Skip to main content
Loading Explore guides…
Sperm Whale Phonetic Alphabet · Acedra
Home
Explore
Biology
Sperm Whale Phonetic Alphabet
Sign in to save
Chat
If YouTube asks you to sign in, open the video on YouTube.
Watch on YouTube
Reload player
Transcript
37 segments
Copy
Hide video
0:00
I recently found this story in my news feed about a phonetic alphabet proposed for sperm whale vocalizations, and in this video I’ll take a look at the new research this is based on. The topic of animal communication raises fascinating questions about language and what it is, phonetics, meaning, and the possibility of actually communicating with other species. I have no specialist knowledge in this area, I’m just going to be sharing my understanding of,
0:30
and reactions to, this research as a phonetician. So apologies in advance for any misinterpretations or errors, which are of course my own. The sperm whale’s binomial is physeter macrocephalus, where physeter means blowpipe or blowhole, and macrocephalus means large headed. It has the biggest brain on earth, five or six times bigger than a human brain. Now the ratio of brain size to body size
0:57
isn’t so impressive, but apparently the whale brain’s composition, including spindle neurons, suggests emotional and social intelligence. Sperm whales are known for their complex social behaviour. They collaborate to raise their young and go foraging at incredible depths. You can easily find lots of wonderful videos of them hanging out or taking naps together. Their brain certainly isn’t the reason for the enormous head. That’s largely filled with a waxy
1:30
substance called spermaceti that sadly they’ve been hunted for, that includes Moby Dick. The whales use it to control buoyancy and to help transmit the clicking sound they make. As I understand it, sperm whales use one nostril for breathing -- remember they’re mammals and breathe air like us -- and the other nostril for making these clicks, with a pair of phonic lips. If you’re not squeamish you can see Dr Joy Reidenberg dissecting
2:00
a beached sperm whale and exposing the actual articulators, sometimes called monkey lips. The click sounds made by these lips are reflected off the back of the whale’s head, then amplified and beamed out from the front end. At their loudest they’re the loudest sounds produced by any animal we know. However, claims that their loudest clicks can stun or kill turn out to be highly exaggerated.
2:33
In fact this whole area is one where exaggeration and misleading reports are prone to happening. The whale phonetic alphabet story got a lot of coverage, as you can see here on Ground News, which is where I first heard about it. I’ve made videos before on the dodgy reporting of linguistic and phonetic topics, so my priority here was to sort through the less helpful stuff and find the real content. I know from Ground News that NBC News is a generally High Factuality source,
3:03
but in this case it was leading people to think that a whale ‘language’ had just been discovered, which isn’t true. So, going deeper, I found the story in National Geographic, the Smithsonian Magazine, and here, bingo, the actual paper itself, from Nature. If you follow my channel you’ve heard me talk about Ground News before. I was a subscriber before they sponsored any of my videos, not just because of its sheer range -- with literally
3:32
thousands of sources worldwide. These are shown to you with their left-right political orientation, assessed independently. And diverse headlines are easy to compare, here’s that ‘whale language uncovered’ line again, whereas these other headlines are more measured . It all makes the plethora of news easier to navigate, and Ground News can tailor content to your interests -- or let you know what you’re NOT reading about. The My News Bias feature is
4:00
fascinating, because it shows where your news diet has been coming from and lets you see how skewed it is. You might find some surprises. Apparently only 5% of my reading was from local sources, but I did read news about these 60 countries. I can even review the actual topics I read about. To see why I’m such a fan, do check it out through the link in the description, ground.news/drgeofflindsey, and you even can get 40% off the Vantage Plan, that’s the one I have
4:26
Sperm whales uses their clicks in various ways. One is to produce a continuous buzz which I believe is a kind of sonar to check out objects in their vicinity. But the kind of clicking that we’re interested in consists of shorter bursts of clicks lasting a second or two. These vary in the number of clicks they contain, and the rhythmic pattern they make. The bursts are known in the literature as “codas”,
5:00
a musical term which is potentially confusing as it refers specifically to the END of a piece of music. But nowadays “coda” is just the accepted term for these self-contained bursts of clicks. As we'll see, the new paper, by Sharma et al, proposes that there’s more structure in these codas than previously thought, enough complexity to justify analogy with the categories of the International Phonetic Alphabet.
