Final stop devoicing in Polish: An acoustic and historical account for incomplete neutralization

Bozena Tieszen

Ph.D. Dissertation, 1997

Abstract

This thesis examines the acoustic nature of word-final devoicing process in Polish and provides a possible explanation for its development on the basis of the history of the Polish language. Based on the environments of voicing neutralization at the word junctural level, Poland has been divided into two major dialects: Northeastern (Warsaw) and Southwestern (Krakow). For acoustic investigation, data from these dialects and from the transitional area of Bydgoszcz was considered. The data consisted of nine minimal pairs containing /p,t,k, b,d,g/, preceded by each of the vowels /i,a,u/ and followed by an initial voiceless obstruent in one case, and a vowel in another. All of the test words were put into carrier sentences, and some of them were embedded in a reading passage. Five monolingual speakers from each dialect area read the passage first and then the sentences. The following voicing cues were examined: duration of the preceding vowel, consonant closure duration, and glottal pulsing during the closure. The results revealed that word-final devoicing is not complete in Polish, and that the dialect regions influence the kind of voicing cues which maintain the voicing distinction phonetically. For speakers from Warsaw in all environments and all places of articulation, the duration of glottal pulsing into closure was significantly longer for underlyingly voiced stops than for voiceless. Speakers from the transitional area maintained voicing distinctions phonetically, in the environment of initial voiceless obstruent, via consonant closure duration and glottal pulsing into closure duration. Preceding an initial vowel, only the differences in glottal pulsing during the closure between underlyingly voiceless and voiced word-final obstruents, provided a significant voicing cue. The results obtained from the Southwestern dialect show small, statistically non-significant durational differences in voicing cues, suggesting that the voicing distinction word-finally before following initial voiceless consonant is fully neutralized or that some other voicing cue(s), not considered in this study, are relevant in maintaining the voicing distinction word-finally in this particular dialect. A strong dialect effect on the durational dimensions of the voicing cues was also observed. This dissertation shows that in at least one dialect of Polish final voicing neutralization is not complete.