! REFERENCE THROUGH DEMONSTRATION: SINGULAR TERMS, PROPOSITIONAL ATTITUDES AND FICTION by GEMMA CELESTINO FERNANDEZ Llicenciada en Filosofia, Universitat de Barcelona, 1999 Diploma d’Estudis Avançats, Universitat de Barcelona, 2002 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE STUDIES (Philosophy) THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver) October 2012 © Gemma Celestino Fernández, 2012 ! Abstract This thesis defends original accounts of the semantics and metaphysics of propositional attitudes, their reports and fiction in general. These accounts are unified by being parts of a general view on language, according to which demonstrations play a crucial role in explaining a wide range of phenomena. The basic ideas of the view defended here about propositional attitudes and their reports can be summarized as follows: first, propositional attitudes are binary relations between individuals and structured propositions, which are constituted by individuals and properties, but also possibly by representations of those individuals and properties. Secondly, like utterances of sentences containing singular terms in general, propositional attitude reports containing singular terms express both a descriptive proposition and a singular structured proposition, which are the objects of propositional attitudes. Thirdly, a standard Gricean pragmatic explanation is offered to account for those cases in which reports seem to ascribe an attitude towards a less fine-grained proposition than the ones to which the theory appeals. The proposal on fiction defended here is also derived from more general views on language and metaphysics. The basic idea is that the meaningfulness of fictional discourse, which involves many empty singular terms, is mostly due to the descriptive proposition that any utterance of a sentence expresses in addition to expressing the other, more widely accepted proposition. This is also due, however, to the occurrence of implicit prefixes such as ‘the fictional character’, ‘fictionally’, or ‘a fictional persona’, in addition to the ones that are already discussed in the literature. According to the view defended here, fiction does not represent real possibilities –nor is it intended to. Fictional worlds, while having the same nature as possible worlds, do not represent possibilities for the actual world. Both proposals are based on a semantic view of singular terms presented and defended at the outset, which is a Fregean metalinguistic token-reflexive view. ii ! Preface -A previous version of part of Chapter 5 has been published as Celestino Fernandez (2008) “Fictional Contingencies” ESSLLI’08 Student Session Proceedings. -A previous version of part of Chapter 6 has been published as Celestino Fernandez (2009) “Demonstrating Fictional Names”, in Actas del VI congreso de la Sociedad de Lógica, Metodología y Filosofía de la Ciencia en España, Valencia. (ISBN 978-84-3707655-3) ! """! !"#$%&'(&)'*+%*+,& -#,+."/+&0000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000011 2.%("/%&000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000111 !"#$%&'(&)'*+%*+,&00000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000013 41,+&'(&5167.%,&00000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000003 -/8*'9$%:6%;%*+,&000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000031 <%:1/"+1'*&00000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000= >&?*+.':7/+1'*&0000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000> @&A1*67$".&!%.;,&00000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000B !"#$%&'()$%&)*+,-./0$1-2(-/'$""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""3 !"!$45/2,&6$7,&0(2(-/&6$8,-965:'"""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""#; !"<$=.''566>'$7?5-,@$-A$B5A(/(25$B5'),(C2(-/'"""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""!D !";$E9F5)2(-/'$2-$B5'),(C2(G(':"""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""!H !"I$B5'),(C2(G(':$&A25,$J,(C*5>'$E9F5)2(-/'""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""<< !"K$8,&+:&2()$L(66(&/(':$&/0$M5:&/2()$7N-OB(:5/'(-/&6(':"""""""""""""""""""""""""""""""""""""""""""II C&!'8%*DE%($%=131+F000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000GH <"#$7?5$7-*5/O=5A65P(G5$Q(5N$-A$R/05P()&6'$&/0$B5:-/'2,&2(G5'$""""""""""""""""""""""""""""""""""""IH <"!$E9F5)2(-/'$2-$2?5$7-*5/O=5A65P(G5$Q(5N"""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""K< <"<$S$T,5+5&/$7-*5/O=5A65P(G5$Q(5N$"""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""3! <";$S$8,5'.CC-'(2(-/&6$T,5+5&/$Q(5N$-A$M(/+.6&,$75,:'$"""""""""""""""""""""""""""""""""""""""""""""""""""3; <"I$7?5$M5:&/2()'$&/0$L52&C?@'()'$-A$M(/+.6&,$75,:'$"""""""""""""""""""""""""""""""""""""""""""""""""""""33 I&<%;'*,+."+13%&<1,/'7.,%&00000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000JI ;"#$%&'()$8,-965:'$&/0$1-2(-/'$""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""UI ;"!$S$8,&+:&2()$1-/OL(66(&/$Q(5N$-A$8,-C-'(2(-/&6$S22(2.05$S'),(C2(-/'V$L52&C?@'()'"""H3 ;"<$S$8,&+:&2()$1-/OL(66(&/$Q(5N$-A$8,-C-'(2(-/&6$S22(2.05$S'),(C2(-/'V$M5:&/2()'"""""### ;";$B5&6(/+$N(2?$7,&0(2(-/&6$8,-965:'$"""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""#!D ;"I$E2?5,$=56&250$8?5/-:5/&$"""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""#!K ;"K$E2?5,$Q(5N'$"""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""#<< ;"3$E9F5)2(-/'$&/0$=5C6(5'$""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""#<U G&51/+1'*&"*:&?+,&)'*+1*6%*/1%,&00000000000000000000000000000000000000000000000000000000000000000000000000000000000000>II I"#$8,-965:'$""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""#;; I"!$M-:5$L&(/$Q(5N'$"""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""#3I I"<$B&G(0$W5N('$&/0$X,5+-,@$4.,,(5>'$B5'),(C2(G(':$""""""""""""""""""""""""""""""""""""""""""""""""""""""#3H B&<%;'*,+."+%:&51/+1'*"$&K";%,&00000000000000000000000000000000000000000000000000000000000000000000000000000000000>HH K"#$7?5$M5:&/2()$&/0$L52&C?@'()&6$%&'('$-A$BT1$"""""""""""""""""""""""""""""""""""""""""""""""""""""""""""!DD K"!$B5:-/'2,&250$T()2(-/&6$1&:5'$"""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""!DI K"<$BT1$G5,'.'$2?5$B5'),(C2(G('2$W5N('O4.,,(5$Q(5N$"""""""""""""""""""""""""""""""""""""""""""""""""""""""!!; K";$Y-,,(5'$&/0$S/'N5,'$""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""!!K L&)'*/$7,1'*,&0000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000@IM E%(%.%*/%,&000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000@IC !" List of Figures Figure 5.1: Facebook post ....................................................................................................... 154 ! "! ! Acknowledgments There are many people I want to thank here for many different reasons. I want to thank the members of my supervisory committee: Dom Lopes, Jeff Pelletier and Ori Simchen. The three of them have had an especial impact on me as a philosopher and have influenced my ideas and career. I have learned a lot from them. Since my very first days in Vancouver, Dom pushed me into the realm of analytic aesthetics and professional philosophy. Clearly, he is the one to blame for my having taught courses on analytic aesthetics at UBC, having participated in conferences and meetings on analytic aesthetics, and having written a dissertation that is partly on fiction. Jeff provided me a great support as well. He was extraordinarily generous with his time and created what we ended up calling ‘the JeffLab’, which consisted in weekly meetings with him and graduate students of SFU and UBC to discuss our work in progress. He would always help us understand each other’s work with extremely clear, interesting and helpful explanations of any of the issues that each of us were working on at the moment. Finally, I want to thank Ori for his constant support during all my time at UBC, for his contagious passion for metaphysics and the theory of reference, and for forcing me to think twice, and better, about my views. He made me improve my views, as well as my arguments in their favour, with his criticism and constructive opposition. He very carefully read different drafts of this dissertation and extremely shaped it with his comments. Undoubtedly, without his contribution this dissertation would have been worse. I also want to thank those who went out of their way to discuss with me main ideas of this dissertation in their different stages of development, thereby helping me to improve their content: Manuel García-Carpintero, Josep Macià, Chiara Panizza, Stefano Predelli, Pablo Rychter, Steve Yablo, and Elia Zardini. I also benefited from presenting parts of this work in several conferences and meetings. I thank my audiences in Aberdeen, Barcelona, Buenos Aires, Edmonton, Évora, Hamburg, Santiago de Compostela, Sueca, Valencia, and Vancouver. In addition to these, I want to ! "#! ! thank some people who helped me to learn, understand, form or consider ideas related to my research in some way or another: Roberta Ballarin, Oscar Cabaco, Javier Castro Albano, Esa Díaz León, Ambròs Domingo, Jessica Leech, Dan López de Sa, Manolo Martínez, Genoveva Martí, Laura Natalia Pérez, Manuel Pérez Otero, Agustín Rayo, Sònia Roca, Sven Rosenkranz, Adriana Silva Graça, Giuliano Torrengo, Gabriel Uzquiano, Crispin Wright, and Nicole Wyatt. I also want to thank people who dedicated especial part of their time to this, for which I am very grateful: Maite Ezcurdia, Adele Mercier, Eleonora Orlando, and Ignacio Vicario. I would like to make a especial mention of José Luis Bermúdez, from whom I think I learned a way of approaching philosophy that I like and from which my work has greatly benefited. For all their support while I was becoming a professional philosopher, I want to thank many people in addition to those already mentioned above: Valeria Abusamra, Yuichi Amitani, Mariano Amoroso, Marc Artiga, Saray Ayala, Mily Bachmann, Cristina Balaguer, Eduardo Barrio, Paul Bartha, John Beatty, Vince Bergeron, Sylvia Berryman, Juan Blanco, Joao Branquinho, Paul Boghossian, Iván Camacho, Yolanda Cano, Laia Capel, Josep Capdevila, Isabel Cavallé, José Chaves, Guille Colet, Fabrice Correia, Eric Desjardins, Sofía Di Scala, Jose Díez, Marisol Diezma, Eileen Doherty, Merche Fages, Olga Fernández, Raquel Fernández, Zoe Geva, Mario Gómez Torrente, Anna González, Miguel Ángel González, Carl Hoefer, Oyvind Ihlen, Carles Isach, Josh Johnston, Marta Jorba, Philipp Keller, Manos Kitos, Max Kölbel, Brian Laetz, Chen Lizra, Mireia López, Rosana López, Natàlia Lorenzo, Ana Macias, Teresa Marques, Pepe Martínez, Esmeralda Miras, Mohan Matthen, Reza Memarsadeghi, Sonia Memetea, Ana Mladenovic Williamson, Radamés Molina, Simón Moreira, Paco Murcia, Pedro Nunes, Kaori Otsu, Romina Padró, Joan Pagès, Marco Panza, Pere Pardo, Francesc Pereña, Mirja Pérez, David Pineda, Miriam Poves, Josep-Lluís Prades, Jim Pryor, Daniel Quesada, Cristina Roadevin, Luís Robledo, Alirio Rosales, Joana Rosselló, Paul Russell, Mauricio Rychter, Noelia Sánchez, Gonçalo Santos, Margaret Schabas, Celia Serichol, Shane, Juan Carlos Solano, Chris Stephens, Camilo Suárez, Célia Teixeira, Alan Thomas, Cecilia Valencia, Jordi Valor, Maria Verdaguer, Marta Vidal, Agustín Vicente, Xavi Villar, Timothy Williamson, Ariel Wilner, and John Woods. ! "##! ! Of all the people mentioned above, I want to thank four of them with special emphasis: Manuel García-Carpintero, Josep Macià, Manuel Pérez Otero, and Pablo Rychter. Manolo, Josep and Manolo met me as an undergraduate student when I was 19 years old, and over the time they have been my teachers, colleagues and friends. Manolo GarcíaCarpintero was also my supervisor during my graduate studies in Barcelona, and we kept in touch even when I was far away. I would write him an email to ask him or tell him about a personal or philosophical question, anytime, and he would reply immediately in a helpful way. He educated, supported, encouraged, oriented, motivated me, and made it possible for me to continue working in philosophy since I finished my undergraduate studies. Furthermore, I think he is an extraordinary philosopher from whom I have learned something every time we have met. I feel very lucky to have met the three of them in the context that I have met them. Pablo helped not only by offering his full support and confidence, but also by urging me to write and finish this dissertation. He always listened to me when I needed to explain my ideas and arguments to someone who could understand them. In fact, some of the main ideas in this dissertation were originated and developed in conversation with him. He often made me see ideas, arguments, or interpretations that I was not seeing. His way of doing philosophy continuously motivates me to continue doing philosophy myself. During the time of my doctoral studies, I lived for much time far from my hometown Barcelona, in Vancouver, Cambridge Mass., and Oxford. For their welcoming, hospitality and friendliness, I want to thank Barbara and George Dabrowsky, Olga Fernández, Dom Lopes, Mohan Matthen, Daniel Quesada, Alejandro Pérez Carballo, Agustín Rayo, Katia Vavova, Steve Yablo, Ana Mladenovic Williamson, Timothy Williamson, Merche Fages, and Gabriel Uzquiano. Their hospitality made a big difference. For hosting me during the years of my doctoral studies, I want to thank several institutions. First, I want to thank my home institution, the Department of Philosophy of the University of British Columbia. Secondly, I am also thankful to the LOGOS Research Group, the Department of Logic, History and Philosophy of Science of the University of ! "###! ! Barcelona, and the Department of Linguistics and Philosophy of the Massachusetts Institute of Technology. My research was partially funded by a fellowship from “la Caixa” and the Canadian Studies Foundation; a graduate entrance scholarship, two university graduate fellowships, tuition awards, and a research assistantship from UBC; the Spanish research projects HUM2004-05609-C02-01 and FFI2011-29560-C02-01; and the consolidated and funded LOGOS Research Group 2009 SGR 1077. ! #$! ! ! ! ! ! ! ! ! "#!$%!&'()*+,!-./'!'*0!1'+2(3!'*0!+#!$%!42,/'*0!5'/.#! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! "! 1. Introduction In this thesis, I defend an original account of the semantics and metaphysics of propositional attitudes and their reports, as well as a new account of the semantics and metaphysics of fiction in general. These two accounts are unified by being parts of a general view on language, according to which demonstrations play a crucial role in explaining a wide range of phenomena. The basic ideas of the view that I defend on propositional attitudes and their reports can be summarized as follows: first, propositional attitudes are binary relations between individuals and structured propositions, which are constituted by individuals and properties, but also possibly by representations of those individuals and properties. Secondly, like utterances of sentences containing singular terms in general, propositional attitude reports containing singular terms express both a descriptive proposition and a singular structured proposition, which are the objects of propositional attitudes. Thirdly, a standard Gricean pragmatic explanation is offered to account for those cases in which reports seem to ascribe an attitude towards a less fine-grained proposition than the ones to which the theory appeals. My proposal on fiction is also derived from my more general views on language and metaphysics. The basic idea is that the meaningfulness of fictional discourse, which involves many empty singular terms, is mostly due to the descriptive proposition that any utterance of a sentence expresses in addition to expressing the other, more widely accepted proposition. This is also due, however, to the postulation of implicit prefixes such as ‘the fictional character’, ‘fictionally’, or ‘a fictional persona’, in addition to the ones that are already familiar in the literature. My view is that fiction does not represent actual modal facts –and it is not intended to; and thus, fictional worlds are alike in nature to possible worlds, except that are not possible. 1 Both proposals are based on a semantic view on singular terms that I defend at the outset, which is a Fregean metalinguistic token-reflexive view. It is Fregean in that according to it, singular terms express a sense that determines their reference and gives part of their meaning. It is metalinguistic in that the sense that singular terms express mentions language; in particular, the tokens produced of the singular terms in question. And, finally, it is token-reflexive in that the semantic properties that singular terms have are properties of the tokens, rather than types, and derive from the properties and relations that the tokens themselves have. It is worth noticing at this point that although the present proposal is Fregean in certain respects, it departs from the classical Fregean conception in certain other important respects. In particular, it takes into account the insights of Saul Kripke and other new theorists of reference in accepting that singular terms are rigid designators, and that their only contribution to the main propositions expressed by simple sentences containing them is just their reference. In addition, it does not include definite descriptions among the singular terms. My proposals are also based on a metaphysical view on the individuation of proper names inspired by Kaplan’s, according to which the semantic properties of proper names are essential. Within my proposal regarding fiction there is an account of the semantics of certain descriptive phrases, which, according to this proposal, occur often in fictional discourse, but are also outside of it. I offer an account of how these phrases work in general, inside and outside fictional discourse. The phrases in question are those constituted by a description followed by a proper name, such as ‘the Spanish writer Javier Marías’ or ‘the fictional character Sherlock Holmes’. I propose that within these contexts proper names behave differently: they do not refer -whether they are empty or not. Proper names in these specific descriptive phrases function as demonstrata. I would like to emphasize at this point that even if the proposal I make about fiction is descriptivist, it is relevantly different from the descriptivist views defended by David 2 Lewis and Gregory Currie. Unlike theirs, the present account does not claim that fictional names are disguised abbreviated definite descriptions or that they are used as such. As I have just said, I defend original views on different linguistic phenomena. By ‘original views’, of course, I do not mean views that do not share or take ideas that other philosophers or linguists have contributed to the literature on each of the topics of this dissertation. On the contrary, my views result from the historical and collaborative work of a given philosophical community of the kind that analytic philosophy takes to be fundamental for the progress of philosophy and knowledge in general. My views are original first, in that, as far as I know, no one has defended them before. And secondly, in that they do contain original ideas, as well as some already familiar ideas, put together in an original way. The defense of the general semantic and metaphysical view of language that I offer in this dissertation, and of each of its constituent parts, is based in showing how it accounts for all the relevant data. I think that in what follows I provide a good defense of my views, but by no means do I pretend to refute the competing views that may exist. Refutations rarely occur in philosophy. My main goal in this dissertation is to make a more positive than negative contribution. I am aware that there are views that I do not consider, and that I do not address all the subtleties of the views that I do consider in this dissertation. However, in most cases I at least explain what my reasons are for not buying those views. In what follows, I briefly describe the contents of each of the five main chapters that constitute this work. In Chapter 2 (‘Singular Terms’), I briefly explain those ideas, theses and views in relation to the semantics of singular terms that are relevant to and useful for understanding the views that I defend in later chapters. Among these are the basic ideas, theses and views on the semantics of singular terms and the metaphysics of propositions, such as Millianism, Rigidity, Russellianism, Direct Reference, Fregeanism, and Possible World 3 Semantics, as well as the classic discussion on definite descriptions. In addition, I carefully discuss Descriptivism, including the more sophisticated versions of it that appeared after Kripke’s well-known objections. At the end of the chapter, I also consider the pragmatic strategy adopted by Millian theorists such as Nathan Salmon and Robert Stalnaker, and Semantic Two-Dimensionalism. In Chapter 3 (‘Token-Reflexivity’), I present and defend the view on the semantics and metaphysics of singular terms that I favor. An important part of what I do is to consider the token-reflexive view of indexicals, as well as possible extensions of this view to proper names. I defend the view from objections that have been raised against it, and then I offer an assessment of a possible straightforward Fregean extension of it to proper names, and of Manuel García-Carpintero’s particular version of the token-reflexive view. Finally, after considering David Kaplan’s account of the individuation of words, I defend the version of the token-reflexive view on the semantics and metaphysics of singular terms that I prefer. In Chapter 4 (‘Demonstrative Discourse’), I discuss the semantics and metaphysics of propositional attitude ascriptions, indirect speech, ‘so-called’ sentences and quotation, as well as the informativeness and cognitive value of identity statements and simple sentences. I defend a view of all of these phenomena that explains them, as well as singular terms and fictional discourse, as manifestations of the same kind of linguistic mechanisms. These linguistic mechanisms have the following in common: they involve a demonstration; and they contextually contribute (at least) part of the semantic content of expressions as something that appropriately relates to something that serves as a demonstratum. I call this general view on language ‘The Demonstrative Discourse View’ (‘DD’, for short). In Chapter 5 (‘Fiction and Its Contingencies’), I present some important familiar problems and views of the semantics and metaphysics of fiction and fictional discourse. In addition, I explain the descriptivist view of David Lewis and Gregory Currie in detail and argue against it, paying special attention to what distinguishes it from the 4 descriptivist view that I defend in the next chapter. I also consider Frederick Kroon’s pragmatic pretense-based view and his arguments against both the familiar Russellian treatment of true negative existentials involving definite descriptions that do not denote and the ellipsis strategy, because these are a treatment and a strategy that DFN makes use of. In Chapter 6 (‘Demonstrated Fictional Names’), I propose a theory about the semantics and metaphysics of fiction and fictional discourse that is based on the particular thesis about the meaning and metaphysics of proper names, and singular terms in general, defended in Chapter 2. This proposal extends to an account of descriptive phrases in general, such as 'the Catalan writer Jaume Cabré' or 'our son John'. In particular, although on this proposal fictional names do not refer to anything at all, descriptive phrases in which they sometimes occur (such as ‘the fictional character Sherlock Holmes’) do have denotation, and we manage to make true claims about the fiction by using them. The proposal presented here is, in part, that we manage to do this by using fictional names as demonstrata within the descriptive phrases in which they occur. More specifically, fictional names are sometimes used as demonstrata that help contribute the individual that the descriptive phrase in question is about. 