5:26
The research is a collaboration by the MIT Computer Science and AI Lab with Project CETI and the Dominica Sperm whale project, and their analysis is based on nearly 9000 codas recorded from a clan of whales called Eastern Caribbean 1. The researchers identified 18 coda types that have between 3 and 10 clicks, in patterns that are a bit reminiscent of the letters of Morse Code. We all know SOS, dit dit dit,
5:54
dah dah dah, dit dit dit. But whereas Morse Code has two basic units, short and long, or dits and dahs, sperm whales, as far as I’m aware, only have one basic unit, the click. But they can make just as many patterns as the Morse Code system, by making the gaps between the clicks short, or long, or anything in between. If you were going to signal SOS with just dits, I guess it would be ditditdit, dit, dit, dit, ditditdit. Though
6:21
in fact if humans want to send messages with just a single short sharp noise like that, the more common method is what’s called Tap Code, the kind of thing that’s often been used by prisoners. Tap Code works by using groups of between 1 and 5 taps, all the same distance apart. Whereas the whales are able to space out their clicks with various different time intervals. For example, the East Caribbean whales can consistently produce
6:48
three different five-click codas. And two different seven-click codas. And so on. Think about that from the articulatory point of view. The most similar thing in human articulation that I can think of is probably vocal fry, where the vocal folds act like the whale’s phonic lips, producing a series of pops that’s slow enough that we can make out the individual pops. Right. –Oh my God But just try controlling your vocal fry to reproduce rhythm patterns like these consistently.
7:24
So it’s pretty impressive how much variation the sperm whales are producing when apparently the basic sound they produce is just this one click. But they can do a lot more with this click-timing, multiplying considerably the inventory of codas. And this is the main point of the paper from the MIT Computer Science and AI Lab. The 18 coda rhythms that they list can each be produced at different tempos, or tempi if you prefer, just like you can play a piece of music faster or slower.
7:55
But tempo isn’t chosen randomly. Computer analysis shows that they fall into five discrete speeds, ranging from a third of a second to a second and a quarter. You can get some idea of what this means by just looking down there to the right. If you click the cog wheel icon for the settings menu, YouTube currently offers 8 discrete tempos for playing videos, from a quarter of normal speed to twice normal speed So for example we could take our three different five-click codas,
8:26
and hear each one at say Tempo 1, or Tempo 4. If you multiply eighteen rhythms by five tempos, you’ve now got dozens of consistently repeated codas, all from an articulation that consists of one click. Now, as we find in human phonetics, some combinations are far more common than others, and some of the logically possible combinations don't turn up at all, at least not in this dataset. So if we look at our three five-click codas, this rhythm doesn’t occur at the two quickest tempos,
8:56
and this one doesn’t occur at the two slowest tempos. Also these two five-click codas are by far the most common out of the eighteen, accounting between them for over a third of all the sperm whale codas in the database. Above all this one, which seems to be kind of a theme tune for this clan. It's known to sperm whale researchers as one plus one plus three. So despite the predominance of some codas, there are still dozens of rhythm-tempo
9:33
combinations that turn up repeatedly in the dataset. And that’s not all. Another new discovery in this research is that when they’re producing sequences of codas, whales shift from one tempo to another in a particular way. Now purely for illustration and with no claims whatsoever about meanings that the whale codas might have, let's imagine a whale saying a sequence of HELLOs. Hello They tend to produce codas at about four second intervals Hello usually with a lot of repetition Hello
10:06
Often another whale will join in: Hello, what’s up The first whale might now switch to the other pattern: What’s up Now so far I’ve played ‘what’s up’ lasting longer than hello. But remember the different speed possibilities. Here's the first whale choosing a slower tempo. Hello, what’s up And here's whale 2 choosing a faster tempo. Hello, what’s up The authors plot this kind of diagram at right angles, and obviously with clicks instead of
10:59
my letters, and they call it an exchange plot. Now for this video I wanted to show an animated exchange plot, so I was very excited to find this video by computer artist Philip Meyer, who read the same research and decided to generate artificial sperm whale conversations. So we have two whales, one on the left, one on the right. But let’s actually bring this guy down a little more. There we go. So now you can hear them separately, this one’s the bassier whale.
11:35
And especially for my viewers, Philip has made this animation of an actual whale exchange from the article. Now as I said, the whales will often change from one tempo to another during an exchange. And the researchers found that they do this smoothly, gradually, and they do it in parallel.
12:04
The authors choose the term ‘rubato’ for this elasticity in tempo. This smoothly undulating rubato reminds me of intonation in human language. Of course the thing that’s changing isn’t the same. Human intonation is the rising and falling pitch of the voice, superimposed on the words being said. Sperm whale rubato is the slowing down and speeding up of tempo, superimposed on the rhythm patterns the whales are producing.
