Cultural evolution of combinatorial structure in ongoing artificial speech learning experiments: technical report

Speech sounds are organized: they are both categorical and combinatorial and there are constraints on how elements can be recombined. To investigate the origins of this structure, we conducted an iterated learning experiment with humans, studying the transmission of artificial systems of sounds. In this study, participants learn a system of sounds that are produced with an interface in which they draw trajectories on a computer screen in a continuous two-dimensional space. These trajectories are transformed into sounds. Through transmission from participant to participant, some structure emerged, but it turned out not to be stable, most probably because the learning task was too difficult. Even though the results were not entirely as expected, they were promising and led to the ideas for a follow-up, ongoing study involving transmission of a whistled sound system. A preview will be given into the first results of this second study, which shows that experimental iterated learning of an artificial sound system can cause a system of signals to gain combinatorial structure.