Effects of social structure on establishing lexical conventions in a computational model of task-oriented primeval dialogue

Abstract

In the field of language evolution, the only way of obtaining empirical data for most of its parts is with the help of computational models and simulations. As a consequence, a large number of different modelling approaches have been introduced in the area over the past two decades, with the level of detail in these models ranging from highly abstract formalisms to experiments with humanoid robots. However, and to some degree precisely because of this discrepancy in the modelling approaches that are spawning in the field, the results obtained within different research projects remain largely incomparable to each other due to the unbridgeable discrepancy in the assumptions made by the respective models. Furthermore, many of these assumptions, such as the availability of telepathic communication, to name the most striking one, are not strongly supported by the findings from the corresponding fields, leaving whole models that have been built on top of these open to questioning.