Learning Opaque and Transparent Interactions in Harmonic Serialism
Jarosz, Gaja. 2016. Learning Opaque and Transparent Interactions in Harmonic Serialism. Proceedings of the 2015 Annual Meetings on Phonology, Vancouver, BC.
This paper presents initial modeling results comparing the relative learnability of four basic types of process interactions: bleeding, feeding, counterfeeding and counterbleeding. There are two main contributions. The first is a proof of concept: the paper presents a novel learning model that successfully copes with the hidden structure learning problem instantiated by serial derivations, including opaque derivations, in a variant of Harmonic Serialism. The second contribution is to illuminate some of the learnability consequences of opaque and transparent process interactions, with the principle finding being that ease of learning does not simply depend on the interaction type – how evidence for the processes and their interactions is quantitatively distributed in the learning data is also a principle factor driving learning.