A colleague in my phd program went on the job market this year with research that is related to this, especially to the connection between writing and walking discussed in the final paragraph.
In a nutshell, his theory and experimental evidence suggest that we use the same cognitive mechanisms to process real maps as we do abstract maps. In particular, he shows that we often have mental "landmarks." He focuses on number lines as maps, and suggests that the round numbers behave like landmarks. He shows that all of the known cognitive signatures of the way we process landmarks on, say, a road-trip are present for these abstract landmarks.
He shows the same thing with textures in the context of anchoring and adjusting (we anchor on an initial texture and use it as a cognitive landmark.)
To me, his theory is extremely compelling and explains a lot. It makes sense that our brains, as our environments have become more complex, have adapted to use spatial cognition to process the abstract.
EDIT: Here is a link to his CV: http://marketingphdjobs.org/sites/default/files/Gaurav%20Jai.... The paper I'm talking about is called "Presence of Numerical Landmarks and Their Effects on Judgments." He hasn't posted it on the web, I'll see if I can get a copy from him and post it (if he's ok with me doing so.)