Prosody as diagonalization of syntax. Evidence from complex predicates

Chomsky and Halle (1968) derive prosody and phonological domains by an algorithm that recursively operates on sister constituents in a surface tree structure: the transformational cycle. Prosodic phonology, on the other hand, phrases an output string into a universal prosodic hierarchy based on conventions that map certain types of syntactic constituents into certain types of prosodic constituents, e.g via XP-Edge Marking (Chen, 1987; Selkirk, 1986) or XP-Alignment (Selkirk, 1995; Truckenbrodt, 1995, 1999). An asymmetry in the assignment of prosody is established that connects syntactic relation, linear order, and prosodic structure. The data serves to illustrate why XP-edge marking does not provide a viable model of the syntax–phonology interface, and is used to argue in favor of a syntax–phonology mapping closer to the transformational cycle, along the lines of recent proposals in Cinque (1993) and Arregi (2002). 2. Prosodic Asymmetry This section presents evidence for the following generalization about prosodic asymmetry: (1) Prosodic Asymmetry • When a projecting element A precedes its complement B, a sequence of two prosodic domains that are on a par is derived: ´A ´B. The last domain provides the ‘nuclear stress’. • When a projecting element A follows an element from the complement do-main B, A is subordinated: ´B A (unless A is focused or B is old information) In the following, I present evidence in favor of (1) from different dialects of West-Germanic (Dutch, English, German), involving predicates with infinitival and nominal complements