self-knowledge and intensification

recall in a previous post, i tried to elicit judgments on the occurrence of the adnominal ER with a non-de se instance of a pronoun. it’s always been slightly mysterious to me why the de se reading of such sentences becomes so strong when you throw in that adnominal ER. well, here’s hintikka’s (1970) answer to this observation:

Summing up, we might thus say the following: Earlier, I have explained the peculiarity of “he himself” statements by saying that in them we are speaking of the individual in question, not of whoever happens to be referred to by a term. Thus, e.g., “a knows that he himself is b” becomes “(∃x)(x = a & K_a(x = b))” while “a knows
that a is b” will be “K_a(a = b).” There is no need to modify this point in the slightest. However, it has to be complemented by an account of the principles of individuation on which the bound variable x relies. The point I have tried to make is that in the case of “he himself” these principles are typically (though not always)
situational (“individuation by acquaintance”).

when you consider what we intuitively might guess is a semantics for the adnominal ER, this really stops becoming mysterious. our theory will say something about how the adnominal ER is presupposing that its neighbor NP’s referent is some kind of scalar endpoint–in order to place the referent here on any kind of scale, though, we’re going to need to know exactly who the NP is referring to.

for more, see hintikka’s on the attributions of “self-knowledge”