Effects of prosodic position and phonetic context on Spanish coda stops

Abstract

The paper reports on a study of Peninsular Spanish stops in coda position. The voicing and duration of the stops [b p d t g k] was measured in initial (V.CV) and syllable final (VC.C) contexts to study the possible effects of prosodic position and segmental phonetic context on the realization of the stops under study. Results show that in initial position, stops overwhelmingly surface with their underlying voicing, whereas in coda position there is evidence of weakening processes. Thus voiceless stops have voiced outcomes in 56.4% of the cases. However, weakening cannot account for the high percentage of voiceless outcomes for both underlying voiceless and voiced stops in coda position (43.6% and 31.7%, respectively.) The majority of these voiceless outcomes can be accounted for by regressive assimilation. The results of the study also show voicing differences between stops followed by obstruents, which surface as voiced in 52.1% of the cases, and the stops followed by sonorants, which surface as voiced 80.6% of the time. This difference provides limited support for Steriades [manuscript (1999)] proposal that voicing is licensed by particular phonetic cues and not by prosodic position [L. Lombardi, Natl. Lang. Ling. Theory 13, 39–74 (1995)].