On the relation between speech perception and loanword adaptation

Abstract

Loanword adaptation has been claimed to provide a unique window onto the relation between speech perception and the phonological grammar. This paper focuses on whether the ‘illusory vowel’ effect—in which the presence/absence of a vowel is poorly discriminated within an illicit cluster—is sufficient to explain why vowel epenthesis is the preferred repair for medial clusters in Korean loanword adaptation. A cross-linguistic discrimination experiment revealed a causative role of the stop release burst (or other audible frication noise) in the perception of an illusory vowel; in some cases, perception alone explains vowel epenthesis in loanword adaptation. A follow-up, identification experiment showed that Koreans’ perceptual similarity judgements do not match up with the adaptation pattern for stop-nasal clusters (e.g. pakna), although they do for fricative-stop and stop-stop clusters (e.g. paska, pakta). This finding is problematic for a purely perceptual account of loanword adaptation. The paper sketches a Bayesian account of Korean speech perception that integrates top-down phonotactic likelihood and bottom-up acoustic match and is able to explain the experimental results. It closes with some speculation on the role of the Preservation Principle versus perception in loanword adaptation.