5 2. Singular Terms In this chapter I briefly explain the ideas, theses and views in relation to the semantics and metaphysics of singular terms that are relevant to and useful for understanding the views that I defend in later chapters. As explained in the introduction, in those later chapters I defend a general semantic and metaphysical view that is constituted by three more particular semantic and metaphysical views: one about the meaning and individuation of singular terms, another about propositional attitude ascriptions and other related contexts, and finally, one about fictional discourse. As we will see, the semantics and metaphysics of singular terms is central for the semantics and metaphysics of the other two linguistic phenomena. That is why it is essential to start this thesis with the semantics and metaphysics of singular terms, and consider the major proposals that have been made about it. Before proceeding, let me insist that I do not intend to offer a historical reconstruction of the development of the views, but a systematic presentation of them. That is to say, the presentation mainly focuses only on the ideas themselves, rather than their origins or other historical details. In Sections 1 and 2, I explain what singular terms are and present the theses and problems that started the debate. Then, in Section 3, I explain Russell’s theory of definite descriptions (as well as Strawson’s and Donnellan’s alternative views) and the way it solves problems that motivate Fregeanism. In Sections 4 and 5, I discuss the classic objections to Descriptivism and some descriptivist proposals which may seem to have survived these objections and which are close to the one I favor in certain respects (although I do not think they are correct in the end). Finally, in Section 6, I end the chapter by explaining the main solution that Millian or Direct Reference proponents have provided to some of the problems they initially face. I also consider Semantic Twodimensionalism here. ! "! 2.1. Basic Background Notions In this section I present the basic central theses on the meaning of singular terms: Millianism, Rigidity, Direct Reference, and Fregeanism. I also present the basic central views on the nature of propositions: Russellianism, the Possible Worlds view, and Fregeanism. The presentation of all of these theses and views, together with the discussion of Russell’s theory of descriptions in the next section, will prepare the ground for the discussion about the meaning of singular terms in the rest of this chapter and the next. Singular terms Singular terms are those expressions of language that purport to refer to individuals: proper names such as ‘David’ and ‘Susana’, indexicals such as ‘I’ and ‘now’, and demonstratives such as ‘that’ and ‘those’. Those who think that definite descriptions such as ‘the man in black’ purport to refer to individuals include definite descriptions in the list as well. There are different theories and theses about the meaning of singular terms that have been proposed, some of which I consider in what follows. ‘Meaning’ should here be understood pretheoretically as what competent speakers of the language understand in virtue or their competence when using the language. This competence of speakers is in turn assumed to be explainable in terms of the linguistic conventions they follow. Meaning understanding can be thought of as propositional understanding: that is, to understand the meaning of a linguistic expression X is to understand that X means Y. In what follows, I consider the main theses and theories about the meaning of singular terms that have been proposed, knowledge of which is necessary to understand the views that I defend in this dissertation. ! #! The Millian view of singular terms According to Millianism, the only meaning of singular terms, or at least of proper names, is just their referent. Thus, for instance, the proper name ‘John Stuart Mill’ refers to John Stuart Mill and so, John Stuart Mill himself, that is, its referent, is its only meaning. The Rigidity view of singular terms Rigidity is the thesis that singular terms refer to the same individuals at all possible worlds in which those individuals exist.1 Utterances of sentences containing them are evaluated at all possible worlds with respect to the same individuals that are their referents. In other words, the evaluation and, therefore, the truth-value of those utterances at all possible worlds depend on one and the same individual. Thus, for instance, ‘Saul Kripke’ refers to Saul Kripke at all possible worlds in which he exists, not just in the actual one. An utterance of ‘Saul Kripke is at CUNY’ is evaluated at all possible worlds with respect to Saul Kripke and not someone else. To evaluate an utterance at a possible world w does not imply that the utterance exists in w. Rather, it is to see whether what the utterance actually says is true at w. The Direct Reference view of singular terms The Direct Reference thesis2 is the thesis that the only contribution of singular terms to the proposition utterances of sentences containing them semantically express is their referent. Thus, for instance, the only contribution of ‘David Kaplan’ to the proposition semantically expressed by an utterance of the sentence ‘David Kaplan is at UCLA’ is David Kaplan himself. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! 1 Cf. Saul Kripke (1972). 2 Cf. David Kaplan (1989). ! $! Propositions are that which utterances of sentences express and that which propositional attitudes are directed to. The three theses about singular terms that we have just considered, and those that are considered next, relate to the different views of propositions in important respects. So let us now consider two main views of propositions. The Russellian view of propositions On the Russellian view, propositions are complex, abstract and structured entities whose constituents are individuals, properties, relations, and other propositions. The constituents of a proposition expressed by utterances of sentences are usually considered to be the semantic contents of the expressions that make up those sentences. Propositions so conceived are usually called ‘Russellian propositions’.3 Thus, for instance, the Russellian proposition expressed by ‘Russell liked Montblanc’ is a structured entity that has Russell and Montblanc, with all its snowfields, as well as the relation of liking as constituents. And it may be characterized as an ordered pair as follows <like, <Russell, Montblanc>> The Possible Worlds view of propositions Utterances of sentences have truth-conditions and according to this view, propositions are sets of just those possible worlds at which what the utterances actually say is true. Meaning and truth-conditions of utterances of sentences are thus identified.4 !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! 3 Cf. Bertrand Russell (1910-1911). 4 Cf. Robert Stalnaker (1978). ! %! Thus, for instance, an utterance of the sentence ‘Robert Stalnaker is Robert Stalnaker’ expresses the proposition that is the set of all possible worlds at which what this utterance actually says is true. Since identity is a necessary relation, utterances of this sentence express a necessary truth and so, the proposition they express is the necessary proposition, that is, the set of all possible worlds. It is a consequence of this view that no utterances of sentences that express necessary truths differ in meaning. Thus, according to this view utterances of ‘Robert Stalnaker is Robert Stalnaker’, for instance, mean the same as utterances of ‘David Lewis is David Lewis’. Analogously, all utterances of sentences that express a necessary falsehood express the same proposition, that is, in such a case, the empty set, since they are true at no possible world; and so, they all mean the same as well. Note that utterances of ‘Robert Stalnaker is Robert Stalnaker’ express a different Russellian proposition than utterances of ‘David Lewis is David Lewis’ due to the fact that the former is constituted by the referent of ‘Robert Stalnaker’, whereas the latter is constituted by the referent of ‘David Lewis’. Singular versus general propositions In general, propositions expressed by utterances of sentences that contain a singular term, or that are partly constituted by individuals, are called ‘singular propositions’, in opposition to those called ‘general propositions’. Examples of general propositions are the propositions that all men are mortal, that there is more than one black raven, or the proposition that when it rains, it rains. ! &'! The Fregean view of singular terms and propositions According to the Fregean view, singular terms express a sense in addition to referring.5 Sense and reference are both part of their meaning. The notion of sense was introduced !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! 5 As I said above, my aim is not to offer here a historical presentation of these ideas – that is, a presentation that pays special attention to the letter of exactly what philosophers said when they introduced them. Rather, here I am offering a characterization of what has been taken to be the traditional Fregean view, in order to discuss the ideas that matter for the views that I defend in this thesis. However, let me point out that what I have just said above -that is, that according to the Fregean view, singular terms express a sense in addition to referring- is in fact compatible with two theses concerning what Frege really thought or said that seem to be in dispute. On the one hand, the thesis that for Frege it is senses that refer and linguistic expressions refer only by expressing senses that refer. On the other, the thesis that for Frege reference is a relation that holds between referential expressions and their referents, and not between senses and the referents that they present or determine. According to this latter hermeneutic thesis, for Frege it is not senses that refer, even if it were true that Frege thought that the relation between senses and references was in some sense prior to the relation between linguistic expressions and references. That the former relation is prior does not mean that it is a relation of reference. Moreover, notice that this latter thesis is also compatible with the idea that for Frege, the relation of reference between linguistic expressions and referents was mediated by senses. Again, this does not mean that it is senses that refer. To illustrate how controversial these issues may appear to be, it is useful to consider the following passage from Michael Dummett (1996): Frege held that it is the thought that is primarily said to be true or false, the sentence being called true or false only in a derivative sense; and, since for Frege the reference of the sentence is its truth-value, this means that it is the sense of the sentence that primarily has the reference, and the sentence only derivatively. He laid little emphasis on the generalization of this principle to all expressions, but he did acknowledge it as correct: hence, for example, it is the sense of a proper name that primarily refers to the object, rather than the proper name itself. In practice, however, Frege never conformed to this order of priority when expounding the distinction between sense and reference. He never first introduces the notion of sense, subsequently explaining reference as a feature of sense: he speaks first of the expression as having reference, and proceeds either to argue that it also has a sense or to say something about what its sense consists in. This order of exposition is actually demanded by his conception of the sense of an expression as the way in which its reference is given: for it follows from this conception that the notion of sense cannot be explained save by appeal to that of reference, and so we must first have the notion of reference. Now if we have the notion of reference before we have that of sense, we cannot construe reference as a property of the sense, but only of the expression. It follows that Frege’s thesis that it is the sense to which the reference is primarily to be ascribed is incorrect. (p. 8) I think that in this quote Dummett makes a bad inference from the thought is primarily said to be true or false and the reference of the sentence is its truth-value to the thought has reference. As I said above, it may perfectly well be the case that it is the thought that primarily is true or false ! &&! by Frege (1892) as the notion of something with certain properties, which satisfied certain roles: senses were thought of as inter-subjective modes of presentation of objects; modes of presentation that individuate those objects and which explain the cognitive value of linguistic expressions that express them. Objects may be presented in different ways, some of which individuate them, and senses are precisely of this kind. That is, senses are individuating modes of presentation such that each sense is a mode of presentation of only one object. That is, two senses may constitute ways of presenting one and the same object, whereas no two objects may be presented by one and the same sense. Thus, for instance, being the morning star that so and so and being the evening star that so and so (the ‘that so and so’ standing for a more precise characterization of the traditional individuating properties of Venus) provide two distinct modes of presentation of the planet Venus, and individuate it, since no other object different from the planet Venus exemplifies these two properties6. Frege made an analogy with a telescope directed at the moon: the moon is objective, like the reference of a singular term. Second, the image of the moon on the retina of the observer is subjective, like a mental representation of the reference that speakers may have in their minds. And finally, the image of the moon reflected in the telescope is intersubjective: equally accessible to different observers. Our use of singular terms reflects this feature of our cognitive relations to objects by expressing a sense in addition to referring and in such a way that the individuating sense expressed by a singular term is what determines its reference. Since objects are usually presented in different ways, distinct co-referring singular terms may express different !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! and that the reference of a sentence is its truth-value, while it is only linguistic expressions, and not thoughts, that refer. 6 In fact, neither does the planet Venus satisfy them, since it is not a star but a planet. But I will henceforth assume that it does. ! &(! senses. Thus, for instance, ‘Hesperus’ and ‘Phosphorus’ are proper names of the planet Venus; they co-refer, but express different senses. In contrast, since senses individuate objects, no two singular terms that share their sense may diverge in reference. Thus, for instance, ‘Hesperus’ and ‘the morning star’ could perhaps express the same sense and if so, could not refer to different objects, but must refer to the same one. Thus, co-referring proper names such as ‘Hesperus’ and ‘Phosphorus’ may present their referent, i.e. Venus, in different ways, i.e. as the morning star that so and so, and as the evening star that so and so, respectively. They do so in ways that individuate their referent and so determine it, since only one individual satisfies those properties, and that is Venus. Senses thereby explain the cognitive value of linguistic expressions that express them. At least part of what it is to understand the meaning of a singular term is to know its referent, that is, to know what object it refers to. But our access to objects is mediated by the ways they present themselves to us, and so, to know the referent of a singular term we need to know some mode of presentation that individuates it. Thus, different singular terms may present their referents in different ways; that is, they may express different senses, and so, we may express and understand different things by them. Likewise, utterances of sentences containing different singular terms may thus have different cognitive values and express different things; and if propositions are what utterances of sentences express, different propositions as well, even if the singular terms refer to the same object and are the only expressions that distinguish the sentences. Propositions that utterances of sentences express are, on the Fregean view, made up of the senses of their constituent expressions. Propositions are complex senses of sentences and Frege called them ‘thoughts’. Propositions so conceived are also called ‘Fregean propositions’. For instance, one does not understand the meaning of ‘Hesperus’ if one does not know the object it refers to, but one cannot know that object unless one knows one mode of presentation that individuates it. ‘Phosphorus’ refers to that same object, but may present it in a different individuating way than ‘Hesperus’, and in this way, utterances of ! &)! sentences that differ only in that they contain one of these names instead of the other may differ in their cognitive value and express different propositions. 2.2. Central Traditional Problems In this section, I present the central traditional problems that any theory of singular terms has to confront: the problem of the informativeness of identity statements, the problems created by meaningful empty names, and the problems that arise in both indirect and direct discourse as well as propositional attitude reports. In later chapters I show how well the view that I propose in this work fares with respect to these problems. Identity statements Pairs of identity statements such as the following illustrate the problem of the difference in informativeness and cognitive value of identity statements such as (1) and (2), which only differ in their containing different co-referential proper names. (1) Hesperus is Hesperus (2) Hesperus is Phosphorus While (1) and (2) seem to say the same about the same thing, i.e. that Venus is identical to itself, (2) is informative and (1) is not. If, as according to Millianism, the meaning of proper names is just their referent, it is not clear how the difference in informativeness is to be explained. If, in contrast, as Direct Reference has it, the only contribution of a proper name to the proposition semantically expressed by utterances of the sentence in which it occurs is its referent, then it is not clear either. Finally, if, as in a possible worlds view of propositions, the proposition semantically expressed by the utterance of a sentence is the set of possible worlds at which the sentence is true, since (2) is necessarily true, it is true at all possible worlds and hence, uninformative. ! &*! However, a view that sees meaning as a more complex property of words, such as the Fregean view, has a ready-made plausible explanation to account for this difference. On the Fregean view, (1) and (2) differ in cognitive value because the proper names ‘Hesperus’ and ‘Phosphorus’ express different senses and due to this (1) and (2) express different propositions. If the sense expressed by ‘Hesperus’ is the meaning of a long and complex definite description abbreviated by ‘the morning star’, whereas the sense expressed by ‘Phosphorus’ is the meaning of a long and complex definite description abbreviated by ‘the evening star’, then (1) expresses a long and complex non-informative proposition simplified as the proposition that the morning star is the morning star, whereas (2) expresses a long and complex informative proposition simplified as the proposition that the morning star is the evening star. Those competent speakers who do not know that ‘Hesperus’ and ‘Phosphorus’ refer to the same thing may acquire knowledge from (2), but they would not from (1). Empty names Empty names, i.e., names that do not seem to refer –such as ‘Vulcan’, ‘Phlogiston’, or ‘Anna Karenina’– are apparently meaningful notwithstanding their apparent lack of reference.7 Sentences with empty names seem to be used with meaning, to say something, even if what they say might perhaps lack truth-value. The following are examples: (1) Vulcan is a planet. (2) Anna Karenina suffers. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! 7 It is controversial whether fictional names such as ‘Anna Karenina’ are intuitively empty and it is even more controversial whether they are so after theoretical reflection. However, it is useful for our present purposes to assume that they are, and so I will for the time being. ! &+! (1) is an example of a non-fictional sentence while (2) is a fictional one. Sentences with empty names such as these initially pose a problem for Millianism and Direct Reference views. According to Millianism, the only meaning of a proper name is its referent and according to Direct Reference views its referent is its only contribution to the proposition expressed by utterances of sentences containing it. Fregeanism, however, can easily account for them by appealing to the senses these empty names express. Senses are modes of presentation of objects and there are no objects that these names refer to, but this does not prevent them from expressing a sense.8 Empty names are exceptions: it is just that there happens to exist nothing that exemplifies those otherwise individuating properties that are the senses of these empty names.9 We are not infallible and sometimes we are wrong in thinking that there is some object that satisfies certain otherwise individuating properties, just as in the case of Vulcan, or even if we know that there is no such an object, we sometimes enjoy pretending that there is, as in the case of Anna Karenina and fiction in general. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! 8 It is worth clarifying a point here. One might reasonably wonder in what sense these are individuating modes of presentation, given that modes of presentation are modes of presentation of things, and here there is nothing to be presented. I think there are two ways in which this worry might be met. One option is to consider that these are not real modes of presentation, since they do not present anything, but nevertheless call them ‘modes of presentation’ because they are of the same kind as modes of presentation that do present things. Another option is to consider that these are modes of presentation, because these are of the same kind as modes of presentation that do present something and, further, it is clear what they would have presented if things had been different (including ways things could not have been). Perhaps one might say that these are defective modes of presentation. Both options seem feasible to me. It also seems that the difference between the two is merely terminological. A similar worry arises with respect to names that do not name: in what sense are they names? My answer to this worry is analogous to the one just discussed. 