Self-organization in Sound Systems: a Model of Sound Strings Processing Agents

Several typological universals of sound systems of human language are assumed to be a result of self-organization in a population’s communication performance. This has been shown for human vowel system in diverse studies (c.f. de Boer, 2000b; Jäger, 2008). In these studies computational models were designed of agents communicating with single vowel sounds. In our study we present a computational model where agents communicate with concatenations of single sounds forming complex expressions. The goals of this study are i) to examine decisive factors that contribute to the emergence of realistic sound systems in artificial societies of interacting agents, and ii) to discuss ways to evaluate artificially emerged sound systems.