The logic of fiction

Tags: Harrison Ford, truth conditions, strange visitor, Peter, Sherlock Holmes, sentences, semantics, Dr. Watson, University of British Columbia, Dekker, natural language, Goran Sundholm, literary semantics, Peter Alward, John Woods, natural language semantics, The four horsemen, logic of fiction, University of Istanbul Logic, Oxford University Press, application conditions, Alward, fiction, intuitive semantics, Leo Tolstoy, formal semantics, Bill Sykes, English sentences, Cambridge University Press, Parmenides, English speech, Brandom, Parmenides Law, Holmes and Watson, University Press
Content: FINAL VERSION June 12, 2015 UNILOG LOGIC SCHOOL 2015 University of Istanbul Logic and Fiction John Woods* "`I see nobody on the road', said Alice. `I only wish I had such eyes,' the King remarked in a fretful tone. `To be able to see Nobody! And at that distance, too! Why, it's as much as I can do to see real people, by this light.'" Lewis Carroll 1. The four horsemen of the logic of fiction We will consider the impact of four ontologically parsimonious principles purporting to regulate the logic of fiction and the efforts of others to produce an intellectually satisfying natural language semantics for literary discourse. Parmenides' Law: There is nothing whatever that doesn't exist. Frege's Law: No singular referring expression refers unless there exists a referent for it. The Particular Quantifier Law: Nothing is something unless it exists. The Fiction Law: There exists no object that any object of fiction is.1 The objects of fiction don't exist. Although each of these horsemen has gained considerable traction among philosophers of language and logic, it is prudent to keep in mind how deeply at variance the first three are from actual linguistic practice and common sense. For the time-being, we'll stick with these counterintuitive fiats, if only for the occasion they afford us to consider whether or not they facilitate, rather than preclude conscientiously satisfying accounts of fiction. We'll begin with four attempts to answer the former question affirmatively, in each case only lightly sketched: an imaginist semantics a psychology-first semantics * The Abductive Systems Group, Department of Philosophy, University of British Columbia, [email protected], www.johnwoods.ca 1 All four fail Russell in Principles of Mathematics , second edition, London: Allen and Unwin, 1937, first published in 1903, but they take hold in "On Denoting", Mind, N. S. 14 (1905), 479-493. In truth, the first three are horsemen of mainstream logic itself, and the fourth is purpose-built for fiction. I cite Russell only as an example. 1
an objectless truths semantics a no-ambiguity semantics. Unfortunately, we won't have time to consider something I consider essential for an understanding of reference and truth in natural languages, namely, a semantics for the dead. NOTE ONE Imaginism Reflections on a paper by Otбvio Bueno2 1. Dual-aspectism. In the 1982 movie "Blade Runner", Harrison Ford plays the central character Dekker. There are some odd goings-on in situations of this sort. We know that what we see on the screen is not Ford, but only cinematic images of him, and that what we hear are only recordings of his voice. But when Ford himself speaks and moves, and these doings are cinematically captured for the audience's rapt attention, everybody in the audience is in two unconflicted unified states at once. (1) They experience these doings as Dekker's. (2) They know concurrently and full well what induces these experiences are actually induced by Harrison Ford's doing something else, namely acting. Indeed, (3) They know that there is no one whomsoever who Dekker actually is, hence nothing at all that he does. This tells us something important about what it is to be engaged by fiction. Movie-goers experience themselves as observing Dekker's doings, notwithstanding their uncluttered understanding that there is no Dekker to do them. Taken together, (1), (2) and (3) bespeak a dual-aspect notion of fictional engagement: (4) We experience ourselves as knowing that Dekker did these things, knowing that that couldn't be true, and yet knowing it without an iota of cognitive dissonance.3 2 "Interpreting science, interpreting art", Philosophy Colloquium, University of British Columbia, Okanagan campus, Kelowna, Friday, March 6, 2015. The other speakers were Rob Wilson of the University of Alberta and your obedient servant. 2
Or for those who like structural schematization (4) E[(K (Dekker did these things )) K(~(Dekker did these things ))], where E is the experiencing-as operator, and K a knowledge operator. The scope of E indicates what I mean by "two unconflicted unified states at once." The fact that these apparently opposite states are experienced without conflict is prima facie evidence that what we are experiencing ourselves as knowing is not a contradiction. Again more informally, (4) We experience ourselves as knowing that Dekker did these things and as knowing that he didn't. This, we might think, is a primary and motivating datum for a theory of fictional engagement. Supposing the double aspect hypothesis (DA) to be true, an obvious question now presses for attention. (5) How in the world do we manage to do it? 2. Imaginism There is an already good-sized literature suggesting that one plausible answer to (5) is that we do so by applying the resources of imagining.4 In Otаvio Bueno's Kelowna talk, he proposed a formula that might capture at least the gist of the imaginist approach to "Blade Runner". When the movie engages us, (6) We imagine an image of Harrison Ford as standing for Dekker. 3 Not, of course, in Leon Festinger's original sense, as when the purchaser of a new stove keeps checking to see whether the proud acquisition had been purchased at best price. 4 There is a multiplicity of imaginist approaches to fiction. In addition to Bueno's own "Representing and picturing: Approaches in the sciences and the arts", American society for Aesthetics Newsletter, 34 (2014), 8-11, see Stacie Friend, "Imagining fact and fiction", in Kathleen Stock and Katherine Thomson-Jones, editors, New Waves in Aesthetics, pages 170-187, London: Palgrave Macmillan, 2008, as well as her talk at UBC, Vancouver, March 27, 2015, "Reality in fiction"? Although there are numberless other manifestations of this way of proceeding, I stay with these particular cases for the recency of their stimulation. Many of those numberless others are in the slipstream of Kendall Walton's make-believe or pretense approach. I've never been convinced that make-believing-that and imagining-that are sufficiently resembling to warrant an attribution of identity or even of close conceptual equivalence. Nor can I reconcile myself to Kripke's proposal that the names that occur in fiction aren't names at all, but only "pretended names". See here Walton, Mimesis as Make-Believe: On the Foundations of the Representational Arts, Cambridge, MA: Harvard University Press, 1990; Timothy Schroeder and Carl Matheson, "Imagination and emotion", in Shaun Nichols, editor, The Architecture of the Imagination: New Essays on Pretense, Possibility and Fiction, pages 19-39, New York: Oxford University Press, 2006; and Saul Kripke, Reference and Existence: The John Locke Lectures, New York: Oxford UNIVERSITY PRESS, 2013; p. 29. The book is an essentially unrevised record of the original 1973 lectures. 3
This might be so. Let's suppose it is. Then something else is true as well. (7) There is something that we imagine an image of Harrison Ford as standing for. The ease of the passage from (6) to (7) tells us of the affection natural languages have for a simulacrum of the formal rule of existential generalization (EG). It is a root and branch fondness for the idea that when some proposition is true, there is something it is true of. On the face of it, imaginists have landed themselves in a pickle. On the one hand, they believe (8) There is nothing that Dekker is, and yet, if we give English its sway, (9) There is something that Dekker is after all. He is the right-most relatum in the standing-for relation of (6) and the very thing that (7) quantifies over. But (9) contradicts (3) Imaginists want to make merry with Dekker without the nuisance of anyone having to be him. As we have it now, the cost of the merry-making of (6) is that Dekker himself is its object. 3. Remedies and loops I suggested at the Kelowna conference that, upon reflection, one form of possible remediation wouldn't actually work. It is a suitably adjusted reapplication of (6) to itself. The point of (6) was to make it ontologically and epistemologically respectable to believe: (10) In doing what he did on camera, Harrison Ford made it the case that someone was doing what Dekker does in the story. The trouble is that (10) in English gives in turn (11) There is something of which (10) is true. Suppose that we treated (7) in the cleansing way we tried to get (7) itself to cleanse (6). Doing that might give us something like (12) When we imagine an image of Harrison Ford as standing for Dekker we imagine that same image as standing for someone. But given the inferential leanings of English, we also have it that (13) When we imagine an image of Harrison Ford as standing for someone, there is something that we imagine that image of Ford as standing for. 4
Whereupon the original problem comes full circle, nipping the imaginist's ontically motivated parsimony in the tail. There is no strict problem of inconsistency here, not anyhow an inescapable one. Every time we get we something like (7) we have a point of release in the likes of (12), crafted in the spirit of (6). But each time we avail ourselves of it, we get something like (13) in return, again in contradiction of (3). We might think, however, that it's not an enduring contradiction, because we can always call upon a re-casting measure in the manner of (6) to restore at least momentary consistency. The trouble each time is that this is consistency-restoration just one step ahead of a fit, as the old saying has it. What this suggests is the futility of thinking that the easy way to get rid of objects is simply to paraphrase them away. Paraphrases often bring new quantifiers in their train, and with them the objects they range over, which in the cases presently review are the very objects we wanted to get rid of in the first place. This is the basic lesson of ontological commitment realistically construed. Avoidance of their mention is no guarantee of an object's nonexistence. On the other hand, see below note three, section 5. Some people have suggested that it might repay imaginists to seek relief in some or other member of the free logic family. Let's look into that now. 4. Free semantics In some of its mainline variations, quantifiers in so-called "positive" free logic are allowed to range over two classes of objects, undistinguished as to type except by domainresidency. Intuitively speaking, existent objects would fall into one domain and nonexistent ones into the other. Reference is allowed within the latter domain, but quantification isn't. In particular, the EG rule fails in that domain's referentially successful contexts. "Sherlock Holmes" is allowed to refer but nothing that is Sherlock Holmes can be a value of a variable bound by the existential quantifier. This would appear to free the imaginism of (6) from the perils of looping. (6) can stay, but (7) cannot. The EG rule is inoperative here. Even so, in several cases free logicians try to uphold the four horsemen laws, maintaining that while "Holmes lived in Baker Street" is true, it cannot be true that Holmes exists.5 5. Models and regimentations The question I'd like to raise here is whether or to what extent it is plausible to model the semantics of English in the formal semantics of a free logic, or to submit natural language discourse about the nonexistent to the regimentations that free logics offer. I admit to thinking it unwise to exercise the modelling or regimentation options simply on the uncritical assumption that since we've got a successful formal semantics for a system that handles abstract representations of fictions, it would be advisable to adapt the semantics of our own literary discourse in the requisite ways. For one thing, English is attached to EG at the hip. 5 Another work of note is R. M. Sainsbury, Reference Without Referents, New York: Oxford University Press, 2005, and Fiction and Fictionalism, London: Routledge, 2010. However, Sainsbury's is not a "positive" dual-domain free semantics, but rather a "negative" free semantics in which, among things, all atomic sentences of the form t = t¬ are necessarily false, as are sentences such as "x (`Vulcan' refers to x iff x = Vulcan)". A further difficulty is that in negative free logics all non-existents are indiscernible. 5
