Rimzhim, A. 1, 2 , Fowler, C. A. 1, 2 , Feldman, L. B. 1, 3 , Ernestus, M. 4, 5 & van Hell, J. G. 5
1 1Haskins Laboratories, New Haven, Connecticut, USA
2 2Department of Psychology, University of Connecticut, Connecticut, USA
3 University of Albany, SUNY, New York, USA
4 Centre for Language Studies, Radboud University, Nijmegen, NL
5 Max Planck Institute for Psycholinguistics, Nijmegen, NL
Transcription analyses of the speech of Dutch and English show that Dutch speakers, unlike their English counterparts, categorically devoice coda obstruents (both stops and fricatives). Thus Dutch bad (Dutch for bath) is pronounced bat. A gradient phonetic tendency to devoice final obstruents is observed in many languages, but, in Dutch and some other languages (e.g., German), it is a categorical, phonological process. We are investigating Dutch accent in English using a naming task in which monosyllabic English words are visually presented to native speakers of English and Dutch (who have English as their L2). Acoustic measurements of these utterances reveal that, in both native English speech and in Dutch-accented English, vowel durations are longer and closure durations are shorter before voiced than voiceless stops and fricatives. The two languages show similar durations of voicing in stop closures; with more voicing in the closures of voiced than of voiceless stops. Interestingly, native Dutch speakers showed significant voicing in closure for final voiced stops despite the suggestion in the literature that a feature of Dutch-accented English is final devoicing. We conclude that highly proficient Dutch speakers of English do not apply the phonological devoicing process of Dutch to spoken English words.