The Vocabulary of Sperm Whales

It's well known that whales and dolphins can communicate using sound, sometimes over long distances. In some cases, this can involve sophisticated "whalesong" that animals can use, for example, to identify each other, and that may impart other information, too. But there are a great many species of cetacean, and what is true for one won't necessarily be true for all of the others, just as what's true of chimps won't always be true for baboons or marmosets.

We'd expect the most complex messages to come from those species with the most complex social lives. In these cases, it would be useful for the animal to identify not only its gender, fitness, sexual status, and so on, but also which other whales it might associate with, and where it stands in the local hierarchy. Most (but not all) cetaceans live in pods although, for many, we haven't gotten very far in identifying how those are structured. They all, however, have relatively large brains and it can be worth asking, for any given species, just how complex their communication really is.

Sperm whales (Physeter macrocephalus) are the largest of all the toothed whales, sufficiently distinctive that, in modern schemes, they are considered the only living species in their "family". As toothed whales, they are more closely related to dolphins than they are to the toothless whales that produce true whalesong - although, in fairness, they aren't particularly close to the dolphins, either. 

Regardless, sperm whales do not "sing" in the way that humpback and other toothless whales do. Instead, their calls consist of simple clicking sounds, similar to those they use for echolocating prey in the inky depths of the ocean, but slower and with a deeper tone. These are strung together in "codas", series of up to forty rapid clicks lasting no more than a couple of seconds and separated by longer pauses.

The individual clicks within each coda have the same tone and duration, but the coda is structured by how far apart the clicks are. So there might be a regular beat, or a series three far apart followed by two increasingly close together, and so on. (And, since the whole coda is only a couple of seconds long, even by "far apart" we're only talking a fraction of a second). 

Significantly, sperm whales have complex societies, with individual pods arranged into higher-order clans. And each of these clans has a distinct dialect, allowing scientists (and, presumably, the whales) to determine which clan a particular animal belongs to by the sequence of clicks in its codas. Through analysis of the patterns used, we have so far identified 150 different codas across the world, although far fewer are in use in any given geographic area. That's pretty good compared with most other mammals, but it's hardly what you'd call a large vocabulary from our perspective.

Given the otherwise complex nature of sperm whale behaviour, this has led some scientists to ask: "Really? That's it?"

One way to try and answer that would be to take a closer look at a range of codas produced by a single clan and to see if they have further hidden structure that might vary depending on the context of how and when they are used. This is what a recent study did, examining a previously recorded set of 8,719 calls produced by a particular clan of sperm whales.

The clan in question is called the Eastern Caribbean 1 clan, or EC-1 for short. They come closest to shore near Guadeloupe, Dominica, and St Vincent and the Grenadines, with the distinct EC-2 clan favouring Martinique and St Lucia. The clan comprises 400 whales, spread amongst several pods and has been reported to use 21 of the worldwide coda types known from previous studies.

The coda types are built by varying the tempo and rhythm - that is, how long the coda lasts and how the clicks within it are spaced out. However, the analysis showed that two other features are added to the codas, increasing their complexity. Firstly, when the whales produce a string of similar codas, sometimes they add an additional click to one of them - usually, but not necessarily, at the end. Secondly, there is what the researchers called rubato.

"Rubato" is a musical term that refers to slowing down or speeding up part of a piece to convey emotion - it's particularly associated with Chopin, but can apply to any situation where the performer is putting their own expression on the music. Here, it refers to a sequence of codas where the tempo increases and/or slows as the sequence progresses, smoothly altering to add an additional layer of information on top of that provided by the codas alone. In a way, this could be seen as similar to the way that a given language is constructed from a relatively small number of sounds (34 or so in English, depending on what you count) that can be built into an almost infinite array of words.

Significantly, when whales reply by copying a sequence of codas, they imitate the rubato as well as the specific rhythm and tempo, indicating that they are paying attention to all three factors, and it isn't simply some random change. While the additional clicks added into regular codas are not copied in this manner, whales hearing them tended to change their replies in other ways, either starting up a chorus or pausing or finishing one. So, again, they are paying attention to these, even if they're only a "stop/start" instruction, that's still additional information being provided that we didn't previously know about.

Again, this isn't random. Particular patterns of rhythm and tempo are more likely to be combined with rubato and/or extra clicks than others, and many potential patterns just aren't used at all. This also resembles the way that individual phonemes are used in human languages, with no language containing every consonant that it's possible to make (consider the Spanish 'j' or the Welsh 'll') even before we consider the greater complexity of vowels - and how, exactly, you're supposed to pronounce words like "bath". Similarly, you can't expect to randomly string sounds together and always produce a real word, and the same seems to be true of sperm whale codas.

Putting all of this together, the researchers conclude that there are not just 21 "words" in the vocabulary of this particular sperm whale clan. In fact, by their count, there are 143. We don't, of course, know what any of these "words" might mean, or, if, indeed, they all convey something unique. 

But if meaning really is encoded in how individually meaningless codas are strung together that would be something we have never seen in non-human animals before.

[Photo by Vitaly Sokol, from Wikimedia Commons.]