Propositional attitude ascriptions display something which is usually called “de re/de dicto” ambiguity. For example,
(1) John believes that Frank is ugly.
can be true in two very different situations:
(1d) John has the belief [Frank is ugly].
(1r) John has the belief [that guy in the red shirt is ugly], where the guy in the red shirt is Frank.
Normally, the candidate de dicto belief entails the candidate de re belief, so you might think this is just a case of under-determination. The evidence for genuine ambiguity is that, in a situation where (1r) is the case but (1d) is not the case, there is, intuitively, a way to read (1) in which is it is true, and also a way to read (1) in which it is false. (Other evidence has to do with non-existants… but this is more complicated.)
What’s going on here? More or less, when a name appears in the “scope” of a propositional attitude ascription to X, its ambiguous whether this name should be read as component of X’s beliefs, or part of the speaker’s description of X’s beliefs. Put another way:
Perspectival Ambiguity. De Re/De Dicto ambiguity in propositional attitude contexts is a function of an option to take either the the ascriber’s perspective or the ascribee’s perspective when evaluating the target proposition.
That’s my first point. (I’m not making any claims about the originality of these points– I’m just making them.) My second point is that the ambiguity is completely ubiquitous within propositional attitude contexts, and not at all restricted to names. Here’s an illustration with respect to color terms. Suppose (following Fodor) that Granny’s favorite color is white, but John doesn’t know this. And suppose John believes that all swans are white. Then there is a sense in which it is true to say:
(2) John believes that all swans are Granny’s favorite color.
(If you’re suspicious of the alleged true reading here, a bit of context may help… I’ll try to think of something convincing.)
So too for verbs… You and I discover that falling is the same thing as accelerating towards the earth. John doesn’t know this. Yet both of the following could be true:
(3a) John believes the apple fell.
(3b) John believes the apple accelerated towards the earth.
By the way, if you think that beliefs in the head are quite language based, then any attitude ascription in English to someone who doesn’t speak English will be a good example of this…
Finally, I’d like to link all this up with an interesting phenomenon involving tense. (This is the third point.) Suppose we encounter:
(4) At noon yesterday, John believed that Mary was ill.
This is ambiguous: did John believe Mary was ill at noon yesterday, or did he believe she was ill at some time prior to noon? I’d like to suggest that the same perspectival ambiguity is at work here too, with respect to tense. Taking the speaker’s perspective, the past tense of “was” is the same past tense as “believed”; taking John’s perspective, the past tense of “was” was something he believed at that time.
Of course, working this out in formal detail is another matter…