February 11, 2005

Factivity and "Must"

Mark A.R. Kleiman, faced with a claim of the form "X must be shown that p," asks, "But how can he be shown that when it is demonstrably not true?"

In philosopher-speak: "show that" is factive. Granting (which I do) Kleiman's claim that p is false, it is impossible to show that p.

Except the factive locution is inside a modal. So the original speaker might be saying the following: p must be made true, and X must be shown that p.

It'd be interesting to work out the semantics of this. Even if q entails p, Must(q) seems like it's an odd way to express Must(p). If Grad Student says to Advisor "In two weeks, you must realize that I've finished my dissertation," that seems like a very roundabout way to say "I must have my dissertation finished in two weeks."

However, in the case under discussion, the party that would make p true is the same party that would be showing X that p. So in that case it may be more natural to use Must(X is shown that p) to express Must(p). My instinct is that the passive voice doesn't help, though, because it doesn't make clear exactly on whom the onus is to show X that p. Must(Y shows X that Y is not F) would be much clearer. For Grad Student to say to Advisor, "By Friday, you must be shown that I'm ready to defend" might make it seem as though someone else had to show Advisor something that's already true. For Grad Student to say "By Friday, I must show you that I'm ready to defend" makes it sound as though Grad Student knows that readiness to defend is part of the issue.

The moral here, I guess, is that when you're assigning responsibility stay away from the passive voice. Hoary but true.

Posted by Matt Weiner at February 11, 2005 02:15 PM