Some Thoughts on Dolphin Language

by Simo

Having it occur to me that the main barrier to communication with dolphins, excluding scientists who are still not prepared to find that dolphins may have mental abilities comparable to humans (I have a pet name for this type of "scientist". I call them JADAs) is just common ignorance.

For example:

Dolphins produce many whistles. The function of many dolphin whistles is unknown, but, at least half the whistles are signature whistles. A dolphin's signature whistle ranges between 5 and 20KHz and last about a second. Dr. Lilly called these "slash calls", and determined their duration ranges from 100mS to 600mS. (Mind of the Dolphin, p. 156) Naturally, some so-called "scientists" latched onto this as "proof" dolphin phonations are as limited and as instinctive as a dog's barking. All these folks are marine biologists who don't know what they're really observing. I am not a biologist; my forte is electronics, and has been since I built my first crystal set when I was eight or nine.

I instantly recognized what these whistles really are. I've seen this before. These signature whistles consist of a sinusoidal carrier with a center frequency of 12.5KHz that is frequency modulated by a baseband signal consisting of a sawtooth wave with a frequency varying from 1.67Hz to 10Hz. This is not a "signature whistle"; it is FMCW sonar. The identical principle is utilized in the radar cops use to nab speeders, and the radio altimeters pilots use to determine height above ground while shooting an ILS approach. At least Lilly was on the right track when he called it "long wave length sonar", or "solid angle sonar" (another term he's used) however he over stated its capabilities. (Mind of the Dolphin, p. 156)

FMCW is useful for determining range and speed only. Yet the situation where it's most often observed, a dolphin in familiar surroundings, range and speed is all he'd need to know anyway. So, of course, half the dolphin's whistles are FMCW sonar to keep from running (so to speak) into things. Naturally these whistles are "stereotypical"; they have to be to minimize errors. The reason these whistles vary from one dolphin to another is in order to avoid interference. It is also likely that dolphins superimpose intelligence on the basic FMCW sonar signal so that they can communicate with one another while using their sonar. They may copy another dolphin's FMCW sonar, not as a name, but as a "Hey you!" type of general alert. After all, the profile of a dolphin's FMCW sonar is the first thing another dolphin would know about him.

How does this happen? Would a marine biologist be expected to know anything about sonar or radar? This is clearly out of his field. He makes, what seems to him, a reasonable assumption out of ignorance. And he guesses wrong. On the other hand, an honest "I dont know" has yet to win anyone a prestigious professorship, or one dime in grant money.

It doesn't get any better. The methodologies used by Herman, Lilly, Dreher, etc for dolphin communication research are quite flawed. Once again, we have individuals who are trying to grasp concepts that are way out of their field of expertise. Writer Robert McNally (So Remorseless a Havoc) understood this. He comments on D. W. Batteau's experiment with Maui and Puka:

Unless the translated whistles happened to copy the same sound characteristics and arrangements dolphins use to make words, they would respond to the translator's whistles as cues, not as words (So Remorseless a Havoc, p. 53)

Later in the chapter McNally concludes about dolphin language research in general:

Research into the language ability of the dolphin has assumed that the language, if it exists, must be like human language: sounds in units, which we call words, put into sequences, whose meaning depends upon learned rules, which we call grammar. Dolphins, however, could have a wholly different way of getting their meaning across. (So Remorseless a Havoc, p. 68)

These are a couple of very astute observations that have eluded every dolphin language researcher, regardless of whether or not they are trying to prove the existence, or nonexistence, of a natural dolphin language. What makes it even more remarkable is that McNally isn't all that sympathetic to the idea dolphins may have a natural language, or an advanced intelligence. He may not know why he is correct, however he is right.

Consider what we do know about dolphin-speak. In 1961 Dr. Lilly performed an experiment involving two dolphins' having to use an intercom-like device to communicate. He discovered the optimum passband to be 2.0KHz to 80KHz. With a passband of 5.0KHz to 28KHz, they stopped using the intercom. The optimum bandwidth of dolphin phonations is 78KHz (at least). Biologists fail to take this into consideration (or even to realize that it is a consideration at all).

Dr. Sheri Gish has stated: "The exact information theyre [dolphins] transmitting is unknown to me It doesnt appear to have a syntax or any other attribute of language" (Smithsonian, p. 78) Jacques Cousteau, Rene- Guy Bushnell and his faithful side kick Albin Dziedzic, and the Caldwells have said much the same thing. As McNally observed, they all expected to find something not unlike human language, and, when they didn't find it, voila the dolphin must be just another dumb animal.