Besides, a proof of formal representability would be helpful here, but I see none presently on offer and none in prospect. In its more general sense, modelling X as Y is thinking of X in ways it's known not to be, and thinking so for the benefits that flow from so doing. But without some specification of the hoped-for benefits, and some account of how these benefits are actually brought about by such misrepresentations, modelling could turn out to be little more than flying on a wing and a prayer. Whereas modelling is virtuous distortion, regimentation is virtuous reformation. Regimentation has a different motivation. Let Renglish be a regimentation of English. Then, like English itself, Renglish is also a natural language, a reformed version of the former. Reforming measures normally include cleansings of the English lexicon, and enrichments by way of neologisms and other notational contrivances. Of at least equal importance are restructurings of the home grammar, sometimes in ways designed to render the regimented language fit to be engaged by some or other pre-selected model-theoretic structure or frame. In which case, the semantics of Renglish would not be the nativist semantics of English. It would be a formal semantics of a non-natural notational system in application to the muchgerrymandered semantics of the regimentation's lexicon and grammar. In application also, to the relations in which the regimentation's expressions stand to the abstract entities of the interpretation-structures on which the formal semantics rests. Think here of the perestroika favoured by Quine, in which a true scientific theory would pass the test of philosophical tenability only if truth-preservingly recastable as a classical first order theory. Quine had no difficulty in specifying at least some of the benefits of this preferred kind of face-lifting. It would expunge intensional idioms and eliminate any need of abstract objects, beyond the mathematically everyday ones we find in axiomatic set theory. No one should think that Quine's project for science is F.O.B.6 Think here of Quine's own reluctantly conceded recognition of the first order impenetrability of Quantum Mechanics. The point generalizes. It is strikingly difficult to get even well-individuated fragments of English, made up of simple declarative sentences and compounds of them effected by what English Grammarians used to call conjunctions ­ e.g., "not", "and", "or", "if then", and "if and only if" into good enough shape for the regimentations on offer in the school-boy propositional calculus. This leaves us with the two questions currently in view. (14) What are the sought-for benefits of the regimentations of our mother tongue? (15) By what means are these benefits effected by them? Following close-by is a third question: (16) How do we know that they do? Let's sum this up with a rule: The Due Diligence Rule: Before modelling X as Y or restructuring E as RE, due diligence should be performed. 6 The reference is to imported goods available for collection free of all excise on board the ship that brought them. 6
If, on the other hand, we looked for a treatment of fictional discourse in the nativist semantics of English, the unanswered question of (5) would still await. If it were true that such semantics would have to reflect the dual-aspectness of our ties to fiction, there would be the question of whether it could be provided for in any of the ways of imaginism in which the EG rule were gold-standard. Mind you, since the theory is ours to build, perhaps we could simply override EG on policy grounds. We could say that we're sacrificing EG for the preservation of (3). We could say that the semantics of English is internally inconsistent as witness (9) and (3) and, being so, the theorist of meaning has no option but to legislate. Think here of Russell in relation to sets, and Tarski in relation to truth.7 I have nothing principled to say against this, but I'd like to see a little due diligence first. There are other possibilities to consider. We could opt for quantifiers that range over nonexistent but otherwise well-individuated objects, but not before overriding the particular quantifier law.8 Or we could accept the challenge to write a load-bearing semantics for an English that honours the four horsemen but flourishes without EG. Each of these ways has its costs and benefits, always in my experience with more of the former than the latter. With the help of Jill Isenberg who was a doctoral student at the time, I've given option two a try, but I am far from resolved to give up on option one. Meanwhile, I await with anticipation how today's imaginists might fire-proof natural languages theories of fictional engagement in which EG has only selective sway at best, and (5) is favoured with a plausible answer. Let's turn to this in Note Two, after a short detour into meaning. 6. Meaning Here is where another dificulty arises. It lies in the very idea of a nativist, or intuitive, semantics for English. What the word "semantics" means in everyday English is a theory of meaning, whereupon "a semantics for English" likewise means a theory of English meanings. It is a mistake to think that a formal semantics is just a species of a theory that tells us the meanings of English, except in this kind of case the language is not English or any other of its ilk. The language is L, a notational system, none of whose expressions carry anything like what English carries as meanings. Consider a simplified case: 1. The rod is heated the rod expands 2. The rod is heated 3. So, the rod expands. 7 Disturbing moves in each of those cases. For details see my "Does changing the subject from A to B really provide an enlarged understanding of A? A puzzle and a muddle", to appear. 8 See, for example, Richard Routley, "Some things do not exist", Notre Dame Journal of Formal Logic, 7 (1966), 251-276; John Woods, The Logic of Fiction: A Philosophical Sounding of Deviant Logic, The Hague and Paris: Mouton, 1974; second edition, with a Foreword by Nicholas Griffin, volume 7 of Studies in Logic, London: College Publications, 2007; Terence Parsons, Nonexistent Objects, New Haven: Yale University Press, 1980; and Dale Jacquette, Meinongean Logic: The Semantics of Existence and Nonexistence, Berlin: de Gruyter, 1996. See also my overview paper, "Fictions and their logic", in Dale Jacquette, editor, Philosophy of Logic, a volume in Dov Gabbay, Paul Thagard and John Woods, editors, The philosophy of science, pages 1061-1126, Amsterdam: Elsevier, 2007. It is reprinted as an appendix to the second edition of The Logic of Fiction. For some more recent state-of-the-art work, Woods, editor, Fictions and Models: New Essays, with a Foreword by Nancy Cartwright, Munich: Philosophia, 2010. 7
The antecedent and consequent of (1) are English sentences substituting for atomic wffs p and q of the propositional calculus. The formal semantics of that logic will tell us a lot about this structure. It will tell us that it instantiates a valid argument form, made so by its engagement of the logic's modus ponens rule. It will also give us the truth conditions for (1), and it had an atomic valuation function it would tell us whether (2) and (3) are true, and also that if (1) and (2) are true so, of necessity, will (3) be. But one thing that this formal semantics won't do is give us the slightest idea of what these English sentences mean, just as it gives no idea of what "p" and "q" mean in L. How could it do? Atomic letters are meaningless in L. Perhaps when we speak of adapting the intuitive semantics of English to the requirements of a theory of fiction, we mean only a theory of reference, truth and inference for English, adapted for fictional discourse. If that were our intention I wouldn't mind a bit, beyond calling it what it isn't. It would not be a theory of meaning for fiction, hence not a semantics either. All the same, it is far from clear that there's anything cobbled together by linguists that could convincingly be called a theory that assigns to each sentence of English the meaning(s) it has in English, any more than we have ways to specify for each sentence of English the very conditions under which they would be true. (Citing Convention T here would just be a joke.) In a way, this might not matter. It has only comparatively been at the centre of any semantical agenda for fiction to arrive at an understanding of what "Holmes lived at 221 B Baker Street" means.9 Surely what it means is just like what "Obama lives at 1600 Pennsylvania Avenue means. Suppose there actually were a wholly general theory of the meanings of English expressions. Then (17) If a theory of meaning for fiction were called for, then M would be the right call for fiction written or translated into English. Truth, however, would be a different matter, and reference too. We'll explore these a bit more in the note to which we now turn. NOTE TWO Semantics psychologized A note on a psychology-first paper by Woods and Isenberg10 1. Affective response When we read of Bill Sykes' savage murder of Nancy in Dickens' Oliver Twist, we are shocked and angered. When we await to see whether the Soviet agent Karla will finally cross from East Berlin to the West, where his vanquisher George Smiley awaits, the heart races and 9 Although Sainsbury is an exception of sorts. All sentences carrying singular terms in referential position are false if they have no referents. Yet "Holmes lived in Baker Street" is intelligible. Sainsbury wants to know why and how. He delivers his answer via reference condition into the manner of a Davidson explanation of meaning via truth conditions. 10 John Woods and Jillian Isenberg, "Psychologizing the semantics of fiction", Methodos: Savoir et Textes, online April 2010. 8