9 Of course, these otherwise individuating properties cannot be “properties in re” –i.e. properties that exist in the things that exemplify them. I would like to remain as neutral as possible with respect to the nature of properties and not commit myself to a view in which uninstantiated properties may exist. Then, if properties turn out to be in re, things such as the mode of presentation expressed by ‘Vulcan’ will not be real properties after all. Perhaps they will be something constructed out of simpler properties in re. ! &"! By the appeal to senses, Fregeanism can thus explain the meaningfulness of utterances of sentences containing empty names: utterances of sentences such as (1) and (2) express a thought. They lack truth-values, however, because they contain expressions without reference. However, there are other utterances of sentences containing empty names that seem to have a truth-value and these prove to be initially difficult for Fregeanism too. These are utterances of sentences such as, for instance, the following: (3) Phlogiston does not exist. (4) Sherlock Holmes is a detective rather than a cook. (5) Superman is a fictional character. (6) Tolstoy created Anna Karenina. I discuss these other cases in Chapters 5 and 6. Indirect and direct discourse and propositional attitude reports Indirect discourse and propositional attitude attributions also present special difficulties for semantic theories. Consider the following two pairs of sentences: (1) Lois Lane says that Superman is brave. (2) Lois Lane says that Clark Kent is brave. and (3) Lois Lane believes that Superman can fly. (4) Lois Lane believes that Clark Kent can fly. ! &#! Suppose that the Superman story is real, rather than fictional. According to common intuitions, utterances of (1) and (2) may differ in truth-value, and utterances of (3) and (4) may do as well. That is the case, even though ‘Superman’ and ‘Clark Kent’ are proper names of one and the same individual; and these sentences only differ in that where one of them contains one of the names, the other sentence contains the other. Millianism and Direct Reference entail that utterances of (1) and (3) semantically express the same propositions as utterances of (2) and (4), respectively. For, as I already said, the only thing these sentences differ in is that they contain distinct co-referential proper names in the same position. According to these views, the referents of proper names are their only meaning or their only contribution to the proposition that sentences containing them semantically express. Therefore, utterances of (1) and (3), on the one hand, and utterances of (2) and (4), on the other, must express the same proposition and hence have the same truth-value. Fregeanism can naturally explain these data by appealing to senses again. Frege’s view was that expressions changed their reference when they appeared in these contexts of indirect discourse and propositional attitude attributions –i.e. when they occur in their that-clauses. In such contexts, expressions have an indirect reference and refer to their usual sense. Thus, if the proper names ‘Superman’ and ‘Clark Kent’ express different senses even if their usual reference is the same, their indirect reference is different and so, in utterances of (1)–(4) they refer to their distinct senses. This difference in reference accounts for the difference in truth-values of these pairs of utterances as well as the difference in the propositions they express. For due to the nature of senses, expressions that have a different reference must also differ in sense. So, if they have an indirect reference, they must also have an indirect sense. Direct discourse is usually treated as an exceptional special case, in which expressions clearly behave differently, but the Fregean thesis of change of reference may analogously help explain these other cases as well. Consider the following pairs of sentences involving direct discourse: ! &$! (5) Lois Lane said: “Superman is brave”. (6) Lois Lane said: “Clark Kent is brave”. and (7) ‘Frege’ is Frege’s family name. (8) ‘Gottlob’ is Frege’s family name. Intuitively, utterances of sentences (5) and (6), on the one hand, as well as utterances of sentences (7) and (8), on the other, express different propositions and may differ in truthvalue, respectively. However, again, they only differ in that they contain distinct coreferential proper names in the same position. In the case of (7) and (8), both ‘Gottlob’ and ‘Frege’ are names of the philosopher Gottlob Frege, and so, corefer. Frege applied his thesis of change of reference to these cases too. This time, though, the indirect reference of expressions appearing in contexts of direct discourse are the expressions themselves. That is, expressions in these contexts refer to themselves. Thus, for instance, the name ‘Frege’ in (7) refers to itself, i.e. ‘Frege’, and the name ‘Gottlob’ in (8) refers to itself, i.e. ‘Gottlob’, thereby explaining the difference in meaning and truth-value of the pair of utterances of sentences (7) and (8). Likewise, utterances of the sentences ‘Superman is brave’ and ‘Clark Kent is brave’ in (5) and (6), respectively, would refer to themselves and this would explain the difference in meaning and truthvalue of the pair of utterances of sentences (5) and (6) as well. Whereas Fregeanism provides a unified account of these phenomena, Millianism and Direct Reference can only treat quotation as a special, exceptional case in which proper names are not used to refer but are just mentioned. I consider the issue of quotation in more detail in Chapter 4. ! &%! In short, by appealing to senses Fregeanism provides natural explanations of distinct phenomena which are otherwise problematic, which makes it at the very least an interesting theory to consider. Throughout this work I discuss and defend more sophisticated Fregean views. Before that, however, it will be useful to introduce Russell’s theory of definite descriptions, a version of which is also part of the views that I defend. This is what I do in the next section. 2.3. Russell’s Theory of Definite Descriptions Before Russell, definite descriptions were thought to be singular terms. Russell (1905) proposed a theory of definite descriptions according to which definite descriptions are not singular terms, but general terms that can only be defined in context and behave as quantifiers. According to Russell’s theory of definite descriptions, statements of the form: (1) The ! is ". are analyzed as having the following logical form: (2) There is exactly one x such that !x; and (furthermore) "x. And since ‘exists’ is either not a (first-level) predicate or a predicate that applies to everything, existential statements of the form: (3) The ! exists. are analyzed, in accordance with Russell’s theory, as having the following logical form: (4) There is exactly one x such that !x. ! ('! The following logical form (5) There is exactly one x such that !x; and (furthermore) x exists. would be redundant. Thus, according to Russell’s view, utterances of sentences containing definite descriptions and no singular term express a general proposition rather than a singular one. Russell’s theory of definite descriptions may also easily explain the cases of sentences that initially pose a problem for Millianism and Direct Reference when the sentences contain definite descriptions instead of singular terms. In the remainder of this section I show how it does this. Identity statements with definite descriptions Pairs of identity statements that only differ by containing in the same position different definite descriptions that denote the same individual, such as the following: (6) The morning star is the morning star. (7) The evening star is the morning star. differ in informativeness and cognitive value precisely because they express different general propositions, which on the Russellian view of descriptions are, respectively, the following: (8) There is exactly one x such that x is a morning star and there is exactly one y such that y is a morning star and x=y. ! (&! (9) There is exactly one x such that x is an evening star and there is exactly one y such that y is a morning star and x=y. Non-denoting definite descriptions Definite descriptions that do not denote anything have an obvious meaning in context and sentences containing them express general propositions as well. For instance, (10) The king of France is bald. expresses the following general proposition: (11) There is exactly one x such that x is a king of France; and (furthermore) x is bald. (10) is meaningful even if the description ‘the king of France’ does not denote anyone, and so understood, it is false, because France is not a monarchy. Scope ambiguities Russell’s theory of definite descriptions can also explain sentential ambiguities as scope ambiguities in a nice way. Thus, for instance, consider the negation of (10): (12) The king of France is not bald. Which, according to Russell’s theory, is ambiguous and has the two following readings: (13) There is exactly one x such that x is a king of France; and (furthermore) x is not bald. ! ((! (14) It is not the case that there is exactly one x such that x is a king of France; and (furthermore) x is bald. Again, given that France is not a monarchy, these two differ in truth-value: (13) being false, while (14) is true. The reading expressed by (14) would be the one that explains the true proposition that one usually intends to assert by uttering (12) (whether or not one succeeds in the intent, and whether or not one does so by pragmatic means). Scope ambiguities arise not only with negation but also from the interaction of definite descriptions and other expressions such as indirect discourse, propositional attitude verbs, and modal adverbs. These scope ambiguities account for differences in cognitive value and truth-value of pairs of utterances of sentences of indirect discourse, propositional attitude ascriptions, and modal statements which only differ in that where one contains a singular term, the other contains a definite description that denotes the same individual as the one the former refers to, such as the following: (15) She says that Juan Carlos is sassy. (16) She says that the king of Spain is sassy. (17) He knows that Scott is Scott. (18) He knows that the author of Waverley is Scott. (19) Necessarily, nine is nine. (20) Necessarily, the number of planets is nine. Whereas (15) would have only one reading, (16) would have the following two: (16a) She says that there is a unique king of Spain and he is sassy. (16b) There is a unique king of Spain such that she says that he is sassy. It is easy to see that there are situations in which (15) and (16b) would be true, while (16a) false. ! ()! Analogously, whereas (17) has a unique reading, (18) is ambiguous and has the following two readings: (18a) He knows that there is a unique author of Waverley and that he is Scott. (18b) There is a unique author of Waverley such that he knows that he is Scott. It is again easy to see that there are situations in which (17) and (18b) would be true while (18a) false. Finally, and also analogously, whereas (19) has a single reading, (20) is ambiguous and has the following two: (20a) Necessarily, there is a unique number of planets and it is nine. (20b) There is a unique number of planets such that necessarily it is nine. It is as easy as in the above cases to see that there are situations in which (19) and (20b) would be true while (20a) false. Negative existentials with definite descriptions True negative existentials containing non-denoting definite descriptions such as: (21) The king of France does not exist. would not be ambiguous because ‘exists’ is not a discriminating predicate and hence the final Russellian analysis of existential statements does not entail any scope ambiguity but predicts it to have a single reading, which is as follows: (22) It is not the case that there is a unique king of France. ! (*! and which is true. Note that Russell’s theory of definite descriptions helped to explain all of these problems away when they were presented by sentences containing definite descriptions rather than proper names, indexicals, or demonstratives. In fact, he thought that proper names were disguised definite descriptions and treated them as such, thereby also solving these problems when they are presented by sentences containing proper names instead. Peter Strawson (1950) disputed Russell’s theory of descriptions, defending an alternative view and thereby initiating a lively debate in the literature. Strawson’s view is relevant for several of the problems I am discussing, so let us briefly consider it before proceeding further. Strawson’s presuppositional account of definite descriptions Strawson contended, against Russell’s theory of descriptions, that the existential condition involved in uses of definite descriptions is presupposed rather than asserted, while understanding presuppositions as conditions whose satisfaction is necessary for the truth-evaluability of utterances carrying them. So that utterances of a sentence such as: (10) The king of France is bald. presupposed that there is exactly one king of France and since there is no such individual, those utterances are neither true nor false. Utterances of sentences containing definite descriptions that do denote, such as the identity statements: (6) The morning star is the morning star. (7) The evening star is the morning star. ! (+! would, however, also carry existential presuppositions that nonetheless would be satisfied and so those utterances would be truth-evaluable. Although I adopt a Russellian theory of descriptions, it is useful to bear Strawson’s criticism in mind. In fact, Strawson’s conception of presuppositions is adopted by a view on singular terms that I discuss in the next chapter, which is close to the one I defend in certain respects. Strawsonian presuppositions are not the only presuppositions this view adopts, however. In spite of its initial plausibility and independently of Strawson’s criticisms, it might seem that Russell’s theory does not give a correct account of certain uses of definite descriptions. Keith Donnellan (1966) drew attention to what he called ‘referential uses of definite descriptions’ and argued that these were distinct from the uses that Russell’s theory captured, which he called ‘attributive’. I discuss this in what follows. Referential versus attributive uses of definite descriptions Donnellan accepted Russell’s theory for some uses of definite descriptions, but not for all. He argued that definite descriptions were used in two distinct ways: attributively and referentially. When used attributively, definite descriptions behave in accordance with the analysis provided by the Russellian view, but in their referential use, they behaved as singular terms. Donnellan argues for the existence of referential uses of definite descriptions with examples such as the following: (23) Smith’s murderer is insane. –where ‘Smith’s murderer’ is understood as the definite description ‘the murderer of Smith’– ! ("! (24) Who is the man drinking a martini? (25) Bring me the book on the table. uttered having a particular individual in mind and intending to refer to it by using the definite description in question. According to Donnellan, there are contexts in which one succeeds in saying something true about an individual by uttering sentences such as (23)– (25), even if the individual in question does not satisfy the definite description used to pick it out. Thus, one may manage to say something true about a man that did not kill Smith by uttering (23), ask about the identity of a man that is drinking water rather than martini by uttering (24), and receive a book that was not on the table but beside it by uttering (25). However, in defense of Russell’s theory of descriptions, Kripke (1977) argued that what Donnellan categorized as referential uses of definite descriptions were only what we would call ‘pragmatic uses’ rather than semantic ones. As such, they are easily explainable in terms of Gricean conversational implicatures10 and Kripke’s distinction between speaker’s reference and semantic reference. By ‘pragmatic uses’ I mean uses that are different to the uses that strictly follow linguistic conventions. I will call ‘semantic uses’ those that are conventional. For instance, I may use the expression ‘the ham sandwich’ to speak about a person who ate a ham sandwich, rather than the ham sandwich itself. My use would not merely follow the linguistic convention that governs the use of that expression. Thus, this use would count as pragmatic. In Kripke’s terms, the person who ate a ham sandwich, and that I managed to speak about with my use, is the speaker’s reference. In contrast, the conventional use of ‘the ham sandwich’ is the use made in order to speak about a ham sandwich instead. This use is what I call ‘its semantic use’, and the ham sandwich is, in Kripke’s terms, its semantic reference. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! 10 For the notion of Gricean conversational implicatures see Grice (1967) and my explanations below. ! (#! According to Kripke, utterances of sentences such as (23)–(25) of the kind contemplated by Donnellan in which the definite descriptions contained are used referentially, semantically express what Russell’s theory judged them to express, and only pragmatically express what the speakers mainly intend to communicate. Such utterances semantically express something about the individual that satisfies the definite description in question, namely its denotation, and this is what Kripke calls its ‘semantic reference’, whereas pragmatically they express something about the individual that the speaker has in mind and mainly intends to say something about, and this is what Kripke calls ‘the speaker’s reference’. Kripke argues that analogous cases exist but involving sentences containing proper names instead of definite descriptions, and that it is clear that they do not show proper names to be ambiguous either. Kripke illustrates his point by considering the following dialogue between two people who see someone called ‘Smith’ at a distance but who they mistake for some other individual called ‘Jones’ A: What is Jones doing? B: Raking the leaves. Both speakers have Smith in mind and pragmatically convey something about him, which is what they intend to do. Hence, Smith is the speakers’ reference. However, the name that the speakers use is ‘Jones’, and since that’s not the name of Smith but of Jones, they semantically say something about Jones, who is the semantic reference. The views that are defended in this thesis are in agreement with Kripke on this particular point and thus do not consider definite descriptions among the singular terms. ! ($! 2.4. Objections to Descriptivism In this section I present some objections that many consider to have refuted the view known as ‘Descriptivism’. Descriptivism is the view that competent speakers of proper names, in virtue of their semantic competence, associate with them a definite description, the meaning of which determines their referent. Fregeanism is broadly regarded as a descriptivist view. The Fregean senses that singular terms express are supposed to be the meaning expressed by attributive uses of definite descriptions. Russell’s view that proper names are disguised definite descriptions is taken to be another descriptivist view. Since I defend a descriptivist view like the Fregean in certain respects in later chapters, I deem it necessary to discuss these objections in detail beforehand. Kripke (1972) objected to descriptivist views of proper names by offering three major arguments: the semantic argument, the epistemic argument, and the modal argument. I will discuss them in turn. Kripke’s semantic argument The semantic argument is also known as the argument from ignorance and error. The argument proceeds as follows: first, Kripke argues that speakers cannot give any description the meaning of which individuates the referent of most names that they are competent with. Further, Kripke argues that in many cases in which the speakers can indeed provide detailed descriptions, the meanings of which could individuate an individual, they are mistaken and these descriptive meanings do not individuate the referent of the name. Finally, Kripke argues that for any proper name and any definite description that competent speakers might associate it with, a situation in which the bearer of the name does not satisfy the associated definite description is conceivable. This would show that the meaning of proper names is different from the meaning of any definite description, and that the latter does not even determine the former. So Kripke concludes that Descriptivism is false. ! (%! This means, for the Fregean view, which holds that the meaning of a definite description determines the reference of proper names and that their reference is part of their meaning, that the whole meaning of proper names is not determined by the meaning of any definite description. In particular, the name’s reference is not determined by the meaning of any definite description. In the case of Russell’s descriptivist view, Kripke’s semantic argument would show that proper names cannot be disguised definite descriptions which is precisely what they are according to Russell’s descriptivist view; descriptions are not referential expressions, and thus, proper names do not even refer. Kripke illustrates his argument with the following case. Consider Gödel, namely the person who discovered the incompleteness of arithmetic. Many would associate that individuating description with his name, i.e. ‘Gödel’. However, Kripke argues, a situation in which we learned that it was not Gödel but someone else who discovered the incompleteness of arithmetic is conceivable and it would not be if the name ‘Gödel’ meant the same as the definite description ‘the person who discovered the incompleteness of arithmetic’ or if the meaning of the latter determined the referent of the former. Furthermore, if we were actually to learn such a thing, we would not call that other person who really discovered the incompleteness of arithmetic ‘Gödel’ just because of this. I share Kripke’s intuitions; but, like many other people, I do not accept the general conclusion that he draws from them. As I argue below, Kripke’s semantic argument can be resisted. Kripke’s epistemic argument Kripke also argued that if Descriptivism were right, utterances of sentences containing proper names that attributed the individuating property expressed by the definite ! )'! descriptions associated to proper names to their referents would be analytic and therefore a priori, when in fact they are not. For instance, consider utterances of sentences such as: (28) If Hesperus exists, then Hesperus is a bright celestial body seen in the evening. (29) If Phosphorus exists, then Phosphorus is a bright celestial body seen in the morning. (28) and (29) express truths that are a posteriori rather than a priori. This is so even if the descriptions that competent speakers associate with the names ‘Hesperus’ and ‘Phosphorus’ are ‘the bright celestial body seen in the evening’ and ‘the bright celestial body seen in the morning’ respectively. (28) and (29) cannot be justified independently of experience: one needs experience to believe them with justification. I agree with Kripke with regards to these cases; but again, like many other people, I think that the general conclusion can be resisted. More sophisticated descriptivist views, developed after Kripke presented these arguments, provide more interesting descriptions to play the role that Descriptivism assigns to descriptions. I discuss them in later sections. Kripke’s modal argument Kripke argues that utterances of sentences containing proper names do not have the same modal properties as utterances of sentences containing definite descriptions in the same syntactic positions instead. This is so even in cases in which the proper name and definite description in question are such that the proper name is introduced by means of the description, which fixes its referent, such as the proper name ‘Julius’ introduced to name whoever was the inventor of the zip. To see what this modal difference is, consider the following pair: (30) If Julius exists, Julius invented the zip. (31) If the inventor of the zip exists, the inventor of the zip invented the zip. ! )&! The intuition here is that whereas utterances of (31) express a necessary truth, utterances of (30) express a contingent truth. For even if by stipulation Julius cannot but be the actual inventor of the zip if such an inventor exists, he could have not invented the zip. Therefore, there are possible worlds in which he exists but did not invent the zip. Utterances of (30), but not (31), would be false at such worlds. This modal difference between utterances of sentences such as (30) and (31) is due to an important modal difference between proper names and definite descriptions, which Kripke captures in his thesis that proper names are rigid designators. That is, proper names refer to the same individuals at all possible worlds in which those individuals exist. This is precisely the thesis that characterizes the Rigidity view presented earlier in this chapter. The idea is that when we consider different possible worlds in evaluating an utterance of a sentence containing a proper name, we do so with respect to one and the same individual, namely the name’s actual referent. This is not so in the case of utterances of sentences with definite descriptions. From the fact that proper names are rigid and definite descriptions are not, Kripke concludes that Descriptivism must be false. Notice also that cases of names such as ‘Julius’ exemplify Kripke’s distinction between fixing the meaning and giving it: even if the meaning of a definite description fixes the referent of some few special names such as ‘Julius’, no definite description gives their meaning, according to Kripke and other anti-descriptivists. The initial intuition on which Kripke bases his modal argument, namely that utterances such as (30) and (31) have a different modal profile, is widely shared. It is also widely agreed that Kripke’s rigidity thesis is correct and explains this initial intuition. However, it is controversial that the argument succeeds in refuting Descriptivism. My own view is that even if the argument works against traditional versions of Descriptivism, more sophisticated versions, some of which I discuss in what follows, may be thought of to resist it. In fact, the view that I defend in later chapters adopts Kripke’s rigidity thesis, and contends that no definite description gives the meaning of a proper name (but gives ! )(! only part of it), even if it is a descriptivist view. I argue that the rigidity thesis and Descriptivism are thus compatible. 