From Form to Formation of Phonetic Structures: An evolutionary computing perspective

Abstract

The purpose of this paper is to explain how evolutionary computing and machine learning open new perspectives in Phonetics and Speech Science. Using these techniques, it is possible to simulate the emergence and the evolution of a common language in a society of speech robots. Experimental results show how simple local rules of interaction between robots may explain some of the universal characteristics of the phonological structure of world's languages. On going work aiming to answer more complex questions, such as language evolution or dialect apparition, is presented. 1 INTRODUCTION Languages have very specific forms, their phonetic structures are not completely arbitrary. Typological studies have shown that the sound systems of world's languages exhibit systematic structural characteristics. For example, the vowel /i/ 1 is present in 87% of world's languages, /a/ in 87% and /u/ in 82% [13, 18]. As regards the consonants, some regularities are also found, for example, the frequen...