This paper examines the learnability of different types of tone sandhi in Structural Optimality, a constraint-based framework that posits hierarchical scales and defines constraints over the scales. Approached as a hidden structure problem, we show that Expectation Driven Parameter Learning can acquire these grammars, but that their properties can make learning difficult.

Phonological generalizations are finite-state. While Optimality Theory is a popular framework for modeling phonology, it is known to generate non-finite-state mappings and languages. This paper demonstrates that Optimality Theory is capable of generating non-context-free languages, contributing to the characterization of its generative capacity. This is achieved with minimal modification to the theory as it is standardly employed.

Whether phonological transformations in general are subregular is an open question. This is the case for most transformations, which have been shown to be subsequential, but it is not known whether weakly deterministic mappings form a proper subset of the regular functions. This paper demonstrates that there are regular functions that are not weakly deterministic, and, because all attested processes are weakly deterministic, supports the subregular hypothesis.