I've occasionally done that in text, now that you mention it. My in-person verbal comprehension has such a high latency that I can't really do it there. (by the time I've worked it out the conversation has moved on) Pre-caching expected misinterpretations may help, if I can anticipate them accurately enough.