The distinction between first-order languages, or object languages, and metalanguages is a familiar one to readers of the NS.  However, over at Philosophy15, Rob Talisse and I try to explain how once we have the distinction, a unique kind of phenomenon occurs within arguments. In effect, the point of the metalanguage is to have a kind of evaluative criterion for arguments, but the trouble is that it keeps getting used as just one more line of first-order argument.  One of the results, we see, is what we call weaponized metalanguages.