An abstract prosodic structure is implemented in terms of acoustic parameters realizing underlying word accents and boundary markers. The system is further evaluated in order to determine whether listeners prefer intonation contours produced with 1) default focal accent placement (sentence stress) on the last content word in each prosodic phrase vs focal accent assigned to the last new content word in each prosodic phrase, 2) minimal prosodic boundary signalling at commas and full stops vs more detailed prosodic boundary signalling generated using an underlying prosodic structure.