the pulse throbs. When we watch Juliette's death dance in the Prokofiev ballet, we cry ­ sometimes we sob and scant seconds later we dab away our tears and whisper to our companion, "Wasn't that simply wonderful!" We respond in these ways in all these cases without ever losing sight of the fact that although it was Nancy who was slaughtered, Karla who was so nervously awaited, and Juliette who was suffused with such grief, there isn't, wasn't, and never will be anyone who is Nancy, Karla or Juliette. While our affective responses were palpable and as real as the tears on our cheek, these engagements with the fictional are dual-aspected. Nowhere in these scenarios was there the least suggestion of cognitive dissonance. What makes such dual-aspected responses possible? Which takes us back to question (5) in the preceding note on Otаvio's Kelowna paper: How in the world do we manage to do it? 2. Narrativity11 DA-affect proceeds from Nancy, Karla and Juliette to us, raising the question of how, if at all, this depends on relations in which we stand to them. Under the ontically austere assumptions in play here, we cannot say that when these affects are happening, they are caused by them or by what happened to them. It can't be on account of Nancy's murder, or of Sykes' bringing it about that our hatred of Sykes and our sadness for Nancy ensues. But, affects also being effects, something or other caused them to arise. To keep things manageable, let's confine ourselves to novels and short stories. In these contexts, it is hard to imagine the causal chain leading to a reader's tears not originating in the printed pages of the story. Whereupon a further question: (18) However can typeface on the paper of a text be causally implicated in readerly grief and distress? We would achieve another useful economy if we could put to one side, for treatment another time the peculiarities executed by modernist rescaffoldings of mainstream storytelling. I intend not the slightest disrespect to the creative rhythms of modernity. But I would think that we won't have much of a chance in getting Malone Dies right until we make some real headway with Sherlock Holmes or, if you prefer to go upscale, with the Anna of Vronsky's seductions. Thus restricted, the stories for which we seek some semantic clarification are read with narrative effect. They flow in the same general way as a Times report of events in Aleppo would read. If, like the fiction-maker, the war correspondent has a byline, he too might intervene from time to time in the narrative flow, in the exercise of his auctorial authority. In each kind of case, it is the authority of an on-the-scene observer wellstocked with backstory. The difference is that Doyle is the author of that which his or the story-teller's authoritative observations are observations of, whereas the authors of what the Times correspondent observes are the combatants in Aleppo. A real difference, metaphysically huge. A difference, and some would say a problematic one, between Doyle and Tolstoy, is that whereas Tolstoy is the narrator of War and Peace, it is Dr. Watson, not Doyle who narrates 11 Derek Matravers' Fiction and Narrative, New York: Cambridge University Press, 2014, has interesting things to say about the role of narrativity in how authors tell their stories and we ourselves speak about what goes on in them. 9
the Holmes stories. I myself don't think the first part of this claim is actually true. It is true that Tolstoy wrote War and Peace narratively, and did so in a way that made it the case that someone was narrating the story. Doyle, on the other hand, is different. Doyle also writes his Holmes stories narratively, but in such a way as to make it true that the stories' narrator is not anonymous, but rather is the stories' Dr. Watson. Leo Tolstoy is no more the anonymous narrator of War and Peace than is Doyle the narrator of the stories of which Dr. Watson is the narrator. It is true that both authors can inadvertently or otherwise lead their narrators into error. When the error is inadvertent, say a confusion over the date of someone's birth ­ was it the 26th or the 29th of July, 1803 ­ it can be ignored or repaired in some contextually indicated way. But if Doyle contrived Watson's getting something wrong, he did get it wrong, and getting it wrong could be considered part of the story. Whatever else we might say of the 26th29th mixup, it does not become part of the story that it's anonymous narrator confused those dates .12 3. Telling as causal The Times' reports are tellings of what is the case. When the correspondent reports that P, his intention is to convey that P is the case, that is, that "P" is true. Sometimes these tellings are wrong. When the Times gets it wrong, it is because the reporter got the facts on the ground wrong. When the fiction-writer gets it wrong, it's because he's lost track of the facts of his own creation.13 This too is a real difference, a metaphysically huge one. Even so, there is one stable similarity that bestrides the chasm between facts of the matter made so by the world and facts of the matter made so by Doyle's creative stipulations. It is that (19) For very wide ranges of cases, being told things in reportorial fact-stating ways triggers a pre-existing causal disposition to believe what you've been told as factual.14 For some people, this won't work for Doyle's reportage of Holmes and Watson. But in light of the sheer fact of dual-aspect responsiveness, it might also be thought that it is not less than plausible to suppose that when someone is told of the doings of Holmes and Watson in The Hound of the Baskervilles, something like this happens if he knows that the teller is the creator of things he himself or his story's narrator tells. It strikes me as helpful to consider handling the question of fiction-induced belief in the same sort of way we've tried to handle fictioninduced affect. In which case, not only would we have a profitable distinction between grief and DA-grief; we would have an equally attractive one between belief and DA-belief, which latter is a state someone is in when he believes the things that Holmes got up to while perfectly 12 I am grateful to Chris Mole for helpful prodding on this point. 13 In some of the Holmes stories, Dr. Watson's war wound is located in his leg, and in others it is located in his arm. But no knowledgeable reader of these stories would take it that Watson had been twice-wounded. Rather, it was Doyle who briefly forget where. 14 This is but a particular instance of a much more general fact about the causal impact on belief simply by being told things, not confined to fiction-writers and foreign affairs reporters. See here John Woods, Errors of Reasoning: Naturalizing the Logic of Inference, volume 45 of Studies in Logic, London: College Publications, 2013; especially chapter 9, "Being Told." 10
mindful that there was no Holmes on hand to do them accompanied throughout by not a flutter of cognitive dissonance. Whereupon we would have it that (20) There is no state-of-mind type of which Jane's belief that Dov lives in Muswell Avenue and John's belief that Holmes lives in Baker Street are tokens.15 Whether in affective form or doxastic, these DA-states would be states in which we have been put by tellings. These would be things that happen to us when we read literary narratives. But, by virtue of their double-aspectness, what these states are responses to is not the things that caused them. 4. The text, the full story and the world Everyone knows that printed texts understate the stories they tell, from which it must follow that some of the sentences that tell the full Holmes story are not of Doyle's pennage. For at least twenty-five years a good many literary semanticists have made themselves at home with a tripartite distinction drawn by Gregory Currie, separating such sentences into the fictive, metafictive and transfictive. Fictive: Sentences expressly occurring in the text; e.g. "Holmes waved our strange visitor into a chair." Metafictive: Sentences that can be inferred from the text; e.g. "Holmes was a wonderful abducer". Transfictive: Sentences about the fictional having no occurrence, express or inferential, in the story the text tells; e.g. "Othello is not the main character in Othello."16 My own inclination is to absorb this trio into a larger classification, but not before some rebaptizing. Explicit: Sentences occurring expressly in a fiction text. ("Holmes waved our strange visitor into a chair.") Inferred: Sentences having an implicit occurrence in the text, arising from inferences drawn by the reader. ("Holmes had a spine.") External: Sentences expressing the observations and speculations by readers about the goings-on in a story. ("Othello is not the main character in Othello.") 15 Unless possibly, there is some more generic state of mind ­ not generic belief ­ but a state of which both the state that Jane is in and the state that John is in could be types. Perhaps something along the lines of a Quinean disposition to assert. 16 Currie, The Nature of Fiction, Cambridge: Cambridge University Press, 1990. 11
Intentional: Sentences reporting relations in which we ourselves stand to the objects and events reported by sentences of the first two classes. ("Agatha Christie admired Holmes more than any other detective.") Cross-over: Sentences registering cross-story comparisons. ("Holmes was certainly more intelligent than Li'l Abner.") Rest-of-the-world: Sentences true of the world of the story that are not themselves part of the story. ("London is thousands of miles east of Moose Jaw.") A key part of this semantics is what we might call (21) The Anti-Closed World Assumption: Except for contrary indications in the stories, fictional works inherit the world. Save for those author-sourced exclusions, the world of the story is the actual world; everything true in the actual world at the time of the story is true in the world of the story. Why do we need this assumption? We need it because there is always more to a story than what the sentences of the text directly express. Some of those further sentences will be derived by inferences, all of whose premisses are expressly in the text. Others however, will be drawn from premisses at least one of which is nowhere to be found in the texts or in solely text-based inferences. For example, the inference that leads to "Holmes had a spine" in the absence of a premiss that tells us that human beings have spines, would be imperilled, and readers would be left in a state of non-negotiable uncertainty over what's true of Holmes and what's not. The anti-closed world assumption turns the trick by supplying a large number of those premisses. It does so by ruling that whatever is typical of the human animal is also true of Holmes. But we don't want that sentence itself to be part of the story. We now begin to see a semantically significant interplay between our inferreds and rest-of-the worlds. (22) The-rest-of-the-worlds furnish premisses for a large class of a story's inferreds.17 Here are some further points worth noting: These distinctions of sentence-type mightn't be exhaustive and certainly not exclusive. (There is no reason to think of the intersection of implicits and externals as empty.) The explicits and the inferred constitute the text's full story. Externals and intentionals are markedly different. Externals owe their reference to the prior provisions of the full story (FS). 17 Even "London is thousands of miles east of Moose Jaw" stands ready for premissory service. In conjunction with "Holmes lived in London", it gives "Holmes lived thousands of miles from Moose Jaw", which although not part of the story, is nevertheless true. 12
