Explaining the origins of complexity in language: A case study for agreement systems

Abstract

The complexity observed in human languages (structural complexity, system complexity, size complexity, individual diversity and language diversity) can only be understood by taking a functional and evolutionary viewpoint. The functional viewpoint insists that language is a tool for communication and hence language users expand their language systems to increase expressive power, but this must be balanced with minimizing cognitive effort, e.g. by re-using existing elements for new purposes or by limiting combinatorial search. The evolutionary viewpoint emphasizes that a language community has to collectively find within the open-ended space of possible languages a candidate that satisfies their purposes. Nobody has a global view or telepathic insight in the brain of another person. So speakers and hearers make guesses and adjust these based on the outcome of their communications. Individual variation arises naturally and language diversity is unavoidable. I will report on a number of computational simulations and robotic experiments of language games, in which these functional and evolutionary viewpoints have been explored. The experiments allow us to pin down precisely the different factors contributing to the complexity of human language. I will use case studies coming from studying how (internal) agreement systems may arise.