This paper presents work done on the syntax-intonation interface used in the Edinburgh University Centre for Speech Technology Research (CSTR) text-to-speech system, a linguistically sophisticated morph-hased speech-output system for processing unrestricted English text [1]. The first part of the paper discusses the need to identify phonological domains for intonation synthesis and the usefulness of hierarchical information, and the second part shows how an approximation to hierarchical metrical structure can he constructed from a very basic syntactic analysis without recourse to expensive tree-building algorithms.