Donnellan’s paper raises the question of specifying the circumstances under which one succeeds in referring, by using a description, to the person or thing one intends to refer to.

    Russell holds, or at least his writings suggest, the view that one can succeed in referring (by using a description) at most to what we called the semantic referent of that description.  Donnellan claims to provide counterexamples to that view.  In order to assess the situation, let us review some terminology:

x is the SEMANTIC REFERENT (SR) of description D on occasion O  =_df  x uniquely satisfies D on O.

x is the SPEAKER’S or UTTERER’S REFERENT (UR) of D on occasion O  =_df  x is the object, if any, to which the speaker who used D on O intended to call to the attention of his/her audience.  (The utterer’s referent is the object that the utterer takes her/himself to be talking about.)

x is the ACTUAL REFERENT (AR) of D on O  =_df x is the object which the speaker on O actually succeeded in making a statement (asking a question, issuing a command, etc.) about, by using D.

    The next question we can ask of a given case is that of whether in that situation the SR and the UR are identical or distinct.

CASE I:  Suppose the SR = the UR.  This is the normal case in ordinary speech on ordinary matters among ordinary people who are not mistaken about anything relevant.  And in such a situation, the AR = the SR (= the UR).

CASE II:  I am sitting in a bar, brooding on the state of the monarchy.  Outside, a crowd of spectators is watching a royal parade.  The King is passing by.  Actually, all the spectators believe him to be a usurper--he is a detested tyrant who seized the throne in a rather unorthodox way.  Moreover, the spectators and virtually everyone else in the kingdom are completely justified in believing this.
    There is a rigid custom--a convention, if you like--among the people, prohibiting the use of ‘the King’ to refer to the putative usurper.  Everyone justifiably believes that the real (rightful) King is a deposed prince who now washes tankards in the bar I am sitting in.
    Now it happens that everyone is wrong about this.  The man in the parade is no usurper, even though he is a tyrant.  His ascent to the throne was legal, due to a technicality.
    I am new in town, and I know none of the usurper story.  I am also ignorant of the custom of never deigning to call the ruler “the King.”  However, I have seen that the King is a noxious despot, and as the parade passes by I turn to the bartender and say, “The King is a fink!”  Hearing this, the other patrons leap on me and hack me to pieces with broken bottles.  It is useless for me to protest (while they’re at it) that I meant to refer to the tyrant outside; and the fact that that man really is the King seems irrelevant.  Despite my good intentions and the unknown fact that the tyrant really is the King, I have failed to refer to him and referred instead to the tankard-washer, calling him a fink.  Here it seems the SR = the UR, but the AR is distinct from both.

    Now what if the UR = the SR?  Three subcases should be distinguished under this heading, I think.

CASE III:  In a standard Donnellan “near miss” example, the AR = the UR.  (E.g., a courtroom scene in which I remark to my companion that “Smith’s murderer”--the man in the dock, who did not in fact do the murder--looks guilty.)

CASE IV:  Sometimes the AR = the SR rather than the UR.  Suppose at an auto race I say to you, “I bet you $100 that the winner will be over forty years of age.”  I am thinking of Dale Earnhart, completely confident that he will win, and I have him very much in mind, clear mental image and all.  But Earnhart does not win; he comes in second to Fat Freddy Phreak, who has got loose again.  Fat Freddy is only twenty-two.  I owe you $5.

    Finally, I want to suggest that even if we continue to maintain an independent notion of the “AR,” there are cases in which there is no AR.  In such a case, the SR = the UR, and all the facts are in, without determining any nonarbitrary choice of which to identify with the AR.  Take CASE IV, and suppose my betting opponent is a well-meaning person who does not want to cheat me.  He takes me seriously when I protest that I meant to refer to Earnhart and that my bet was based on a misunderstanding.  We may not want to insist here that the AR = the SR.  On the other hand, we certainly would not insist that my opponent pay up on the grounds that I had succeeded in referring to Earnhart.  Perhaps my opponent is being a sap, but perhaps I just failed to refer to anyone in particular even though my use of the description was intended referentially rather than attributively.

    That seems to suggest that whether or not there is an AR depends in part on my hearer’s reaction to my utterance and to the discovery of misfire.  I would not want to be committed to that, since the AR if any is supposed to be a feature of the utterance itself at the time it is produced.  Possibly what makes the difference (if we agree there is a difference) between this revised example and our original IV is simply the existence of the convention again.  But this raises the general question of how much power convention carries in overruling the intentions of speakers; we must admit that it carries some, but presumably it does not always take precedence.  (Also, as I said, the convention may not be a linguistic one but rather a special custom about betting designed to keep gamblers’ lives unfair, fun and exciting.)