2.5. Descriptivism after Kripke’s Objections Kripke’s objections, together with other arguments presented against Descriptivism, have been thought by many to be conclusive. However, some have developed improved versions of Descriptivism and offered replies to the critics’ arguments. In this section I discuss two kinds of such versions, which I will call ‘Metalinguistic Descriptivism’ and ‘Rigidified Descriptivism’, respectively. The reason why I discuss these two versions of Descriptivism is that my own view on singular terms, to be presented in Chapter 3, is a version of Descriptivism as well. In particular, it is a metalinguistic descriptivist view. I think that it is worth considering other versions of Descriptivism in order to better situate and evaluate the one that I favor. Also, these views make important contributions to understanding the semantics of singular terms. Metalinguistic Descriptivism Metalinguistic descriptivist views have been defended by Brian Loar (1976) and Kent Bach (1981), among others. I focus here on the versions of Loar and Bach, which I find especially interesting. First, I provide a detailed presentation of Loar’s complex view in order to be able to assess it correctly, which I do immediately afterwards. After discussing Loar’s view, I turn to Bach’s, which, as we will see, is close to Loar’s. In my opinion, both are powerful versions of Descriptivism, even if they are not ultimately correct. On Brian Loar’s view, the conventional function of singular terms (including definite descriptions) is to introduce an individual concept into what is meant or expressed by their particular uses. According to this view, singular terms literally express a normally ! ))! non-individuating concept that is a specific referential qualifier. That is, singular terms express a concept that qualifies their referent and this is what Loar calls ‘a singular term’s referential qualifier’. In addition, in each particular use, singular terms express an individual concept that is intrinsic to what the speakers mean by utterances of sentences containing them, and which sometimes is not a logical restriction of the singular term’s referential qualifier. In those cases in which the individual concept expressed is not a restriction of the singular term’s referential qualifier, the latter is not essential to what the speakers mean. Let us consider an example. The description ‘the descriptivist theory’ literally expresses the referential qualifier being a descriptivist theory and may express different individual concepts through particular uses. One such individual concept could be being the descriptivist theory that I have been considering this morning, which is a logical restriction of its referential qualifier. However, another individual concept that it may express could be being the theory that I have been considering this morning, which is not a logical restriction of its referential qualifier. Examples of the referential qualifiers that other kind of singular terms express are the following: the concept being a theory, which is expressed by the complex demonstrative ‘this theory’; the concept being a male person, which is expressed by the indexical ‘he’, and finally, the concept being called ‘Brian Loar’, which is expressed by the proper name ‘Brian Loar’. According to this view, all singular terms, including definite descriptions and proper names, have attributive as well as referential uses. In their referential uses, on the one hand, the individual concepts that they express are identifying and not logical restrictions of their referential qualifiers, which thereby are not essential to what the speaker means. In attributive uses, on the other hand, those individual concepts are non-identifying, or generalizing, but logical restrictions of their referential qualifiers. ! )*! Thus, for instance, one may express the individual concept being the theory that I have been considering this morning by using the definite description ‘the descriptivist theory’ referentially and succeed in expressing something about the theory that satisfies the individual concept even if it is not a descriptivist theory and so, does not satisfy the definite description’s referential qualifier. Which individual concept is essential to what the speaker means is somehow restricted by the singular term’s referential qualifier, even in cases in which the former is not a logical restriction of the latter. What makes certain concepts the referential qualifiers of certain singular terms is the following convention: the speaker expects that it is mutually believed by the participants in the conversation that the individual concept meant is satisfied by an individual who or which also satisfies the referential qualifier and, further, that the hearer will believe that this is mutually believed as well. For the particular case of proper names, the view is in general that for any proper name N, N has as its referential qualifier the concept being called N and its conventional meaning is the same as the conventional meaning of the definite description ‘the thing or person called N’. Often, on using a proper name N, the individual concept that is essential to what is meant is identifying as well as a logical restriction of the one expressed by an attributive use of such a description. Moreover, according to this view, names are not ambiguous, since there is only one conventional meaning for every proper name, and only through their particular uses do they contribute particular distinct individual concepts. Sometimes though, proper names are used so that their referential qualifier is not essential to what is meant. Loar offers the following examples. (1) Jenkins is off balance. uttered by someone leaning over a balcony, so that the fact that the person is called ‘Jenkins’ is not essential to what the speaker means. The speaker might even be mistaken ! )+! and that person may not even be called ‘Jenkins’. In these cases, the individual concept expressed which is essential to what the speaker means is probably an identifying perceptual one such as the person that one can see in that direction or the person that I am looking at. These are cases in which the proper name is used referentially. The other example is: (2) Scotus here reasons badly. uttered by someone who is examining a text attributed to Duns Scotus, meaning the author whoever it was reasoning badly. Neither in this other case is the proper name’s referential qualifier, which in this case is being called ‘Scotus’, essential to what the speaker means. Unlike the previous example however, this time, the individual concept expressed, namely the author whoever it was, is not identifying either. Thus, it is neither a case of an attributive use, nor a case of a referential use. Now, if the case was rather one in which the person who utters (2) and the audience are simply interested in authors’ names, having no clue of who this particular author was, the use of ‘Scotus’ may express the non-identifying individual concept the author called ‘Scotus’, which is a logical restriction of the name’s referential qualifier and the name would thus be used attributively. According to non-descriptivist views, there is an important distinction between two semantic relations that hold between terms and individuals: reference and denotation. On these views, only proper names, indexicals, demonstratives and, perhaps, referential uses of definite descriptions –if there are such uses– refer; whereas only attributive uses of definite descriptions denote –that is, they are about the individuals that uniquely satisfy them. According to a descriptivist view like Loar’s, this distinction does not really apply, since all singular terms denote individuals who uniquely satisfy the individual concepts that the terms contribute to what the speakers mean. However, Loar proposes a characterization of reference as the relation that holds between singular terms and ! )"! individuals, when the individuals instantiate some identifying concept which both is intrinsic to what the speaker means and has the widest scope. Understood this way, nonetheless, reference is not opposed to denotation. Thus, according to Loar’s view, uses of singular terms that refer semantically express an identifying individual concept, which determines their reference, and which is what the singular term contributes to what the speaker means. (I take it that according to this view, what the speaker means is the proposition literally and semantically expressed by the utterance of the sentence containing the singular term. Loar also talks about sentences having a conventional and literal meaning that is different from the propositions utterances of them end up semantically expressing on particular occasions of use). Which individual concepts are identifying is, according to Loar, relative to our classificatory interests. Any individual concept may be identifying in the following sense: for any F, one can specify a set of interests and circumstances relative to which learning that x is F is learning who or what x is. Loar, however, considers the following kinds of individual concepts as usual cases of identifying individual concepts: perceptual individual concepts whose form is like the such and such which we see (saw, perceived) to be such and such; comprehensive individual concepts of a form similar to the person or object who or which approximately fits such and such comprehensive dossier; and finally, reference-directed individual concepts whose form is similar to the such and such, which is called N, or the such and such referred to by x. As I said above, several arguments were launched against Descriptivism. Let us consider now how Loar’s view fares against them. Loar defends his view against Kripke’s objections by highlighting the following facts regarding it: different uses of singular terms may express different individual concepts; individual concepts are often complex and partly implicitly known; there is always at least one individual concept specially connected with the speaker’s communicative intentions; proper names normally take a wide scope over modal operators; and the situations considered in the modal argument are conceivable not because proper names are not synonymous with definite descriptions, ! )#! but because the identifying individual concepts that these definite descriptions express are concepts of contingent properties of the individuals thus identified. Loar considers, and replies to, the arguments against Descriptivism that I am considering in this chapter. In what follows I explain how I would reply in defense of Loar’s view. I present my own version of how Kripke’s semantic argument applies to Loar’s particular view and how I would reply to Kripke’s argument in defense of Loar’s view. Consider Kripke’s semantic argument again, presented as an objection to Loar’s particular descriptivist view this time. According to Loar’s view, each use of a proper name expresses an individual concept. Thus, it might be, for instance, that when using the name ‘Gödel’, the speaker means the individual concept expressed by ‘the person who discovered the incompleteness of arithmetic’, but that in another different use, the speaker means another different individual concept. Thus, if the speaker were to discover that Gödel did not in fact discover the incompleteness of arithmetic, the speaker would have been wrong or have been talking about someone else, when using the name meaning what the former definite description means. However, after discovering that Gödel was not in fact the discoverer of the incompleteness of arithmetic, the speaker would still be able to use the name ‘Gödel’ meaning a different individual concept instead that Gödel did satisfy (for instance, the metalinguistic one being the person called ‘Gödel’). Let us consider this in more detail. As applied to Loar’s view, Kripke’s semantic argument should be that for each use of a proper name N, and each individual concept that on Loar’s view this use of N contributes to what the speaker means, there is a conceivable situation in which the referent of N does not satisfy the individual concept allegedly meant. Thus, consider again the particular example of Gödel and a particular use of the proper name ‘Gödel’ that contributed the identifying individual concept expressed by an attributive use of the definite description ‘the person who discovered the incompleteness of arithmetic’ to what the speaker means. The objection is then that if this were the way in which we use the proper name ‘Gödel’, we would not be able to conceive of a situation in which we were mistaken and Gödel did not in fact discover the ! )$! incompleteness of arithmetic. But we can and do conceive of such a situation. At most, the argument continues, we would be able to conceive of a situation in which we would discover that the person we had in mind when using the name was not in fact Gödel. With this version of Kripke’s semantic argument, however, it can be argued that in describing that conceivable situation, the name ‘Gödel’ is used in a relevantly different way to the particular way it was assumed that it was used at the beginning of the argument, and in that sense, illegitimately. From the point of view of Loar’s theory, it may be argued that the proponent of the semantic argument is using the name ‘Gödel’ as meaning the same as ‘the individual called ‘Gödel’’ at least in the end, when stating the conclusion of the argument. In other words, according to a view like Loar’s, using the name in the same way in which it is illegitimately used in the objection when uttering: (3) Gödel did not discover the incompleteness of arithmetic. we might be expressing something such as: (4) The individual called ‘Gödel’, who we were taught discovered the incompleteness of arithmetic, did not in fact discover the incompleteness of arithmetic. which is false, but describes a conceivable situation. However, using the name contributing the individual concept expressed by an attributive use of ‘the person who discovered the incompleteness of arithmetic’ to what we mean –as assumed at the beginning of the argument– upon uttering (3), we express a contradiction, namely that: (5) The person who discovered the incompleteness of arithmetic did not discover the incompleteness of arithmetic. and the situation thus described is not conceivable. Therefore, this does not show this particular descriptivist view to be wrong. ! )%! In short, Kripke’s semantic argument is very convincing as an argument for the conclusion that for any semantic use u of any proper name N that individuates one and the same individual, be it by referring or by denoting, there is no single attributive use of a definite description that is synonymous with u in every context. For every context of use in which u occurs and every descriptive meaning, there seems to be another use u’ of N with a different meaning that makes it conceivable that the denotation or reference of the name in such a use does not satisfy that descriptive meaning. Loar’s view does not claim that things are otherwise and hence, Kripke’s argument does not seem to work as an argument against it. Note that an analogous argument could be applied to cases in which the individual contributed concepts were logical restrictions on the name’s referential qualifier (and so include being the individual called ‘Gödel’) and in which we discovered that Gödel was not in fact called ‘Gödel’. The only such situation that would be conceivable is the one in which we discovered that in fact Gödel had not been originally baptized with the name ‘Gödel’ and so we would stop calling that person ‘Gödel’, or would continue to call him that name because this naming practice started at some point anyhow. In contrast, a situation in which we discovered that Gödel was not in fact called ‘Gödel’ by anyone is obviously not conceivable at all. For after all, some of us do call Gödel ‘Gödel’. In any case, the argument fails when applied to these cases as well. Let us now move to Kripke’s epistemic argument. This argument does not fare any better against Loar’s view. Consider an utterance of: (6) If Gödel exists, then Gödel discovered the incompleteness of arithmetic. This utterance would be as a priori as an utterance of: (7) If the person who discovered the incompleteness of arithmetic exists, the person who discovered the incompleteness of arithmetic discovered the incompleteness of arithmetic. ! *'! in which the descriptions were attributively used, if both uses of the name ‘Gödel’ in (6) contributed the individual concept expressed by an attributive use of the description ‘the person who discovered the incompleteness of arithmetic’. In contrast, an utterance of (6) in which both uses of the name ‘Gödel’ contributed the individual concept expressed by an attributive use of the description ‘the individual called ‘Gödel’ who we were taught discovered the incompleteness of arithmetic’ would be a posteriori. But this does not indicate any problem for the view. Again, the view is not committed to there being a single definite description associated with each use of a proper name and so it can interpret those uses as meaning the individual concepts expressed by different definite descriptions, which avoids the problem that the argument presents. So far we have seen that Loar’s view can stand up to Kripke’s semantic and epistemic arguments. Let us now see how it deals with the modal argument. Loar defends his view against Kripke’s modal argument by arguing that proper names normally take a wide scope over modal operators. So that utterances of: (8) Julius might not have invented the zip. (9) It is possible that Julius had not invented the zip. (10) Possibly Julius did not invent the zip. are usually understood as utterances of sentences in which proper names take a wide scope, that is, as utterances of, for instance: (11) There is a unique inventor of the zip and he might not have invented the zip. (12) There is a unique inventor of the zip and it is possible that he had not invented the zip. (13) There is a unique inventor of the zip such that possibly he did not invent the zip. for the case of the use of ‘Julius’ as the very much discussed descriptive name; or, for instance, as utterances of: ! *&! (14) There is a unique individual who such and such and who is called ‘Julius’ and he might not have invented the zip. (15) There is a unique individual who such and such and who is called ‘Julius’ and it is possible that he had not invented the zip. (16) There is a unique individual who such and such and who is called ‘Julius’ such that possibly he did not invent the zip. where ‘such and such’ is further specified. As for the necessity or contingency of the truths expressed by utterances of sentences such as: (17) If Julius exists, Julius invented the zip. (18) If the inventor of the zip exists, the inventor of the zipper invented the zipper. again, on this particular descriptivist view, it depends on which particular individual concepts the uses of the name ‘Julius’ and the definite description ‘the inventor of the zipper’ contribute. To summarize, a view such as Loar’s can successfully counter Kripke’s three major arguments. In spite of this, one might consider, as I do, that the fact that proper names are very infrequently (if at all) interpreted as having a narrow scope provides enough reason to think that proper names do not work as Loar advocates. I return to this once I have presented my own view in Chapter 3. Now, in addressing Metalinguistic Descriptivism, Kripke presented an additional objection which I will call ‘the circularity objection’. Kripke proposed the following as a constraint on any descriptivist theory of names, and argued against metalinguistic versions of the descriptivist view that they did not satisfy it. ! *(! (C) For any successful theory, the account must not be circular. The properties which are used to determine the name’s referent must not themselves involve the notion of reference in a way that it is ultimately impossible to eliminate. Let us now evaluate this objection of Kripke’s. There are several descriptivist views that would clearly violate condition C. That would be the case of, for instance, a view that claimed that a metalinguistic description such as ‘the individual called N’, ‘the bearer of N’, or ‘the individual I [or someone] refers to by N’ is just by itself what fixes the referent of the name N –that is, without being accompanied by a relevant act of baptism of any kind. A view like this could be the view that claimed that one could refer to something or someone just by uttering ‘the referent of this very phrase’. However, it seems to me that a view that claimed that the metalinguistic description which a speaker associated with a name N expressed what the name meant on that particular occasion and determined the name’s referent, would not be circular. Therefore it would not violate this condition if it did not also claim that that metalinguistic description by itself is what fixes the referent of the name. Therefore, Loar’s view is not circular and does not violate Kripke’s constraint. For what determines the name’s referent in each particular use is the fact that the referent is the individual that the identifying individual concept expressed in that use individuates, which in fact does not always include the concept being called N. In any case, it does not matter if it does include it, for when the individual concept intrinsic to what the speaker means includes it, it is just another property that the referent satisfies. The reason why the referent satisfies this property is simple: at some point they