See here (especially Kai von Fintel's comment), here, and here for previous discussion of evidentiallity.
Briefly, an evidential marker is one that indicates what kind of evidence exists for a statement. "Must" appears to be an evidential marker. From Brian's post (ultimately from Kai, I think): If you see someone come in from outside dripping wet, it's OK to say either of these:
(1) It must be raining
(2) It is raining
But if you're standing outside in the rain, (1) seems odd. (Unless perhaps, as argued by Mike in Brian's comments, someone is arguing with you about whether it's raining.) "Must" only seems to work when your evidence for the statement is somehow indirect.
This post is prompted by the following insight, which I bet is not new to me: "Clearly" is evidential in the same way that "must" is. Take (3):
(3) Clearly, it's raining.
When you see someone come in from the rain, (3) is acceptable. When you're out in the rain yourself, it sounds weird. "Obviously" lines up the same, I think. It is perhaps not surprising that these are evidentially marked, since 'clear' and 'obvious' suggest aspects of the arguments that lead up to the conclusion. (Or something like that.)
I also suspect something analogous to Keith DeRose's comments here and here; when standing outside, we wouldn't respond to (3) by saying "It's raining, but it's not clear that it is." (Whereas, if someone inappropriately said, "Surprisingly, it's raining," we might respond "It's raining, but it's not surprising that it is.") So "clearly, it's raining" might be true but carry a false implicature.
Posted by Matt Weiner at February 2, 2006 10:00 PMHi Matt,
there is some work on "it is clear that ...":
http://semanticsarchive.net/Archive/TY0OWM1Z/barker.taranto.clarity.pdf
http://www.amazon.com/gp/product/0415976081/
Posted by: Kai von Fintel at February 3, 2006 04:28 AMhey, thanks for the unfogged info.
also, if i wake up hours before i want to on a friday and have class alllll day (stupid friday afternoon makeup), i'd say "it must be raining."
Posted by: matty at February 3, 2006 05:30 AMThe most not disliked person in the blogosphere very recently (past twelve hours) made a point based upon the used the "must be raining" model, citing it explicitly. Maybe he's reading Weatherson, since he's certainly a sometimes Crooked Timber reader.
Posted by: washerdreyer at February 3, 2006 10:03 AMi wonder if being the least objectionable poster at volokh isn't a pretty good gig, for blogistan.
Posted by: matty at February 3, 2006 05:48 PMThanks (as always) for the comments!
Kai, I see that Baker and Taranto explicitly make the connection between 'clearly' and 'must'. On a quick reading I'm not convinced by their analysis of 'clearly'. In particular, I think that the problem they're trying to solve may not be a problem; sometimes we need to point out to people things that are entailed by things that are known, and that doesn't necessarily require assuming that the hearer is an idiot. But I don't have an account that I can give here (and I'd need to give this more thought to come up with a good account).
matty, do you say that because the rain wakes you up, or are you just inferring it from the suckosity of everything else? Either way, it seems like you're inferring it; so 'must' is OK. And I think you're right about Kerr; he's the one who seems to me to uphold the site's old rep for reasonableness.
w/d, it seems to me that he's distinguishing the epistemic and normative uses of 'must', although the evidentiality of 'must' only applies to the epistemic use. Then, it's because of 'must' that you could tell that in the original case he was inferring what the court did rather than reading it directly off their opinion.
Posted by: Matt Weiner at February 4, 2006 04:27 PM