Dolphin Language Discovery? - The Fail Whales of September

by Rich Mallon-Day - 9/26/2016 7:14:55 PM

Closing out a busy week of Whale and Dolphin story lines on September 11, 2016; beached whales and dolphins in various locations, the beginning of the annual Japanese mass dolphin slaughter at the infamous Taiji cove, Worldwide condemnation of the continuing Japanese "Scientific whaling" program, calls for stricter monitoring of Super Trawlers after numerous suspicious dolphin deaths, Orcas starving in the Pacific North-West due to the depletion of Salmon stocks, the delisting of many endangered Humpback Whale stocks, the final article was a piece from The Telegraph claiming that a Ukrainian scientist has "overheard" a conversation between two captive dolphins. I sighed. With all these important topics from the last week this nonsense is going to be the only story that gets coverage.

After reading The Telegraph article, authored by "Science Editor" Sarah Knapton, my B.S. meter was registering a 5/10. I located the source paper The study of acoustic signals and the supposed spoken language of the dolphins by V.A. Ryabov. Reading this paper I was not impressed by study, it seemed to be a rehashing of other researchers work from the last 50 years. But then I hit a tell-tale sign of a problem, "To determine the role of NP's(packs of mutually Noncoherent Pulses), we analyzed the form and spectum of 50 pulses." Wait, what? 50 pulses? You should be able to record 50 pulses in roughly 3-4 minutes. This paper's conclusions are based on 50 pulses? My B.S. meter jumped to a 7/10. Checking the References section of his paper I noticed a few other things, firstly Dr. Ryabov is the sole author on all of his publication history, meaning he doesn't work with any other researchers, secondly his abstract from the 27th Conference of the European Cetacean Society in April 2013 is the same study with the same 50 analyzed pulses! In the last 3 years he has shown no progress with his study. My B.S. meter is now pegged at 10/10. The only new piece to the study is his call for Humankind to create translation devices to communicate with "the first intelligent inhabitants of the planet Earth".

Reading Scientific Papers 101:

If you haven't ever read a Scientific Research paper let me give you a quick summary of the standard structure of such a paper.

First, there is, as you would expect, the Introduction. In the Introduction the author will discuss the Question or Hypothesis (note I did not say Theory, nothing in Science is called a Theory until it has passed overwhelming and rigorous testing), the author will reference other studies that relate to the Question and why he/she has decided to explore this question.

The second part of the paper will be the Methods section, this is where the author will describe the design of his/her experiment, how they are going to collect data and eliminate potential biases.

The third part describes what happened during the experiment, commonly the author will say something like "We ran this experiment from January through April and observed 500 individual events." Sometimes, the author will comment about events, expected or unexpected, that occurred during the experiment.

Of course, the next section would cover the results and analysis of the data acquired.

The next to last part is the section for Conclusions, this is where the author will discuss what was learned, did the data support or refute the initial question, how this study can be improved and what are the next steps for fixing or extending the study.

Lastly, there is the Reference section, this is a list of all the other research matterials that the author used to build the study.

Keep this basic structure in mind because it provides clues to the problems with the Ryabov paper that any competent Science Journalist should have spotted.

The Ryabov paper: A layman's view

Ryabov starts out stating that "Most species of dolphins produce two types of sounds, which possibly play the role of communication signals in their social relationships. These are packs of broadband pulses and ‘whistles’". Ok, this is commonly accepted by dolphin researchers.

Ryabov then notes that a few dolphin species have never been identified using whistles so that they "do not produce whistles and may communicate by pulsed sounds". What are "Pulsed Sounds"? This is the clicking that is commonly associated with the dolphins sonar.

Ryabov then discusses the difference between whistles and pulsed sounds and what is currently understood about these sounds and their function in dolphin communication. Ryabov comments that most dolphin communication research has fixated on the whistles as the most likely vocal structure for communication simply because the whistles are within the hearing range of humans. Remember, Ryabov is ultimately claiming that he alone, out of all the researchers who have been studying dolphin vocalizations for the past 50 years, has discovered the dolphin language. This is where Ryabov starts making claims...

CLAIM #1: "acoustic signals were recorded using equipment with insufficient dynamic range" So, according to Ryabov, all previous researchers have been using the wrong equipment.

CLAIM #2: "the position of the dolphins relative to the hydrophone (the animals were swimming freely) were not taken into account" Ryabov's conclusion? "Perhaps that is why the authors of these studies failed to clearly identify which acoustic signals of the animals could be regarded as communication." For these reasons dolphin communication researchers have concentrated on the lower frequency ranges, not because of a supposed lack of equipment with sufficient range.

At this point Ryabov describes the basic setup of his experiment, two dolphins stationed side by side with a hydrophone on either side of the pair.

