@digs and @solost and maybe @Treppenwitz –
(and any other Dopers who are interested)
Based on your recent remarks in another current thread, this article may be of interest, in addition to being of potential interest to readers of this thread:
Can Animals Think? by Eugene Linden, Time Magazine, March 1993.
Discusses how, and how much, animals may be communicating among themselves (and with their humans).
Also mentions Alex the Parrot, Koko the Gorilla, Clever Hans the Horse, and others.
UPDATE, 22 years, 1 month after OP:
Science is on it:
In 2021, Gruber officially launched the CETI Project (Cetacean Translation Initiative), a global interdisciplinary collaboration devoted to cracking the code on interspecies interpretation and communication, starting with the resident population of sperm whales in Dominica, in the Eastern Caribbean. The list of partner institutions includes the Dominica government, and about 50 cryptographers, linguists, technologists, and biologists who are heavyweights in their own fields. There’s natural language processing expert Michael Bronstein from Oxford University, Harvard University roboticists who specialize in extremely gentle technology for humane animal research, and Roger Payne—the American biologist who, 56 years ago, discovered for the first time that humpback whales sing, sparking the marine conservation movement. Tech superpowers like Google, Amazon and Microsoft are on board, too.
More at the Daily Beast. Why Translating Whale Speech Might Help Us Talk to Aliens One Day
A National Geographic article on dolphins said that after decades of research we have not figured out anything resembling a language in their communication. It pointed out that dolphins can engage in complex coordinated behavior without any kind of communication that we can decipher. Are we starting out any further along with whales?
Here’s a link to what I think is the underlying scientific article, which I’ve only skimmed. https://www.sciencedirect.com/science/article/pii/S2589004222006642
It builds on other animal communication work, including that of dolphins. I couldn’t find post-2020 work on Zipf’s Law, so I’m not sure whether Humpbacks or Sperm whales have potentially higher communication content than dolphins. I’m not quite convinced that we, “Haven’t found anything resembling a language”, among dolphins. From the article:
A comparatively long list of skills required for language learning in humans has been demonstrated among cetaceans (whales, dolphins, and porpoises), who share many social characteristics that are strikingly similar to our own. Whales and dolphins are among a few animals capable of vocal production learning (the ability to copy novel sounds as well as to vary those to produce individually distinctive repertoires) in addition to some birds, bats, pinnipeds, and elephants (Janik and Slater, 1997, 1998; Poole et al., 2005). Of those, only a few species, including parrots and dolphins appear to use arbitrary, learned signals to label objects or conspecifics in their communities in the wild (Balsby and Bradbury, 2009; Janik and Slater, 1998; King and Janik, 2013; Tyack and Sayigh, 1997; Wanker et al., 2005). Dolphins can use learned vocal labels to refer to and address each other when they meet at sea (King and Janik, 2013; Quick and Janik, 2012). This sort of vocal recognition system mediates highly dynamic societies among cetaceans, which involve social relationships lasting decades as well as regular interaction with strangers (Bruck, 2013; Connor, 2000; Gero et al., 2015, 2016a; Tyack, 1986).
Dodging the paywall, here’s an excerpt from the 2015 National Geographic article:
Dolphins make a large variety of noises, but “despite a half century of study, nobody can say what the fundamental units of dolphin vocalization are or how those units get assembled.” Some scientists may suspect that dolphins have a language, but they do not have hard evidence for it, and others doubt whether one exists at all:
“There is also no evidence that dolphins cannot time travel, cannot bend spoons with their minds, and cannot shoot lasers out of their blowholes,” writes Justin Gregg, author of Are Dolphins Really Smart? The Mammal Behind the Myth.
I don’t find that conclusive. The multiple forms of dolphin vocalization are surely doing some sort of communication, though whether it’s a language is unclear. I don’t see evidence of the contention that if they had a language, it would be obvious. We do know that attempts to teach dolphins English have been unsuccessful, which I guess puts some sort of cap on their capabilities.
The new work is hoping to take advantage of large language learning models.
