When possible this is best, but some people at SIAI (cough Vassar cough) have conversational styles that are very fast so as to convey the most information in the shortest time, and it's hard to do real-time transformations from ultra-abstract statements to reasonably-precise internal models and back as information is exchanged and people build up their ontologies on the fly. (Which is pretty awesome when it happens -- one of the joys of being a Visiting Fellow. And of talking to Michael Vassar.)