I do a lot of work with augmented/virtual reality and it usually involves trying to make the computer understand a space or object in some way so that it can then superimpose/blend virtual objects with that space. Most of the tricks for this come from the world of robotics (because robotics has NASA/DARPA money behind it so there is funding to develop fundamental techniques, the rest of us just pick up the scraps they leave behind) and there was a big shift in robotics/Artificial Intelligence from earlier symbolic/language/discursive reasoning based techniques of the 1950s-80’s to systems like subsumption architectures (used in things like the roomba vacuum cleaners and the video game The Sims) that shifted away from models like “and then the robot will form a mental model of the world using a symbolic language we provide and perform feats of reasoning upon it to decide its course of action” to things like “the robot responds to sensor data from the world and acts upon it using reflex like actions and then reasons upon its actions’. The shift was from descriptions of objects being the primary way of knowing the world to direct interactions with objects being the main focus.  A good non technical introduction to these non language/symbol focused approaches is Rodney Brooks’ paper “Elephants don’t play chess” http://rair.cogsci.rpi.edu/pai/restricted/logic/elephants.pdf

The other approach near and dear to me would be Antonin Artaud’s writings on the “Theatre of Cruelty” and his reaction against the text/language focus of most theater. That instead of a stage being a place where lines are read to an audience, it is a place where things happen to an audience. Mise en scene becomes the primary focus of the Theatre of Cruelty. I’m especially fond of his writings because this is where the term “Virtual Reality” is first used, he describes the “la réalite virtuelle” of the world a theatre production creates and how the technologies of lighting, sound, and stagecraft bring that into being.

A final approach to ‘non linguistic models of knowing objects’ that I’ve been pondering is the “speculative realism” tendency (really hard to call them a group, or a school or even really a Them.) of people like Ray Brassier, Iain Hamilton Grant, Grant Harman, Quentin Melliassoux, etc in general and the “object oriented philosophy” tangent that has emerged from them (Ian Bogost writes relatively jargon free essays/books in this vein). The part that ties most closely with what we are talking about is their rejection of Kantian “correlationism” which claims all we can know of the world is the intersection of thinking and being, usually mediated through language. If any of that sounds at all interesting a good place to start would be the journal “Collapse” which was the source of a lot of early writing from these people and which sponsored some important early conferences. http://roundtable.kein.org/sites/newtable.kein.org/files/C3_Spec_Real.pdf