Every human language consists of a couple dozen or so simple sounds arranged in temporal sequences to convey meaning. Why is that so? It results from the natural limitations in both the human vocal organs and auditory system. Human speech has a bandwidth of about 2.5KHz. This is standard for telephone lines and the IF filters of most short wave receivers (although some rigs use IF filters as narrow as 1.2KHz for voice work). Due to the limited bandwidth of human speech, the intelligence can reside only in the time domain.

These days, most ham operators have switched from AM to SSB. The advantages offered by AM are ease of demodulation and high fidelity. AM sounds good on the air as it preserves "naturalness". The disadvantages are that AM wastes power and bandwidth. Two thirds of the transmit power goes into the carrier, which conveys no intelligence, serving only to demodulate the transmission to baseband. The rest of the power is in the two sidebands, one on either side of the carrier. As these sidebands are mirror images of one another, they carry the same intelligence. This doubles the signal's bandwidth unnecessarily.

If you get rid of the carrier, and one sideband, you conserve radio spectrum. Also, 100% of the transmit power can be devoted to the intelligence-carrying portion of the signal. (Nor do you have those annoying heterodynes from nearby AM carriers.) This gives SSB a theoretical gain advantage over AM of 12.0db (although, in practice, it amounts to more like a 3.0 -- 8.0db advantage).

What you give up is fidelity. Due to the nature of SSB, demodulation is accompanied by severe distortion. "Naturalness" is lost completely. A SSB signal is unmistakable: it sounds like nothing even remotely resembling a human voice. (indeed, in a good many sci-fi movies, they use the sound of SSB for the voices of aliens and/or computers.) And yet, it is entirely comprehensible, even though you can't tell who's on the other end. (It could be a little girl or an old man, but you'll never know just from how the voice sounds.) Nor does this take any experience; you can understand what is being said even if you never heard SSB before. If the time domain is preserved, then so is the intelligence. Even severe distortion of the frequency spectrum of the voice ruins only the fidelity, not the intelligence.

The intelligence of dolphin vocalizations, however, resides in the frequency domain. As Lilly demonstrated, delphinese is very intolerant of distortion. Upset the frequency spectrum and you not only destroy the "naturalness", but also the intelligence.

The only biologist to ever begin to understand this is Peter Marshall:

These differences in our sensory modalities have made it very difficult for men and cetaceans to learn one another's communication systems. Eyesight in humans is a space-oriented distance-sense, which gives us complex simultaneous information in the form of analog pictures but has poor time discrimination. Our auditory sense, however, has poor space perception but good time discrimination. Human languages are therefore comprised of fairly simple sounds arranged in elaborate temporal sequences. The cetacean auditory system is predominantly spatial, like our eyesight, with much simultaneous information and poor time resolution. So dolphin language apparently consists of extremely complex sounds which are perceived as a unit. A whole paragraph's worth of information might be conveyed in one elaborate instantaneous hieroglyph (sic) (Mind in the Waters, p. 58)

This explanation is badly muddled; Marshall uses totally incorrect terminology and clumsy analogies that betray his lack of knowledge. Yet he is correct in principle.

According to Hartley's Law: 78/2.5 = 31.2
Delphinese can potentially convey over thirty times as much information per second as can human speech: A whole paragraph's worth of information might be conveyed in one "elaborate instantaneous hieroglyph" is not too far off the mark. Marshall is also correct in that optimizing any system's frequency domain performance degrades its time domain performance, and vice versa.

Dolphins could be using that bandwidth for another reason. Sending intelligence slowly over a wideband channel offers protection from corruption by noise. This technique is used to retrieve information from deep space probes over distances of hundreds of millions of miles. Might dolphins use the same principle to communicate over long ranges? Do they do both: exchange large amounts of information rapidly at close range, and more slowly over larger distances? Could this account for Kenneth Norris's observation that news spreads through small pods of dolphins faster than through larger ones?

