Neutralization of Coda Obstruents in Korean: Evidence in Production and Perception

The aim of this dissertation is to investigate acoustic and perceptual aspects of coda neutralization in Korean. By extending the Kim and Jongman (1996) study that documents complete neutralization of manner contrasts in coda obstruents, I examine all three types of coda neutralization in Korean (i.e., laryngeal, manner, and palatal) in production and perception. The question of whether coda neutralization is phonetically complete or incomplete is addressed by analyzing acoustic correlates of final obstruents. This study also examines the effect of explicit orthographic cues on coda neutralization by eliciting the stimuli in both spontaneous and read speech and comparing their acoustic properties. A subsequent perception study investigates the perceptual consequences of coda neutralization. In a two-alternative forced-choice identification task, native Korean listeners are asked to identify the correct underlying form of neutralized coda obstruents with or without measurable acoustic differences. The production results reveal that laryngeal neutralization is phonetically incomplete, with its effects stronger in the spontaneous speech than in the read speech. The effects of incomplete neutralization are weaker or absent in the manner or palatal neutralization: the pairs with underlying manner distinctions exhibit acoustic differences in a single measure in one or the other task, whereas the pairs subject to palatal neutralization are completely neutralized in terms of all acoustic parameters in the both speech types. In perception, all three types of coda obstruents are completely neutralized. Regardless of the presence of significant acoustic differences in coda obstruents, the listeners are not able to successfully label neutralized words with the correct underlying forms. These findings suggest that coda obstruents are not always completely neutralized in production, but exhibit complete perceptual neutralization. The asymmetrical patterns of neutralization across the three types of neutralization as well as across production and perception challenge the current exemplar-based approach to neutralization. I suggest that the current account needs to be modified so that it reflects the fact that the effects of neutralization in production and perception can vary depending on the number of categories neutralized into a single output as well as the dimensions of phonological contrasts that are neutralized.