The Current - Turns out whale song and baby talk have something in common

Episode Date: February 12, 2025

Parents may pleasantly puzzle over the “goos” and “gahs” that their babies make, but now researchers say that baby talk shares patterns with the songs of humpback whales. What can that teach us about how children learn to talk, and how language evolves more broadly?

Behind every statistic is a person, a family, a story. This year alone, over 30,000 Canadians will hear the words, you have breast cancer. Breast Cancer Canada is driving groundbreaking research across the country, because every discovery brings us closer to understanding, treating, and one day ending breast cancer. You can join the progress. Donate today at breastcancerprogress.ca.

And if you don't think it will make a difference, we have the science to prove it does. This is a CBC podcast. Hello, it's Matt here. Thanks for listening to The Current wherever you're getting this podcast. Before we get to today's show, wonder if I might ask a favor of you if you could hit the follow button on whatever app you're using There is a lot of news that's out there these days We're trying to help you make sense of it all and give you a bit of a break from some of that news, too So if you already follow the program, thank you And if you have done that maybe you could leave us a rating or review as well

The whole point of this is to let more listeners find our show and perhaps find some of that information that's so important in these really tricky times. So thanks for all of that. Appreciate it. And on to today's show. It's the sound of a humpback whale. Scientists have been gathering underwater recordings of whale song for years trying to decipher the lyrics. While we are no closer to translating the meaning of what you just heard, a new study

published in the journal Science suggests that humpback whale song shares patterns with human language. Male humpbacks sing long, elaborate songs composed of a series of sounds strung together, patterns that sound something like this. You probably don't need me to tell you that, but that's not the sound of a humpback whale. It's a baby human. These new studies brought together marine biologists and human language experts and discovered parallels between humpback whale songs and human baby talk. Inbal Arnon was part of this research team, co-author of the study. She's a cognitive scientist and developmental psychologist in the Department of Psychology

at the Hebrew University of Jerusalem. Inbal, good morning. Hi, good morning, Matt. Thanks for having me. Thank you for being here. This is, the first question is kind of confusing, but it's really interesting. How does a cognitive scientist who studies how babies learn to talk get involved with what whales are saying?

And that's a great question. And I think it's actually part of the beauty of this research. It's truly interdisciplinary. So I study how babies learn how to talk and why it's harder for adults to learn a second language. And what really interests me is why human languages are the way they are. How did humans develop language? How can we learn it? Those are foundational questions that really

interest me. And when I heard, so Simon Kirby, it's this study is co-led by three scientists myself. Simon Kirby from the University of Edinburgh who studies language evolution and Ellen Garland from the University of Edinburgh who studies language evolution and Ellen Garland from the University of St Andrews who studies whales and whale song. And we were at a conference together, the three of us, at a very interdisciplinary workshop about the evolution of language. And we heard each other's talks. And we suddenly after hearing all three of us, we kind of looked at each other and were like, wait a minute, we think we can do something new together. And what we did is to take insight

from how babies learn to talk and apply that to analyze eight years of Havdak whale song. Okay, walk us through this in a way that we would understand and start with what we know about how babies learn to talk. So one of the first things babies have to do to break into language is understand where words end and begin. So speech is continuous. We don't talk pauses between words. That would be very weird. And so babies have to discover what the words are in the language that they're learning. And they have to do that without knowing what they're looking for, right? The baby doesn't know that they should be looking for words. So it turns

out that the way they do this, and this is 30 years of research, is by paying attention to these kind of low-level statistical cues, and specifically to how likely sounds are to appear together. Sounds are more predictable, more likely to occur together if they're part of the same word rather than if they're part of two different words. And maybe I'll give an example to make this clear. So let's say I'm saying a sentence like, look at the pretty baby.

