Perception of incompletely neutralized voicing cues in word-final obstruents: The role of differences in production context

Abstract

Experimental data on final devoicing in languages such as German and Russian usually show that speakers produce incompletely neutralized acoustic differences between words ending in phonologically voiced versus voiceless obstruents (e.g., /kod/'code' vs. /kot/'cat' in Russian) and that listeners can use these differences to identify the underlying specification of final consonants at an above-chance level. The current study examines how the seemingly successful perceptual identification of voicing varies across stimulus items recorded in reading vs. non-reading procedures and with and without full minimal pairs present in the experimental list. Results of a series of identification tasks reveal that Russian listeners' identification responses are more in line with underlying voicing for the stimuli recorded during word-reading and with minimal pairs included among the experimental items. This shows that voicing judgments are strongly influenced by the acoustic differences produced when speakers encounter orthographic forms or lexical competition. At the same time, perceptual neutralization is also not complete for the items recorded without such exposure, which indicates that listeners' ability to recover underlying voicing is not limited to the production contexts involving written forms or minimal pairs.