were “baptized” with that name and that “baptism” successfully initiated a naming practice of calling them N. If a “baptism” is successful, the individual “baptized” with a name N becomes an individual for whom it is true that they are called N. It is not necessary that the view requires that “baptism” to be a typical one in which it is usually babies that get their names. “Baptisms” can take many forms and, it seems to me, naming practices can even be ! *)! started by mistake, or simultaneously with the first use of that name to refer to the individual who is thus “baptized”. Now, that said, all of this is also compatible with it also being a definite description, distinct from the metalinguistic ones considered above, which in fact fixes the referent of names in these acts of “baptism”. The referent of a name could thus be fixed by a description such as ‘the individual I am looking at this exact moment’ said at an act of “baptism” following an utterance of ‘I will call N’ by the individual performing it. In conclusion, Kripke’s circularity objection does not show Metalinguistic Descriptivism to be wrong. Furthermore, Metalinguistic Descriptivism seems to be as circular as Kripke’s own causal theory of reference. Let us now turn to some further positive arguments for Descriptivism. In addition to the traditional arguments that are usually adduced in favor of Fregean or descriptivist views, which I discuss earlier in this chapter, Loar adduces the following additional interesting ones. First, upon every utterance of a sentence containing a singular term, there is some definite description such that the speaker could have expressed what he had meant by using it attributively instead of the singular term actually used. For instance, when a speaker says ‘Vancouver is beautiful’, what she means could also been expressed perhaps by ‘the city in which I am right now is beautiful’, ‘the city in which I started a PhD is beautiful’, or ‘the city where Greenpeace originated is beautiful’. This is of course controversial, but it does seem that there is no use of a name for which there is no description expressing an individual concept that fits what the speaker meant. Secondly, the attributive versus referential distinction not only applies to uses of definite descriptions as Donnellan proposed, but also to uses of all other singular terms, as Kripke also noted. Furthermore, it seems to be a matter of degree, which is not what one would expect if this distinction were a clear-cut one, as non-descriptivists argue it is. ! **! Finally, Loar presents an example of a case in which some mode of presentation of the individual referred to by the use of a proper name is intuitively essential to what is being communicated. Loar describes the case as follows: “Suppose that Smith and Jones are unaware that the man being interviewed on television is someone they see on the train every morning and about whom, in that latter role, they have just been talking. Smith says 'He is a stockbroker', intending to refer to the man on television; Jones takes Smith to be referring to the man on the train. Now Jones, as it happens, has correctly identified Smith's referent, since the man on television is the man on the train; but he has failed to understand Smith's utterance. It would seem that, as Frege held, some 'manner of presentation' of the referent is, even on referential uses, essential to what is being communicated.” (Loar (1976), p. 357) Let me summarize the discussion of Metalinguistic Descriptivism so far. We have seen that Loar’s view can not only counter Kripke’s objections, but can also come up with additional support. This is good news for the view that I defend in next chapter, given that it is similar to Loar’s in certain relevant respects. (However, let me advance that I will draw different conclusions from Kripke’s modal argument and my view will depart from the Fregean tradition precisely because of that.) In any case, I think that this discussion of Kripke’s arguments as applied to more sophisticated versions of Descriptivism sheds some valuable light on them, helping us to better understand their force. Before going on to discuss Rigidified Descriptivism, let us pause to consider a second version of Metalinguistic Descriptivism: Kent Bach’s. As we will see, Bach’s view is very close to Loar’s, but it will be useful for my purposes to discuss some of its differences. Unlike Loar, Bach defends this metalinguistic descriptivist view only for the case of proper names, rather than for all singular terms, and calls it ‘The Nominal Description Theory’. Bach’s view further differs from Loar’s in the following respects: first, he thinks ! *+! that for any proper name N, its conventional meaning is the conventional meaning of the definite description ‘the bearer of N’ and not ‘the individual called N’ or ‘the individual referred to by N’, because he thinks that the last two, unlike the first, fall prey to Kripke’s circularity objection. Secondly, he explicitly contends that what speakers usually mean by uttering sentences containing proper names, as well as the individual concepts that are thereby usually meant, is pragmatically rather than semantically conveyed. Unlike the literal conventional meaning of proper names, which he identifies as their sense, these individual concepts do determine reference in the cases in which singular terms refer. Thus, according to Bach’s particular version, all utterances of a sentence containing a proper name such as, for instance: (19) Bach and Loar agree on many points. or (20) Manolo is named ‘Manolo’. semantically express the same. Their literal meaning is, respectively, something like: (19’) There is a unique bearer of ‘Bach’ and there is a unique bearer of ‘Loar’ and they agree on many points. (20’) There is a unique bearer of ‘Manolo’ and he is named ‘Manolo’. So, sentences containing proper names are, like the names themselves, not ambiguous. They all mean the same. If proper names were ambiguous, Bach suggests, not knowing who all the bearers of ‘Manolo’ are would entail being deficient in linguistic competence; but this does not seem to be the case. Bach argues that the Nominal Description Theory of proper names could treat proper names as unambiguous indexicals whose uses referred to distinct bearers of the name. This can be done by following Peacocke’s (1975) suggestion that definite descriptions ! *"! have demonstrative readings such as that man for some uses of ‘the man’ or that bearer of N for some uses of N and ‘the bearer of N’. However, he rejects this view because he thinks it would multiply meanings beyond necessity, there being no reason for thinking that incomplete expressions require such demonstrative readings. In short, according to Bach’s view proper names are like incomplete definite descriptions that are pragmatically completed in their context of use. What the speakers mean by uttering sentences containing proper names, thus, is not what the sentences literally mean, but a proposition partly constituted by other individual concepts instead, some of which, unlike the name’s literal meaning, determine its reference. In fact, a third difference that follows from this second is that for Bach, but not for Loar, referential uses of proper names are pragmatic. For, as just observed, on Bach’s view, the semantic uses of proper names are all attributive. In order to better understand Bach’s nominal view of proper names, it may be useful to consider his reply to one of Kripke’s further complaints. Kripke complains that there is no more reason to suppose that being so-called is part of the meaning of proper names than of any other word. To which, Bach rightly replies as follows: “There is a reason: there is nothing else for a name to mean! Putting the difference crudely, whereas horses are called ‘horses’ because they each have the property of being a horse, Socrates is called ‘Socrates’ because he has the property of bearing the name ‘Socrates’. He is called ‘Socrates’ because that’s his name.” (Bach (1981), p. 371)11 Moreover, Bach goes on to argue that only a nominal theory can account for the difference between literal and non-literal uses of names; their literal uses being the ones in which they refer to their bearers. However, I think this point of Bach is wrong. The difference could perfectly well be explained by appealing to the causal-historical chains !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! 11 One might wonder why having the property of being Socrates should not be the reason why Socrates is called ‘Socrates’. I think that this suggestion reverses the order of explanation: one is called a name because at some moment of time someone decided to give one that name (and therefore, as Bach puts it, because that’s his name), and not because one has a particular property prior (or in addition) to receiving that name. ! *#! of the uses of names and the acts of “baptism” that originate them (towards which Kripke (1972) brought our attention) as follows: a literal use of a name is one in which it refers to an individual who was at the act of baptism that originated a causal-historical chain of uses of that name. This view does not entail that the meaning of proper names is that of a nominal description and so it is not a version of the nominal description theory of proper names. As Kripke pointed out, something can fix the reference or meaning of a word without giving its meaning. Let me now turn to my own assessment of Kripke’s major arguments as addressed to Bach’s particular metalinguistic view. We already saw above how a view such as Bach’s deals with Kripke’s semantic argument, paying attention to what would be the semantic content of utterances of sentences containing proper names according to that view. (This was at the end of the discussion of a possible reply from Loar to Kripke’s semantic argument, where I considered the metalinguistic descriptions.) I concluded that if being the bearer of a name N entails being originally “baptized” N, we would not call the impostor ‘Gödel’. On the contrary, if being the bearer of a name N did not entail being originally “baptized” N, no situation in which Gödel was not the bearer of the name ‘Gödel’ would be conceivable. Let us now see how Bach’s view stands up against Kripke’s epistemic and modal arguments. In my opinion, the only available explanation of the intuitive a priority of the truth expressed by utterances of some conditional metalinguistic sentences according to Bach’s view (as well as Loar’s view, indeed) is implausible. Consider utterances of the following sentences: (21) If Gödel exists, then Gödel bears the name ‘Gödel’. (22) If the bearer of ‘Gödel’ exists, then the bearer of ‘Gödel’ bears the name ‘Gödel’. It follows from Bach’s theory that both express the same thing semantically: namely: ! *$! (23) If there is a unique bearer of ‘Gödel’, then there is a unique bearer of ‘Gödel’ who bears the name ‘Gödel’. which is a necessary, a priori truth. However, this is contrary to intuitions. For even if both (21) and (22) express a truth that is intuitively a priori, utterances of (21) express a truth that is intuitively contingent, whereas utterances of (22) express a truth that is intuitively necessary. Therefore, Bach’s view needs to appeal to some pragmatically conveyed content of utterances of (21) to give an account of these intuitions. In particular, he needs to appeal to some referential use of the proper name ‘Gödel’ which contributed some individual concept that is not a logical restriction of the proper name’s referential qualifier being the bearer of ‘Gödel’. Such content could be, for instance, that semantically expressed by utterances of the following sentence: (24) If the person who discovered the incompleteness of arithmetic exists, then the person who discovered the incompleteness of arithmetic bears the name ‘Gödel’. which is contingent. Now, any such a possibility interprets (21) as expressing, pragmatically, an a posteriori rather than an a priori truth. But if the intuitions are that utterances of (21) express a truth that is both contingent and a priori, then none of the possibilities available to Bach’s theory (or indeed to Loar’s, for that matter) seem to be able to express such a truth. In fact, (21) seems to me analogous to sentences whose utterances are considered to express contingent a priori truths such as: (25) I exist now. (26) I am here now. (27) I am uttering something now. (28) If Julius exists, then Julius invented the zip. –where ‘Julius’ is the descriptive name mentioned above. ! *%! To this (at least apparent) difficulty, it might be replied that the case is similar to utterances of the following pair of identity statements in a relevant respect. (29) Hesperus is Hesperus. (30) Hesperus is Phosphorus. Utterances of pairs of identity statements such as (29) and (30) are widely considered, at least by non-descriptivist, to express the same necessary truth, even if the first expresses an a priori truth and the latter expresses an a posteriori truth. So it seems that the different epistemic profile of these utterances must be explained by something relative to some difference in the sentences uttered, which is the only thing they seem to differ by. It might then be that the different epistemic profile of utterances of (21) and (24) is explained by something relative to some difference in the sentences used as well. However, there is an important difference between these two cases, which in my opinion makes this explanation a bit odd. In the case of identity statements such as (29) and (30), there is something utterances of both of them semantically express, namely that Venus is identical to Venus, whereas utterances of (21) and (24) would semantically express two quite different descriptive propositions: the proposition that an utterance of (23) would semantically express in the case of (21), and the proposition that an utterance of (24) itself would semantically express in the case of (24), which I repeat below. (23) If there is a unique bearer of ‘Gödel’, then there is a unique bearer of ‘Gödel’ who bears the name ‘Gödel’. (24) If the person who discovered the incompleteness of arithmetic exists, then the person who discovered the incompleteness of arithmetic bears the name ‘Gödel’. Thus, the following analogous explanation does not seem to be equally plausible: utterances of (21) and (24) mean the same contingent truth in certain contexts, but precisely because sentences (21) and (24) are different, their epistemic profiles differ: whereas the truth that these utterances of (21) mean is a priori, the truth meant by these ! +'! utterances of (24) is a posteriori. And this is so, even if all utterances of (21) and (24) semantically express totally different descriptive propositions. What is weird is that the explanation of the a priority of the contingent truth expressed by utterances of (21) appeals to the form of that sentence (or to something similar to what Bach’s view regards as its semantic content) at the same time as the explanation of its contingency appeals to what a sentence with a totally different form semantically expresses. Indeed, its a priority seems to come from something closely related to the necessary truth that utterances of (21) would semantically express according to such a view. If I interpreted Loar’s view correctly, according to it and contrary to Bach’s, such utterances of (21) would semantically express the same as the utterances of (24) and so the case would be like that of the identity statements above in that respect as well. However, the explanation of the a priority of the truth expressed by utterances of (21) available to Loar’s view would be the same and so, equally odd. To sum up, like Loar’s view, Bach’s version of Metalinguistic Descriptivism also survives Kripke’s arguments against Descriptivism. However, in the course of the discussion of Bach’s theory, we have seen that both versions of Metalinguistic Descriptivism have some odd consequences. This does not make them untenable, but provides reasons to look for a better alternative. This is what I do in Chapter 3. As I mentioned at the beginning of this section, Metalinguistic Descriptivism is just one of the descriptivist views that seem to survive Kripke’s and other anti-descriptivists’ attacks. Another important view is Rigidified Descriptivism, to which I turn now. ! +&! Rigidified Descriptivism Loar and Bach’s Metalinguistic Descriptivism claims that the conventional meaning of proper names is the meaning of a definite description that is actually satisfied by different individuals and hence, not rigid. This view would thus contradict Kripke’s view that proper names are rigid (explained at the beginning of this chapter) as regards conventional meaning. This is so even if, according to the metalinguistic view, particular uses of proper names often do contribute individual concepts that are expressed by attributive uses of rigid definite descriptions. However, by identifying the meaning of proper names with the meaning of definite descriptions containing the adjective ‘actual’, one gets a descriptivist view that is compatible with Kripke’s rigidity thesis. For then, proper names and these definite descriptions that give or at least fix their meaning become rigid. In fact, this is the kind of view defended by David Lewis (1984) and Frederik Kroon (1987)12. It is the inclusion of this specific adjective that rigidifies the descriptions associated with names. For instance, the meaning of the proper name ‘Frederik Kroon’ could be identified with the meaning of the rigidified description ‘the actual Australian and descriptivist philosopher of language who is a professor at Auckland University in New Zealand nowadays’. So that sentences containing the name and sentences that only differed from them in containing the rigidified description instead of the name, such as: (25) Frederick Kroon participated in the sixth Barcelona workshop on issues in the theory of reference. (26) The actual Australian and descriptivist philosopher of language who is a professor at Auckland University in New Zealand nowadays participated in the sixth Barcelona workshop on issues in the theory of reference. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! 12 ! For discussions of this view see also Alvin Plantinga (1978) and Scott Soames (1998). +(! would all be evaluated at other possible worlds with respect to one and the same individual, i.e. Frederik Kroon, even if that individual did not satisfy the description if it were uttered in another possible world. It is worth noting that this version of Descriptivism does not preclude rigidified nominal descriptions, such as ‘the actual individual called N’ or ‘the actual bearer of N’, from being the ones meant on using proper names. This allows for a mixture of the two descriptivist views that we are discussing. In fact, even if Loar’s and Bach’s Descriptivism could ultimately be defended from Kripke’s objections, Kripke seems to be right in distinguishing proper names from non-rigid definite descriptions in relation to their modal behavior. Loar’s and Bach’s view may perhaps provide right interpretations by interpreting proper names as definite descriptions taking a wide scope over modal operators. The fact, however, is that utterances of sentences containing proper names are evaluated with respect to one and the same individual at all possible worlds, and this shows a difference between proper names and non-rigidified definite descriptions. For utterances of sentences containing non-rigid definite descriptions are not so evaluated. Moreover, as Lewis points out, if proper names meant or were used to mean the same as what non-rigid definite descriptions mean, two speakers “who have attached the same term to the same referent by means of different descriptions” would mean, contrary to intuitions, something that had different truth-values when uttering modal sentences containing it. Consider: (27) Possibly, Frederick Kroon did not participate in the sixth Barcelona workshop on issues in the theory of reference. uttered by someone who associated ‘the philosopher called ‘Frederick Kroon’ who wrote the paper “Existence in the Theory of Definite Descriptions”’ thereby meaning: ! +)! (28) Possibly, the philosopher called ‘Frederick Kroon’ who wrote the paper “Existence in the Theory of Definite Descriptions” did not participate in the sixth Barcelona workshop on issues in the theory of reference. and by someone who associated ‘the Australian descriptivist speaker who participated in the sixth Barcelona workshop on issues in the theory of reference’ thereby meaning (29) instead, for instance. (29) Possibly, the Australian descriptivist speaker who participated in the sixth Barcelona workshop on issues in the theory of reference did not participate in the sixth Barcelona workshop on issues in the theory of reference. Whereas utterances of (28), in which the definite descriptions are attributively used, say something true on both readings, utterances of (29), in which the definite descriptions are attributively used too, say something false in one of its readings. In particular, in that reading, these last utterances say that a contradiction is possible: namely that it is possible that there is a unique Australian descriptivist speaker that participated in the sixth Barcelona workshop on issues in the theory of reference that did not participate in the sixth Barcelona workshop on issues in the theory of reference. Before ending this section, let us take stock. First and foremost, I conclude from the discussion in this section that Descriptivism is not dead. Furthermore, that in spite of Loar’s and Bach’s powerful defense of Descriptivism, proper names are rigid whereas “non-rigidified” definite descriptions are not. So, Descriptivism should be take proper names’ rigidity into account, and that is what I try to do in later chapters of this work. 2.6. Pragmatic Millianism and Semantic Two-Dimensionalism Part of what supports a given view is how it fares in comparison to its rivals. So, having discussed some post-Kripkean descriptivist views and the ways they deal with objections, ! +*! in the remainder of this chapter I explain how Millianism handles one of the traditional problems I explained at the beginning: the informativeness of identity statements. Also, at the end of this chapter I very briefly say something about Semantic Two-dimensionalism, a view that will be relevant in later chapters. Pragmatic Millianism Despite initial appearances, it looks as if Millian views may also explain the informativeness of identity statements, and give an account of propositional attitude reports by appealing to pragmatics (this second issue is discussed in more detail in Chapter 4). This is precisely what Robert Stalnaker (1978), Nathan Salmon (1984) and Scott Soames (1987) defend. Their account of the informativeness of identity statements is as follows. Consider again: (1) Hesperus is Hesperus. (2) Hesperus is Phosphorus. It follows from Millianism that utterances of (1) and (2) semantically express the same: a necessarily true proposition. According to the possible worlds view of propositions – which is the one that Stalnaker defends– this proposition is the set of all possible worlds, whereas on the Russellian view of propositions –which is the one that Salmon and Soames argue for– it is the Russellian structured proposition constituted by the identity relation and the planet Venus, which may be characterized as the following ordered pair: (3) <=, <Venus, Venus>> According to their view, the informativeness of utterances of (2) is due to the fact that in addition to semantically expressing this proposition, they pragmatically convey a different contingent proposition, which is informative. This other proposition is the proposition that utterances of: ! ++! (4) The evening star is the morning