Cetacean chatter
« previous post | next post »
Well, I'm not so sure about this:
What I find much more interesting and compelling is the video embedded in this article:
Humpback whale songs are structured like human language
Languagelike patterns in whale songs could make them easier for whales to learn
By Alexa Robles-Gil, Science (6 Feb 2025)
I strongly encourage LL readers to watch / listen to the video. What you will hear are eerie, enchanting humanoid sounds: eructations, belches, burps, whistles, squeaks, croaks, groans, melodic glides; repeated bilabials, velar stops, and other phonemic signals that, at least to my ear, may convey meaning.
The mysterious grunts and moans of the humpback whale have long captivated humans—so much so that we put recordings of them onto the Voyager spacecraft to convey the sounds of Earth to other life forms. A new study published today in Science reveals an unexpected similarity between human and humpback vocalizations: The songs have a statistical structure similar to that of human language.
“This is a really elegant paper,” says whale biologist Shane Gero of Carleton University, who was not involved with the work. “When we listen long enough and when we look, we find more complexity in these animal communication systems.” The findings suggest this structure might exist because whale songs, just like human language, are communication systems transmitted through social learning, Gero says.
Some animals, such as dogs, make their vocalizations instinctively—they don’t need to learn how to bark. But like human language, humpback whale song is culturally transmitted. Male humpbacks learn the songs, thought to be used to attract mates, from other males. Also like language, humpback whale songs have patterns and structure—individual “elements,” such as a single grunt, combine to form phrases, strung together into “themes” that make up a song, which can last 30 minutes.
So does whale song have some of the features that make it easier for human babies to learn language? To find out, whale biologist Ellen Garland from the University of St. Andrews and her team turned to babies for inspiration. Infants, confronted with a stream of nonstop language, must figure out where the boundaries of words are. They learn to discern individual words by detecting statistical patterns. The sounds within a given word are repeated often, making this chain of sounds predictable—but it’s less predictable which word will come next, so these “dips” in probability hint at a word boundary. Garland and her team segmented recordings of whale song using the same technique.
…
When Garland’s team applied the method to 8 years’ worth of songs from a humpback population in New Caledonia, she was “dumbfounded” to find that whale song structure aligns with a pattern found in human language. Across different languages, researchers have found a predictable relationship in how often common and rare words appear in language: For instance, the most common word in English (“the”) appears twice as often as the second most common word (“of”). This statistical pattern—called Zipf’s law—is thought to make language easier to learn. And the humpback whale song showed a similar pattern. This suggests Zipf’s law might emerge in any complex, culturally transmitted communication system.
The findings don’t suggest whales have a language, where combinations of sounds have fixed meaning and join together in grammatical structures, Garland emphasizes. But the research offers scientists an “amazing window” into how this core property of human communication appears in other species.
To me, these whale sounds are not meaningless, but I have no idea what they mean.
Selected readings
- "A new look at sperm whale communication" (5/8/24)
- "Sperm whale talk" (5/15/23)
- "Orca emits speech-like sound; reporters go insane" (1/31/18)
- "Moby Zipf" (6/1/19)
- "Dolphin naming?" (5/9/06)
- "Dolphins using personal names, again" (1/23/13)
- "Cetacean needed" (3/27/24)
[h.t. Cynthia Hagstrom]
Philip Taylor said,
March 15, 2026 @ 12:56 pm
I really wish that I had not waded through the comments to the original article — Homo sapiens appears capable of trivialising (or mocking) anything and everything …
Barbara Phillips Long said,
March 15, 2026 @ 1:42 pm
One musician’s reaction to hearing humpback music when he was in the U.S. Coast Guard, and the way he reciprocated. (No whale sounds.)