I believe that Clever Hans was shown to be pretty much a parlor trick, with the owner indicating correct responses by subtle moves, sounds, gestures, etc. some maybe even inadvertent. Turned out it was more like Clever Horse Owner.
There’s a good article on the OP’s topic from Nov 2020 by Kershenbaum et al entitled, “Shannon entropy as a robust estimator of Zipf’s Law in animal vocal communication repertoires”, which unfortunately is over my head. One of the building blocks of information theory is Shannon entropy: the intuition is that unexpected sounds convey more information than expected ones. From Wiki:
The core idea of information theory is that the “informational value” of a communicated message depends on the degree to which the content of the message is surprising. If a highly likely event occurs, the message carries very little information. On the other hand, if a highly unlikely event occurs, the message is much more informative. For instance, the knowledge that some particular number will not be the winning number of a lottery provides very little information, because any particular chosen number will almost certainly not win. However, knowledge that a particular number will win a lottery has high informational value because it communicates the outcome of a very low probability event.
Kershenbaum et al note that efforts to assess animal language complexity struggle with small datasets outside the human domain. They outline 3 recommended testing methodologies depending upon whether the communication is high or low entropy and whether it appears to have a high or low vocabulary. Their introduction characterizes the existing scientific consensus (notwithstanding my previous post on the CETI Project (note pun: SETI - CETI)) . Emphasis added:
All animal species have evolved communication systems to suit their ecological requirements, but these systems vary greatly in their complexity, and the volume of information that can be transmitted, received and interpreted. Humans appear to be an extreme case, with the capability of conveying essentially unlimited information in our language, while relying on a finite set of signal elements. Non-human animals, on the other hand, are generally thought not to possess any true language (Cheney & Seyfarth, 1998; Fitch, 2005)… Comparative studies between human language and non-human animal communication systems continue to be important for identifying the conserved features of these systems and explaining any differences that emerged (Fitch, 2005).
Cheney & Seyfarth’s paper is entitled, “Why Animals Don’t Have Language”. It emphasizes ape studies, though it covers a few dolphin ones as well. They argue that while some species use labels, there’s little evidence for creating new calls and new labels for novel objects. Similarly, while they can be taught sentences, they don’t spontaneously create new sentences on their own. I suspect that this sort of thinking is what TriPolar’s National Geographic article was alluding to. I speculate that CETI got funding because larger whales haven’t been subjected to these sorts of inquiries to the extent that dolphins have.
The Kershenbaum paper uses datasets from 11 species including humans, orangutans, baboons, orcas, bats, and mockingbirds. It wasn’t a comprehensive study: it didn’t include dolphins, gorillas, or chimpanzees. I struggle to interpret figure 8 but AFAICT bats are language geniuses based upon their entropy. Humans cover the k=~-1 range, as do a few species of birds depending upon the metric. But again, I don’t quite grasp this paper. (Also, I haven’t figured out how to post an image here- I tried.)
I’ll observe that just as language is not universal among primates, it’s conceivable that it could exist among some whale species and not others.
Something that bothers me about the entropy based studies is that we don’t have a good grasp on the variations possible within the signal that can carry information versus background. Humans have terrible entropy in the spoken word if you just measure the emitted sound. We need kilobits per letter. Within the bandwidth and reasonable signal to noise of human speech on a phone line we can encode tens of kilobits per second. We don’t converse at those rates.
We really don’t have a good grasp of how a cetacean might encode meaning over the top of a base carrier. A dolphin clearly can encode a huge amount of information into its sonar system. It can modulate the emitted sound over a wide range in order to optimise its location and imaging capabilities. It would be reasonable to imagine that, if it communicated, it would do so with a language based upon its sonar capabilities. Maybe emitting a clip stream that is characteristic of the form of: sound used to optimise location of something - even when it isn’t there, or some characteristics of the reflected sound from something. It could have a grammar based on object morphology. The grammar might not be sequential in nature.
We would struggle to determine what and where in the sound the information lives. It could be encoded in all manner of ways. Some idea that we are going to find language and grammar in the form we are generally used to is likely naive. Chompsky not withstanding.
