Prior works on music composition with grammars tend to focus on the generation of sequential structures, such as melodies, harmonic progressions, and rhythmic patterns. While the natural form of the sentence is a sequence, this does not have to be reflected in the chronology of the piece. Shape grammar researchers have long internalized this perspective and use grammars to distribute geometric entities across multiple spatial dimensions. We take inspiration from these works and allow free placement of our musical entities on the timeline. With overlapping entities we can model chords, preludes, drum patterns, voices—polyphony on small and large scales. Beyond a foundational discussion of functional techniques for polyphonic composition with non-deterministic context-free grammars, we present the implementation and practical application of an automated composition system developed on these principles.