Bob Zentz singing Ocean Station Bravo:
https://www.youtube.com/watch?v=SfTH9cOJsUY&list=PL4v8pmOOhaUStPbeutHARPvil1rJc9ypw&index=3
Philip Taylor said,
March 15, 2026 @ 2:09 pm
This (https://www.youtube.com/watch?v=f3uwKCF-pyo) was my introduction to whale song (somewhere around 1970, listened to with my then-girlfiend in her flat in the Jewish Vegetarian Centre, Golders Green, London). It moved my to tears then, and it does so to this day.
AntC said,
March 15, 2026 @ 6:41 pm
Before we get too dewy-eyed, the "structured like human language" means no more than that the pattern of sounds follows Zipf's distribution/power law. Then there are bazillions of examples of distributions quite remote from human language that follow the same pattern, and that suggest nothing about ease of learning for babies, such as book sales, sizes of files on your hard disk. See wikipedia Zipf's Law 'Statistical explanations' for random text-generators whose output also follow the Law.
We've often discussed on the Log whether 'word' is an applicable concept in all human languages. I'd have no predisposition to think whale-song consists of words.
Without probing into whether any of these authors know anything about Linguistics, perhaps what they're segmenting into are more akin to phonemes than words?
maidhc said,
March 16, 2026 @ 3:33 am
Scientist says we’ve got whale song all wrong
Eduardo Mercado believes humpback whales are singing as sonar, not to find love
https://www.cbc.ca/radio/quirks/why-whales-sing-9.7022970
Michael Vnuk said,
March 16, 2026 @ 5:15 am
The 'Science' article says: 'Infants, confronted with a stream of nonstop language, must figure out where the boundaries of words are. They learn to discern individual words by detecting statistical patterns.'
That might be one of their tools for decoding language, but most people I see talking to infants will often emphasise key words (and often with movement or pointing, or a clear object or person nearby).
'Do you WANT to have some APPLE? APPLE?'
'Time to GO in the CAR. Let's go in the CAR!'
'SAY 'Ta-ta' to NANNA. TA-TA, TA-TA!'
Once they know the main words, other words can be found by subtraction of known words. But I'm not sure that infants have actually decoded these extra words. Does any infant know what 'to' or 'the' mean? It's hard enough for adults to define them. I'm inclined to think that infants gradually learn to fit templates.
Later, perhaps when they are young children not infants, they might start to use more sophisticated means to decode language as described above.
David Marjanović said,
March 16, 2026 @ 11:38 am
What is "this"? A Facebook video? All I can see is a blank line.
Peter Grubtal said,
March 16, 2026 @ 1:34 pm
As AntC intimates, the screaming headline "decoded whale language", "translated songs" is not just totally overblown, it's just plain wrong. The full text of the article makes that clear.
Barbara Phillips Long said,
March 16, 2026 @ 1:35 pm
The Associated Press has an article about the earliest known humpback recording, found at Woods Hole and dating from the 1940s:
Just as significant is the sound of the surrounding ocean itself, said Peter Tyack, a marine bioacoustician and emeritus research scholar at Woods Hole. The ocean of the late 1940s was much quieter than the ocean of today, providing a different backdrop than scientists are used to hearing for whale song, he said.
The recovered recordings “not only allow us to follow whale sounds, but they also tell us what the ocean soundscape was like in the late 1940s,” Tyack said. “That’s very difficult to reconstruct otherwise.”
Barbara Phillips Long said,
March 16, 2026 @ 1:37 pm
Link to AP article:
https://apnews.com/article/ocean-whale-song-recordings-oldest-867e200d248f5d55453797d8b8dcf935
Philip Taylor said,
March 18, 2026 @ 2:13 pm
"Eduardo Mercado believes humpback whales are singing as sonar, not to find love" — I ordered Professor Mercado's book two days ago, it arrived yesterday, and I read the first chapter in bed last night). Although I found some of the idioms that he used ("squat", meaning nothing; "no-pants dance"; "bro chants", …) somehow inappropriate and off-putting, overall I feel that the book is well written and look forward to reading the remainder.