A transformation of michi-annai-bun, written directions, into a michi-annai-zu, map, on a display device is described. It involves three distinct processes: sentence analysis, internal code generation, and map drawing. The sentence analysis establishes case frames for the incoming sentences; the internal code generation produces a sequence of matrices, each containing a motion or scene from the directions; and the map drawing computes a map by interpreting the contents of the matrices. A complete understanding of a michi-annai-bun is achieved when the linguistic process is integrated with common sense knowledge, and spatial knowledge of the real world situation, i. e. , when the map is followed.