Hysteresis in Language Emergence and Evolution

A system is said to show hysteresis when its evolution depends on its history. Language is such a system. One of the consequences of this is that when a system changes in one direction, it may end up in a different state than when it changes in the other direction (all else being equal). The relevance of this for language evolution and emergence is that newly emerged languages tend to have different properties (e.g. variability, iconicity) than more established languages, even if the social properties (group size, network structure) of these languages are similar. Current models that have been proposed to explain the link between properties of a population and properties of the language of that population do not explicitly take hysteresis into account. This paper proposes a very basic and abstract model, based on statistical physics and maximum entropy, that has hysteresis. The model has two parameters, one of which is related to the mean value of the linguistic property in the population, and the other appears to represent that there exist different strategies (e.g. iconic versus conventional communication) to achieve the goal of language (e.g. communication, learnability) and that a language needs to compromise between these. However, the precise interpretation of this parameter in terms of cognitive and social factors is still an open problem, as is the question of how empirically testable hypotheses can be derived from this model. Nevertheless, it is hoped that this work can draw attention to the importance of hysteresis in the emergence and evolution of language.

ISBN:2666-917X

Jaar van publicatie:2022

Trefwoorden:Language Evolution, Hysteresis, Computational Modeling

Toegankelijkheid:Open