star. semantically express; which, following Russell’s theory of descriptions, would be analyzed as: (5) There is a unique evening star and a unique morning star and they are identical. Moreover, this informative pragmatic content, which according to these Millian views utterances of sentences such as (2) convey, would also account for their appearance of contingency. Utterances of sentences such as (2) seem to express contingent a posteriori truths. Notwithstanding, they express a posteriori but necessary truths. They appear to be contingent, the explanation goes, precisely because they pragmatically convey this other proposition which is contingent. Stalnaker’s pragmatic explanation of the informativeness and appearance of contingency of identity statements such as (2) is Gricean. The explanation is that this informative and contingent proposition pragmatically conveyed by utterances of identity statements such as (2) is a Gricean conversational implicature that is pragmatically conveyed due to the recognizable violation of some Gricean maxim of conversation. In particular, since the proposition that utterances of identity statements such as (2) semantically express is clearly uninformative, hearers will understand that in uttering them, the speaker meant something different and informative –which in the case of (2) is something such as (4). For otherwise, the speaker would be violating the maxim of quantity, namely the maxim that says: “Make your contribution as informative as is required (for the current purposes of the exchange)”. This explanation is offered against the background of a picture of the nature of assertion. According to such a picture, informative content excludes certain possibilities but not all, and the essential effect of an assertion is to exclude possibilities in this sense as well. Thus, what one does by asserting something is to exclude some possibilities by ! +"! partitioning the space of possible worlds between those at which what is asserted is true and those at which it is false. For these reasons, according to such a view, any informative content as well as anything that one asserts must be contingent. Stalnaker also develops a two-dimensional framework and calls the descriptive proposition pragmatically conveyed ‘the diagonal proposition’ –which, within that framework, is the proposition that for each world w it is true at w if and only if the proposition expressed in w is true at w. The two-dimensional framework is a framework that captures the two ways in which the truth-values of utterances of sentences depend on facts: on the one hand, it is facts that provide their content; on the other, it is also facts that make this content true or false. Thus, using our example again, an actual utterance of (2) semantically expresses a necessary proposition about Venus, which is true at all possible worlds. However, a possible utterance of (2), made in another possible world, may semantically express a necessarily false proposition not about Venus, but about two distinct celestial bodies, and this other proposition will be false at all possible worlds. Semantic Two-Dimensionalism Although Stalnaker and Kaplan developed the two-dimensional framework and put it to service in their defense of Millianism, it may also be used to defend a Fregean view and this is precisely what David Chalmers (2006) and Frank Jackson (1998), for instance, do. The resulting view is what may be called ‘Semantic Two-Dimensionalism’. According to Semantic Two-Dimensionalism, expressions have two distinct semantic contents: singular terms thus have both sense and reference, and utterances of sentences semantically express two propositions. In the case of sentences containing singular terms, ! +#! one of them is a descriptive proposition, which is primarily expressed, and the other is a singular proposition, which is secondarily expressed. The view that I offer in this thesis shares with Semantic Two-Dimensionalism the idea that utterances of sentences semantically express more than one proposition13. In the following chapter, before defending the view I favor, I discuss another view on which my proposal bears: the Token-reflexive view of indexicals and its possible extension to proper names. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! 13 It is worth mentioning that Perry (1988) already distinguished two propositions in relation to the semantics of utterances of sentences: a proposition expressed and a proposition created. Perry identified the cognitive significance with the latter. ! +$! 3. Token-Reflexivity Having discussed singular terms in general, in this chapter I am more concerned with indexicals and demonstratives. This is because the view that I start presenting in this chapter makes use of the idea of token-reflexivity; an idea that is most plausible in the context of explaining the meaning of indexicals and demonstratives. In Section 1, I present the token-reflexive view of indexicals and David Kaplan’s (1989) alternative view. In Section 2, I argue for the token-reflexive view by offering what I take to be plausible replies to objections raised against it. In Section 3, I consider the straightforward Fregean version of this view, extended to proper names as well, in order to clearly distinguish it from the Fregean versions that I consider in later sections. In Section 4, I then discuss Manuel García-Carpintero’s (1996, 1998, 2000) Fregean version of the token-reflexive view of singular terms, including proper names. Finally, in Section 5, I present the Fregean version of the token-reflexive view of singular terms that I favor and that will be assumed in the rest of this work. In this last section I also discuss Kaplan’s (1990) view on the individuation of words in order to introduce my own. 3.1. The Token-Reflexive View of Indexicals and Demonstratives Conventional meaning and the type–token distinction Meaning must be attached to word-types. Words are meaningful because of regularities in the way the members of a community use words in their language. Such regularities are linguistic conventions that provide words with a conventional meaning, which must then be attached to repeatable expressions; that is, to word-types. Word-types are repeatable abstract entities that may be exemplified by word-tokens, which are the ! "#! concrete instantiations of words that speakers utter or inscribe. It is thus, for instance, the word-type ‘semantics’, which different speakers use on different occasions, that conventionally means the study of meaning, rather than just one particular word-token of that word that a particular speaker uses on a particular occasion. Indexicals and demonstratives The type–token distinction seems essential to understanding indexicals and the tokenreflexive view of them that I discuss in this section. In fact, a useful characterization of what indexicals are, which I will adopt, makes use of such a distinction. This is the characterization offered by García-Carpintero’s (1998): “an indexical is an expression such that any given instance of it is not linguistically required to share truth-conditional import with any other instance (except for anaphoric uses)”. Indexical expressions, thus, include not only those expressions that are regarded as pure indexicals in at least some of their uses, such as ‘I’, ‘today’, ‘here’, and ‘now’, but also those regarded as true demonstratives in at least some of their uses, such as ‘you’, ‘she’, ‘there’, and ‘then’. What distinguishes these two kinds of indexicals is that whereas pure indexicals obtain their reference without recourse to demonstrations, true demonstratives could not. There is discussion on what these demonstrations are (whether they are pointing gestures or directing intentions), but for my purposes it is not necessary to introduce it here. Thus the token-reflexive view of indexicals is to be understood not only as a view of pure indexicals, but as a view of true demonstratives as well. This view has been defended by Mark Crimmins (1995), John Perry (1993, 1997 and 2001) and Manuel GarcíaCarpintero (1996, 1998, and 2000), but has its origin in Hans Reichenbach (1947). The token-reflexive view of indexicals is the view that indexical-types are associated with a (conventional) linguistic semantic rule that attributes semantic properties such as reference and truth-conditions to tokens, rather than to types, and that, furthermore, it attributes these properties to tokens in relation to properties of the tokens themselves. ! $%! Thus, for instance, the rule associated with ‘I’ is something like: (1) For any token of ‘I’ t, t refers to the utterer of t. so that a token of ‘I’ in an utterance of: (2) I am a woman. refers to whoever uttered it, in virtue of the fact that she bears the appropriate relation to the token itself (that is, she has the relational property of being the one who uttered it) and that this is precisely what the type of the indexical ‘I’ conventionally means. Analogously, the rule associated with ‘now’ is something like: (3) For any token of ‘now’ t, t refers to the time at which t is uttered. And finally, to illustrate it in the case of a demonstrative as well, consider the demonstrative ‘that’, for instance. The rule associated with ‘that’ could then be something such as: (4) For any token of ‘that’ t, t refers to the most salient thing in the context in which t is produced. It is important to notice that it is token events rather than token objects which these semantic properties are ascribed to, and that this is so in virtue of their relational properties rather than the relational properties of token objects. This is important because one and the same token object may be used in different token events with different semantic properties. Take a written note that is used on more than one occasion such as, for instance, a note a teacher may hang on the classroom door saying ‘we are in the middle of an exam’ to keep other students from coming into the classroom too early, every time a class is having an exam. Each time the note is used, it means different ! $&! things1; and this is so despite the fact that the token object of the sentence ‘we are in the middle of an exam’ that is used in the note, like the note itself, is always the same one. For on each occasion, the times and the students referred to by means of the note are different. Also, as García-Carpintero (1998) points out, not all token events constitute full-fledged linguistic acts and the word ‘utterance’ suggests that such a full-fledged linguistic act has taken place. However, in this work I continue using the word ‘utterance’, while keeping this limitation in mind. Having presented the token-reflexive view, let me now very briefly introduce David Kaplan’s (1989) alternative view of indexicals. Kaplan’s theory is relevant for my purposes because it is the major rival view to the token-reflexive view, and in the next section I consider objections raised to the token-reflexive view which have to do with the differences between the two. In contrast to the token-reflexive view, according to Kaplan’s view, it is types-in-context and not tokens, that have semantic properties. Kaplan calls these types-in-context ‘occurrences’. According to this view, indexicals have character and content. Indexicaltypes have a character and, in context, (that is, their occurrences) obtain content. Characters are functions from contexts to contents, whereas contents are the individuals2 which the indexical-types-in-context refer to and contribute to the propositions that sentences containing them express. Characters represent the conventional linguistic meaning of the indexical-types. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! 1 The fact that these different tokens of the same type mean different things is compatible with the fact that it is types rather than tokens to which conventional meaning attaches. It is precisely in virtue of the types’ conventional meanings that token events refer and so, acquire their own particular non-conventional meaning. 2 In this presentation, I am identifying the content of an expression with its extension, following Braun (2007). However, it is worth mentioning that in Kaplan’s formal system, the Logic of Demonstratives, contents are intensions (i.e. functions from possible worlds to extensions). In the case of indexicals, these intensions are constant, having as their value the same individual for each possible world taken as an argument. This is how the formal system represents indexicals as rigid. ! $'! For instance, the character of the indexical-type ‘I’ would be something like: (5) For any context c, the indexical-type-‘I’-in-c refers to the agent of ‘I’ in c. Notice that I am focusing here on singular terms. The case of predicates is different. The content of a predicate is a property rather than an individual, and the intension that the formal system attributes to it is not constant. 3.2. Objections to the Token-Reflexive View In this section I discuss four main objections to the token-reflexive view. Two of them are due to Kaplan (1989), one to David Braun (1996), and the final one has been advanced recently by Stefano Predelli (2006). I argue that none of these objections is convincing. The logic of indexicals problem Kaplan (1989) objects to the token-reflexive view of indexicals on the grounds that it is not able to account for their logic. Arguments that we consider valid consisting of sentences with indexicals, even of the simple semi-formal form ‘P, therefore P’ (for instance, ‘I am dancing now. Therefore, I am dancing now.’) would no longer be logically valid if it were tokens and not types-in-context that had the semantic properties of having reference and truth-conditions. This is so because since different tokens are usually uttered at different times, the different tokens of the indexicals involved in the argument exist in different contexts and so, could acquire different content. Different tokens pertain to different contexts. Whereas if it is types-in-context that have content, different tokens of the same type could be paired with a single context. ! $(! Thus, the argument ‘I am dancing now. Therefore, I am dancing now.’ would be logically valid if it is interpreted with respect to one unique context, so that the indexical-type-‘I’ that appears twice in it (and indirectly, both tokens) is assigned one and the same content when interpreted in that single context. Analogously, the same applies to the indexicaltype ‘now’. Notice that Kaplan’s view may assign semantic properties to tokens too. The difference between the two views consists of the fact that whereas the token-reflexive view assigns semantic properties to tokens that are not properties of the types they instantiate, Kaplan’s view does so derivatively. That is, according to Kaplan’s view, the semantic properties of tokens are the properties of their types-in-context. This objection, however, is not sound. I think it is linguistic data that premises or conclusions of arguments that instantiate the same sentence-type in which different tokens of the same indexical-type appear may acquire different interpretations. This is so precisely because such premises or conclusions contain different tokens of the same indexical-type and so those tokens occur in different contexts. Indexicals do not require it to be the case that any tokens of them receive one and the same interpretation. This is how indexicals work. In contrast, logical constants do require that any tokens of them receive one and the same interpretation. Indexicals do not behave like logical constants do in formal languages. One could think of keeping a logic of indexicals by restricting interpretations of all premises and conclusions of an argument to one single context; but that, in my view, would be like treating them not as what they are, but as something different: as if they were formal logical constants. In any case, this is a possibility that is available to both theories: on the token-reflexive view one needs only to abstract and consider the different tokens as if they had been uttered in one and the same context. Moreover, as García-Carpintero (1998) argues, Kaplan’s objection is due to an equivocation between the two senses of the expressions ‘logical truth’ and ‘logical validity’: the sense of formal logical truth and formal logical validity as truth or validity in virtue of logical form alone; and the sense of analytic truth and analytic validity, that ! $)! is, the sense of truth or validity in virtue of meaning alone. Among arguments that consist of sentences with indexicals, there are only analytically valid arguments such as ‘if that is older than that, then the former is older than the latter’, but no logically valid arguments in the formal sense. Analogously, there are only analytically true instances of the logical schema ‘a = a’ in natural language such as ‘Aristotle is identical to himself’, but no logically true instances in the formal sense (‘Aristotle is identical to Aristotle’, for instance, does not require an interpretation such that both tokens of ‘Aristotle’ refer to the same individual). These examples are analytically rather than formally valid or true, because they are not true in virtue of logical form, but in virtue of meaning alone. My conclusion is then that there are no logically formal valid arguments involving indexicals because there are no two tokens of indexicals that require a single interpretation. Kaplan’s objection is therefore unsuccessful. The lack of expressions problem David Braun (1996) poses another objection to the token-reflexive view: that the tokenreflexive view does not provide enough interpretable expressions for all the possible interpretations. He illustrates this point by considering the long sentence formed by conjoining the sentence ‘I exist now.’ with itself a trillion times and a context which has him as its agent and which lasts less than one minute. There is no actual or metaphysically possible utterance of this long sentence-type with the content that this sentence-type in the context considered would be assigned. I think this second objection is not successful either. The reason is that although there is no such metaphysically possible utterance, as García-Carpintero (1998) notes, there is such a conceptually or logically possible utterance and that is enough. Like scientific theories, semantic theories abstract away from things that are irrelevant to the explanations they aim at. In the example put forward by Braun, a conceptually possible ! $"! token is not metaphysically possible due to features that are irrelevant to semantics, such as the physical properties of token production. In a semantic analysis, however, one can consider this conceptually possible token event and assign some interpretation to it, despite its not being metaphysically possible. As García-Carpintero (1998) says: “It is as if Braun objected to Galileo on the grounds that his account was given for frictionless circumstances, and frictionless circumstances are not metaphysically possible” (p. 539). ‘I say nothing’ objection According to Kaplan, and later to David Braun and Stefano Predelli, it is clear that there are sentences such as ‘I say nothing’ that are true in some contexts, and only in those, in which they are not uttered. So any correct semantic theory should entail this. The tokenreflexive view does not, however: according to the token-reflexive view, it is tokens that are interpreted and are true or false, and hence it is not possible for a sentence to be true if not uttered. So they conclude that the token-reflexive view is not correct. In my view, this third objection is no more effective than the first two. I think the objection would be correct if there were linguistic data consisting of a pretheoretical intuition of ordinary speakers that a given sentence-type-in-context is true. If that were the case, any correct semantic theory should account for such data; but as GarcíaCarpintero convincingly argues, there is no such linguistic data because there is no such intuition. It seems clear to me that ordinary people have intuitions about what they mean when using words and the truth-value of their utterances, but they do not have intuitions about unuttered types-in-context. Types-in-context are theoretical entities about which a pretheoretical intuition is not possible. Predelli’s analyticity objection I now close this section by discussing in more detail an objection to the token-reflexive ! $$! view that was raised recently by Predelli (2006) and is related to the ‘I say nothing’ objection just considered. I argue that this objection, like the others, fails. Predelli (2006) understands the ‘I say nothing’ objection as having to do with pretheoretical intuitions of ordinary speakers about issues of meaning. That is, as indicating issues intimately related to the intuitive meaning of sentences such as the one Kaplan considers, i.e. ‘I say nothing’. For this reason, according to Predelli, it is related to issues of analyticity. Following this interpretation of Kaplan’s objection, Predelli argues that the tokenreflexive view attributes analyticity to non-analytic sentences, which no correct semantic theory should do. In particular, he claims that traditional token-reflexive views are committed to the analyticity of the following non-analytic sentences –and that non-traditional tokenreflexive views are committed to the analyticity of the non-analytic sentences (3) and (4): (1) I am uttering something now. (2) A token exists now. (3) Either a token exists now, or it has existed in the past, or will exist in the future. (4) Something either exists now, has existed, or will exist. As Predelli points out, on the token-reflexive view, the interpretation of indexicals and thereby the interpretation of the sentences containing them, guarantees the existence of utterances of these same indexicals or sentences in their context of interpretation. This is guaranteed in virtue of the conventional meaning (i.e. the token-reflexive rule) that this view ascribes to the indexical-types. According to the token-reflexive view, the conventional meaning of ‘I’, for instance, requires its interpretation to be that precise individual who is the speaker of the utterance, and the conventional meaning of ‘now’ requires its interpretation to be that exact moment that is the time of the utterance. This means that without any utterance, there would be no interpretation. Therefore, their ! $*! interpretation guarantees the existence of utterances of them or of sentences containing them in any context in which they are interpreted. In virtue of this, sentences (1)–(4) are true in all contexts in which they are interpreted. Since this is so in virtue of the conventional meaning of the indexicals involved, Predelli identifies this feature with the phenomenon of analyticity (i.e. truth in virtue of meaning alone) and objects that from the token-reflexive view it follows that these sentences are analytic, when intuitively they are not. According to Predelli, intuitively, sentences (1)–(4) are not analytic because their intuitive conventional meaning (and the intuitive conventional meaning of the indexicals that appear in them, in particular) does not require the existence of any utterances at all. Intuitively, according to Predelli’s view, for instance, the conventional meaning of ‘I’ only requires the interpretation of ‘I’ to be an individual, and the conventional meaning of ‘now’ only requires the interpretation of ‘now’ to be a moment of time. They do not further require, he argues, that the individual assigned to ‘I’ be the speaker of any utterance or that the time assigned to ‘now’ be the time of any utterance. Hence, intuitively, there should be contexts of interpretation of them in which sentences (1)–(4) are false, and these are contexts in which they are not uttered. Therefore, Predelli