Configuration of the experiment (a) and photograph of the quasi-stationary position of the dolphins during the recording of sounds (b). Positions 1 and 2 are dolphins Yana and Yasha, respectively; 3 and 4 are the hydrophones of the first (I) and the second (II) channels; 5 is the walkway; 6 and 7 are the long and the short sides of the pool; 8 is the pool bottom.

CLAIM #3: "This technique has allowed for the first time to ascribe each signal to a specific animal". This claim has several problems; firstly the dolphins are in a small concrete tank which will bounce their vocalizations all over the place, secondly the dolphins are stationed at the surface with their heads partially exposed, the water surface will both reflect their sounds but will also absorb a percentage of the sound as well, thirdly the hydrophones are to the side of the dolphins, dolphin pulse vocalizations are highly directional and the best records are only acquired directly in front of the dolphin, this hydrophone placement is going to give you a highly distorted recording. One solution to this problem is to attach the hydrophone to the dolphins forehead with a suction cup.

CLAIM #4: The results of the studies give reason to regard all acoustic signals of dolphins as sounding signals of not one sonar (as discussed earlier) but at least six different sonar types. Wait! What? Ryabov is claiming he can differentiate dolphin pulsed sounds into six different sonar types? This is a major claim that he offers no evidence of. But here we come to the most telling part, to me anyway

CLAIM #5: "To determine the role of NPs, we analyzed the form and the spectrum of 50 pulses." Recording 50 pulses with an average duration of 1/4 second should take less than 3 minutes. Is he really basing this on such a small data set? Ryabov states that he did a spectrographic analysis of these 50 pulses, but he doesn't show them to illustrate how he is distinguishing these six different types of sonar.

CLAIM #6: "However, no identical pulses were discovered among them. This suggests that, most likely, each pulse in the NP packs is a word of the dolphin’s spoken language, and a pulse pack is a sentence, i.e., some kind of message." Well, this is quite a claim, with no evidence given.

CLAIM #7: "the dolphin pronounces all the phonemes of a word simultaneously" WAIT! WHAT? This is where we get to the heart of the problem of using pulsed sounds for a language, their duration, these sounds average only 1/4 second and yet Ryabov claims that each of these is an individual word. Again, without any spectrographs illustrating his claims, no one can be expected to believe this. There is simply no precedent for pronouncing all of the phonemes of a word simultaneously, this would just make the sound simple noise.

CLAIM #8: Yana and Yasha don't talk at the same time This claim implies that the dolphins are carrying out a conversation with syntactical structure that indicates when one dolphin has completed it's sentence and the other dolphin can reply. It has long been noted that dolphins don't echolocate at the same time, nor do they echolocate directly at another dolphins head. Firstly the echolocation of one dolphin would be confused if another echolocated at the same time, echolocating at another animals head would be pretty rude, akin to screaming in their faces. However, the fact that the dolphins are stationed side by side it is easier to imagine a simple non-verbal signal between the two dolphins is controlling this.

CLAIM#9: "Since the spoken language of the dolphin consists of spectral extrema that act as phonemes, we can hypothesize that it has both phonological and grammatical structures, so dolphins can create an infinite number of words from a finite number of spectral extrema, which can in turn create an infinite number of sentences." As we have seen every part of this statement is unsupported by any evidence provide in this study. In addition, the lack of any follow-up in the 3 years since this study was first presented indicates that even the author can't justify it's conclusions.

Is this study completely worthless? Not at all, ultimately, it is not the nature of Science to make such judgements. It is entirely possible that Ryabov is correct, it is interesting that some species of dolphins don't use whistles, but there is a tremendous amount of basic work that needs to be done before conclusions of this magnitude can be made.

If this study had remained in the realm of the scientists working in dolphin communication there would be no issue, the paper would be discussed, it's short comings would have been noted, perhaps Ryabov would have taken this information and refined his experiment to eliminate the perceived problems and maybe something interesting would have come from it. However this is not what happened. Somehow the Science Editor of The Telegraph got ahold of this study and in the interest of a quick article did a completely inadequate reading of the study, she then dashed off her article with no clear understanding of what the study was saying. Additionally she included information about other studies, some of which are highly suspect in the eyes of other researchers.

The Telegraph article.

For the most part Sarah Knapton simply repeats, without question, the statements and conclusions made by Ryabov. Except she made the comment that "Yasha and Yana could create sentences of up to five 'words', but scientists still do not understand the content." Nowhere in Ryabov's paper is this claim made. Ms Knapton is apparently confusing the work of Dr. Lou Herman, who created an artificial "language" with simple syntax to test the dolphins comprehension ability. Dr. Herman found that dolphins could correctly understand sentences up to five "words" in length. Dr. Herman's work is a classic example of how a good study would be carried out. He started by assigning sounds and trainer hand gestures to common objects in the tank with the dolphins, eg "ball", "hoop", "basket", "float"... He also included actions, "jump", "take" etc. during the early stages he would have the trainer signal the dolphin "Ball take basket", this was correctly interpreted if the dolphin took the ball and put it into the basket. The problem of non-verbal cueing was controlled by having the trainer wear goggles that effectively blindfolded the trainer, so there was no chance the a glance from the tainer would show the dolphin what to so. As time went on the dolphins showed facility of understanding novel sentences of up to five words. To remove the cuing problem even more Dr. Herman removed the trainer from the poolside and placed them in a studio with a camera feed to a television viewable in an underwater widow in the tank. Ms. Knapton then references four other studies with no context and limited understanding of their findings as if these support what Ryabov is claiming.

