Effects of Prosody on Articulation in English

Taehong Cho


This dissertation investigates how phonetic realizations are conditioned by various prosodic conditions (i.e., sentence stress, level of prosodic domains, position-in-domain), by examining articulation in three prosodic ally strong locations: accented syllables, domain-initial positions, and domain-final positions. In particular, this dissertation aims to understand how articulatory strengthening that may arise from these prosodically strong locations is manifested in the articulatory maximum positions, V-to-V coarticulation, movement kinematics, and mass-spring dynamical parameter settings. To accomplish this, three articulators, the tongue, the jaw, and the lips were examined concurrently, collected from six speakers of American English using an electromagnetic articulography (EMA).  Results were remarkably similar across speakers.


The results regarding accent show that accented vowels are strongly articulated, having larger maximal jaw and lip openings, extreme maximal tongue positions (lower for /a/ and fronter for lif), greater V-to-V coarticulatory resistance, and larger, longer, and faster movement. As for boundary effects, the results show that vowels at higher prosodic boundaries are strongly articulated, evident in larger maximal lip opening but not in maximal jaw opening, extreme maximal tongue position (lower for both /a/ and /i/), greater V-to-V coarticulatory resistance, and longer, sometimes larger, but not necessarily faster movement. Although accent and boundary both give rise to articulatory strengthening, these effects are not the same. They differ in the dimension in which the tongue is expanded, the involvement of the lips and the jaw, and the movement velocity.


The results regarding movement kinematics suggest that speech mechanisms are more complex than has generally been assumed by researchers who have adopted a mass-spring gestural model in explaining certain speech phenomena. This dissertation also provides a basis from which the window model can be further developed, accommodating prosodically-conditioned variations.


Overall, the results are interpreted in terms of the contrast maximization principle. It has been proposed that articulatory strengthening makes a sound more distinct from neighboring segments (syntagmatic enhancement) and/or makes the sound distinct from other contrastive sounds in the language (paradigmatic enhancement). Such a phonetic enhancement of linguistic contrast arising from prosodically-conditioned articulatory strengthening is interpreted as an articulatory signature for prosodic information.