Patterns of voicing in American English voiced obstruents in connected speech

Previous research shows that the implementation of obstruent voicing in American English is strongly influenced by adjacent sounds and prosodic factors, but studies mainly focus on words in isolation and carrier phrases. This study examines the voicing of stops and fricatives in phrase-medial position in the connected speech of 37 speakers. Results indicate that stops devoice most often word-initially or next to other obstruents (regardless of voicing). Fricatives devoice word-finally, but are less affected by the adjacent sound. An analysis of where voicing is realized in the constriction interval shows that bleed from a preceding sonorant is common, but voicing beginning partway through the constriction interval is almost non-existent. Aerodynamic and articulatory implications of the results are discussed.