Re-evaluating Phoneme Frequencies.

Jayden L Macklin-Cordes,Erich R Round

doi:10.3389/fpsyg.2020.570895

Abstract

Causal processes can give rise to distinctive distributions in the linguistic variables that they affect. Consequently, a secure understanding of a variable's distribution can hold a key to understanding the forces that have causally shaped it. A storied distribution in linguistics has been Zipf's law, a kind of power law. In the wake of a major debate in the sciences around power-law hypotheses and the unreliability of earlier methods of evaluating them, here we re-evaluate the distributions claimed to characterize phoneme frequencies. We infer the fit of power laws and three alternative distributions to 166 Australian languages, using a maximum likelihood framework. We find evidence supporting earlier results, but also nuancing them and increasing our understanding of them. Most notably, phonemic inventories appear to have a Zipfian-like frequency structure among their most-frequent members (though perhaps also a lognormal structure) but a geometric (or exponential) structure among the least-frequent. We compare these new insights the kinds of causal processes that affect the evolution of phonemic inventories over time, and identify a potential account for why, despite there being an important role for phonetic substance in phonemic change, we could still expect inventories with highly diverse phonetic content to share similar distributions of phoneme frequencies. We conclude with priorities for future work in this promising program of research.

Highlights

Linguistic theorists seek to reveal causal mechanisms which explain the observable diversity of human language
Of greatest interest in this respect is the striking degree of improvement in fit, such that the power law distribution shifts from a largely implausible fit against full phoneme inventories, to a largely plausible fit after we exclude the least-frequent observations from samples. This raises the obvious question of why this might be so. We consider this in our Discussion, after we have examined distributional alternatives to power laws
The difficulty we find in ruling out the power law distribution may reflect this

Summary

Introduction

Linguistic theorists seek to reveal causal mechanisms which explain the observable diversity of human language. Good causal hypotheses are often suggested by the mathematical distribution that a linguistic variable is described by, owing to the fact that the distribution can be understood as an emergent outcome of some underlying causal process, and that a given mathematical distribution will be consistent with only certain mathematical kinds of underlying processes It is important for the development of theory that proposed claims about distributions be as sound as possible. Methodologists have developed more rigorous and secure methods for diagnosing power laws and for distinguishing them from similar but significantly different distributions This creates an opportunity, to re-examine our own putative power law distributions, and by doing so to improve the pathway to sound explanatory theorizing

Objectives

Methods

Results

Conclusion