For more than a decade, Elie observed and recorded the sounds the birds made and classified the calls according to the situation and the bird that made them. She then used machine learning to analyse what and how information was encoded in the calls. Finally, she ran tests that showed the birds agreed with her classification.