Sonority-Driven Stress does not Exist

Abstract

This paper presents new acoustic evidence against sonority-driven stress in Gujarati. Gujarati has been the subject of more stress descriptions than any other sonority-driven stress case, and is one of the very few where stress is reported to be sensitive to multiple sonority levels. However, all these impressionistic descriptions of Gujarati stress disagree with each other in significant ways. In contrast, this paper presents the results of a production experiment. Acoustic measurements show that [a] does not attract stress away from the default position; instead, stress consistently falls on the penultimate syllable. If such a well-described case as Gujarati does not have sonority-driven stress, it is possible that none of the other cases do either, challenging Kenstowicz (1997)'s and de Lacy (2002 et seq.)'s theories that metrical structure can be sensitive to sonority. If so, the lack of such stress systems presents a challenge to the phonological property of symmetric effect found in a number of theories.