Is Referring Opaque?
If I refer to Hesperus with “Hesperus” am I also referring to Phosphorus? If I refer to Clark Kent with “Clark Kent” am I also referring to Superman? Suppose I disbelieve that Hesperus is Phosphorus and that Clark Kent is Superman: is it still true that I am referring to Phosphorus and Superman when I use “Hesperus” and “Clark Kent”? It is generally held that such inferences go through, but I am doubtful: that is, I doubt whether “A refers to x” generates a transparent context at “x”. Compare the verbs “allude to”, “advert to”, “mention”, “mean”, and “list”: if I allude to Clark Kent, do I also allude to Superman, and so on for the other verbs? That sounds wrong: I might protest the imputation and I would be right to do so. In other words, these verbs create an intensional context: we can’t substitute co-denoting terms within it salva veritate. We might be able to report concerning Superman that I referred to him (de re), but we can’t say that I referred to Superman (de dicto). Or consider “A intentionally referred to x”: surely that is an opaque context—I did not intentionally refer to Superman by saying “Clark Kent” if I disbelieve that Clark Kent is Superman. Referring is a mental act and it inherits the opacity common to mental acts (same for “alludes” etc.). Speaker reference thus exhibits referential opacity.
But isn’t semantic reference ultimately dependent on speaker reference? Isn’t speaker reference the basic concept? If so, semantic reference must inherit the logical behavior of speaker reference: the construction “w refers to x” is also an opaque context. So it doesn’t follow from the truth of “’Hesperus’ refers to Hesperus” and “Hesperus is Phosphorus” that “’Hesperus’ refers to Phosphorus” is true. Indeed, we might conclude that it is simply false, given that speakers don’t use “Hesperus” when they want to refer to Phosphorus—they use “Phosphorus”. If we know the identity, we won’t object to using the names indiscriminately, but strictly speaking “Hesperus” refers to Hesperus and “Phosphorus” to Phosphorus, but not contrariwise. To repeat, “refers” generates an intensional context, like “believes” and other psychological verbs. This is a rather startling result, given how the notion of reference (denoting, designating) has been employed in analytical philosophy, yet it seems intuitively correct. But there is a still more startling consequence of accepting the opacity point: reference is not a relation. For genuine relations don’t generate opaque contexts: they hold between things however those things are designated (e.g. “to the left of”). But reference does not hold between things no matter how those things are designated (the subject position of the sentence is transparent, however). So we should not speak of the “reference relation” or model reference on genuine relations; it is no such thing. If there are relations between words and the world (and there undoubtedly are), they are not relations of reference. For example, Phosphorus may indeed cause utterances of “Hesperus”, but it is not denoted by “Hesperus”; and it is not possible to analyze denotation in terms of causal relations, since that would be false to its logical character. No relation could constitute reference. The whole picture of reference as some sort of real relation between words and things has to be wrong. We picture language as over here and the world as over there, with all sorts of relations holding between the two; but the concept of reference doesn’t correspond to any of these relations, because reference is not a relation at all. It is a kind of pseudo-relation.
 It is much the same as with the verb “thinks of”: if I am thinking of Clark Kent, I am not thereby thinking of Superman, and might strongly the reject the suggestion that I am. It may yet be true that I am thinking of someone (Clark Kent) who happens to be identical to Superman, but it doesn’t follow from this that I am thinking of Superman. I think the same is true of perceptual verbs, though here the intuition doesn’t seem so robust: if I see Clark Kent in the office, does it follow that I see Superman in the office? That is, is there a reading of this sentence under which it doesn’t follow? I think there is, but you have to think hard to find it. If so, “sees” doesn’t express a genuine relation either (ditto “remembers”).