concludes that intuitively they are not analytic. Furthermore, Predelli thinks that there are in fact contexts in which (1) and (2) are both uttered and false; that is, contexts in which the indexical ‘now’ is uttered but does not refer to the time of the utterance. He does not explain how this is possible for the cases of (1) and (2), but he argues that ‘now’ in, for instance, an utterance of (5) The allied troops are now ready to strike. during a historical narration, would refer to a different time than the time of utterance. All of this is meant to apply to traditional token-reflexive views. However, Predelli also ! $+! considers non-traditional token-reflexive views in which the rules assigned to the types of indexicals do not talk about speakers of tokens or times in which tokens are spoken. With respect to these non-traditional views, Predelli claims that it no longer follows from these other modified views that (1) and (2) are analytic, but it does that (3) and (4) are, when in fact neither of them is. Thus, Predelli concludes that any token-reflexive view is wrong because such views cannot escape the consequence that if they were correct, certain nonanalytic sentences would be judged to be analytic. Non-traditional token-reflexive views are views that ascribe different conventional meanings to indexicals. Those different conventional meanings would not identify the referent of ‘I’ or the referent of ‘now’, for instance, as the speaker or the time of speaking, respectively; but would continue ascribing reference and truth-conditions to tokens directly rather than to types and so would still guarantee the existence of the utterances in question at the contexts of interpretation. That way (3) and (4) would continue to be true in all contexts of their utterance, even if (1) and (2) would not. The conventional meanings of ‘I’ and ‘now’, for instance, according to these non-traditional views, would be something similar to: (6) For any token t of ‘I’, the referent of t is ia, where i is the index representing the context in which t takes place and ia is the agent of that index. (7) For any token t of ‘now’, the referent of t is it, where i is the index representing the context in which t takes place and it is the time of that index. Let me now explain why Predelli’s arguments are not convincing. First, consider again Predelli’s claim that the indexicals ‘I’ and ‘now’ may refer to an individual who is not the speaker of ‘I’ and a time which is not the time in which ‘now’ is produced, respectively. It seems clear, and Predelli himself realizes this possibility, that the false utterances he envisages are utterances in which the speaker is pretending to be in a different context when speaking and hence, pretending to refer to a different time by saying ‘now’. Pretending to do or be something is not the same as doing or being it, and in these cases ! $#! in particular, the speaker is not in the context she pretends to be in and for this reason, the token of ‘now’ does not refer to the time she pretends it to refer to. So these utterances do not constitute literal uses of these sentences and so, are not cases in which tokens of ‘now’ refer to a different time than the time at which they are uttered as Predelli judges. Moreover, analogous cases for sentence (1) do not even seem possible3. Secondly, it is true that, according to the token-reflexive view, in virtue of their conventional meaning (i.e. the token-reflexive rules) all utterances of sentences (1)–(4) are true in all contexts of their utterance. That does not, however, make them analytic. For it does not mean, and in fact it is not even true, that sentences (1)–(4) are true in virtue of meaning alone. Meaningful utterances of (1)–(4) are false at circumstances of evaluation in which they are not uttered, since the existence of utterances of (1)–(4) is contingent rather than necessary. Therefore, their meaning does not guarantee their truth and they are not analytic. According to the token-reflexive view, utterances of (1)–(4) are not true in virtue of meaning alone; they are not even necessary truths (except perhaps for (4) if there is no empty possible world, but this is a fact that is independent of the theory Predelli is criticizing). What is true is that according to token-reflexive views, utterances of (1)–(4) are true when evaluated at any context in which they are uttered, but are false at other circumstances of evaluation –that is, at some circumstances in which they are not uttered. The fact that there needs to be an utterance for it to be interpreted and evaluated (because it is utterances that are interpreted and evaluated thanks to their own properties and after reflecting on themselves) does not mean that these utterances cannot be evaluated with respect to other circumstances, in particular, ones at which they do not exist –i.e. at which the utterances are not made. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! 3 One might think that there are analogous scenarios for the case of (1) that clearly do not involve any pretense. Consider a family listening to an audio recording of a deceased member of the family, intended to be listened to after her death, in which the dead person recorded an utterance of (1). It seems to me implausible that we could consider this case one in which the token recorded of ‘now’ refers to the moment in which it is being listened to, and I cannot conceive of a scenario like this in which ‘now’ somehow refers to the moment of the listening that does not involve pretense. ! *%! Analytic truths such as all bachelors are unmarried, that are said to be true in virtue of meaning alone, are necessary and a priori; but utterances of (1)–(4) are contingent instead. There are contexts of evaluation at which they are not true: contexts different than the ones in which they exist, in which the sentence-types are uttered and hence, interpreted. It is just that utterances of (1)–(4) can never be false at those contexts in which they exist and are interpreted. But this is not what analyticity means; this is not because they are true in virtue of meaning alone. In fact, they are not. In any case, to think that the conventional meanings of indexicals such as ‘I’ and ‘now’ only require them to refer to individuals or times, respectively, regardless of whether those individuals are the speakers of ‘I’ and those times are the times in which ‘now’ is uttered, is to treat them as a different kind of expressions than the ones they are. It is to think of them as if they were logical constants; and in fact, this is what Kaplan and Predelli do when they consider that arguments consisting of sentences with indexicals should be predicted to be valid and Kaplan complains that the token-reflexive view does not. The way that Predelli argues that the token-reflexive view could also account for the alleged validity of those arguments falls prey to the same mistake; that is, it treats the indexicals in them as if they were not indexicals, as if they were logical constants, when they are not. As Predelli shows, the token-reflexive view has the conceptual apparatus required to make such a move, interpreting indexicals as logical constants; but this does not show that the arguments involving indexicals are valid, or that indexicals are rightly modeled by logical constants. As I said above, the reason why is not appropriate to model indexicals by logical constants is that logical constants require that any tokens of them receive one and the same interpretation. In contrast, indexicals do not require this. In the end, I think that Predelli’s arguments amount to a simple negation of the tokenreflexive view of indexicals. ! *&! 3.3. A Fregean Token-Reflexive View In this section I discuss the straightforward and simple Fregean version of the tokenreflexive view, which I think is not successful. In later sections I consider more sophisticated Fregean views, which are, in my view, on the right track. A Fregean version of the token-reflexive view would have indexical-tokens not only refer, but express a sense as well. The senses expressed by indexicals would then be the properties corresponding to the rules that constitute their conventional meaning. For instance, the sense that a token of ‘I’ would express would be the property being the speaker of the token just produced of ‘I’, expressed by a description such as ‘the speaker of the token just produced of ‘I’’. It is this sense of indexical-tokens instead of the conventional meaning of indexical-types that would determine the reference of indexicaltokens and give the meaning of the indexical-tokens that express it. This sense would thus indicate the way speakers conceive of the referents of indexicals, and the description that expresses the sense of an indexical-token would thereby be straightforwardly synonymous to the indexical-token and so, substitutable salva significatione. Let me consider here a potential objection to the view just described. One might think that if an utterance of ‘I am tired.’ is synonymous to, or somehow expresses the same as, ‘The speaker of the token just produced of ‘I’ is tired.’, then it must be possible to utter the second sentence without uttering the first, and to mean the same. However, the objection goes, it is not: if the first sentence is not previously uttered, then the description ‘the speaker of the token just produced of ‘I’’ does not denote. In reply, I think that it should be noticed that according to this Fregean version of the token-reflexive view, an utterance of ‘I am tired.’ will be synonymous not to any utterance of ‘The speaker of the token just produced of ‘I’ is tired.’ but only to some of its utterances: those utterances in which the definite description ‘the token just produced of ‘I’’ is interpreted as denoting the token of ‘I’ produced in the utterance of the first sentence. But no such utterance will exist without the previous utterance of ‘I am tired.’ ! *'! Therefore, the objection can be met. Fregean extension to proper names Indeed, this Fregean version of the token-reflexive view could be extended to cover proper names as well. According to this Fregean extension, proper names would also be token-reflexive expressions. Proper names would just be like indexicals, a possibility that Bach (1981) considered, but rejected. A natural way of spelling out this extended view would be as follows. Like indexical-types, proper name-types are also associated with a rule guiding the adscription of sense, reference, and truth-conditions to proper nametokens. Every proper name-type N would thus be associated with a rule of the following form: (1) For any token of N, t, t refers to the most salient individual named N at the time t is produced. Thus, for instance, the rule associated with the proper name-type ‘David’ would, according to such a view, be the following: (2) For any token of ‘David’, t, t refers to the most salient individual named ‘David’ at the time t is produced. As explained in Chapter 2, proper names name the individuals that are at the origin of some causal or historical chain of uses of those names in which an act of “baptism” takes place. Typically, proper names are mentioned rather than used in the act of “baptism” that initiates a practice of naming a particular individual with such a name. Thus, typically, no proper name-token is used in these acts to refer to the individual who is being “baptized”. Acts of “baptism” establish naming practices this way. According to the view under consideration, once the practice exists, tokens of proper names are used to refer as indicated by the token-reflexive rule associated with the type they exemplify: that is, they ! *(! refer to individuals who already bear that name and they do so thanks to some act of “baptism” that took place at some previous time. Notice though that, as I also said in Chapter 2, there does not seem to be any problem with the possibility that the first time a name is used to refer to a given individual this happens simultaneously with the act of “baptism” that makes this act of naming possible or successful. Tokens of proper names would then express the Fregean sense expressed by the description ‘the most salient individual named ‘David’’ and would refer to that particular individual. This Fregean sense would thereby determine the referent of the particular token relying heavily on the work of the context of use and the intentions of speakers. Moreover, it is thanks to and via the contextual domain restriction interpretation of the determiner ‘the’ of the definite description that expresses it. According to this view, the same way that there is only one indexical ‘I’, there would be only one name ‘David’ of which there are tokens that we produce to refer to different individuals, such as David Kaplan, David Lewis, etc. However, this view would suffer from the problems raised against plain descriptivist views of singular terms such as Kripke’s objections to descriptivism regarding proper names, which I consider in Chapter 2. This does not, however, show that no Fregean version of the token-reflexive view extended to proper names is successful. In what follows, I consider two more sophisticated and promising views of such a kind. 3.4. A Presuppositional Fregean View of Singular Terms According to García-Carpintero’s (2000) Fregean account, senses are non-purelyqualitative individuative properties that fix the reference of singular terms. They do not give the meaning of singular terms. That is, singular terms are not straightforwardly synonymous with definite descriptions expressing their senses. Singular terms and definite descriptions are not salva veritate or salva significatione substitutable. According ! *)! to this view, the token-reflexive senses that tokens express are essential ingredients of presuppositions of acquaintance that involve particulars, i.e. the tokens themselves. The idea is that utterances of sentences assert or mean something like a Russellian proposition, in addition to carrying presuppositions; and it is the presuppositions, rather than what is asserted or meant, that senses contribute to. Let me elaborate on this first short presentation of the view, starting with the notion of a presupposition. Presuppositions are propositions that are taken for granted rather than (primarily) asserted: which are “inherited” when the sentences carrying them are embedded in other linguistic contexts in which what the sentences assert is not “inherited”, and which can be cancelled in certain other linguistic embeddings. For instance, consider an utterance of: (1) The talk was interesting. It presupposes: (2) There was a talk. and if (1) is embedded in other linguistic contexts, such as in (3) The talk was not interesting. (4) If the talk was interesting, I am the king of Rome. the resulting utterances of these sentences still presuppose (2), even if they no longer state what utterances of (1) state when they do not appear embedded in this way. But the presupposition can be cancelled, for instance, in the following way: (5) If I had not drunk too much, I would say that the talk was interesting. But probably there was not even a talk. ! *"! There are two major accounts of presuppositions, which can in fact be combined: Stalnaker’s and Strawson’s. According to Stalnaker’s (1978) account of presuppositions, presuppositions are propositions that are thought to be already in the conversational background, whereas according to the Strawsonian (1950) account, presuppositions are propositions the truth of which is a requisite for the utterance to have a truth-value. Now, in particular, the presuppositions that are essentially constituted by senses, on García-Carpintero’s view, are Strawsonian semantic presuppositions that state the conventional truth-conditional import of the expressions carrying them. More specifically, they are presuppositions of acquaintance: presuppositions, as Donnellan put it, “of some particular someone or something that he or it fits the description” (Donnellan (1966)). More exactly, the proposition that is presupposed by an utterance of a sentence containing an indexical i is the proposition that there is a unique F to whom/which i refers –where F expresses a property appropriate to be the sense of the indexical i. This proposition may, in turn, be unpacked as: that there is a unique F with whom/which the conversational participants are related by having a dossier for the F (contextually specified to include, in addition, properties such and such) individuating it. For instance, in Loar’s example discussed in Chapter 2, Smith’s utterance of ‘He is a stockbroker’ semantically presupposes the proposition that there is a unique male most salient when the token t of ‘he’ is produced, and t refers to him. Furthermore, in that context, it also carries a non-fully-conventional, pragmatic presupposition that involves nonlinguistic contextual information: the proposition that there is a unique male on the television screen with such-and-such visual aspect when the token t of ‘he’ is produced, and t refers to him. As with the Fregean extension to proper names considered in the previous section, according to this more plausible Fregean extension (see García-Carpintero (1996)), proper names would also be token-reflexive expressions. Like indexical-tokens, utterances of proper names would carry a presupposition of acquaintance such as the ! *$! presupposition that they refer to the most salient individual named with such a name at the time the token is produced. Also, according to this view, the same way that there is only one indexical ‘I’, there would be only one name ‘David’, for instance, of which there are tokens that we produce to refer to different individuals, such as David Kaplan, David Lewis, etc. It is clear that this view does not suffer from objections to traditional Fregean or descriptivist views such as Kripke’s, precisely because it does not identify the proposition meant or asserted by utterances of sentences with Fregean thoughts, but with Russellian propositions. However, the view under consideration is still a Fregean view because it appeals to senses as essential determinants of the references of singular terms, presenting the referents in certain ways, individuating them, being known to competent speakers in virtue of their semantic competence, being part of what is understood in understanding speech acts, and referred to in certain linguistic environments such as contexts of indirect discourse and propositional attitude reports. In what remains of this chapter, I explain the view of singular terms I favor and defend in the rest of this work, which is also a version of the token-reflexive view and close to García-Carpintero’s. In doing this, I discuss Kaplan’s view of the individuation of words, because the view I defend –unlike García-Carpintero’s4– individuates proper names along the same lines as Kaplan’s view. 3.5. The Semantics and Metaphysics of Singular Terms It is common among philosophers to identify words, and proper names in particular, as types, the tokens of which we utter or inscribe when using them. The idea behind this view is that what makes a token of a word the token of the word it is, is its shape, spelling or sound. Thus, a certain token is a token of a certain word if it sounds, is shaped or is spelled in a certain way. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! 4 ! García-Carpintero tells me that he would also do it that way now. **! David Kaplan (1990) argues for an alternative view, according to which words (and proper names in particular) are not types but, like people, created natural objects that evolve and have an origin and a history that individuate them. Like people, words may change their properties: they may change their shape, the way they sound or the way they are spelled, and perhaps even their meaning or their reference. Kaplan does not commit himself to a conclusion on that last point. He does distinguish, though, between the possibility that a word changes its reference and the possibility that the word would have had a different reference than the one it had at the moment of its creation. But he leaves that question open. In Kaplan’s (1990) own words: It would be gratifying to be able to show that the process whereby a common currency name appears to change its referent involves these middling and conflicting intentions in such a way that when we are prepared to say that utterances of a name now have a new referent, there will have been sufficient weight given to the creative side of the intentions for us to claim that a new name has replaced the old. It might thus turn out that names don't in fact change in referent. So one strategy would be this. Try to show that something like Donnellan's referential use is involved in cases of apparent change of referent of a given common currency name. Then try to show that this sort of referential use is sufficiently like creating a new common currency name from the genus of a given name so that by the time a new 'semantic referent' appears, a new name does also. I don't know that such a strategy would be successful. (pp. 117-118) According to this view, words are like worms: their utterances and inscriptions are their stages. Words are neither abstract nor physical, but created natural artifacts which are physically and mentally embodied in their utterances and inscriptions. Stages of the same worm are causally linked in a certain specific way that conforms a causal-historical chain that goes back to the moment of creation; that is, the moment of the introduction of the word. Thus, according to Kaplan’s view, his name is a different name than, for instance, David Lewis’: the two are different created natural objects which were created at ! *+! different moments of time and which have evolved differently. In addition, there is yet another different generic name ‘David’ which is also a created artifact and has a different origin and history than the first two, but does not refer. I will follow Kaplan in calling the former kind of names ‘common currency names’ to distinguish them from the latter kind of generic names. Kaplan compares the relationship between generic names and common currency names with the relationship between species and specimens. Generic names are a distinct kind of words. Although he is not completely explicit about this, he seems to think of them as something we may call ‘super worms’, that is to say, worms that include those worms that are the corresponding common currency names, that consist of stages that are their utterances and inscriptions, plus those utterances and inscriptions of themselves that are not utterances or inscriptions of any common currency name. I agree with Kaplan in that we should distinguish between common currency names and generic names, individuating them in relation to their origin and history. Furthermore, my view is that not only the origin and history of proper names (and perhaps of words in general) individuate them, but their meaning and reference do as well. Meaning and reference are essential properties of proper names and so individuate them. There is no one proper name with more than one meaning or referent, the apparent change in meaning or reference of a proper name is only apparent. Once a token that is causally linked in the relevant way to a previous token of a proper name has a different meaning or reference than the previous one, it is a token of a different proper name. If that happens, a new proper name has been created and so introduced into the language. For such a thing to happen, the presence of creative intentions is not required: creative intentions are not a necessary condition for creation (as opposed to what Kaplan seems to think –see quote above). One example of this could be the case of the proper name ‘Madagascar’ discussed by Gareth Evans (1973). This is a case in which a name seems