Intentionals state (purported) facts about the world, but do so only subject to the condition that, whereas they aren't sentences of the story, they couldn't be true without them. (It couldn't be true that Christie admired Holmes more than any other sleuth, without there having been facts fixed by the sentences of the Holmes stories. If (21) is right, literary semantics has no need of or room for a possible worlds semantics. Indeed, I regard all talk of the world of the story as un faзon de parler. Of course, a writer might write into being a world called Other, radically different from this one (I mean the one and only). But it won't be a world, just so. It will be a world of its author's story. The main point is that fiction does not, as such, introduce any world but this one, albeit as fictionalized in accord with the author's comparatively very modest imposition. Since the world of the story in this world thus modified, it is surplus to need to postulate that the real world thus modified is indeed a different world from this world, crying out for the mercies tendered by Kripke frames. I am minded to see two things wrong with this. If Holmes' world is a world at all, it is not a possible one. There is no possible world in which Holmes waved our strange visitor into a chair. The reason why is that Holmes himself is not a possibile, not an individual capable of actualization.18 Of course, we could always seek refuge in nonnormal worlds. This takes me to my second reservation. Nothing much is known of impossible worlds beyond what is vouchsafed us by nonnormal models. Were we to pursue this option, we'd be back at modelling the good Queen's good English in ways that we've decided to abandon as, to say the least, surplus to need. What (21) also gives is an intelligible distinction between what the story makes true of Holmes and what's true of Holmes' world. As we now see, this is a distinction that motivates our sentences. Accordingly we can define a story's maximal account (albeit not in Parsons' sense).19 A story's maximal account (MA) is the one constituted by the union of the sentences of the full story and the sentences true of the world in which the story occurs. In standard model theoretic approaches, a semantics is an ordered triple R, T, I, where R is a theory of reference, T a theory of truth (and thereby also of consequence), and I is a theory of inference, construed as the drawing of consequences. We are now approaching a double-aspect semantics for fiction, in which these three theoretical elements are retained but the order of their appearance is reversed as follows: (23) The structure of a semantics for fiction: I, T, R.20 The anti-closed world assumption helps us to see how full a full story actually is and also how determinate its world is: 18 See The Logic of Fiction, chapter 3, sections 3 and 4. 19 Terence Parsons, Nonexistent Objects, New Haven: Yale University Press, 1980. 20 This is a departure from "Psychologizing the semantics of fiction", prompted by fruitful correspondence with Matthieu Fontaine. 13
(24) The determinacy principle: Except where expressly provided for by an author's interventions, a story's world is as determinate as the actual world is. (25) The completeness principle: Except where expressly provided for by the author, fictional objects are as complete as are objects of their same type in the actual world. Another benefit of the closed world assumption is that we can now say how a theory of inference would work for the semantics of fiction. It can be seen as a corollary of (17), which tells the same story for meaning: (26) A theory of inference for fiction: Whatever the correct, or best, theory of inference for English itself may turn out to be, it is also the theory that works for stories told in fiction, subject only to disallowances recognized by the anti-closed world assumption. We can now begin to see why a semantics for fiction would start with an account of inference rather than ending with it, as in the classical approach. When it's time for truth to step up to the plate, we'll need a decently specified subject matter for it, something to which the theory's truth-predicate can apply. We wouldn't be able to satisfy this condition without an anti-closed world assumption and the deliverances of (26) 5. Quantifiers de re and de dicto Gladstone was born in 1809. When someone tells us this, we have a belief formulable de re: x (x = Gladstone Belreader (born 1809, x)). Likewise we may say that x (refers ("Gladstone", x)). As long as we stick with our four ontically parsimonious constraints, we can't treat the sentences of a text's FS in like manner. Accordingly (27) The no-reification principle: A semantics for fiction that honours these four constraints ­ especially the particular quantifier one ­ must withhold its true FSsentences from positions de re. Which means that a reader's admiration of, belief that and reference to Holmes can only take de dicto formulation. What this means is that (28) Our four fiats require the suppression of EG in fictional contexts. As we have it so far, when I grieve for Nancy or believe that Sykes is a brute, there is in 14
each case a relation in which I stand to something, but not in either case to Nancy or to Sykes or to his brutal murder of her. The thing to which I stand when in those DA-states is Oliver Twist. Is there a corresponding story to be told about reference? When I experience myself as referring to Nancy or to Sykes is there a DA-state that I'm in, say a state of DA-reference? Perhaps that is actually so, which would mean that being in it causes no trouble for Frege's law. There are two obvious questions about this: (29) How does DA-referring get to be like DA-hating? (30) How do sentences constructed around DA-references get to be true? 6. Truth and ambiguity21 As mentioned, in a good many of the standard approaches a theory of truth for a language arises from a prior theory of reference. In the present case, it will be necessary to reverse this dependency. Our first task, as we saw, was to activate the account of inference so as to provide an application-class for the truth predicate. It is now time to consider what its application conditions are, but not before taking due note of an important kind of ambiguity, The Ambiguity Postulate: If a declarative sentence occurs in an FS, it is made true by its occurrence there. If it appears outside an FS and yet is true, it is made so by other conditions. If has a fictional name in referential position, it is false in contexts apart from its own FS. Accordingly, (31) "Holmes waved our strange visitor into a chair" would be ambiguous with respect to contexts of occurrence. Intuitively, (31) captures the difference between being true in actuality and being true in a story. It is not uncommon for a literary semantics to reflect this difference with (usually unvoiced) sentence-operators, "a" for "in actuality" and "f" for "in the story". There is a further benefit promised by these ambiguities. It would help us see why when we believe that Sykes killed Nancy and yet no such thing actually occurred, we aren't believing a contradiction. That in turn explains the lack of cognitive discourse. On the view we are developing here, the central task for a semantics of fiction is to furnish an implicit definition of "f", by specifying truth conditions for sentences f (S)¬, whose scopes S occur under tacit modification of the a-modifier. These will be different sets of truth conditions. The reason why is that (32) It is a requirement of a DA-semantics for fiction that it reflect the relevant ambiguities in the semantic structure of f-sentences. 21 Heads-up: In Note Four I'm going to say why I think the ambiguities discussed in this note and the one that follows aren't ambiguities after all. They are, so to speak, faзons de parler prompted by an important characteristic of fiction sentence, about which more anon. Still, the faзon de parler is in constant use, and briefly availing ourselves of it until Note Four will do no lasting damage here. 15
(33) The natural way to do it by construing the scope of f (S)¬ as carrying the aoperator. In other words, (34) The explicit sentence "Holmes waved our strange visitor into a chair" comes out in the syntax of the theory as "f (a(Holmes waved our strange visitor into a chair))". Our theory of truth makes this sentence true just in case its scope occurs in an appropriate FS. This means a true f-sentence can have a false scope, whose local falsity is semantically overridden by virtue of its occurrence in the FS. It is a false sentence all right, but when it appears in the story, it isn't false there. 7. Reference If a full story inherits as much of the world as the author's own provisions allow, then in the Holmes' stories it will be true that (35) (f "Sherlock Holmes" refers to Sherlock Holmes) and also that (36) When we ourselves read those stories we are induced to DA-believe its scope. If someone, say Barbara, reads The Hound of the Baskervilles with understanding, (37) She flat-out knows that f (`Holmes' refers to Holmes), but DA-believes that `Holmes' refers to Holmes and also that (38) In uttering "Holmes lived in Baker Street", she knows that she refers to what "f (`Holmes' refers to Holmes" refers to, but DA believes that she refers to Holmes. She also knows, even so, that (39) `Holmes' doesn't refer to Holmes. That's why her belief that it does is a DA-belief. Let's try to sum this up. DA-reference: In uttering "Holmes lived in Baker Street", Barbara DA-referred to Holmes just in case Barbara knows that what she's referred to is what the true sentence "f-(`Holmes' refers to Holmes)" refers to ­ namely, nothing ­ and yet experiences herself as referring to Holmes, even recognizing that her referential experience lacks a referent. 16
Note well: Lacks a referent, but not a cause. In other words, (40) Barbara experiences her situation as if Frege's Law is false, believing all the same that it's not false. Towards the end of note one I floated the idea that unless we got rid of EG it might not be possible to give expression in any natural language to a theory that tells the truth about fiction. I want to come back to this now. This will be part of the business of note three. NOTE THREE Objectless Truths A note on a book by Peter Alward22, 1. Empty Revelations adopts Currie's threefold classification of fictional sentences. I have already indicated a preference for something more expansive, which for ease of reference I'll simply repeat. 1. Explicit: Sentences occurring expressly in a fictional text. ("Holmes waved our strange visitor into a chair".) 2. Implicit: Sentences having an implicit occurrence in the text. ("Holmes had an oesophagus.") 3. External: Sentences expressing the observations and speculations made by readers of the goings-on in a story. ("Othello is not the central character of Othello".) 4. Intentional: Sentences reporting relations in which we ourselves stand to the objects and events reported by sentences of the first two classes. ("Agatha Christie admired Holmes more than any other detective".) 5. Cross-over: Sentences registering cross-story comparisons. ("Holmes was certainly more intelligent than Li'l Abner.") 6. Rest-of-the-world: Sentences true of the world of the story but not part of the story itself. In other words, sentences in the maximal account of the full story. ("London is thousands of miles east of Moose Jaw.") 22 Peter Alward, Empty Revelations: An Essay on Talk About and Attitudes Toward Fiction, Montreal and Kingston: McGill-Queen's University Press, 2012, the subject of an Author Meets Critics panel at the Canadian Philosophical Association, Victoria, June 2, 2013. The other panelists were Richard Vallee, l'Universitй de Moncton, and Jill Isenberg, UBC. 17
