Speech Accent Archive에 나타난 영어 어말 유성 저해음 발화 연구

This study examines the realizations of English word-final voiced obstruents depending on their following sound, word type, and morphological structure. Underlying voiced obstruents in word-final position are expected to surface as voiced ones, but they are actually realized in a variety of forms: voiced, devoiced, deleted, or with vowels inserted. Moreover, it is not clearly known how much target voicing is maintained or to what extent substitute pronunciations occur. Based on the data from Speech Accent Archive, this study analyzes the production of English voiced obstruents in word-final position. The results show that 66.32% of underlying word-final voiced obstruents are realized as voiced, 19.67% as devoiced, 14.00% as deleted, but there are no realizations with vowels inserted. With regard to the three factors―the following sound, word type, and morphological structure―it is also shown that target voicing is more maintained before a pause and in content words; consonant devoicing more occurs before voiceless sounds and in function words; consonant deletion more occurs before voiced sounds and in function words. Morphological structure does not affect the realizations of target voicing, consonant devoicing, and consonant deletion. Along with these results, implications on English pronunciation education are presented.