The sounds pre and T are much more likely to appear together than the sound T and bae, right? The end of one word and the beginning of another. Does that make sense? Yeah, yeah, yeah. No, the babies, the babies are, it's like, like you said, it's statistical kind of analysis. The babies are kind of- Exactly, just statistical. Yeah. Yeah. And so they use the statistics to kind of discover the word boundaries. And so we

took that and we said, okay, if whale song also has parts kind of like words, then the same cues should be useful for segmenting whale song. Okay. And what we did is, yeah, go on. No, I was just gonna say, should we hear some whale songs? Yeah. And then you can explain how this works. Yeah, so if you hear the song,

the part of the song, I think it will, yeah. Okay, let me play this and then I'm gonna get you to explain how you did this. Okay so having heard that, what is it that you heard that would help you understand the way that humpback whales are singing? Yeah, are singing. Yeah, so first of all, it's so beautiful, right? It's moving.

And whales are one of the most evolutionary ancient species to have a communication system that's learned. And this is important. So the baby whales learn the song from their caretakers. Or, yeah. And when you listen to that song, you hear that there are different sounds, right? It's kind of whistle and groan and moan. And those sounds are strung together to make a

longer sequence and what we did is we took a whole song of a certain year and we just calculated how likely are two sounds to appear together so how likely is a grunt in a moan or a moan in a whistle and whenever the transitions were more surprising, we cut the song. We said, okay, maybe this is where a word in quotes ends and another begins. And then we haven't what we have is a song that's now divided into these word like units. Do you know whether you've do you know whether you cut it? How do you know whether you cut it, how do you know whether you cut it in the right place? That's a great question and honestly we don't know for sure because we don't know what the

meaning of these word like units are for the whales. And we can't do the kind of experiments we can do with other species where you can kind of play back a sequence and see how they behave, because these are free wild animals, right? But what we did, what we can do is we discovered that these units, they follow another pattern that is found in all human languages, but has never been found in other species. And this is something that has to do with how frequently we use different words. So in human language we have this interesting weird statistical

pattern where the most frequent word in a language appears twice as often as the second frequent word and three times as often as the third most frequent word and so on. And this is basically a power law which may be familiar to listeners from COVID. So it's an exponential. So if you remember how people were getting infected and it was growing exponentially, this is the same kind of pattern in terms of the frequency. And this has been found across all human languages. And it's kind of weird. Why do we get this specific distribution?

So it turns out from work that we did in my lab that these kinds of distributions actually help humans learn. So babies and children and adults learn better when their input has this kind of distribution. So, now I have to make another kind of transition to say that the work that Simon Kirby has been doing on language evolution is work that shows that a lot of the way language is has to do not with something that's special to humans,

but related to the way that language is learned and culturally transmitted. And what does that mean? It means that we learn language from our parents who learn language from their parents and so on. And for the system to manage to be go from generation to the next, it has to be learnable. For the system to manage to go from generation to the next, it has to be learnable. And so languages actually evolve to become easier to learn. And that can apply not just to humans, but also to whales and perhaps birds and to other species as well. Exactly.

Exactly. So this makes this lovely prediction that we should find the same statistical pattern whenever a communication system is learned and culturally transmitted. And like you said, there aren't that many species but there are more than just humans. So this happens in whales and in dolphins and in bats and in a lot of bird song, songbird species. And also it seems in elephants, this is new. So that was the logic behind the study, that if we find the same pattern in whale song,

this is probably because these two species, the similarity between us and whales is in only in that we have a learned communication system. So what these findings tell us is in only in that we have a learned communication system. So what these findings tell us is not only about uncovering a new structure in whale song, which is really cool and shows it's more complex than we thought, but it also tells us something about human language, right?

Suggesting that some of its properties are there because of the way it's learned and are shared with other species in nature. I have to let you go, but does this get you any closer or get us any closer to understanding what these whales are actually communicating? Not quite yet, but maybe hopefully in the future, but not quite yet, sadly. It's fascinating. I mean, I think just the idea that those properties of language are not exclusive to humans, that some of those rules are shared beyond us is really quite powerful in some ways.

Yeah, yeah, it is. Inbal Arnon, thank you very much. Thank you. Inbal Arnon is a cognitive scientist and behavioral psychologist in the Department of Psychology at the Hebrew University of Jerusalem and one of the co-authors of this study looking at the parallels between humpback whale songs and human baby talk.