Logic and the Generative Power of Autosegmental Phonology

Abstract

Autosegmental Phonology is studied in the framework of Formal Language Theory, which classifies the computational complexity of patterns. In contrast to previous computational studies of Autosegmental Phonology, which were mainly concerned with finite-state implementations of the formalism, a methodology for a model-theoretic study of autosegmental diagrams with monadic second-order logic is introduced. Monadic second order logic provides a mathematically rigorous way of studying autosegmental formalisms, and its complexity is well understood. The preliminary conclusion is that autosegmental diagrams which conform to the well-formedness constraints defined here likely describe at most regular sets of strings.