    Adam’s excellent suggestion may or may not be helpful here.  The suggestion, made in response to the MacKay-type example,<3>  was that we get an AR only when, in the context, the speaker’s intention could (or would?) have been recognized by a reasonable and reasonably informed hearer.  So, in the foregoing reinterpretation of CASE IV, would a reasonable person have recognized my intention to designate Earnhart?  I don’t know; the case is underdescribed.  If I had pointed or even nodded toward Earnhart’s car as I spoke, that would have done the trick, but even here I feel the betting convention pulling in the other direction.  In contrast, if I had been paying scant attention to the race itself and only dimly aware that Earnhart held the lead, it would have been obvious that I was speaking attributively.

    Here is one final case, in which the SR =/ the UR and there seems to be no AR.

CASE V:  A dispute arises in the kingdom as to which of two pretenders to the throne is the real King.  The two hold a press conference, agreeing to take questions from reporters designed to reveal which is the real King and which the fake.  The populace is heavily polarized over the issue.  Exactly half the people believe Fregg the Ferocious to be the King, while the other half accept the claim of Quinn the Querulous.  The two groups call themselves the Fregeans and the Quineans.  The reporters at the press conference constitute a representative sample--half of us are Fregeans and half are Quineans, all mixed up together.  I turn to the reporter next to me and exclaim, “The King is magnificent!”  I am a Fregean and perhaps believe my hearer to be one too.  He is a Quinean.
    Now the Quineans have much better evidence for their view than do the Fregeans.  In fact, the Fregeans are an irrational faddish crowd who champion Fregg merely because he looks so ferocious.  And the Quineans are right.  Quinn is the real King, and anyone with any real sense would know that.  In this case the SR =/ the UR, but there seems to be no determinate AR, no one that I as speaker was really talking about.  Adam’s suggestion predicts that nicely.

    So far in this handout I have been just playing the game suggested by Donnellan’s paper.  But let us take note of two more general points.

    (1)  In my textbook I floated the possibility that there simply is no consistent notion of an “AR.”  One might want to take the position that what I have been calling “semantics” (Russellian propositional semantics) requires the notion of an SR, while the theory of communication (a distinct though related enterprise) requires that of a UR, but the idea of an “AR” is just a confusion of the two based on our failure to see the difference between sentence- or propositional semantics and communication theory.  Then we would have to explain away the fact of our having intuitions about “ARs” in cases like some of the foregoing.  Kripke takes roughly that line, making use of an idea of Grice’s that we shall discuss later.  I now think the line is made less attractive by Adam’s suggestion.  At least, the suggestion both rules out MacKay cases (thus solving a problem for Donnellan) and makes it easier to distinguish AR from UR and SR, as in the variants on CASE IV.  The more I think about it, the better I like the notion of an AR.  Way to go, Adam.

    (2)  We have played very fast and loose with the question of truth- and reference-bearers.  Against Strawson’s advice, we have talked of the truth of sentences and what grounds it; and I distinguished from this the truth or falsity of what a speaker meant to convey.  Strawson and Donnellan both talk of a third truth-bearer, the “statement made” when one uses a sentence (what has sometimes been called the “assertion-product”).  We may or may not want to identify this sort of item with the proposition expressed by the sentence as so used.  There is a large literature on this issue, though we shall not get to it in this course.

1.  A very slippery notion in some areas of philosophy of language, but I’ll take it for granted here.

2.  That is the way Donnellan sometimes puts it, but notice that it’s not quite right.  The speaker may be using the description attributively but nonetheless happen to have a particularly individual in mind that s/he believes fits the description.  I don’t know if this creates any real problem for Donnellan.

3.  To repeat it:  Geoff has formed the insane belief that Keith Donnellan is the illegitimate son of Santa Claus and Margaret Thatcher.  Intending the description referentially, he says, “Mrs. Thatcher's Christmas bastard wrote a classic article on descriptions.”  If you know enough about his weird beliefs, you will pick out the right individual and understand what he meant, but no one could correctly describe him as having said that Keith Donnellan wrote the classic paper.  In this example, intuitively, there is a UR but neither SR nor AR.