Biologists: Dolphins and scientists talk to each other using shared primitive language
http://www.helium.com/items/2095342-dolphin-communications

I have no idea, or strong opinion, of the validity of this.

This article is another example science reporting by regurgitating press releases.

The journal article cited presents no results. It is a discussion of how studies of dolphin communication can inform the search for extraterrestrial intelligence.

(I was initially concerned that Acta Astronautica might be a wackaloon “journal.” Anything associated with SETI brings out the crazies. But it appears to be a respectable journal. You really need a program to tell the legitimate researchers from the loons when it comes to SETI.)

For example, dog growls all sound alike to us, but experimental evidence now shows that dogs can tell the difference between an aggressive growl and a play growl (this is semantics).

I can tell the difference between a play growl and an aggressive growl, at least I can with my own dog. And my observations of other dog owners at the park show that they too are quite adept at interpreting the intentions behind a growl.

But that’s not the point. No one is disputing that animal communication has semantic content; it clearly does. The question is whether there are discrete parts to a growl that when changed alter the semantic content of the message. This does not appear to be the case. Growling appears to occupy a continuum with no discrete signals embedded within it. Growls are also inextricably linked to body language and facial expression, which carry much of the content. Human language is associated with facial expressions and body language too, but the vocal components can be separated and carry the message all on their own. This doesn’t appear to be the case with dogs, who are simply signaling an emotional state, “I’m annoyed,” “I’m afraid,” or “This is fun.”

in most cases we lack the experimental evidence to say whether or not these criteria are present, so we make the assumption that they are not.

As a good scientist should. You don’t assume the existence of something that you have no strong evidence for. Of course, a good scientist should also be prepared to reassess when new evidence is presented. And in the field of animal communication there is clearly a lot that we just don’t know.

Personally, I’d be surprised if there were no other animal species on the planet that exhibited language. But as of now, I haven’t seen any strong evidence that any do.

Elephants seem to mourn the death of one of their herd but does this mean they are aware of their mortality and communicate it to their youngsters? Only human language allows that. Animals just use signs, some auditory; danger, pissed off, fight mode, humping readiness alerts, etc.
Next thing you know people will be claiming they have souls and pet ‘good dogs’ are destined for heaven - a real bummer for wild howling wolves and dingoes.

If you can tell the difference between the aggressive and play barks of your dog, you are to be congratulated. A study of the ability of humans to discriminate between these categories of barks found that the humans could not tell them apart, even though the dogs could tell the difference (Molnar et al. 2006. Behavioural Processes 73: 76-83).

And a study of dog barks found that there were structural acoustic differences between barks produced in different contexts (Yin & McCowan 2004. Anim. Behav. 68: 343-335), just like we have structural acoustic differences associated with our words.

There might indeed be emotional encoding as well as a meaningful message in the barks, but how is that different from saying “I am angry” in a neutral tone of voice, vs. saying “I am angry” in a voice that is shaking with rage? My point is that many people make assumptions and then make categorical statement about the absence of meaning based on those assumptions (by the way, I approve of the ... There does not seem to be.... showing a willingness to consider other possibilities).

Some animal vocalizations are analagous to tonal languages, such as Chinese or Navajo, where small differences in tones or sound frequencies can change the meaning of a word. These tones can be assembled into a short vocalization that might sound like a simple “chirp” to us, but when we analyze the acoustic structure, we find a considerable amount of complexity. For example, prairie dog alarm calls sound like the chirps of a bird, but when we analyze the structure and perform experiments to tease apart the different components of meaning, we find that a single chirp might encode “ tall thin human in yellow shirt.” (Disclaimer: all we know is that the prairie dogs have consistent sounds in their alarm calls that are associated with “tall” “thin” “human” and “yellow shirt” and we get a different set of sounds associated with “tall” “thin” “human” and “blue shirt”, where the “tall” “thin” “human” sounds are the same, but the “blue shirt” sound is different from the “yellow shirt” sound. We don’t know what sort of mental concepts prairie dogs have of any of this).

That’s an interesting study. But since the difference in vocalizations was primarily due to the presence or absence of the owner rather than other conditions, the indication is that it is simply expression of an emotional state, rather than communicating a particular message (such as warning of a stranger approaching). This is akin to differentiating a human scream of terror versus a shout of joy; that’s vocalization, but not language. The study gives some tantalizing hints that there may be more going on, but nothing conclusive. There is a lot there for follow-up research.

When I spoke of being able to distinguish my dog’s growls, I was not separating the growls from the body language and other contextual clues. I’m not sure I could distinguish as well on just the sound alone.

Question regarding the prairie dog research: have researchers taken a different set of prairie dog calls recorded under the same conditions, then, properly blinded, can they distinguish the calls relating to the tall men with different colored shirts? It’s easy to tease out patterns in any random sample, but the real test is going back and finding the same patterns in a new set of data. Has this been done?

No matter how eloquently a dog may bark, he cannot tell you that his parents were poor, but honest - Bertrand Russell

Yes, but given the chance a few dogs and most cats could tell you they were poor because they were honest.

With the prairie dog studies, the researchers originally used Discriminant Function Analysis statistics to determine that the calls for the different situations were statistically significant. Then they gave a set of data to another researcher who did not know how the data were coded. The other researcher used Neural Network Analysis, a non-statistical way of classifying data where the computer decides how to measure, analyze, and classify the data based on the Neural Net algorithm. The same patterns emerged in both the statistical and non-statistical forms of analysis.

As I said we definitely need the help of a pet psychic here - not one who relies on cold reading of the pet’s owners but a genuine one, one who has won the Randi million dollar challenge.

All that shows is that the pattern is real and not a fluke of the analytical methodology, which is good but not sufficient. What you really need is another data set that was not used in the original experiment to show that it is not just a random pattern that exists in the original data. This is the standard problem with data mining that researchers make time and time again. You can always find a pattern that correlates with something of significance in a single data set. You need to find that pattern in successive data sets to show that it is real and not just a random occurrence.

I should have made my sentence more specific: the data set used by the Neural Network Analysis was different from the one used by the Discriminant Function Analysis.

Cool. Thanks.