Earlier I wrote that the central problem in the philosophy of fiction is to find a solution to the following aporetic dyad:
1. There are no purely fictional items.
2. There are some purely fictional items.
The problem is that while the limbs of the dyad cannot both be true, there is reason to think that each is true.
David Brightly comments:
May I offer the following resolution of the paradox? I say that 'purely fictional' does not function as a concept term. Instead, it is ambiguous between two interpretations. On the one hand, it behaves like the pseudo-concept 'inexistent'. To say that Bone is a purely fictional alcoholic is to deny that Bone exists. [BV: Biconditionality seems too strong. If N is a purely fictional F, then N doesn't exist; but if N doesn't exist, it does not follow that N is purely fictional.] The same goes whatever name and concept term we substitute for 'Bone' and 'alcoholic'. This leads us to assert
1. There are no purely fictional items.
On the other hand, I say that 'fictional and 'purely fictional' appear to be concept terms because sentences like
Bone is a purely fictional alcoholic
arise via a surface transformation of
Purely fictionally, Bone is an alcoholic
and inherit their meaning and truth value. We can understand the latter as asserting that
Some work of fiction says that Bone is an alcoholic.
We take this as true, as evidenced by the work of Hamilton, and running the transformation in reverse gets us to
Bone is a purely fictional alcoholic.
Taking 'purely fictional alcoholic' as a predicate, which it superficially resembles, by Existential Generalisation we arrive at
There is some purely fictional alcoholic,
and hence to
2. There are some purely fictional items.
and apparent contradiction with (1).
The idea of a surface transformation may well appear controversial and ad hoc. But the phenomenon occurs with other pseudo-concept terms, notably 'possible'. We have
Bone is a possible alcoholic <---> Possibly, Bone is an alcoholic
Bone is a fictional alcoholic <---> Fictionally, Bone is an alcoholic.
On the left we have 'possible' and 'fictional' which look like concept terms but cannot be consistently interpreted as such. On the right we have sentential operators which introduce an element of semantic ascent which is not apparent on the left. It's precisely because 'possible' and 'fictional' involve hidden semantic ascent that they do not work as concept terms.
I am afraid I don't quite understand what David is saying here despite having read it many times. This could be stupidity on my part. But I think we do need to explore his suggestion that there is an equivocation on 'purely fictional items.' Let me begin by listing what we know, or at least reasonably believe, about purely fictional characters.
First of all, we know that George Bone never existed: that follows from his being purely fictional.
Second, we know or at least reasonably believe that Bone is a character created by its author Patrick Hamilton, a character who figures in Hamilton's 1941 novel, Hangover Square. Just as the novel was created by Hamilton, so were the characters in it. Admittedly, this is not self-evident. One might maintain that there are all the fictional characters (and novels, stories, plays, legends, myths, etc.) there might have been and that the novelist or story teller or playwright just picks some of them out of Plato's topos ouranos or Meinong's realm of Aussersein. I find this 'telescope' conception rather less reasonable than the artifact conception according to which Bone and Co. are cultural artifacts of the creative activities of Hamilton and Co. Purely fictional characters are made up, not found or discovered. It is interesting to note that fingere in Latin means to mold, shape, form, while in Italian it means to feign, pretend, dissemble. That comports well with what fiction appears to be. Of course I am not arguing from the etymology of 'fiction.' But if you have etymology on your side, then so much the better.
Now there is a certain tension between the two points I have just made. On the one hand, Bone does not exist. On the other hand, Bone is not nothing. He is an artifact of Hamilton's creativity just as much as the novel itself is in which he figures. How can he not exist but also not be nothing? If he is not nothing, then he exists.
If Bone were to exist, he would be a human person, a concrete item. But there is no such concretum. On the other hand, Bone is not nothing: he is an artifact created by Hamilton over a period of time in the late '30s to early '40s. Since Bone cannot be a concrete artifact -- else Hamilton would be God -- Bone is an abstract artifact. Thus we avoid contradiction. Bone the concretum does not exist while Bone the abstract artifact does. This is one theory one might propose. (Cf. Kripke, van Inwagen, Thomasson, Reicher, et al.)
Note that this solution does not require the postulation of different modes of existence/being. But it does require that one 'countenance' (as Quine would say) abstract objects (in Quine's sense of 'abstract') in addition to concrete objects. It also requires the admission that some abstract objects are contingent and have a beginning in time. The theory avoids Meinongianism but is quasi-Platonic. London Ed needs a stiff drink long about now.
Now let's bring in a third datum. We know that there is a sense in which it is true that Bone is an alcoholic and false that he is a teetotaler. How do we reconcile the truth of 'Bone is an alcoholic' with the truth of 'Bone does not exist'? There is a problem here if we assume the plausible anti-Meinongian principle that, for any x, if x is F, then x exists. (Existence is a necessary condition of property-possession.) To solve the problem we might reach for a story operator. The following dyad is consistent:
3. According to the novel, Bone is an alcoholic
4. Bone does not exist.
From (3) one cannot validily move via the anti-Meinongian principle to 'Bone exists.' But if 'Bone is an alcoholic' is elliptical for (3), then 'Bone is a purely fictional character' is elliptical for
5. According to the novel, Bone is a purely fictional character.
But (5) is false. For according to the novel, Bone is a real man.
The point I am making is that 'Bone is a purely fictional character' is an external sentence, a sentence true in reality outside of any fictional context. By contrast, 'Bone is an alcoholic' is an internal sentence: it is true in the novel but not true in reality outside the novel. If it were true outside the novel, then given the anti-Meinongian principle that nothing can have properties without existing, Bone would exist -- which is false.
I think Brightly and I can agree that a purely fictional man is not a man, and that a purely fictional alcoholic is not an alcoholic. And yet Bone is at least as real as the novel of which he is the main character. After all, there is the character Bone but no character, Son of Bone. In keeping with Brightly's notion that there is an equivocation on 'purely fictional item,' we could say the following. 'Bone' in the internal sentence 'Bone is an alcoholic' doesn't refer to anything, while 'Bone' in the external sentence 'Bone is a purely fictional character' refers to an abstract object.
We can then reconcile (1) and (2) by replacing the original dyad with
1* There are no purely fictional concreta
2* There are some purely fictional abstracta.
The abstract artifact theory allows us to accommodate our three datanic or near-datanic points. The first was that Bone does not exist. We accommodate it by saying that there is no concretum, Bone. The second was that Bone is a creature of a novelist's creativity. We accommodate that by saying that what Hamilton created was the abstract artifact, Bone*, which exists. Bone does not exist, but the abstract surrogate Bone* does. The third point was that there are truths about Bone that nevertheless do not entail his existence. We can accommodate this by saying that while Bone does not exemplify such properties as being human and being an alcoholic, he encodes them. (To employ terminology from Ed Zalta.) This requires a distinction between two different ways for an item to have a property.
I do not endorse the above solution. But I would like to hear why Brightly rejects it, if he does.