Herein lies the problem. We know what a time domain language is like; we use one every day. Then we project our experience on the dolphins. If no one has found a dolphin language, that doesnít mean there isnít one. They havenít been looking in the right place. (Note: At the Harderwijk Dolphinarium, they tried to analyze what the dolphins were communicating with an oscilloscope: the wrong equipment as it displays the time domain parameters of a signal. They should have used a spectrum analyzer instead.) Every attempt to teach a dolphin an artificial language, such as those of Herman, Lilly, and possibly Batteau (there is some ambiguity as to how his "transphonometer" actually worked) has all used a time domain language. It doesn't matter whether or not these languages use high frequency whistles (although that makes it easier for the dolphins to hear it in the first place) these have all been time domain languages. I am convinced this is the reason their results have been less than spectacular. Consider it from the dolphin's side. It is as hard for a dolphin to imagine a narrowband time domain language as it is for a human to imagine a wideband frequency domain language. How does a dolphin relate to concepts such as words, sentences, and a syntax of temporal sequences? How does a dolphin get used to a narrowband language that takes a whole paragraph to relate the intelligence of "one elaborate, instantaneous hieroglyph"? (It is interesting that one of Dr. Lilly's dolphins, Elvar, would take snatches of human speech and manipulate its frequency spectrum. Did he do this since he was expecting the meaning to reside in the frequencies, not the timing?) Is it any wonder dolphins seem to be at the same level as young children who are just beginning to acquire language? (This is exactly where they are, trying to learn a language that is more foreign, to them, than even the most obscure human language would be to us.) Is this why Louis Herman felt compelled to give up the acoustical approach to artificial language in favor of the gestural approach? (There could also be another reason. Herman has long said that the dolphins won't "talk back". His gestural approach to artificial language absolutely guarantees that. Is he really interested in language research or just in finding ways to give orders to dolphins to create a more versatile and efficient slave one that can't complain or protest its mistreatment?) Perhaps we should be amazed that the dolphins actually understood as much of these artificial languages as they actually did!

This brings up another problem. The ideal tool to analyze dolphin language is the FFT. Unfortunately, it is "fast" only in comparison to doing the calculations by hand. It also requires tremendous amounts of computer memory. Analyzing a signal, in real time, with a bandwidth just a few kilohertz can greatly slow up any personal computer. (As in performing an FFT on the audio output of a SW receiver tuned to a blank frequency to profile its IF filter characteristic. Equipment and programs to do this may be found in the ARRL Handbook.) A signal with a bandwidth of 78KHz might choke even a Cray in real time. This doesn't necessarily mean it can't be done, it just means it won't be cheap. I can already hear the JADAs bleating: "Its too expensive", "No scientific justification". Yada, yada, yada. Would it be as expensive as landing a man on the moon? What did we get for that: a few thousand pounds of rocks and dust, braggin' rights and revenge for Sputnik. Big deal! And we gave up the technical ability to return to the moon almost thirty years ago. Would it be as expensive as an International Space Station no one really wants or knows what to do with once it's built? How do those results compare to the potential benefits of actually communicating with dolphins?

One can justify such "boondoggles" by pointing out the spin-offs. To be sure, the space program, despite the misguided philosophy behind it, despite the mismanagement, has been worth it for the spin-offs alone. Isn't the challenge of breaking down the communication barrier between humanity and another intelligence going to produce unsuspected spin-offs too?

How to proceed? Should we attempt to translate natural dolphin language (i.e. delphinese) directly? I think it would be more practical to ease into that gradually. Start with an artificial language, as D. W. Batteau wanted to do, while avoiding the mistakes Herman has made (if these are mistakes). Once the lines of communication are at least partly open, then we could tackle delphinese. I would suggest first studying delphinese, and not with an oscilloscope (it's the wrong instrument as it displays time domain parameters of signals) but with a spectrum analyzer. Even better would be the FFT. (Since this wouldn't have to be done in real time, perhaps a PC could be used.) This way, we can get a good idea as to enough of the characteristics of the language to enable us to design a decent artificial language that meets the needs of both men and dolphins. This is clearly better than what we've been doing: forcing dolphins to accommodate our time domain languages. That would give the project a decent chance of success.

Even more important might be to keep it out of the hands of the government. The more I learn about the Navy's dolphin program, the more convinced I become that it is essentially a make-work welfare program for incompetents and failures who can't hack it in the private sector much beyond flipping burgers at Mickey D's. Secondly, it is important to keep the marine biologists in their place. From what I can gather, Lou Herman's artificial language program is long on biological experience, yet he doesn't have any electrical engineers, electronics specialists, or computer programmers, or any one else who might know a thing or two about information theory. Is it any wonder his effort is as pathetic as it is? It is essential to screen out the JADAs, as they have nothing to offer except obstruction. As Kenneth LeVasseur noted, they are more interested in belittling the dolphin's intelligence than honestly assessing their mental abilities. If that's their attitude, then t'hellwiddem. Let them read the details of the break-through in the newspaper like everyone else. Communication with dolphins is essentially a technical and engineering problem; it's a matter mainly for the "spark hunters", not the psychologists and behaviorists (who only seem to get in the way).

