Often when it gets late and I start getting ready for bed I’ll launch into a monologue that goes like this, “I’m the sleepiest man in the world. Yep, absolute sleepiest. None sleepier. Look no further for the first place winner of the global sleepiest competition…” and so on. I’ll do this bit regardless of whether anyone is listening, and the more I repeat it the funnier it gets.
It’s a joke, and as with all jokes I feel a linguist’s compulsion to ruin it, so here goes: the phrase “the sleepiest man in the world” is comic because despite being syntactically well-formed it has no denotation. Furthermore, I’d argue that this is not just a problem of determination–i.e. at any given moment there really is a sleepiest man in the world, but identifying him would be impractical–but rather that the idea is incoherent. In order to be sleepy you must still be awake, so if forced to come up with a denotation we’d probably point to some guy right on the cusp on unconsciousness, which we can’t do without drawing an arbitrary bright line between wakefulness and sleep, and even then the actual denotation would be flitting from person to person across the globe on a neurologically fast time scale. The phrase fails as ordinary language, and we are left with no choice but to laugh.
I’d add “the sleepiest man in the world” to that list of non-denoting noun phrases like “the golden mountain” or “the king of France” over which so much ink has been spilled. (Where in the latter case the background assumption is that we are discussing present-day, non-monarchical France.) In fact, I like it better as an example, though I have to think for a bit to determine why. The crux of all of them is that they fail the presupposition of existence and uniqueness required by the definite determiner. The phrases “golden mountain” and “king of France” fail the existence part, while “sleepiest man in the world” fails the uniqueness part. Unless you are willing to undergo some perverse contortions of the sort I lay out in the previous paragraph, the predicate “sleepy” is unable to impose the ordering on possible sleepers that a superlative requires in order to pick out a supremum. Note that this inability is subtly context-dependent. It may be perfectly reasonable to talk about the “sleepiest man in the room” (there are three men in the room; they all look a bit weary, but one keeps nodding off then pulling his head up with a start), but when you expand the salient context to the whole world, “sleepy” becomes too blunt an instrument and gets demoted from graded predicate to a binary one.
It’s this trickiness that makes me prefer “the sleepiest man in the world” as an entry point to Russellesque surface form vs. logical form considerations. To someone not versed in analytic philosophy, the sentence “The king of France is bald” seems completely unproblematic. I mean, there is no king of France, but, big deal, everybody knows that. It takes some hand-holding just to convince someone that they should care. (I still have that reaction to other famous examples.) Similarly, it’s easy to prepend the word “the” to a description of more than one thing, but it’s hard to do this in a way that pragmatics cannot save from paradox. Whereas the weirdness of “the sleepiest man in the world” is hard to explain away and so invites you to wonder if something deeper is afoot.