Cultural evolution creates language-like structure: from humans to humpback whales and beyond

Abstract

All known languages are made up of statistically coherent sequences – words – whose frequency distribution follows a power law known as a Zipfian distribution. Despite the ubiquity of these features across languages their origins are poorly understood. In this talk, I will argue that they arise because they facilitate learning and therefore emerge through the process of cultural transmission of language. I will present a set of results on the learnability sources and consequences of such distributions in human language, looking at infants, children, and adults. I will then summarise results from an iterated learning study in which non-linguistic sequences evolve as they are transmitted from generation to generation of participants. We draw on insights from infant speech segmentation to develop analytic pipelines for analysing the sequences and observe the emergence of Zipf’s law over generations. This work makes a prediction that we should find Zipfian distribution of statistically coherent sequences wherever systems culturally evolve, including in other species. In the second part of the talk I will turn to the culturally evolving song of humpback whales and apply the same analytic technique to 8 years of whale recordings. Together with Ellen Garland and Simon Kirby, we show, for the first time in another species, that these characteristic statistical properties are indeed present in whale song. By doing so, we demonstrate a deep commonality between two species separated by tens of millions of years of evolution but united by both having culture. Throughout, I will highlight open questions at the intersection of developmental psychology, language evolution, and comparative cognition, and point to ways in which cross-species and cross-method collaborations could promote our understanding of the origin of complex communication.

Scientific Bio

Prof. Arnon is a developmental psychologist and cognitive scientist (PhD, Stanford, 2011). She is a Full Professor of Psychology at the Hebrew University, and currently a Leverhulme Trust Visiting Professor at the University of Edinburgh. Her work focuses on how children learn language, why they do so better than adults, and how studying language learning can help us understand how human language evolved to begin with. Prof. Arnon has worked extensively on first language learning, developing a novel framework for understanding why children are better language learners than adults, with applied implications for human and machine learning (The Starting Big Approach, see Arnon, 2021 for a review). Her current projects ask how learning and learnability pressures shape language structure and language evolution, using findings from child language learning to inform research on language evolution and animal communication. Doing so creates new ways of explaining how children learn language, why languages look the way they do, and how similar they are to other communication systems in nature.