Rhetoric (Ῥητορική?) in general offers many "signposting" oral structures so that one may (in a 1-D temporally streamed medium) reliably communicate some of the nested arboreal complexities which writing (somewhat 2D, and amenable to re-reading) tends to communicate more clearly.
How is writing 2d, just because you can't fit a given text on a piece of paper and need to text wrap? That doesn't make it 2d.
I think they mean that, with writing, it’s easier to parse complex syntax trees. Think fractal dimension and finite vs stack memory.