists search for language-like structures in nature and invites cross-disciplinary approaches combining linguistics, acoustics and computation.
Project CETI and AI tools
Project CETI (Cetacean Translation Initiative) aims to decode sperm whale communication using AI, machine learning and field data. The project builds large audio libraries paired with behavioral observations to link sounds to social context. See Project CETI for methodology and field approaches.
According to funder reporting on The Audacious Project, CETI’s models separate whale clicks from background noise with up to 99.5% accuracy, sort click patterns at ~97.5% accuracy, and distinguish dialects at roughly 95.3% accuracy — metrics that help uncover patterns imperceptible to human listeners.
How the research was done
Researchers combined acoustic analysis, linguistic theory and computational models to examine spectral and temporal patterns in codas that resemble vowel traits in human phonology. The study cross-checked acoustic features against models of human speech to assess similarity claims; see the UC Berkeley summary for methodology notes.
Project CETI’s field teams in Dominica record sounds alongside detailed behavioral, social and interactional data, enabling attempts to link specific vocal patterns to observable events — a crucial step toward tentative interpretation.
Cultural and social variation among whales
The study reinforces prior observations that codas vary by group, akin to human dialects. Such differences imply cultural transmission — young whales learning vocal patterns from elders — and require analytic tools sensitive to both shared structure and local variation. For project details, see Project CETI.
Broader scientific implications
If sperm whales use continuous, vowel-like signals, the implications are broad: researchers may need a new blueprint for identifying language-like systems beyond humans. This could lead to technologies that interpret or partially translate whale vocal behavior and reshape debates about animal cognition and communication ethics.
Ethical questions: as decoding advances, scholars and policymakers must consider frameworks for interacting with species that may hold complex social knowledge, guarding against intervention that harms whales or their cultures.
Implications for the Inland Empire, California
Though inland, the Inland Empire sits within Southern California’s broader science and tech ecosystem. Findings like these can spark local interest in marine science, AI and conservation, leading to partnerships, workforce development and public engagement.
Economic impact
Local universities, community colleges and tech firms could pursue grants and applied projects in cetacean research and AI, potentially creating jobs in data science, robotics and conservation outreach.
Political consequences
Riverside and San Bernardino officials might emphasize STEM funding and workforce programs tied to marine biology and AI, and consider policy questions related to ocean protection and scientific partnerships.
Social effects
Schools can build curricula around animal behavior, language and machine learning; museums and nature centers may add exhibits and talks that translate complex science for families and adult learners.
Cultural relevance
The concept of another intelligent species with structured communication can resonate with the Inland Empire’s diverse communities, inspiring events that blend science, storytelling and the arts.
Practical applications for residents
- Citizen science projects and coding workshops using open datasets to teach AI basics.
- Community college short courses on acoustic analysis and data annotation.
- Nonprofit partnerships with coastal groups to raise awareness and support conservation.
Sources and attribution
This article draws on research and reporting from UC Berkeley, Project CETI, and Project CETI’s funder page on The Audacious Project.