As to how to fund such a project, the well-heeled techies of Silicon Valley are privately funding Project SETI. These hackers are obviously curious and open to new ideas. May be they'd be interested in searching for "extra-terrestrial" (i.e. not of the land) intelligence a little closer to home?

Does this mean that men and dolphins will talk to each other? It is highly doubtful you will one day have a real time chat with Flipper. The problem isn't with the required hardware. Digitizing a signal with a pass band of 2.0KHz to 80KHz really isn't all that demanding. A real time FFT presents some difficulties, to be sure, yet computer chips keep getting faster, denser, and cheaper all the time. If delphinese can be translated, there will be a demand for hardware that can get the job done. What we are unlikely to see for the foreseeable future is a language translator. Except for strictly technical material, all language has its subtleties, its nuances, and its shades of meaning that can only be ascertained by "reading between the lines", as it were. This is something no software as ever been able to accomplish. Even between English and Spanish, word-for-word translations will fail. These are two European, Latinate languages that are often quite similar, that share a common origin: the Latin spoken by the Romans. Here is one example:

  1. Su hija es bonita.
  2. Su hija esta bonita

What does this mean? "Your daughter is pretty" Word-for-word:

Su= your
Hija= daughter
Es and esta= is
Bonita= pretty

Seems pretty straight forward doesn't it? However, to a Spanish speaker, #1 is a complement; #2 is an insult. Unlike English, Spanish has two verbs that both translate to "to be": ser and estar. The use of the word "es" implies that the prettiness of the daughter is an inherent characteristic. "Esta" implies that the daughter's prettiness is a temporary condition. She may be pretty right now; otherwise she isn't. In fact, saying "Su hija esta bonita" will be taken as a greater insult than if you had just said the poor girl was ugly, as it sounds sarcastic. (Hopefully, the man or woman to whom you said this would realize what you really meant; however, you will have identified yourself as a gringo.) And that is just one simple example. Spanish is full of these little traps. Other statements, when translated word-for-word, turn into incomprehensible nonsense. The title of a Carlos Santana song is Oye Como Va literally: "I hear how you go". It's been a long time since I took Spanish, so I can't tell you what it means, but it couldn't possibly mean, literally: "I hear how you go". (A good guess would be something like: "I hear you're leaving". The meaning could be determined from other contexts where "oye como va" occur.) The software that can keep this straight has yet to be written and they've been trying since the 1950s. The same Spanish text, unless it's a technical manual of some sort, when translated into English, won't read exactly the same when done by different translators, as there is more to translation than just converting words. Each translator will also try to capture the "flavor" of the Spanish writing, and that is mainly a judgment call. (It is especially bad when religious writings are involved. There probably are as many English versions of the Qu'ran or the Tao Te Ching as there are scholars who've attempted such translations. Nor are there any guarantees that the translator's preconceived notions concerning Islam or Zen don't color the translation. As in journalism, objectivity is an unrealizable ideal.) Computers are incapable of this. And this is with two Latinate languages. I guarantee you: delphinese is not a Latinate language. Just getting a literal translation will be a formidable challenge. Next comes the hard part: analyzing its true meaning. We will be able to translate delphinese texts: stories, legends, mythology, etc., as these can be analyzed at leisure. (Hopefully, the translation will convey at least the gist of the meaning.) Real time communication will still require some sort of artificial language. Ideally, it would be a language that could evolve meanings beyond "Fetch the Frisbee" and so forth. However, that might take years to occur.

Will men communicate with dolphins? I certainly think so. Will men talk to dolphins? Probably not, at least not in the science fiction-ie way envisioned by, for example, Star Trek's Universal Translator™.


Lilly, John C.
Mind of the Dolphin. (Garden City: Doubleday and Co.), 1967.

McNally, Robert.
So Remorseless a Havoc. (Boston: Little, Brown and Co.) 1981.

McIntyre, Joan. Mind in the Waters. (New York: Charles Scribnerís Sons.).

Parfit, Michael.
"Are Dolphins Trying to Say Something, or is it All Much Ado About Nothing".Smithsonian (October, 1980). pp. 73 -- 80.

Walter, George.
The Potential for Communication with the Intelligent, Social Dolphins of the Ocean. (July 27, 1993).

Click here to return to top