Segmental and prosodic effects on coda glottalization

This paper examines effects of segmental context and prosodic phrasing on the occurrence of coda glottalization in American English. On the basis of acoustic data from six female speakers, it is argued that the main effect of following segmental context on the occurrence of coda glottalization is due to anticipatory coarticulation, with anticipation of following sonorant consonant articulations favoring the aerodynamic conditions for glottalized voicing. The cross-dialect propensity to have high coda glottalization rates before sonorant consonants can then be understood to arise as a phonetically natural consequence of normal coarticulation processes. This coarticulatory account is supported by the fact that when an intonation phrase (IP) boundary intervenes between the coda consonant and a following sonorant, the contextual effect on glottalization rates is weakened. Our data also suggest that, all else being equal, presence of an IP boundary favors the occurrence of glottalization. These patterns are compatible with an account in which a glottal constriction goal on the coda consonant is optional in the linguistic representation. The optionality of this goal is somewhat problematic for many models of phonetic knowledge, but the coarticulatory explanation of the sonorancy effect helps to insightfully simplify any account of the distribution of glottalization.