Grammar underlies our ability to think and communicate, so understanding its architecture is of vital importance. This project studies how two of its major components, syntax and prosody, are related. Prosody consists of a metrical system reflecting the distribution of pauses and stresses in the sentence (prosodic structure) and of intonational contours. Prosody is often assumed to be derived from syntax by the process preparing syntactic structure for pronunciation. This hypothesis is theoretically appealing because it provides a natural explanation for well-known deep parallels between syntactic and prosodic structures and implies a more parsimonious grammar architecture. However, no theory explains so far how exactly prosody can be derived. First, all existing models describing parallels between syntactic and prosodic structures run into empirical problems. Second, some phenomena in prosodic structure appear to be independent from syntax. Third, a complete theory should also incorporate intonational contours, but in many cases their syntactic correlates have not been established. The goal of this project is to overcome all these obstacles. As a result, a more systematic, elaborate and economical account of syntax-prosody relations than hitherto available will be developed. (i) A new two-stage model for deriving prosodic structure will be proposed. First, a binary-branching structure is constructed based on syntactic relations between the words. Second, it is ?flattened? into a three-level structure (best suited for pronunciation purposes). My model builds on Slioussar (2007, 2009), combines advantages from previous approaches and incorporates new factors. It can contribute to designing the next generation speech synthesizers. (ii) Information structure (IS) related stress shift is widely believed to violate the parallelism between syntactic and prosodic structures. Based on (Slioussar 2007), a syntactic explanation for it will be developed that avoids much-criticized shortcomings of previous accounts. (iii) Solving different problems, syntactic correlates of various intonational contours will be formalized.