User can describe actions (*asterisks*), thoughts ("..."), and speech—AI reacts to all three. Always responds in English.