A statistical measure of theme and structure

Abstract

This paper describes a computer-assisted analysis of semantic patterning in William Blake'sThe Four Zoas and considers the way in which such patterns contribute to the structure and meaning of the work. The analysis involves examining combinations and recombinations of images across the text for concentrations of images and images groups, recurring images, and patterns in the distribution of individual images and clusters of images. Statistical correlation routines were used to determine the degree of correlation among images across the extire text as well as in specific text segments. Principal components analysis enabled identifying thematic clusters of images, and the distribution of these clusters across these text were in turn examined to determine their patterning. Finally, time series analysis and Fourier analysis were used to find and verify patterns in the distribution of images across the text. Fourier analysis revealed striking patterns in the distribution of imagery in theZoas, which suggests that Blake may have used such patterns to help convey the poem's powerful thematic statements.