Abstract
This article uses computational modeling and large-scale pattern detection to develop a theory of global textual transmission as a process of turbulent flow. Specifically, it models stream-of-consciousness narration as a discrete set of linguistic features and rhetorical elements and uses this model to track the movement of this modernist technique across generic boundaries (from anglophone modernism to more popular genres) and linguistic ones (from English to Japanese). Oscillating between statistical models and moments of close reading, the article shows how a quantitatively scaled-up approach, rather than reinforcing an image of global textual flows as singular and monolithic, illuminates world literature as a system constituted by patterns of divergence in structure and of difference in sameness.
