Process‐based modelling shows how climate and demography shape language diversity

Abstract

Aim Two fundamental questions about human language demand answers: why are so many languages spoken today and why is their geographical distribution so uneven? Although hypotheses have been proposed for centuries, the processes that determine patterns of linguistic and cultural diversity remain poorly understood. Previous studies, which relied on correlative, curve-fitting approaches, have produced contradictory results. Here we present the first application of process-based simulation modelling, derived from macroecology, to examine the distribution of human groups and their languages.Location The Australian continent is used as a case study to demonstrate the power of simulation modelling for identifying processes shaping the diversity and distribution of human languages.Methods Process-based simulation models allow investigators to hold certain factors constant in order to isolate and assess the impact of modelled processes. We tested the extent to which a minimal set of processes determines the number and spatial distribution of languages on the Australian continent. Our model made three basic assumptions based on previously proposed, but untested, hypotheses: groups fill unoccupied spaces, rainfall limits population density and groups divide after reaching a maximum population.Results Remarkably, this simple model accurately predicted the total number of languages (average estimate 406, observed 407), and explained 56% of spatial variation in language richness on the Australian continent.Main conclusions Our results present strong evidence that current climatic conditions and limits to group size are important processes shaping language diversity patterns in Australia. Our study also demonstrates how simulation models from macroecology can be used to understand the processes that have shaped human cultural diversity across the globe.

Drivers of geographical patterns of North American language diversity

“…Thus, locations that support more humans per unit area can also support more languages. The model accurately predicted the diversity of Australian languages [49], and here we apply it to North America. After running the model, replicated 120 times, we used the simulated geographical distribution of language ranges to summarize the model's prediction in the 300 Â 300 km grid of North America.…”

“…Prior studies seeking to identify factors linked to language diversification have been almost exclusively based on correlative analyses [1], in which no causal story is modelled [52]. Recently, a relatively simple mechanistic simulation model explored causal explanations for language diversity in Australia [49]. The model reproduced the spatial pattern of language diversity in Australia assuming only that carrying capacity varies over space as a function of the environment, and groups have maximum size limits (i.e.…”

“…The model reproduced the spatial pattern of language diversity in Australia assuming only that carrying capacity varies over space as a function of the environment, and groups have maximum size limits (i.e. carrying capacity with group size limits) [49]. However, the carrying capacity with group size limits mechanism remains untested in other regions of the world.…”

The evolution of language families is shaped by the environment beyond neutral drift

“…Other studies investigated global linguistic diversity in relation to geographic and demographic data (see Gavin et al 5 for a review). Predictors of linguistic diversity include latitude 6,7 , altitude and rugosity 4 , temperature and rainfall [7][8][9][10][11][12] , political complexity, and subsistence strategy 13 , as well as island size in the Pacific 11 .…”

Robust, Causal, and Incremental Approaches to Investigating Linguistic Adaptation

“…However, the advantage of creating this causal graph is that it breaks the investigation down into smaller links, and each of these links can be investigated in its own right with the most appropriate methods and data. While physical acoustics and laboratory phonetics methods can be applied to the initial parts of the chain, there are more appropriate methods for the later parts including computational modeling (Kandler and Steele, 2008 ; Gavin et al, 2017 ), artificial language learning experiments (Tamariz, 2017b ), historical corpus analyses and historical computational techniques such as phylogenetic ancestral state reconstruction (Gavin et al, 2013 ; Honkola et al, 2013 ).…”