Incomplete neutralization in Dutch opacity

Abstract

In Dutch, syllable‐final plosives are devoiced so /bad/ ‘‘bathe’’ is pronounced [bat]. However, despite being resyllabified to syllable‐initial position by a clitic in /bad+er/ ‘‘bathe her,’’ the /d/ is still devoiced: [ba.ter]. This is referred to as phonological opacity because a generalization syllable‐final devoicing is obscured or rendered opaque by another process—resyllabification. In this experiment, the acoustics and perception of this opaque relationship will be investigated. The voice onset time, release burst, and consonant duration of plosives for native Dutch speakers in phrases such as [ba.ter] will be compared to those in a phrase such as [ba.der] ‘‘bather’’ and /prat+er/ ‘‘speaker’’ [pra.ter]. Also, a perception experiment will be conducted testing whether native Dutch speakers are able to distinguish these phones. Based on a similar study of German [R. Port and M. O’Dell, J. Phonetics 13, 455–471 (1985)], it is expected that the [t] in [ba.ter] is not identical to the [t] in [pra.ter]. If, indeed, this is the case, and the [t] in [ba.ter] instead bears some of the acoustic properties of [d] in [ba.der], this would provide a useful set of perceptual cues enabling speakers to observe the devoicing generalization through the opacity.