<Rant>Why do so many arguments for contextualism about this that and the other only ever consider simple sentences containing the allegedly context-sensitive term? A thorough study of the data should include, at a minimum, a glance at how the term behaves in the presence of temporal operators (as in (2) above), in conditionals, in the scope of quantifiers and, though this is often going to be hard work, speech reports. My new methodological principle is to not accept any contextualist argument that doesn’t include discussion of at least two of these categories, preferably all four.</Rant>

To be fair, Keith’s arguments for contextualism don’t just look at simple sentences, because he makes a point of looking at embeddings under negation, and in fact contextualists often have a simpler story to tell than their invariantist rivals about the behaviour of the term in question under negation. That’s a good start, but we BloodY-Minded Bull-headed Invariantists (BYMBIs) might well hold out for more. Of course, if we insist contextualists produce a full theory of speech reports along with a theory of how contextually variable terms behave in them, we may be accused of setting the bar a wee bit high. Maybe the name BYMBI will be earned then.