ABSTRACT
In the early years of the development of prosodic phonology, the prosodic hierarchy was highly constrained by the Strict Layer Hypothesis: A constituent C n could only dominate constituents of the immediately lower level, C n-1. This restriction was soon found to be too strong, and various modifications have been proposed, most frequently involving skipping levels in the hierarchy and recursion. Weakening a strong restriction on any system, prosodic phonology being no exception, runs the risk of introducing excessive options along with desirable ones, unless new limitations are established. Two recent approaches to re-restricting the prosodic hierarchy are considered: Match Theory, which places stringent requirements on the mapping of prosodic constituents, and the Adjunction Approach, which additionally restricts the distribution of stray elements within prosodic constituents, and it is demonstrated that while both provide important insights, they also have a number of drawbacks. The Composite Prosodic Model, which includes a distinct constituent between the phonological word and phonological phrase – the composite group, is thus advanced as a more nuanced model of the prosodic hierarchy that recognizes three different subparts according to the nature of their interface with other grammatical components, while also unifying these parts with a set of strong restrictions that apply throughout the hierarchy.