As was the case with Currie, my explicits are nearly enough Peter's fictives, my externals his metafictives, and my cross-overs his transfictives. But it is also clear that Peter's book makes plentiful use of my implicits, (41) "Holmes wears underpants", but is not as relaxed as I am with world-sentences, e.g., (42) "It is true in the world of the Holmes stories that Julius Caesar had a determinate (natural) number of full-grown hairs on his head when he died." 2. ER-sentences: Consider now the sentences that registers the claims of Peter's own book. I want to focus on the sentences in which expressions such as "fictional world", "Fictional character", "fictional happening", "fictional fact-teller" and "narrative fact" occur in grammatically referential position. Take care to include those cases in which ERsentences also embed, in referential position, terms such as "Holmes", "Moriarity", "flat-mate of Holmes", and so on: "Holmes is a fictional character" "John H. Watson was Holmes' fictional flatmate" "Sherlock and Mycroft's brother are one and the same". To keep things even simpler, narrow the focus to those sentences that aren't negative existentials, such as "Sherlock Holmes doesn't exist". Call these "Alward's ER-sentences". Suppose that the ER account is true, hence that its ER-sentences are true. By the theory's own provisions, the following would also be true: (43) In none of these sentences is there anything whatever to which such expressions refer. Here is Peter on this point: "Please note: as I use it, talk of worlds is a useful bookkeeping idiom without ontological import [T]he view on offer here is unabashedly anti-realist; any appeal to 18
an ontology of fictional entities in either the account of fictional discourse or of fictional names (or elsewhere) is eschewed." Accordingly, (44) Although all those sentences are true, there is nothing whatever of which they are true. If this is a tolerable conclusion, there is a large and non-trivial class of declarative sentences of English that are about nothing, say nothing of anything, are true of nothing at all, yet are true. If Alward's account is true, it might well be a significant piece of scholarship. If so, it would be a significant piece of scholarship whose value and importance is uncompromised by the circumstance that there is nothing whatever about which it gives its revelations give readerly instruction. 3. Truth: Obvious questions now call for attention: (45) What are the truth conditions of these true sentences? (46) What notion of truth would such conditions embody? It is easy to see that it couldn't be anything like the correspondence concept of truth, or any that admits of a (Fregean) designationalist semantics. Perhaps we could consider taking a coherentist approach. In that case, the truth of Alward's true ER-sentences would be nothing more and nothing less than their coherence with the totality of antecedently acknowledged true beliefs of the community of the author's co-linguals. If this were our choice, we would inherit coherentism's standard baggage. No one seems to quite know what coherence is. Besides, others complain that coherentism is a slippery slope to idealism. A more particular difficulty is that (47) The ER-theory does not cohere with what virtually everyone else believes; e.g., that there are lots and lots of things that don't exist and Holmes is one of them. But Peter is a true four horseman. He won't give up on l'idйe fixe that there is nothing whatever that doesn't exist. 4. Inferentialist and proof-theoretic approaches A third possibility is some or other suitably adapted theory from the meaning-withouttruth school of thought,23 or the languages-without-ontology camp.24 Another related one is to seek out a partner in the inferentialist semantics community. It is a varied community. For 23 Stephano Predelli, Meaning Without Truth, New York: Oxford University Press, 2013. 24 Jody Azzouni, Talking About Nothing: Numbers, Hallucinations and Fictions, New York: Oxford University Press, 2010. 19
example, under Michael Dummett's inferentialist provisions, intuitionism stands to logic as philosophy stands to anti-realism. A realist thinks that there exists an independent metaphysical reality in virtue of which every proposition is either true or false, never mind what we ourselves might think or do; that is to say, independently of our recognizing their respective truth values. Anti-realists are differently minded. Truth in mathematics and everywhere else is either an honorific by-product of the methods by which a proposition is proved, or anyhow a property that merely supervenes on them.25 My reservation in the first instance is that Empty Revelations is not a book about meaning, unlike Sainsbury's Reference Without Referents, which is all about meaning. There is no mention of meaning in the Alward index.26 Nor is Alward's task is to say how FSsentences would acquire meaning if none of them was true, but rather to say how such sentences, whatever the details of its meaning-status, get to be revealing. But, my particular interest is in how Alward's own ER-sentences get to be true. As far as I can tell, Empty Revelations simply presupposes my (17). Of course, there might be something defective about the assumption that if the ERsentences are true, there is no telling of how they get to be true independently of the objects they are true of. A related possibility might be some or other variation of a proof theoretic semantics. Initially proof theory had nothing to do with semantics. Indeed, the basic idea was that proof theory is less about what the following-from relation is than it is about the ways in which, starting from A, we arrive at B. In coming to understand how showing that all those Bs follow from all those As, we come to an understanding of what it is for one thing to follow from another. We come to see that the meaning of "follows from" is fixed rather by the conditions under which it is shown that this follows from that. The idea here is that semantics is best handled in terms of them. In a somewhat later development, proof theoretic semantics is understood to be the semantics of proofs themselves, rather than about semantics in terms of proof, in which meanings are determined not by truth conditions but by proof conditions. Although the two branches can be intertwined, for our purposes only the latter or in-terms-of approach need be considered. 27 25 Michael Dummett, The Logical Basis of Metaphysics, London: Duckworth, 1991. See also Robert Brandom, Making it Explicit: Reasoning, Representing and Discursive Commitment, Cambridge, MA: Harvard University Press, 1994; Articulating Reasons: An Introduction to Inferentialism, Cambridge, MA: Harvard University Press, 2000; and Between Saying and Doing: Towards an Analytic Pragmatism, New York: Cambridge University Press, 2008. 26 Nor of Dummett or Brandom nor of Predelli or Azzouni. However, Frege, the earlier Kripke and the earlier Kaplan make the cut, but not the later Kaplan of (somewhat) inferentialist leanings. (Neither is the later Kripke mentioned, understandably so since Empty Revelations preceded Reference and Existence by a year.) For some of Kaplan's later thinking, see "Afterthoughts", in Joseph Almog, John Perry and Howard Wettstein, editors, Themes From Kaplan, pages 565-614, Oxford: Oxford University Press, 1989; "Words", The Aristotelian Society, 64 (1990), 93-119, and "What is meaning? Explorations in the theory of meaning as use", brief version, draft 1 Ms. 27 Proof theoretic semantics is now a burgeoning field, ramifying all over the place, what with Hilbertian proof theory, Gentzen proof-theory, inferentialism, operative logics. Gentzen semantics, natural deduction with higher level rules, constructive type theory, logic programming, and heaven knows what else. Modern developments of significance include several contributions by Dag Prawitz, beginning with Natural Deduction: A Proof-Theoretical Study, Stockholm: Almqvist & Wiskell, 1971. Another productive flow originates with Peter Schrцeder-Heister's "A natural extension of natural deduction", Journal of Symbolic Logic, 49 (2009), 1284-1300. Unjustly neglected is Per Martin-Lцf's Constructive Tye Theory, originating in his Intuitionistic Type Theory: Notes by Giovanni Samboni of a Series of Lectures given in Padua, June 1980, Naples: Bibliopolis, 1984; repair of its neglect can be found in a major paper four years later by Aaren Ranata, "Propositions as games as types", Synthese, 76 (1988), 377-395, and more recently by Nicolas Clerbout and Shahid Rahman, Linking Game-Theoretical Approaches with Constructive 20
The real question here is whether the four horsemen can be made sense of in the antirealist climate of inferential semantics. Or are they at bottom designationist principles with realist intent? Take Parmenides' Law for example. If we say that "There is nothing that doesn't exist" holds only for a proof-theoretic "exists", the principle would fail. There are lots of things that can't be proved. Goldbach's Conjecture might be one of them. But interestingly enough, the Fiction Law might be better served when expressed as "Nothing is an object of fiction unless proved so by the methods that created it." Proof theoretic/inferentialist semantics arose in contexts for the formalization of constructivist mathematics, and of the intuitionist logics that cater for them. Whatever its levels of prosperity in those formalist environments, it is open to question whether this approach adapts well, in a load-bearing way, to a semantics for English, notwithstanding attempts to pull it off. In other words, as mentioned earlier in respect of free logic semantics, if we wanted to model native semantics in an intuitionist/proof-theoretic environment, would we have at our disposal a suitable formal representability theorem? I must say that at this stage of our semantic enlightenment I have my doubts. 5. Brandom In his approach to the philosophy of language, Robert Brandom wants an inferentialist account of "true" and "refers to", freed from the representationalist idea that the proper role for language and thought to provide a "transcript of reality". He shares with pragmatists its denial that truth is a substantive metaphysical reality, and that the truth about "true" and "refer to" is revealed in the normative regulation of our actual day-to-day linguistic practices, especially those that have to do with reasoning and interactive cognitive engagement. In Making it Explicit we find a systematically formulated normative pragmatics situated in an "anthropological" naturalism in the manner of Cheryl Misak28 (and me too29). On this approach we understand concepts of philosophical interest by keeping an eye on how they are instantiated in practice. Of equal importance, Brandom also adopts a kind of vocabularyrelative stance, in which talking this way rather than that way is the best way to talk of matters in a given context, and that no way of talking should be privileged for all context or made canonical across the board.30 No regimentation here.31 Type Theory: Dialogical Strategies, CTT demonstration and the Axiom of Choice, soon to appear with Springer in the Springerbriefs series. Papers that try to adapt to natural languages include Nissim Francez and Roy Dyckhoff, "Proof-theoretic semantics for a natural language fragment", Linguistics and Philosophy, 33 (2010), 447-477, as well as Goran Sundholm, "Inference and consequence in an interpreted language", a talk at the workshop Proof Theory and Philosophy, Groningen, 2013. Hilbert's breakthrough came as a talk in 1904, reprinted as "On the foundations of logic and arithmetic", in Jean Heijenoort, editor, From Frege to Gцdel, pages 129-138, Cambridge, MA: Harvard University Press, 1967. Gentzen's influence dates from his 1934/35 paper "Untersuchungen ьber das logische Schliessen", reprinted in M. E. Szabo, editor, The Collected Papers of Gerhard Gentzen, pages 68-131, Amsterdam: North-Holland, 1969. Given our present interests, we should take note of Bartosz Wieckowski's "Predications in fiction", in M. Pelis, editor, The Logica Handbook 2007, pages 267-285, Prague: Filosophia, 2008. See also Luca Tranchini, "Truth from a proof-theoretic perspective", Topoi, 3 (2012), 47-57. 28 Cheryl Misak, editor, Pragmatism (= Canadian Journal of Philosophy, Supplementary Volume 24), Calgary: University of Calgary Press, 1999. 29 Errors of Reasoning: Naturalizing the Logic of Inference, especially chapter 3, section 3.2 ff. 30 Between Saying and Doing. 31 As far as I know, Mathieu Marion was the first to suggest a link between Brandom's pragmatic inferentialism and dialogue logic in, for example, his "Between saying and doing: From Lorenzen to Brandom and back", in Giuseppe 21
