Text-setting in Kaytetye

Abstract

Singing is a universal human activity. Across the vast range of song traditions throughout the world, native speakers have consistent intuitions about how the syllables in a given line of song text should be set to the tunes and/or rhythms within their various song traditions. This paper presents an Optimality Theoretic analysis of text-setting in a set of ceremonial songs traditionally sung and passed on orally by groups of Kaytetye-speaking women in Central Australia. Australian Aboriginal songs are renowned for the degree to which they diverge from speech. For our analysis, we use a computational method to exhaustively generate all permitted ways sung forms may diverge from their spoken equivalents along with all possible ways each form may be set to rhythm. We show that the seemingly idiosyncratic nature of text-setting strategies in this song set can be accounted for through a relatively generic set of constraints (even when thousands of competing candidates are considered), reflecting many of the fundamental processes that govern the interaction of language, meter, and music.