Can CHAT translate dolphins’ clicks and whistles into intelligible words?
One of the fundamental lines we’ve drawn between ourselves and the rest of the animal kingdom is that we have a complex, abstract language. Big data is helping researchers comb through the vast amount of information that we can’t decipher – like grunts and clicks – to get to the repetitions that form the basis of mammalian language. The Cetacean Hearing and Telemetry (CHAT) is a wearable device that researchers like Denise Herzig of the Wild Dolphin Project are developing, with the goal of eventually translating dolphin speech. Recently, CHAT has achieved an important milestone toward this goal: it detected a live dolphin “word” for the first time.
However, the language wasn’t entirely the dolphins’ own. It sometimes takes a distinctive bit of data – a kind of shibboleth – to make sure that you’re using the right methods to pick through the whole mass. In this case, Herzig taught the dolphin pod with whom she has been working for 25 years a whistle that signified ‘sargassum,’ a type of seaweed. When the CHAT picked up the pattern for ‘sargassum,’ the device played back a recording of Herzig herself saying the word. Though the whistle was of a higher frequency than the version that the researchers had taught the dolphins, the pattern was the same, and the translator still picked it up.
Other researchers are naturally skeptical.
“It sounds like a fabulous observation, one you almost have to resist speculating on. It’s provocative,” Michael Coen, a biostatistician at the University of Wisconsin-Madison, told the New Scientist.
It’s hard to know whether the dolphins really knew what they were saying. Certainly, more information will have to emerge before any significant conclusions can be drawn from the research. However, Tharner’s algorithms have discovered eight components of 73 whistles and matched them to certain behaviors, such as interactions between mother and child.
Aside from the fun dolphin implications, data scientists are intrigued by this work because it supports their hypothesis that repetition is a basis for meaning. Algorithms, potentially measured by wearable devices, could one day find the signature of any activity where information is transmitted, and our connections to other species could become clearer than ever.