Some see the core of Brandom's notion of inference in communicative contexts in Goran Sundholm's parsing of a remark of J. L. Austin's in "Other minds".32 When I say "Therefore", I give others my authority for asserting the conclusion, given theirs for asserting the premisses. I myself would re-read this slightly as: When I say "Therefore" I invite others to assert my conclusion, given their invitation to me and others to assert their premisses.33 Perhaps the strongest reason for Alward to ally with Brandom is that Brandom is a disquotationalist about truth. For him, the celebrant "Snow is white" is true is simply an anophoric device for saying that Snow is white. So Brandom, I think, would resist my suggestion that if "Snow is white" is true there need be an object of which it is true. Putting disquotational truth together with Brandomian inferentialism, generates (or at least suggests) a theory in which truth is what supervenes on negotiation settlement via the norms that regulate human speech and communication. My own view is that in an adroit and circumspect adaptation, this might be made into a reasonably entertaining semantics for sentences of and about a story. The Hilbertian roots of this movement sprung from philosophical reservations about the reliability of the notion of truth in the foundations of mathematics, leaving little room at the time to anticipate later proof-theoretic contributions to semantics. However, by the time we get to analytic pragmatism, there is no trace of Hilbert, and rather more of Dewey's naturalism, and also of Quine, Sellars and Rorty. It might strike us, in particular, that a naturalized pragmatism in the manner of Brandom could be a happier framework for Alward's empty revelations. It would stir none of the formalized representability doubts arising from free logic, or constructivistly motivated intuitionist logics, or of proof-theoretic inclination. Brandom's inferentialism is a semantics purpose-built for natural language. But, even so, here too I have doubts. I have them not so much about Brandom as I do about a Brandom semantics harnassed to an Alwardian austerity. For one thing, (48) The ER-sentence "It is impossible to refer to Sherlock Holmes of Baker Street" is Primiera and Shahid Rahman, editors, Acts of Knowledge: History, Philosophy and Logic, Essays Dedicated to Goran Sundholm, volume 9 of the Tributes series, London: College Publications, 2009. 32 Aristotelian Society Supplementary Volume 20 (1946), 148-187. 33 I adopt the idiom of invitation in favour of the clumsy idiom of authorization. I owe my preference for invitation to Robert Pinto. See Robert Pinto Argument, Inference and Dialectic: Collected Papers on Informal Logic, a volume in the Argumentation Library series, edited and with a Foreword by Hans V. Hansen, Dordrecht: Springer, 2001; especially chapter four, "The relation of argument to inference". 22
not derivable from any empirically vouched-for regulatory apparatus of English speech. In other words, (48) The invitation to assert the impossibility of referring to the nonexistent hasn't been taken up by the human speech community at large. Brandom could be an attractive fit for fictional discourse, but it would be a good deal less so under three of those four-horsemen conditions that Peter is pledged to honour. Where are we then? (49) As far as I can see, Alward's book offers no principled guidance concerning the truth conditions of its own ER-sentences. (50) What is more, it is not clear to me what those truth conditions would be. If Alward accepted the Brandomian option, he could dismiss (49) and (50) as irrelevant. But if he declined the option, (49) and (50) would still stand. Of course, no literary semanticist to date has produced a head-on semantics for his own book. So I'm not suggesting that in that respect Peter is in any way out of line. Perhaps this is a matter to which he would consider devoting his next book. I, for one, would be first in line to read it. But for the present, it might be interesting to speculate on how this follow-up account might go. 6. English and other mother tongues Let me start these ruminations by emphasizing that Peter's theory of fiction is well laidout in plain English, modestly supplemented by a standard sort of philosophical vocabulary. No doubt we will find these same virtues in the second book. The language of Peter's exposition is English and, like every other native tongue of my acquaintance, English is a language which simply refuses to be bound by the constraints Peter imposes on discourse about fiction, our by now infamous three of four laws. And I think that it may come to matter that he imposes them with what I myself take to be a careless generality. Here they are again: i. Parmenides' Law: There is nothing whatever that doesn't exist. ii. Frege's Law: No singular referring expression refers unless there exists a referent for it. iii. The Particular Quantifier Law: Nothing is something unless it exists. iv. The Fiction Law: There is no object that is an object of fiction. The objects of fiction don't exist. I take it to be an empirical fact of long-established usage that English is structured 23
around a kind of the fiction law and existential generalization rule. But it shows no tolerance of the particular quantifier law. Any speaker of English not contaminated by philosophical anxieties would freely grant that there are lots of things that don't exist: (51) Santa Claus doesn't, Othello doesn't, and, some even say that God doesn't. Another way of saying the same thing is that the English Language, and any other like it, harbours an intrinsic attachment to objectual quantification, but no like attachment to existence-dependent reference. In fact, I think that it has no tolerance either for Parmenides Law and EG as Peter interprets them. That is, (52) English dislikes existence-dependent quantification every jot as much as it disdains existence-dependent reference. In other words still, (53) Since English doesn't want to interpret the "existential" of "existential quantification" asexistence-imputing, it doesn't want to obey the particular quantifier law. If the Parmenides-Frege-Particular Quantifier package were right for English, but not in the way that Peter (and nearly everyone else since Frege) interprets them, they would be right under the right interpretation, roughly as follows: i *Parmenides' Law*: There is nothing that isn't something. ii *Frege's Law*: If `n' refers, there is something to which it refers. iii *The Particular Quantifier Law*: If something Fs there is something that Fs. iv. *The Fiction Law*: There are such things as fictional objects even though they don't exist in actuality. In like fashion, we could have v *Existential Generalization*: If an individual a F's, something F's. Before moving on, let's not overlook a point of importance. The real fly in the ointment of the logic of fiction is not so much EG, but rather the Particular Quantifier Law. Even if read as iii "If something Fs then there exists something that Fs" the fact remains that in English (54) "There exist things that don't exist, e.g. Santa" is neither inconsistent nor untrue 24
It's the first occurrence of "exist" is quantificational and the second predicative. A quirk of English. It is the same way in French, with "il y a" doing duty for "there exists", as "es gibt" also does in German. But with a difference: "exist" has no quantificational function in French or German. The language in which Peter formulates his theory of fiction has an inherently existenceneutral quantificational structure. In languages such as English, there is an irresistible quantificational pull on singular attribution, with quantifiers occurring in positions de re. If Caesar crossed the Rubicon in 49 BC, (55) There is someone of whom "crossed the Rubicon in 49 BC" is true. It is no different with Vulcan. Since Le Verrier hypothesized its presence to account for the orbital perturbations of Mercury, here too there is something of which this is so.34 The plain fact is that humanity's mother tongues cannot abide our four ontically motivated constraints. Human languages respond to them as the living respond to the four horsemen of the apocalypse. Mind you, there are legions of philosophers who make their livings by scolding natural languages for all this naughtiness, for their sheer semantic and ontic confusion. I have some advice for these philosophers. I advise them to be circumspect in their abuse. For if English is the language in which they themselves hurl it, the hurling will not comport well with the hurled. Admirers of the original Parmenides-Frege-Particular Quantifier package are left with no recourse save an aggressive eliminationism, or some or other form of totalitarian regimentation of their mother tongues. It is an abiding feature of those who, as a matter of considered theoretical principle, give up thus on natural language have no means of registering their disapproval except in one of those languages whose abandonment they seek. Frege is notorious for thinking that not one declarative sentence of German has a truth value, yet ventured to say so in German. Tarski notoriously thought that the natural language predicate "true" had a null extension, yet pressed his case by saying in Polish (then German, then English) the very thing he took to be true. Quine was a rare example of a like-minded totalitarian who actually cottoned on to this: Quine insisted that no languages passed philosophical scrutiny, save those having the requisite first-order extensional structure, and said so in a language of irreducibly non-extensional character. Yet it was not entirely lost on Quine that such second-class languages must be tolerated, even prized, for their "conversational ease." Various people have something to say about this: "Who does Quine think he's kidding?", some say. Perhaps that is a trifle harsh. It could be that Quine had beaten Alward to the punch. Might not Quine's more considered opinion have been that English itself ­ not just Alward's theory of English Fictional discourse ­ is replete with revelations that chance to be empty? That is, empty of contents connecting to what they're revelations of? Empty Revelations brims with evidence of this attachment to English's quantificational signature. Page after page, there are load-bearing passages such as these (56) "Fa(p) iff Revealedn (NS, p) 34 Of course, Le Verrier got it perfectly right for Uranus in relation to certain other astronomical oddities. 25
