04.03.2025 Mélissa Berthet

Abstact: Human language is extensively compositional: meaningful elements are combined into larger structures (words or sentences) whose meaning is determined by the meanings of the parts and the way they are combined. Compositionality can be trivial (the meaning of a combination is derived from adding the meaning of its parts, as in “blond dancer”) or non-trivial (the meaning of one element modifies the meaning of the other element, as in “bad dancer”).
Are humans the only species whose communication heavily relies on compositionality? And if not, are other species capable of non-trivial compositionality too? Recent research suggests that animals can produce compositional structures, but these are overwhelmingly confined to instances of trivial compositionality and, to date, no comprehensive quantitative investigation of the compositional capacities of nonhuman species has been performed.
Here, we conducted a holistic investigation of the meaning of the vocal repertoire, followed by a step-by-step quantitative analysis adapted from linguistics to explore the compositional capacities of wild bonobos. We found that bonobos produce four compositional structures and that every call type is used in at least one of these four combinations. Most strikingly, three of these compositional structures exhibit non-trivial compositionality. These findings indicate that compositionality is a prevalent feature of the bonobo vocal system, which may display more parallels with human language than previously thought.