Brian Weatherson blogs on the difference between must-p and p. This came up at the informational session on epistemic modals at the APA Eastern, which I was chagrined to miss.
I blogged something similar to this case just before Thanksgiving. Kai von Fintel pointed out that the data I observed had been remarked on before, and then kindly put my post on his list of deli.cio.us links on his blog, subtitled "Matt Weiner discovers that 'must p' is weaker then 'p'."
Which raises the question, perhaps related to Brian's thoughts on 'discover': Is this a sense of 'discover' in which the discovery can be restricted to a group or domain, who did not know something that was known elsewhere. (Analogous to Columbus's discovery of America, when the Native Americans had known about it for a while.) Or was Kai being funny?
(Thanks to Minjoo Kim for pointing out Kai's link, back in November.)
Posted by Matt Weiner at January 5, 2006 11:30 AM"Many years later, as he faced the firing squad, Colonel Aureliano Buendia was to remember that distant afternoon when his father took him to discover ice."
From your earlier post:
"must" suggests a lack of direct access to the fact being asserted
Do you take this suggestion to be part of the truth-conditions of the "must"-statement?
In cases where the "must" statement seems ill-chosen because of direct access (e.g. "It must be raining" said while personally standing in the pouring rain), the corresponding "might not" statement seems even worse ("It might not be raining"). This seems some reason to suspect that the badness we sense may be a matter of a false implicature rather than the statement itself being false, for it seems unlikely that "Must-P" and "Might-not-P" are both false.
Posted by: Keith DeRose at January 20, 2006 02:37 PMI agree that the evidentiality doesn't seem like part of the truth-conditions of the "must"-statment; I wonder how that compares with languages with fully grammatacized systems of evidential markers (as Kai put it in his comment on the old post).
My tentative proposal that "might" and "must" are suited to conversations in which the participants are trying to figure out some matter (see comments on my old post and Brian's) might be able to account for why this would be an implicature. If you use "must" it implicates that you are trying to figure it out, and so you don't have a warrant that settles the question -- by the Maxim of Manner, I guess. Though I'm not totally convinced my proposal can handle all the data (see comments at Brian's again).
Posted by: Matt Weiner at January 20, 2006 07:15 PM