I think we may exaggerate the difference between “speech” and other animal communication, like we’ve exaggerated “tool use”. Crows don’t build computers or combines, but neither do i. And crows do fashion small tools to fish ants out of containers. And parrots can be taught to initiate zoom calls with other parrots.
Animals don’t just say “hey baby” and “this is mine”. They also say “watch out, there’s a hawk”, “here’s food we can share”, “I’m not going to hurt you” and various other things we’ve translated. Apes, parrots, and crows can asks humans for specific toys or food by name.
Do whales talk? I dunno, and I’m sure it varies by species. But it’s worth studying.
Thanks for looking up the Nat Geo article. I only saw it briefly while waiting at a doctor’s office. The paragraph you included is what I remembered. It is clear that dolphins to communicate somehow, but I don’t think it follows that what we consider a language is necessarily the means. Formal language is certainly not the only way humans communicate.
Indeed, I recall some study where it was stated that meerkats had different alarm calls for snake or hawk (which of course is simple and obvious), but apparently were also discovered to be making consistently different sounds to tell each other about the various humans who were conducting the study - and it wasn’t as simple as snake/hawk/human - they had different sounds for each individual human approaching - so language more like “it’s that guy with the blue shirt again”
“It’s the useful one”
I’m bumping this thread with this recent article. Dolphins coo at their babies.
I find the concept of animal sentience utterly fascinating. Just this morning my cat Lucy “chirped” at me to let her out, then returned about fifteen minutes later with a live lizard for her son, Ricky, to hunt and chase.
Animals absolutely have thoughts, feelings, and opinions. I think our ability to communicate is an evolutionary adaptation that has allowed humans to thrive, but I also think that it’s undeniable that other animals have the rudiments of the same ability.
I think that understanding how Clever Hans did his tricks makes it more impressive, not less. I can do simple arithmetic problems. But I can’t read humans as well as Hans did.
--Golf clap-- Very nicely played.
I feel like this thread is incomplete without a link to:
Because he spoke to fish in the creek
He tried to tell us that the animals could speak
Who knows? Perhaps they do
How do you know they don’t
Just because they’ve never spoken to you?
— John Denver
I was swimming in the ocean yesterday in roughly waist deep water. As often happens a school of a couple hundred slender fast 6" fish zipped back and forth nearby. Sometimes enveloping me in their school, other times veering off when a foot away.
Perhaps they were saying to me: “There’s a shark nearby”.
More ominously, perhaps they were saying to the shark: “Eat him; there’s a lot more fresh meat there than all of us put together. And he’s slow.”
Fish are bastards.
The waggle dance is particularly interesting because it’s like a combination of language and democracy at the same time - the more bees there are reinforcing the same instructions (i.e. because there is a good nectar source so more returning bees are talking about it), the more of the other bees will believe it.
Back to the OP though, I think it’s always arguable that a language could exist that works in a completely foreign way to our notions of the working of languages; we might be looking for some sort of consistency that occurs in the presence of specific objects, indicating that the language has nouns, except it just might not work that way. It might not have words that mean ‘rock’ or ‘water’ on their own; it may have none of the same constructs as any form of language we have or can conceive.
Or it may have those concepts, but they might be hidden beneath a layer of something like compression or encryption that is inherent to the mechanics of communication; by way of illustration: take two text documents including the phrases “This is a red pencil box” and “I would like some red apples”; compress the text files then transmit them over an analogue modem - the data will be converted into electronic warbling; no matter how carefully you can examine the audio waveforms, you are not going to find consistency between the two that relates directly to them both containing the word ‘red’ - it’s in there for sure, but it’s not there to see/hear - unless you can decode the compression, the language behind it is completely impenetrable.
You could measure the Shannon entropy though. Across a certain time span at least: such spans may differ among species. Shannon entropy is used in discussions about file compression and animal language. Higher information value is related to the unexpectedness of the next sound: see my post number 26. But then again, bat squeaks have information dense Shannon entropy, but presumably for the purposes of echolocation rather than communication.