Extensive compositionality in the vocal system of bonobos
Compositionality, the capacity to combine meaningful elements into larger meaningful structures, is a hallmark of human language. Compositionality can be trivial (the combination?s meaning is the sum of the meaning of its parts) or nontrivial (one element modifies the meaning of the other element). Recent studies have suggested that animals lack nontrivial compositionality, representing a key discontinuity with language. In this work, using methods borrowed from distributional semantics, we investigated compositionality in wild bonobos and found that not only does each call type of their repertoire occur in at least one compositional combination, but three of these compositional combinations also exhibit nontrivial compositionality. These findings suggest that compositionality is a prominent feature of the bonobo vocal system, revealing stronger parallels with human language than previously thought. One hallmark of human language is the combination of elements into larger meaningful structures, a pattern referred to as compositionality. Compositionality can be trivial, in which the two parts are added together to give meaning, or nontrivial, in which the meaning in one part modifies the meaning in the other. Recent research has found the presence of trivial compositionality across a number of species, but it has been argued that nontrivial compositionality is unique to humans. Berthet et al. used a large dataset of bonobo vocalizations in conjunction with a distributional semantics approach and found that not only did they display compositionality, but three of the four types were nontrivial. ?Sacha VignieriCompositionality, the capacity to combine meaningful elements into larger meaningful structures, is a hallmark of human language. Compositionality can be trivial (the combination?s meaning is the sum of the meaning of its parts) or nontrivial (one element modifies the meaning of the other element). Recent studies have suggested that animals lack nontrivial compositionality, representing a key discontinuity with language. In this work, using methods borrowed from distributional semantics, we investigated compositionality in wild bonobos and found that not only does each call type of their repertoire occur in at least one compositional combination, but three of these compositional combinations also exhibit nontrivial compositionality. These findings suggest that compositionality is a prominent feature of the bonobo vocal system, revealing stronger parallels with human language than previously thought. One hallmark of human language is the combination of elements into larger meaningful structures, a pattern referred to as compositionality. Compositionality can be trivial, in which the two parts are added together to give meaning, or nontrivial, in which the meaning in one part modifies the meaning in the other. Recent research has found the presence of trivial compositionality across a number of species, but it has been argued that nontrivial compositionality is unique to humans. Berthet et al. used a large dataset of bonobo vocalizations in conjunction with a distributional semantics approach and found that not only did they display compositionality, but three of the four types were nontrivial. ?Sacha Vignieri
doi: 10.1126/science.adv1170
