Repairing grounded artificial ontologies with language games

I am interested in language games as a tool for repairing communication between agents that co-exist within some environment but that do not understand one another's language. Moreover, where these agents perform a co-operative task that benefits from understanding communication. I plan to capture the main dimensions of the problem, introduce key sub-problems, and suggest at the form of solutions to these problems with a multitude of half-baked ideas.