Reconstructing the diffusion of Middle English schwa deletion

Abstract

This paper presents a method for assessing whether schwas in Middle and Early Modern English final unstressed syllables were pronounced or not. This is a difficult problem as schwas were represented by -e- in writing even after they had long disappeared in speech. We approach the issue by using models of logistic growth calibrated to phonologically interpretable poetry data in order to extrapolate time-dependent frequency estimates for the realisation of final-syllable schwas also in prose corpus data. Our models produce estimates that conform to accounts in the phonological literature and can at the same time be used to estimate the probability of schwa realisation at specific points during the long period of its gradual loss.