where the subscripted `a' and `n' denote respectively the actual world and the narrative world." (p. 154). Leaving the question of notational perspicuity to one side, the formula is intended to represent the structure of sentences such as (57) "The narrative informant of The Last Chronicles of Barset revealed that Joseph Crawley was accused of stealing twenty pounds." Another (unformalized) example is: (58) "Finally, the fourth theme is that reader access to the fictional worlds generated by the works they read is mediated by a non-actual fact-teller: the narrative informant." (p. 171) If Peter wants any kind of first-growth Parmenides-Frege-Particular Quantifier semantics for the language of the book in which he ventures to tell the truth about fiction, he will have a formidable regimentation problem on his hands. I have no idea of how its untangling might go. But however it does go, its goings cannot be achieved save in the unexpunged parts of the language whose abandonment is sought. Everyone has known of this since (philosophical) time immemorial. Let's call it: (59) The Bootstrapping Problem: Beings like us, speaking the only languages we are fit to speak or think in, are unable to bootstrap ourselves from the grasp of those languages even for the purposes of giving up on them. Historically, the bootstrapping problem draws one or other of at least three responses from philosophers. Pretend to pay it no mind. Make light of it ("conversational ease", "helpful heuristics"). Mysticism, which is Parmenides' own solution and, later, Wittgenstein's. 7. Mysticism Wittgenstein's reflections on the mystical occur at 6.44, 6.45 and 6.52-6.54 of the Tractatus,35 where he says that the book that he's just written is nonsense. The majority view is that the end of this most extraordinarily rigorous exposure of the structure of the world, Wittgenstein is indulging himself in a burst of Viennese cafй eccentricity. I belong to the minority, which sees in these closing reflections the mediations of Konstantin Levin (nearly enough Tolstoy himself) in Part VIII of Anna Karenina: 35 Ludwig Wittgenstein, Tractatus Logico-Philosophicus, translated by C. K. Ogden, London: Routledge & Kegan Paul, 1922. A reprint of the 1955 printing was released by Dover in 1999. In some ways the McGuiness & Pears translation is more faithful to German, but less so I think to Wittgenstein's German. I read the Tractatus as a philosophical poem. 26
6.52 We feel that even if all possible scientific questions be answered, the problems of life have still not been touched at all. Of course there is then no question left, and just this is the answer. 6.522 There is indeed the inexpressible. This shows itself; it is the mystical. 6.54 My propositions are eludidatory in this way: he who understands me finally recognizes them as senseless, when he has climbed out through them, on them, over them. (He must so to speak throw way he has climbed up on it.) He must surmount these propositions; then he sees the world rightly. The mysticism that I advance here for Alwardian consideration amounts to this. It is the view that there are true propositions, propositions which seriously advance our understanding of things, of which the following is true: (60) They can be thought, understood and believed. (61) They cannot be consistently voiced or sententially expressed. (62) Yet in a great many cases, perhaps typically, they can't get to be thought or entertained, much less than understood or believed, except as voiced or sententially stimulated. (63) They are, as we might say, empty revelations. (64) That is, they lack expressible semantic content. One of the charms of Empty Revelations is its careful and subtle treatment of fiction's empty reflections. Its onus is to show how it is possible to expose revealing truths occasioned by theories of fiction without there being anything of which they are true. This is a smallscale mysticism, purpose-built for the fictional. What I am suggesting is that its more general applicability to any language like English once lashed to the first-growth Parmenides-FregeParticular Quantifier constraints. I say that those constraints are beyond fulfillment by any language a human being is fit to speak or think in. Here, too, the revelations of such speech ­ whether of Caesar and the rest of our inexistent dead, Zeus and those other false gods, and never-there planets such as Vulcan, and made-up ones like Krypton ­ run foul of this fact if fact it be. Of course, the old options are available for consideration. Pay this inexpressibility no mind. Or give it the brush off. Or hitch to the wings of an Alwardian mysticism. It's not at all clear to me how to proceed with this generalized mysticism. But it had better not be rejected out of hand. The alternative could hardly be less dire. There would be no consistently tellable story about fiction for Peter or anyone else to tell. There would be no consistently tellable story about Caesar and Zeus, and Vulcan too, for the humanities to tell. For the present, never mind Caesar and the others. What matters here is Sherlock. Under 27
current assumptions, Peter can't tell the truth about Sherlock. "Whereof one cannot speak, thereof one must be silent" was Wittgenstein's parting shot at Tractatus 7. You might think that he would have thought that Peter's better course would have been silence, or a career in real estate. In fact, Wittgenstein would have thought no such thing. He didn't think it of his own book, and wouldn't think it of Peter's. Wittgenstein thought that the Tractatus was full of empty revelations occasioned by its own inconsistent tellings. And I rather think that he would think it of Peter's book too. The question is: What does Peter think? Indeed, what do the rest of us think? It is now more than time for another change of pace, beginning with an attempt to rid ourselves of the fiction (no pun intended) that the sentences of fiction are systematically ambiguous. NOTE FOUR Fictional Discourse is Not Systematically Ambiguous A note on a recent paper by John Woods36 1. Strategic ambiguation At the heart of note 2 we find the ambiguity postulate, and it resonates across most of the post-1969 fictionalist literature and has had a generous presence in note three. It would be a substantial disturbance to the research programmes of fiction if the ambiguity postulate were shown to be groundless. It might also radiate with good effect to the rest of philosophy. Ambiguity, whether present or merely invoked, is a preferred instrument in the philosophical problem-solver's tool box. When trouble brews and conflict flares up, a hopeful and much pursued question arises: (65) Wouldn't it be lovely if there were something of which the contested sentence is satisfyingly true and something else of which it is satisfyingly false?37 What would it take to bring this about? How could we get a philosophically fevered sentence to be simultaneously true and false? An answer much favoured by philosophers is that (66) A given single sentence S can be simultaneously true and false when it concurrently carries two different meanings. This works beautifully for lots of English sentences: (67) "I'll meet you by the bank." (68) "Lucille's coat is quite dear." 36 John Woods, "How robust can inconsistency get?", IFCoLoG Journal of Logic and its Applications, 1 (2014), 177-216, especially sections 2.3-2.6. 37 This is what in Paradox and Paraconsistency, Cambridge: Cambridge University Press, 2003, I called the Reconcilation Strategy; see pages 80-90 and 151-154. 28
(69) "Visiting relatives can be boring". (Chomsky) The first two of this trio are lexically ambiguous, where differences in sentence-meanings arises from the ambiguities of the words "bank" and "dear". A good many philosophers follow certain theoretical linguistics in characterizing (69)'s ambiguity as syntactic, arising from the grammatical "deep structure". There is no time here to go into the contested complexities of transformational grammar. The first importance of ambiguities such as (69) lies in the impact it has on a compositional semantics for English. But, assuming that we "buy into" the idea of ambiguities embedded in deep sentential structure, we are free to contemplate the source of the alleged ambiguity of (70) The detective who resided at 221B Baker Street was both respected and disliked by the Metropolitan Police especially when compared with (71) The sisters who taught Joan's mom at Loretto Abbey were both respected and disliked by their pupils. It is clear upon inspection to any fluent speaker of English that there is no more lexical ambiguity in (71) than there is in (70), and none at all that would single out fictional sentences as grammatically distinctive. Syntactic ambiguity might be less easily discerned than the lexical variety, but I know of no empirically sensitive linguistics for English in which (70) embeds anything like the syntactic structure of (69). I am inclined to think that whatever its syntactico-semantic character, (70)'s is not the same as (69)'s. It simply can't be true that just any old semantics that works for (69) will also work for (70). Insisting that a Chomskian semantics for (68) will turn the trick for (69) strikes me as wishful thinking of a kind so abundantly present in philosophy as to motivate another false principle: The Ambiguation Strategy: Deadlocked philosophical problems seek and attain relief by postulation of independently unevidenced ambiguities. Of course, postulations sometimes strike pay-dirt. For example, the intractable warfare between epistemological justificationists and anti-justificationists prompted us to notice some genuine ambiguities in words like "justify" and "justifiable", but also trapped us with meanings that had no prior presence in English. The moral of this story is that we should be careful not to over-strategize our philosophical zeal for conflict resolution. In particular, we should be sensitive to the fact that, inattentively resorted to, (72) The ambiguation strategy has a tendency to make things up. When it does, its meanings are fictions. There is a related point to note: 29
(73) Ambiguities are much more frequently invoked than shown to exist. And a corollary: (74) Showing should precede invoking. I think it might fairly be said that the ambiguity thesis for fiction, in the forms in which we've had it for more than forty years, has been advanced in the creative spirit of the Ambiguation Strategy just above and heedless of the advice sounded by (74). The idea that (70) has a grammatically deep structure in which the intuitive distinction between "true in the story" and "true in the world" is brought to principled heel by the postulated distinction between "f (The detective who resided at 221B Baker Street was both respected and disliked by the Metropolitan Police)" and "a (The detective who resided at 221B Baker Street was both respected and disliked by the Metropolitan Police)" is not an idea upheld, or even advanced, in a transformational grammar for "Visiting relatives can be boring". From whence then does the imputed ambiguity arise? On the face of it, it arises from what happens to "true" in apposition, respectively, to the qualifications "in the story" and "in actuality". In which case, we come upon the idea that there is no sentential ambiguity in"f (The detective who resided at 221B Baker Street was both admired and disliked by the Met)", still less so in"a (The detective who resided at 221B Baker Street was both admired and disliked by the Met)". The imputed ambiguity lies in the predicate that quite rightly applies to the former and quite wrongly applies to the latter. If really so, this takes us straight back to lexical ambiguity, not for sentences in the "object language" of fiction, but rather in its "metalanguage" (stretching each time the proper meaning of the two quoted expressions). Which leaves us to ask: (75) Is the predicate of "true" lexically ambiguous in English ? And with it a follow-up question: (76) Is there a known theory of meaning for the English language in which (75) is answered affirmatively? The moral of all this is that although both unambiguously true, the unambiguous sentences "Holmes lived in London" and "It is not the case that Holmes lived in London" have different kinds of truth-makers. Different truth-makers making one and the same ambiguous sentence both unambiguously true and uambiguously false, require no ad hoc plea of heretofore undiscerned ambiguity to explain how this could be so. 2. Whereabouts It is often observed that the meaning of a sentence is not uniquely determined by its terms' lexical meanings and its own syntactic meaning. When the clerk in London's Registry comes upon "Sherlock Holmes lives at 221B Baker Street", he will rightly exclude it because 30
