Vocal labeling of conspecifics in African elephants

Using learned vocalizations to label referents such as objects or concepts is a critical feature of human speech, yet the evolutionary origins of this ability are poorly understood. Personal names are a type of learned vocal label in which the referent is another individual; thus, analogs of names in non-human animals could provide insight into the evolution of vocal labeling more generally. Elephants are among the few mammals capable of learning to produce new sounds, but it is unknown how they use this ability in the wild. I am investigating the hypothesis that African savannah elephants (Loxodonta africana) use their vocal learning ability to vocally label individual conspecifics.

I work in Samburu and Buffalo Springs National Reserves, a semi-arid savannah ecosystem in northern Kenya. The elephant population at this site has been continuously monitored for over two decades by my postdoctoral supervisor Dr. George Wittemyer and the non-profit conservation organization Save The Elephants. All of the elephants in this population are individually identified by the unique shapes of their ears, and we track their movement patterns, social affiliations, and familial relationships. I recorded elephants in Samburu, noting for each call the identity of the caller, the identity of the receiver, and the behavioral context. I also collaborated with Dr. Joyce Poole to obtain recordings from Amboseli National Park in southern Kenya.

Preliminary acoustic analyses of elephant calls from Samburu and Amboseli suggest that the acoustic structure of a call can be used to predict to the identity of the individual to whom the call was addressed at significantly greater than chance levels. Calls produced by the same caller and addressed to the same receiver appear to be more acoustically similar than calls produced by the same caller to different receivers, suggesting that a given caller may use different call variants when addressing different individuals. Analyses are currently ongoing to determine whether different callers use similar labels to address the same individual.

Elephant rumble vocalization

I am also currently conducting playback experiments in which I present wild elephant families with two calls from the same caller that were originally addressed to different receivers. If the calls are vocal labels specific to a given receiver, then the individual to whom the call was originally addressed should be more likely to respond to the playback.