Alan Gibbard in his 1990 book suggests a tempting route for developing a systematic treatment of language for expressivists and non-factualists. Developing that suggestion leads to some difficult choice points for the expressivist. I will focus on a few of these challenges: two challenges about quantification, one about normative logic, and a challenge that arises when we consider the role of context in expressivist utterances. This final challenge generalises to a wider class of views: I will argue that it causes trouble for the “assessment relativism” recently championed by John McFarlane, Andy Egan, and others.