Cross-linguistic variation in conceptual construal has been attributed to both innate biological factors and socio-cultural conventions shaped by cultural evolution. While nativist approaches struggle to account for widespread variability, the relativist view emphasizes learned social conventions as the source of linguistic diversity. However, the origins of this variation remain unclear: is it entirely stochastic, or are there identifiable factors driving culture-specific linguistic structures? Two key candidates are innate cognitive biases and cultural evolutionary dynamics. Recent research suggests a combination of implicit learning, processing biases, and interactional dynamics contribute to diachronic language change. Correlational evidence increasingly suggests that language adapts to its ecological niche, with links observed between morphological complexity and social variables (population size, L2 learners), phonetic variability and diet, and environmental factors influencing lexical and sound inventories. Spatial referencing showcases significant cross-linguistic variation; industrialized communities favor egocentric frames (left/right), while rural communities often use geocentric frames referencing environmental features (uphill, downriver). However, cross-sectional studies, often based on small samples, cannot establish causality, confounding environmental factors with sociocultural ones (subsistence, education, language contact). This study employs an experimental approach to test the hypothesis that linguistic conventions are contingent on environmental affordances – features of the environment enabling or promoting certain actions. The Maze Game, a collaborative task requiring communication of spatial locations, provides a controlled setting to investigate how environmental affordances influence the emergence of linguistic conventions.

The study reviews existing literature on linguistic relativity, contrasting nativist and relativist approaches to cross-linguistic variation. It examines previous research on the role of innate cognitive biases and cultural evolutionary dynamics in shaping language structure. The literature review also summarizes correlational studies demonstrating relationships between linguistic features and environmental variables, such as morphology and social structure, phonetics and diet, and spatial referencing and topography. It highlights the limitations of cross-sectional studies in establishing causality, emphasizing the need for controlled experiments. The review concludes by introducing the Maze Game as a suitable experimental paradigm for investigating the interaction between environment and language.

The study adapts the Maze Game, using a written chat system instead of spoken communication. Thirty-three participant pairs (66 individuals) were randomly assigned to one of three conditions, each featuring differently structured mazes: irregular, stratified, and regular. Irregular mazes presented various figurative shapes, stratified mazes had prominent horizontal lines, and regular mazes were grid-like, without salient features. Participants collaboratively navigated the mazes, communicating the positions of switches and gates. Each trial ended when both participants reached their goals. The study analyzed the types of descriptive strategies used by participants. Four description schemes were identified: Figural (referencing salient shapes), Path (describing a path from a reference point), Matrix (using a coordinate system), and Line (referencing horizontal lines). Data analysis involved multinomial regression models to assess the influence of environment and time on the choice of strategy, examining whether specific environments selected for specific strategies and whether these choices increased over time. To assess conventionalization, a final trial involved a neutral maze; the study tested whether participants maintained their previously established strategies. Additional analysis using discrete-time Markov chains examined the transition probabilities between strategies and their stability within each condition.

The study found strong evidence supporting the hypothesis that environmental layout motivates different linguistic strategies. Irregular mazes strongly selected for Figural descriptions, stratified mazes for Line descriptions, and regular mazes for Path descriptions. The use of these favored strategies increased over time within their respective conditions. Contrary to expectations, the Matrix strategy was not predominantly selected by any condition; however, it exhibited high stability once adopted, indicating its potential efficiency. Regarding conventionalization, participants in stratified and regular conditions retained their condition-specific strategies on the final neutral maze, suggesting a degree of conventionalization. This was less pronounced in the irregular condition, where the Figural strategy, being more concrete and context-dependent, was less likely to be generalized across different environments. Analysis of Markov chains confirmed these findings, showing that Figural, Line, and Path strategies acted as stable attractors in their respective conditions, while Matrix exhibited high self-transition probabilities across conditions.

The findings demonstrate that environmental affordances significantly influence the emergence and selection of linguistic strategies in a collaborative task. This supports the hypothesis that environmental factors causally affect linguistic conventions. The study's results resonate with large-scale cross-sectional studies correlating linguistic features and environmental variables, providing experimental evidence for the underlying mechanisms. The observed conventionalization, stronger for abstract strategies (Line, Path), highlights how successful interaction and alignment solidify linguistic choices, eventually overshadowing the influence of immediate environmental affordances. The less robust conventionalization of the Figural strategy underscores its context-specific nature and lower transferability. The study emphasizes the role of interactive processes in shaping linguistic conventions and their evolution.

This study provides experimental evidence supporting the hypothesis that environmental affordances influence the emergence and selection of linguistic strategies, contributing to cross-linguistic variation. The findings highlight the interplay between environmental constraints, interactive processes, and the conventionalization of linguistic conventions. Future research could explore the influence of community effects, systematically varying group size and interaction dynamics, to further understand the transition between localized adaptations and widespread conventions. Investigating the influence of other environmental variables beyond maze structure on linguistic strategies would also enhance our understanding of language evolution.

The study's artificial environment might limit generalizability to real-world language use. The use of written communication, while replicating earlier findings and simplifying data collection, may differ from spontaneous spoken interactions. The sample size, while adequate, could be expanded for greater statistical power, especially to discern finer-grained effects. Future studies could address these limitations by using more ecologically valid settings, incorporating spoken communication, and increasing participant numbers.