12:39
Now if you’re like me, you may be thinking, ‘Oh, this rubato is probably just because the whales are slowing down or speeding up as they move through the water’. But no. The tags that were used to record the whales’ audio were also equipped with accelerometers and gyroscopes. And the researchers found no significant correlation between rubato and body motion. And one final discovery. Not only do the whales have something a bit like intonation -- at least,
13:09
I think of it that way -- they also have something that looks, to me at least, a bit like punctuation. The authors call it ornamentation. Four percent of the codas in the dataset have an extra click on the end that doesn’t fit the overall pattern. And these extra clicks or ornaments tend to be at the beginning or the end of coda sequences. To use more linguistic terminology, they seem to be boundary phenomena. Like punctuation. Or maybe emojis.
13:42
In human phonetics, many languages have features associated with utterance edges. In Spanish, for instance, the sounds written b d g are usually weakened but are true plosives utterance initially. In English the ejective sounds that some speakers use as in ‘like’ are most common phrase finally, and so on. The exchange we looked at before contains one ornament, here. Notice that immediately after, the other whale stops, skipping a coda, then starts again.
14:21
All these variables we’ve looked at, according to the researchers, add up to a sperm whale phonetic alphabet. Like the consonant chart of the human International Phonetic Alphabet, the proposed chart for sperm whales has two main axes. The IPA has manner of articulation and place of articulation. The SWPA has rhythm type, that’s the Morse Code-like aspect, and the 5 typical durations or tempos. In each cell of the IPA chart,
14:49
you can have a voiceless or voiced sound, and in each cell of the SWPA chart, we can choose a coda with or without rubato, and with or without ornamentation. The fundamental thing here is combination, or combinatoriality. According to the authors, previous work suggested that sperm whale vocalisations have a maximum information rate of 5 bits per coda, but their combinatorial analysis increases this to 8 bits per coda.
15:21
Which leads us to the unanswered 64 thousand dollar question of what information is being transmitted, in other words the question of meaning. Remember that the categories of the IPA are ones that are capable, in human linguistic terms, of changing meaning. For example, changing place of articulation can change map into nap, which have two clearly different meanings. Manner of articulation can turn tell in sell, and voicing can turn safe into save.
15:53
Now those are differences in word meaning. In human language, sounds that are meaningless in themselves combine and recombine in this way to produce a vast vocabulary of words with clearly different meanings. And then there’s the completely distinct syntactic level where words can be combined in entirely different ways, including recursively, to generate an indefinitely large number of sentences.
16:19
The researchers present this table, making analogies between sperm whale vocalizations and different levels of human linguistic structure. They suggest that the whales’ rhythm, tempo, rubato and ornamentation are building blocks of sound, analogous perhaps to phonemes; these combine to make codas, which they compare to words, a bit like my illustration earlier with hello. And the codas combine into sequences that they compare to sentences.
16:47
However, we don’t as yet have anything like a dictionary of codas with distinct definitions for each one. In fact codas are often said to signal caller identity -- I mentioned before that 1+1+3 clan theme tune -- rather than to refer to things in the world. And when codas are put together, we have no idea whether different sequences can have clearly different meanings, the way Harris defeats Trump means something quite different from Trump defeats Harris,
17:20
or for that matter the way that in some human languages those word orderings would have the opposite meanings. The authors have previously presented a roadmap that they hope will lead to understanding sperm whale communication. But for now that seems very speculative. Of course it’s possible that language isn’t the best human analogue for whale vocalizations at all. I've been referring throughout this video to a different human
17:46
activity. The calls of whales are widely referred to not as words or sentences I am receiving whale song. –Put them on speakers. Human music is a complex combinatorial system, where relatively few notes are combined into bigger and bigger structures, but without language’s duality of patterning, the clear difference that language has between word level and sentence level. Music is associated with emotion and social behaviour
18:19
and like the whales’ coda sequences, music is full of repetition, and we like hearing the same music again and again and again. But music’s meaning is as hard to put into words as the vocalizations of other species. We don't have dictionaries for music, and we can't put music into Google translate, but it means something to us for sure. My favourite quote on musical meaning is this one by the great composer Felix Mendelssohn.
19:03
People usually complain that music is so ambiguous, whereas everyone understands words. For me, it’s the exact opposite, whether for whole speeches or individual words. They are so ambiguous, so vague and so easily misunderstood, compared with music, which fills the soul with a thousand things better than words. The thoughts expressed to me by the music I love are not too indefinite to be put into words, but rather too definite.
Here to help you learn
Sign in to keep a private conversation with your course tutor.
Ask me anything about the material…
Sign in to chat