In lieu of an abstract, here is a brief excerpt of the content:

Reviewed by:
  • The origins of vowel systems by Bart de Boer
  • Alan C. L. Yu
The origins of vowel systems. By Bart de Boer. (Studies in the evolution of language 1.) Oxford: Oxford University Press, 2001. Pp. 168. ISBN 0198299664. $21.95.

What is the role of the population in explaining the universal in human vowel systems? This book, based in part on the author’s 1999 dissertation, offers an enticing and intriguing answer to this question through the use of computer simulation working from an agent-based self-organizing model of complex [End Page 880] systems. Originally aimed at computer scientists and researchers in the field of artificial intelligence, the present monograph has been adapted to an audience that has less knowledge of computer modeling and programming.

The book consists of seven chapters. Ch. 1 outlines the structure of the monograph, while Ch. 2 sets the stage by summarizing the universal properties of human vowel systems and the explanations advanced in previous research. In Ch. 3, B introduces the phenomenon of self-organization found in many complex systems in nature. Following work by Luc Steels (e.g. ‘The synthetic modelling of language origins’, Evolution of Communication 1.1–34, 1997) B argues that language is an open, complex, adaptive system, and that the mechanisms of self-organization might be responsible for its origin. Ch. 4, arguably the core of the book, describes the computations behind the simulation. Briefly, the model simulates the emergence of a vowel system in a population of agents that learn to imitate each other as successfully as possible with an open system of vowel sounds. Each agent is equipped with an articulatory vowel synthesizer, a model of perception, and a memory to store a list of vowels. In the imitation games, which are played between two agents, the initiator synthesizes the acoustic signal of a vowel randomly chosen from its inventory. The imitator finds the closest prototype in its inventory and synthesizes the acoustic signal corresponding to this prototype, with noise added. The initiator determines whether the imitator’s output matches the vowel intended. Success or failure is communicated to the imitator through nonlinguistic means. Depending on various factors, the imitator may attempt to improve the matching, by fiddling with different parameters of the output synthesis, or may add a new vowel to the inventory that closely matches the observed acoustic signal.

Ch. 5 summarizes the results of the simulation, arguing that the emergent vowel systems not only are nonrandom, but also are close to human vowel systems. B further demonstrates that a stable vowel system can be maintained even if the population of the agents changes over time based on the results of varying the basic simulation which has a static population. Ch. 6 connects the findings to other attempts to model various linguistic phenomena (e.g. syntax and semantics) computationally, demonstrating the viability and the contribution of this approach to linguistic research. This book ends on a speculative note in Ch. 7, suggesting possible directions of expansion based on the results of this work.

This book is clearly written and should be of interest not only to linguists concerned with the origins of language, but also to phoneticians and phonologists alike, given the growing interest in applying self-organizational models to phonological research (as evidenced by the work of such scholars as Catherine P. Browman and Louis Goldstein, Andrew Wedel and Alexei Kochetov). While additional justifications or elaborations for certain methodological choices are desirable at various points in the book (e.g. the choice of a prototype-based model of perception, the role of historical processes, and articulatory ease in shaping vowel systems), the results of this research nonetheless leave much food for thought.

Alan C. L. Yu
University of Chicago
...

pdf

Share