it's false. But when official of the Bootmaker's Club in Toronto responds to a question about Holmes' residency, he will reply that he lived at 221B Baker Street, and what he says will be perfectly true. What matters here is the whereabouts of an unambiguous sentence that is unambiguously true in one place and unambiguously false I another. This launches the suggestion that (77) Whereabouts differentially distribute truth values over lexically and syntactically unambiguous sentences. There is nothing surprising in what (77) tells us. (77) is true of any contingent sentence subject to the vagaries of time. If Rome's registrar of births and deaths in the year that Caesar crossed the Rubicon were to list him among Rome's living what he'd have said would have been true. But it would have been false the day after his murder on Rome's Senate steps. Similarly when it is 10 a.m. in Vancouver it is 7:00 p.m. in Istanbul. When "It is 10 a.m." is true in Vancouver, it is false in Istanbul. When "It is 7 p.m." is true in Istanbul, it is false in Vancouver. But no one thinks that either sentence is ambiguous. Logicians of various stripes have wanted to invest "Caesar is a resident of Rome" with ambiguities undetectable in linguistic theories of ambiguity, what with embedded temporal operators and inflections of tense. It all strikes me as surplus to need. The intuition we've wanted to preserve is that of what's true in the world and what's true in the story. Both are places where sentences are true and false sometimes differentially so without ambiguity. Besides, when we add sentences operators such as "f" for "in fiction", "n" for "now", "p" for "before now", and a different "f" for "later", we draw down all the heavy-equipment technologies that have accreted since 1959 for modal logics. In 1974, I advanced the idea that the "f" prefix for fiction is a modal operator.38 I've now changed my mind. (78) "Sherlock Holmes lived at 221B Baker Street" is not a modally structured sentence, notwithstanding its sensitivity to whereabouts. At this juncture it might be natural to ask what I think whereabouts we are. Aren't what I call whereabouts what just everyone else calls contexts?39 My answer is that, long since, context has been considered a prime source of ambiguity. If we could free "context" from this bondage, I wouldn't mind a bit. But I will be the first to say that I have no worked-out theory of whereabouts. 3. Truth conditions It is doubtlessly so that the true sentence "Sherlock Holmes lived at 221B Baker Street" has truth conditions of a different sort from the truth conditions for the true sentence "Mr. Obama lives at 1600 Pennsylvania Avenue". It is commonly said that the first is made true by the story and the other is made true by the world. This natural way of speaking has abetted in turn what I take to be an error as unfortunate as it is widespread. It is the error of thinking that truth conditions are what specifies meanings in a natural language. The confusion arises from 38 The Logic of Fiction, especially chapter 5. 39 My colleague Chris Mole among them in an email of 30 April, 15. 31
Tarski's original expropriation of the word "semantics" by which he meant a model theory for formalized notational systems, from which meanings are wholly absent.40 It is true that certain of these notations are "defined". For example, (79) v (~A) = T iff v (A) = F Is said to define, and thus to give meaning to, the connective "~". This could not have been said convincingly were we not also told that we are at liberty to pronounce "~" as "not". But the language L of this logic, say the propositional calculus, is not a language for speaking. There is nothing pronounceable in L. Of course, we see what's going on. In L"~" is a formal representation of "not" in English. Which takes us right back to the earlier question of whether there is much of a chance of striking paydirt in structuring our interest in truth and reference in fiction in a model theory adapted to English. One reason to think not is this: (80) Meanings influence truth conditions in English. Meanings do not perform this role in the model theories of languages like L. There are no meanings in L.41 4. Losing a distinction In "How robust can inconsistency get?" I suggested that the made-true by the story and the made-true by the world distinction was a false contrast. It is true that fictional sentences have different kinds of truth conditions from the others, but what I was suggesting was that the by-the-world/by-the-story distinction doesn't successfully capture it. Here is why. The truth of "Sherlock Holmes lived at 221B Baker Street" is brought about by whatever Sir Arthur Conan Doyle in the course of writing his stories, augmented by the conventions that generate the FSs and MAs of the Holmes canon. These comings to pass are events in the world, not Holmes' world, but Doyle's. In other words, ours; the world. The sheer commonality of facts that make fictional sentences true should not deflect us from their specialnessness. "Analysis is true in number theory" is a sentence made true by facts of the world of a discernibly different kind than the facts that make "Humanity is beset by strife" true. But no one would think that the meaning that the analysis sentence has is of a different kind from the meaning the humanity sentence has. No one would think that what the word 40 See here John P. Burgess, "Tarski's tort" in Burgess, Mathematics, Models and Modality: Selected Philosophical Essays, pages 149-168, New York: Cambridge University Press, 2008. 41 A closer look at (79) will help drive the point home. It is customary to read (79) as saying the same as (81) Not-A is true just in case A is false. This is the wrong reading. The right reading is this: (82) The value of ~A¬ is T just when the value of A is F. In (82) there is no mention of truth or falsity. Truth and falsity are properties respectively expressible by the English predicates "true" and "false", neither of which is satisfied by any item of L. On the other hand T and F while called "truth values" are undefined abstract objects, and values of valuation functions on atomic wffs of L in partnership with the rules for assigning T and F to compound wffs. Calling these things "truth values" is as tendentious as calling model theory "formal semantics". 32
"meaning" means in English is typewise ambiguous as between the analysis sentence and the humanity sentence, still less than the predicate "true" is in respective application thereto. This leads me to propose that (83) "Sherlock lived at 221B Baker Street" and "Barack lives at 1600 Pennsylvania Avenue" are both made true by the world, but are so by different types of fact. Part of this difference in type is caught by the concurrent truth and falsity of the first but not the second. Note, however, (84) "Sherlock lived at 221B Baker Street" is concurrently true and false, made so by different facts in the world. But the facts that make it true are not themselves at any odds with the facts that make it false. In other words, they are compossible facts. "But how can this be?", I hear it cried all about. "How does this comport with the Law of Noncontradiction?" Must we now become Hegelians (or Brazilians or Australians)?" 5. Contradiction and respects The Law of Noncontradiction is one of Aristotle's great contributions.42 In Metaphysics he offers three inequivalent formulations of it, in the following order: (85) The doxastic formulation: No one can believe that the same thing can (at the same time) be and not be. (1005b 19-20) (86) The ontological formulation: It is impossible that the same thing belong and not belong to the same thing at the same time. And in the same respect. (1005b 19-20) (87) The logical formulation: The most certain of all basic principles is that contradictory propositions are not true simultaneously. (1011b 13-14) LNC's logical formulation rules the roost today. This is unfortunate. It understates Aristotle's fullest statement of the law, the ontological formulation. It is in this formulation only that rules at against concurrent and unequivocal truth and falsity in the same respects. In so doing, he accommodates the fact that different compossible facts of the world can make unambiguous sentences concurrently true and false. It also discloses the inadequacy of LNC's doxastic formulation. It may well be, that aside from a sprinkling of dialectical materialists and dialethic logicians, no right-thinking individual will believe that A and "not-A" are concurrently and unambiguously true in all respects, it is left wide open that they might be unambiguously and concurrently believed in concurrently compossible but different respects. I have no general theory of respects. No one does as far as I know. We might think it worthwhile to try to think one up. Perhaps en route we could secure some theoretical purchase on non-ambiguating whereabouts. 42 Although Aristotle is not its originator. See Plato: "The same thing clearly cannot act or be acted upon in the same part of or relation to the same thing at the same time in contrary ways." (Republic, 436 B) 33
6. Dual-aspectness again The full formulation of Aristotle's LNC also accommodates the commonplaceness of dual aspectness. When we thrill to the hoped-for arrival of Karla, and are happy to see it when it happens (88) It is true in respect of the facts that brought Smiley's People into effect that Karla arrived and (89) It is true in respect of the other facts constitutive of the goings-on that night at Checkpoint Charlie that no such thing happened. These propositions make it intelligible, more so than in the Woods & Isenberg paper, that we can concurrently thrill to the prospects of Karla's arrival and yet know that no statement of the form "x arrived safely" can make the cut in the historical record of Checkpoint Charlie arrivals when Le Carrй's Karla is the value of x. However, (90) If we found ourselves drawn to a respects theory of truth for fiction, we'd have a difficult time to continue to side with the four horsemen of the logic of fiction. My advice would be (91) Not to worry; why martyr English to the ontic anxieties of some philosophers, the fastidious high priests of lessness? So (92) Farewell to at least the first three horsemen. Not that these suggestions are entirely problem free. Consider an example I owe to Chris Mole. (93) "There were several planets in Le Verrier's theory, of the Solar System. The one that turned out not to exist was Vulcan. The truth-makers for the first of these claims are Le Verrier's theory, but for the second are the heavens. "And", as Mole observes, "the two are anophorically linked." The question raised by the anophoric link point generalizes to a bigger one. (94) Does a dual-aspect theory have a tellable story about anaphoric contact between one aspect and another? 34
My first inclination is to say that anaphora works there in the same way it does in single-aspect settings, just as do meaning and inference. But I see that I've now exhausted my time. More anon in due course. Acknowledgements: For astute and helpful comments on earlier drafts I warmly thank Nicolas Clerbout, Shahid Rahman, Christopher Mole and Matthieu Fontaine. I also thank Shahid Rahman and Clerbout for an advance draft of their "A dialogical frame for fictions as hypothetical objcts". 35

File: the-logic-of-fiction.pdf
Author: John Woods
Published: Mon Jun 15 23:21:20 2015
Pages: 35
File size: 0.37 Mb


Before I dead, 10 pages, 0.2 Mb
Copyright © 2018 doc.uments.com