Based on an analysis of the distribution of 33 surface linguistic markers of complexity in English across spoken and written texts, the author argues that discourse complexity is a multidimensional construct, that different types of text are complex in different ways, and that the complexities of spoken and written discourse differ.