This paper describes a new approach to compiling and optimizing array languages for parallel machines. This approach first decomposes ray language operations into factors, where each factor corresponds to a different communication or computation structure. Optimizations are then achieved by combining, or joining, these factors. Because factors preserve high level information about array operations, the analysis necessary to perform these join operations is simpler than that required for scalar programs. In particular, we show how data parallel programs written in the ZPL programming language are compiled and optimized using the factor-join approach, and we show that a small number of factors are sufficient to represent ZPL programs.