Science Journalism is a specialized branch of journalism which requires a huge quantity of natural skepticism. The mantra of Science Journalism should be Carl Sagan's famous quote...

"What counts is not what sounds plausible, not what we'd like to believe, not what one or two witnesses claim, but only what is supported by hard evidence, rigorously and skeptically examined; Extraordinary claims require extraordinary evidence."
The Ryabov paper is loaded with extraordinary claims but is lacking even the most basic evidence.

What's the harm?

You might question whether this deserves such an in-depth drubbing, but let me quote Dr. Richard Connor of the University of Massachusetts-Dartmouth

“The biggest problem, is that now when people make real scientific discoveries on dolphin communication, the public, having been exposed to this nonsense, will not be impressed because they will think Russian researchers already showed that they have language.”

An event like this can have more important ramifications, in the early 1990's a researcher I knew was studying the dolphins near Hilton Head, South Carolina, who used a previously unknown technique of feeding. These dolphins would rush the tidal mudflats in the back channels pushing a school of fish ahead of them. The fish would be easy pickings for the dolphins who had slid out of the water onto the mud flat with the fish. This feeding behavior has become known as "Strand Feeding". As it turned out a well known marine science Cinematographer was in Hilton Head and heard of this behavior, my associate was contacted and he took the Cinematographer out to film the behavior. The Cinematographer filmed a few occurrences of the behavior over an hour or so and then left. Several months later this film appeared in a nature documentary, with no mention of the original researcher, additionally a claim was added that the dolphins coordinated their rush to the beach with a vocalization. My associates funding committee pressed him hard as to why this Cinematographer was able to discover something in one hour that he had not discovered in months of field studies. He didn't discover it because it was deliberate fraud to make a better documentary.

In Stark Contrast

Two weeks after the Russian dolphin language story broke, the CBC published a wonderful documentary Darkwave - Underwater languages at the brink of extinction. This documentary shows what true scientific research is learning about whale and dolphin communication and is based on tens of thousands of hours of work spanning decades of time.

The Fail Whales of September

From here a number of other "reporters" picked up The Telegraph article and started all manner of misquoting. The principle concern is that until the end of the week it is apparent that no one went back to the original paper to see if the claims were valid, Knapton DID include a link to the Ryabov paper so finding it was not an issue. Several of these "reporters" even changed the location of the facility from Ukraine to Russia. Many of the "Reporters" conflated the other studies that Knapton had mentioned into the Ryabov study and came up with bizarre mélanges of the available facts. Only a very few real reporters reached out to other dolphin researchers to see if this study’s conclusions were warranted. Jason Bittel of The National Geographic and Madeline Gressel of Nautilus wrote excellent articles interviewing other researchers and questioning the study and it's conclusions.

Here is the list of those other Reporters, in chronological order of their publication, I'm nominating for a "Fail Whale" award for the "Outstanding Neglect of Basic Science Journalism - Extraordinary Claims require.. Oh I can't be bothered".
Sarah Knapton - The Telegraph (
HELIUMRAT - Drudge Report (
Vittorio Hernandez - International Business Times (
James Ried - The NewDaily (
Will Worley - Independent (
Shane McGlaun - Slash Gear (
Hannah Osborne - International Business Times (
Linda Hervieux - Newser (
Lauren Scrudato - Laboratory Equipment (
Jaymi Heimbuch - Mother Nature Network (
Ryan Whitwam - Geek (
Emma Prestwich - The Huffington Post Canada(
Josh Davis - IFL Science (
Maria Gallucci - Mashable (
Andrew Tarantola - engadget (
Ben Westcott - CNN (
Naia Carlos - Nature World News (
Jenny Marchal - BABW News (
Gordon Hunt - Silicon Republic (
R. Siva Kumar - Headlines & Global News(
Bob Yirka - R&D Mag (
Dan Seitz - UPROXX (
Eric Grundhauser - Atlas Obscura (
Erika Owen - Travel and Leisure (
Cliff Jenkins Scott - Capital Berg (
Tim Sandle - Digital Journal(
Tex Dworkin - Care2 (
Grace Baldridge - Pop Triggered - The TYT Network (
Karl Grovenor - Techly(

And an especially depressing "Fail Whale" to veteran reporter...
Charles Osgood - The Osgood File ( @OsgoodFile

At the last minute I decided that the original Fail Whale shouldn't be tarnished by association with these failures so I give you the appropriate version!