Segmental and prosodic effects on intervocalic voiceless stop reduction in connected speech

Abstract

Descriptions of English and other languages have claimed that voiced and voiceless intervocalic stops are often lenited to fricatives and approximants in connected speech. Few acoustic analyses of factors that affect this reduction have been reported for American English [cf. Lavoie (2001), Tucker and Warner (2011)]. In this analysis, intervocalic voiceless stops produced in bisyllabic words during story reading are examined (participants N = 19). The first result shows that speakers never lenite voiceless stops to approximants, except when /t/ is produced as the approximant implementation of a flap. This shows that voiced and voiceless stop reduction behave differently. Second, stress and vowel reduction play an integral role: 33% of stops are produced as fricatives when stress is on the preceding syllable (e.g., “taco”); 4% when stress is on the following syllable (e.g., “account”). The rate of reduction is significantly lower when stops are surrounded by two full vowels and higher when they are followed by schwa. Third, fricative reduction is most common for /k/, since full closures may be most difficult in the velar region. This study will be compared to Bouavichith and Davidson's (2013) examination of voiced stops to provide a fuller picture of intervocalic stop reduction in American English.