to have changed its reference: Marco Polo picked the name ‘Madagascar’ from natives who used it to refer to a part of the mainland, with the intention of using it in the same way as the natives, but used it instead to refer to the island just off the African continent, as we ! *#! do nowadays. It follows from my view that in this case a new name was introduced by mistake; that is, without the intervention of any creative intention. In any case, although my view is like Kaplan’s in certain important respects and he considers his to be a rejection of the type/token distinction as applied to words, I think these proposals are better thought of as modifications of the type/token model rather than a rejection of it. So, this third middle view about the individuation of proper names that I defend could perhaps be one that identifies proper names with types but distinguishes proper name-types with respect to their meaning, reference, origin and history as well as, and not only with respect to their sound, shape or spelling. Thus, tokens that sound the same or share their shape or spelling, but have a different meaning, reference, origin or history –like the common currency names used to refer to David Kaplan and David Lewis– would be different tokens of different types of proper names. Tokens of proper name-types individuated this way would be causally connected by a causal–historical chain of uses that originated at an act of baptism. The idea is that in the same way that ‘A’ is a token of the letter-type ‘a’ but also of the capital-letter-type, italic-letter-type and bold-letter-type, or even of the more specific capital-italic-bold-‘a’-letter-type; the token of the common currency name of David Kaplan, ‘David’, is a token of different types: the generic-name-‘David’-type, the common-currency-name-of-David-Kaplan-‘David’-type, the name-starting-with-a-‘D’type, or even the English-proper name-‘David’-type. This view I favor sees proper names as created types and since types are abstract, it sees proper names as created abstract objects; this might seem objectionable, since abstract objects are supposed by many not to be creatable because of their not being in space and time. However, this potential objection is clearly wrong since many abstract objects are created abstract artifacts, such as fictional characters, novels and musical works. In any case, if we can talk of tokens of proper name-types individuated not only by shape, spelling or sound, but also by meaning, reference, origin and history, and distinguish in ! +%! this way tokens of names that are homophones and homographic (and even synonyms) as tokens of different names, then we can think of the token-reflexive view extended to proper names as follows. Now here is the version of the extended token-reflexive view that I prefer: the conventional linguistic rule associated with each proper name-type N has the following form: (1) For any token of N, t, t refers to the individual named N. Correspondingly, for instance, the rule associated with the common-currency-name‘David’-type is: (2) For any token of ‘David’, t, t refers to the individual named ‘David’. As in the versions discussed above, proper names are mentioned at acts of “baptism” to start a practice of calling a certain individual with that name, but no token is used in those acts to refer. Acts of “baptism” establish naming practices this way. Once the practice exists, tokens of proper names are used to refer as indicated by the token-reflexive rule associated with the type: they refer to individuals who already bear that name thanks to the act of “baptism”. Tokens of proper names then express the Fregean sense expressed by the description ‘the individual named ‘David’’ –where the name ‘David’ mentioned in it is the token of the particular common-currency-name-‘David’-type in question, which is expressing the sense– and refer to the particular individual in question. Let me now make the differences between this view and García-Carpintero’s view described in the previous section more explicit. First, according to García-Carpintero’s view, proper names are much more like indexicals than they are according to my view. According to García-Carpintero’s view, the only difference between indexicals and ! +&! proper names is that the meaning of the latter is more stable than the meaning of the former. Also according to his view, but not mine, proper names are ambiguous expressions: there is only one proper name ‘David’, which like the ambiguous word ‘bank’, has different meanings, whose tokens refer to different individuals. As we will see in later chapters, this will have some important consequences. Secondly, notice that even if both accounts deny that singular terms are straightforwardly synonymous with definite descriptions that express their senses and, in that sense, on both views senses are not thought of as giving the meaning of singular terms, but as fixing their reference; both views hold that senses do give part of the meaning of singular terms. Finally, whereas according to García-Carpintero’s view, senses are constituents of presuppositions, my view remains neutral in this respect, seeing senses instead as something that is somehow indirectly expressed. Before concluding this chapter, let me add a few lines on the content expressed by utterances of more complex expressions: sentences. According to the Fregean view I adopt, utterances of sentences semantically express more than one proposition. And in the particular case of utterances of sentences that express Russellian singular propositions – that is, propositions which contain the individuals referred to by singular terms appearing in the sentences the utterances of which express them– they further express a descriptive proposition that consists of the senses the singular terms appearing in the sentences uttered express. In such cases, the Russellian singular proposition is the one that the utterances mainly or directly express, whereas the descriptive proposition is the one that they indirectly express. For instance, utterances of: (3) Hesperus is Phosphorus. mainly or directly express the following Russellian singular proposition: (4) <=, <Venus, Venus>> ! +'! but also indirectly express the descriptive proposition mainly or directly expressed by utterances of the following sentence: (5) The individual called ‘Hesperus’ is the individual called ‘Phosphorus’. where both definite descriptions are interpreted attributively along the lines of some sophisticated version of Russell’s theory of descriptions. Let me recapitulate. In Chapters 2 and 3, I have examined the semantics and metaphysics of singular terms and I have expounded the view of these matters that I prefer. In the next chapter, I focus on the semantics and metaphysics of propositional attitude ascriptions and of direct and indirect discourse. There, I defend a view of these more complex matters that bears on the view of singular terms I defend in this chapter. ! +(! 4. Demonstrative Discourse In this chapter, I discuss the semantics of propositional attitude ascriptions, indirect speech, ‘so-called’-sentences and quotation, as well as the informativeness and cognitive value of identity statements and simple sentences. I defend a view on all of these phenomena that explains them, as well as singular terms and fictional discourse, as manifestations of the same kind of linguistic mechanisms. These linguistic mechanisms have the following in common: they involve a demonstration; they contextually contribute (at least) part of the semantic content of expressions as something that appropriately relates to something that serves as demonstratum. I call this general view on language ‘The Demonstrative Discourse View’ (‘DD’, for short). I proceed as follows. In the first section, I present some basic problems and notions that are central in the literature about the semantics of propositional attitude attributions. In the next two sections, I offer my account of the metaphysics and semantics of propositional attitude reports. In the fourth section, I show how my view accounts for relevant data. In the fifth section, I extend the view to cover the phenomena of identity statements, simple sentences, indirect speech, ‘so-called’ sentences, and quotation. In the sixth section, I compare my approach to other approaches of the same family. Finally, in the last section, I answer some objections. Before I go on, let me make more explicit the role of this chapter within the whole dissertation. In this dissertation I defend a view on language that explains how many different linguistic expressions semantically work by means of one simple and common linguistic device, that of demonstration. In Chapter 3, I explain how this view that I defend applies to singular terms -after having discussed the relevant theories and arguments that exist in the literature on these expressions in Chapter 2. In the present chapter, I discuss how this view applies to several other, more complex, expressions, in which singular terms, in fact, occur. In Chapter 6 of the dissertation, I present and defend ! "#! how this view should apply to fictional discourse as well –after having discussed problems and theories that exist in the literature on fiction in Chapter 5. Note that singular terms are also an important part of fictional discourse. 4.1. Basic Problems and Notions In this section I explain some of the basic notions around the discussion of propositional attitude reports and some traditional problems that they pose. The principle of substitutivity salva veritate One of the traditional phenomena that theories of propositional attitude reports have to explain is that of the apparent failure of the principle of substitutivity salva veritate of coreferential expressions. This principle says that substituting one expression in a sentence for another that has the same reference cannot change its truth-value1. This principle seems to be well motivated and to follow from yet another plausible principle about language: the principle of compositionality2. However, in the context of propositional attitude reports, the principle seems to fail. To illustrate the phenomenon, consider the following pair of sentences: !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! 1 Notice that this principle is a principle about coreferentiality of terms rather than about merely apparent coreferentiality. This is so, even if the cases in which this principle seems to fail are explained by some theories, such as Frege’s, as cases in which it does not fail. According to these theories, contrarily to appearances, the principle does not fail, precisely because it does not even apply, given that these cases do not really involve coreferential terms. 2 Some think that in fact these cases of attitude attribution together with other cases like ‘socalled’ and quotation show the principle of substitutivity to be false. See Pietroski (1996), for instance, who argues that the principle of substitutivity does not follow from compositionality. I agree with this if we consider the principle as stated above –i.e. as usually stated. But I think, however, that there is some interesting truth behind that principle that is captured by a simply qualified version of it that I discuss later. ! "$! (1) Lois Lane knows that Superman is Superman. (2) Lois Lane knows that Clark Kent is Superman. Assuming of the fictional story of Superman that it is not fictional, utterances of (1) and (2) seem to have different truth-values –i.e. (1) is true while (2) is false. However, if the referents of ‘Superman’ and ‘Clark Kent’ in (1) and (2) are the same, as they seem to be, utterances of (1) and (2) should share truth-value according to the above principle of substitutivity salva veritate. Disquotation and translation principles Other puzzles seem to arise out of some other principles that we follow, such as the principles of disquotation and translation, and independently of the principle of substitutivity salva veritate3: Kripke’s (1979) puzzles of Pierre/London and Peter/Paderewsky. These other principles may be stated as follows (as Kripke does): Weak Disquotation Principle: If a normal speaker, on reflection, sincerely assents to ‘p’, then he believes that p –where ‘p’ is to be replaced by any appropriate standard English sentence4. Strong Disquotation Principle: A normal English speaker who is not reticent will be disposed to sincere reflective assent to ‘p’ if and only if he believes that p –where, again, ‘p’ is to be replaced by any appropriate standard English sentence. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! 3 This is controversial, though, since it is arguable that those puzzles arise out of only the substitutivity principle instead, as David Sosa (1996) argues. 4 In this as well as in the stronger version of this principle and in the principle of translation (below), I follow Kripke in ignoring complications that indexical or pronominal devices or ambiguities would create in order to get a clear and intuitive sense of the principle. ! ! "%! The biconditional form strengthens the simple one by adding that lack of disposition to assent indicates lack of belief, as assent indicates belief. Notice that similar principles could seem to hold for sincere affirmation or assertion in place of assent. Principle of Translation: If a sentence of one language expresses a truth in that language, then any translation of it into any other language also expresses a truth (in that other language)5. These principles are used by Kripke to generate the Pierre/London puzzle, which I explain in what follows. The Pierre/London case Pierre is a Frenchman, ignorant of English, who, after seeing some pictures of London in some French brochures, utters the French sentence (3) Londres est jolie. ‘Londres’ is the only name he knows of the city. Afterwards, he moves to an ugly suburban area of London to live and work there and starts learning English not by translation from French, his native language, -that is to say, not by assigning each English word or phrase the corresponding French word or phrase. He then utters the English sentence (4) London is not pretty. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! 5 As the other principles above, this one also needs qualifications due to indexicals and demonstratives that I am ignoring here to keep the principle clear and intuitive. ! "&! Pierre does not know that the names ‘Londres’ and ‘London’ name one and the same city. From his utterance of (3), and the weak disquotation principle for assertion, we would conclude in French (5) Pierre croit que Londres est jolie. By applying a principle of translation like the one above, we would conclude in English that (6) Pierre believes that London is pretty. But analogously, from Pierre’s utterance of (4) we would conclude (7) Pierre believes that London is not pretty. We would thereby attribute contradictory beliefs to Pierre. Finally, we would also conclude (9) from the strong disquotation principle alone and his lack of disposition to assent to (8). (8) London is pretty. (9) Pierre does not believe that London is pretty. We would thereby contradicting ourselves. The problems that arise are: first, that the case of Pierre is not clearly simply like the case of any irrational man that just has very simple contradictory beliefs, but our practices seem to force us to conclude that it is; secondly, that those same practices seem to force us to contradict ourselves in attributing beliefs to Pierre as well. Does Pierre believe that London is pretty? Ultimately, Kripke insists, this is the question to be answered. ! ""! The same principles are also used by Kripke to generate another puzzle, the Peter/Paderewsky puzzle, which I explain in what follows. The Peter/Paderewsky case Similar problems arise in a case where only English is involved. The Peter/Paderewsky case is similar to that of Pierre -or other cases in which there is an apparent failure of the substitutivity principle, such as the one of the Superman story considered above. The peculiarity of the Peter/Paderewsky case is that it seems to only involve one name, that is, it does not seem to involve two different coreferential names or one name and its translation6. The story goes somehow as follows: Paderewsky was both a famous Polish pianist and politician. Peter learns the name ‘Paderewsky’ as the name of a famous musician, to refer to that famous musician, and assents to (10) from which, together with the weak disquotation principle, we would infer (11). (10) Paderewsky had musical talent. (11) Peter believes that Paderewsky had musical talent. Later Peter learns about someone called ‘Paderewsky’ as a politician, who is in fact our famous Polish politician, and skeptical of the musical talent of politicians, Peter assents to (12), from which, again together with a weak disquotation principle, we would infer (13). (12) Paderewsky had no musical talent. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! 6 Here I say ‘seems’ in consideration of theories of propositional attitude reports such as Larson and Ludlow (1993) which judge this case to be one in which there are two homophonic coreferential names involved rather than just one. For a different account of the individuation of words, see Kaplan (1990), discussed in Chapter 3. In Chapter 3 I also explain the version that I favor. ! "'! (13) Peter believes that Paderewsky had no musical talent. We would thereby attribute contradictory beliefs to Peter. Now, however, since Peter is actually disposed to assent to (10) in certain occasions at least, we cannot clearly derive (14) as we inferred the analogous (9) before in the case of Pierre, by the strong disquotation principle alone and Pierre’s lack of disposition to assent to (8). (14) Peter does not believe that Paderewsky had musical talent. Pierre lacked the disposition to assent to (9), but Peter does not lack the disposition to assent to (10) at all. Indeed, he actually assents to it. This is so even if he at the same time assents to (12). But we could derive (14) from Peter’s assent to (12) and the weak or strong disquotation principle, provided that the subject is a speaker “with the modicum of logicality needed” -as the case presupposes-, and provided that this meant that any such a speaker would not hold simultaneously beliefs that are straightforward contradictions of each other –of the form ‘p’ and ‘~p’. In commenting about the principles, Kripke says: “The simple form [of the disquotational principle] can often be used as a test for disbelief, provided the subject is a speaker with the modicum logicality needed so that, at least after appropriate reflection, he does not hold simultaneously beliefs that are straightforward contradictions of each other –of the forms ‘p’ and ‘~p’. (Nothing in such a requirement prevents him from holding simultaneous beliefs that jointly entail a contradiction.) In this case (where ‘p’ may be replaced by any appropriate English sentence), the speaker’s assent to the negation of ‘p’ indicates not only his disbelief that p but also his failure to believe that p, using only the simple (unstrengthened) disquotational principle.” Kripke (1979), p. 114. ! '(! Since Kripke claims that the same problems arise in both cases and seems to assume just this, I take it that this is the way in which he thinks that we can derive (14) and create the important problem of having to contradict ourselves in the Peter/Paderewsky case as well. Unfortunately, he does not make the point explicit. As before, the problems that arise are first, that the case of Peter is not clearly simply like the case of an irrational man that just has very simple contradictory beliefs but our practices seem to force us to conclude that it is and secondly, that those same practices might seem to force us to contradict ourselves in attributing beliefs to Peter as well. Does Peter believe that Paderewsky had musical talent? Ultimately, Kripke insists, this is the question to be answered. In general, the question is: how can, if at all, a rational individual hold, without any logical inconsistency, simple contradictory beliefs after careful reflection? Summing up, we have been discussing some of the puzzles that any theory of propositional attitudes and their ascriptions has to deal with. Now, it is important to note that the three cases presented so far relevantly involve de dicto rather than de re attributions, a distinction that I discuss at length in the next section. But before getting to the details, let’s see that it is the de dicto rather than the de re reports that generate the puzzles. Consider the following de re attributions to Pierre, which would correspond to the de dicto attributions (6) and (9) involved in the Pierre/London case and considered above: (6*) Pierre believes of London that it is pretty. (9*) Pierre does not believe of London that it is pretty. Whereas the de dicto report (6) is intuitively false, the corresponding de re report (6*) is not –it is, in fact, intuitively true-, and so, the initial part of the puzzle does not arise. The initial part of the puzzle arises out of the intuitive falsehood of a report (i.e. (6)) that an otherwise intuitive theory would entail to be true. Analogously, whereas the de dicto report (9) is intuitively true, the corresponding de re report (9*) is not –it is, in fact, ! ')! intuitively false-, and so, the second part of the puzzle does not arise either. The second part of the puzzle arises out of the intuitive truth of a report (i.e. (9)) that an otherwise intuitive theory would entail to be false. The de dicto and de re distinction There are different distinctions being made by the terms ‘de dicto’ and ‘de re’. ‘de dicto’ and ‘de re’ are applied to propositional attitude attributions and sometimes to the propositional attitudes themselves as well. I use these terms to distinguish attributions and only to distinguish attitudes in a derivative sense: that is, to distinguish attitudes according to the kind of attribution that attributes them. These terms also apply to other cases such as modal attributions of the form it is necessary that S or it is possible that S, but I do not discuss modal attributions here. The following three distinctions are found in the literature: The syntactic distinction: An attribution is syntactically de re just in case it contains a pronoun or free variable within the scope of a propositional attitude verb that is anaphoric on or bound by a singular term or quantifier outside the scope of that verb. Otherwise, it is syntactically de dicto. According to this distinction, (15) is de dicto whereas (16) is de re: (15) Rachel wishes that there were good people in the world. (16) There are people in the world such that Rachel wishes that they were good. The semantic distinction: An attribution is semantically de re just in case it permits substitution of coreferential terms salva veritate. Otherwise, it is semantically de dicto. ! '*! According to common intuitions, (17) is, then, semantically de dicto whereas (18) is semantically de re: (17) Lois Lane doubts whether Clark Kent can fly. (18) Superman is such that Lois Lane doubts whether he can fly7. Let me introduce the notion of semantically semi de re attributions for those attributions that permit substitution of coreferential terms salva veritate with regards to some referential terms appearing in them but not all. According to the view that I propose later on in this chapter, (18) would be semantically semi de re, rather than de re. For even if (18) permits substitution salva veritate of coreferential terms for the proper name ‘Superman’ that occurs at the beginning, one cannot substitute any coreferential term for the indexical ‘he’ that occurs almost at the end salva veritate. For instance, I think that a report like ‘Superman is such that Lois Lane doubts whether Superman can fly’ -that results out of simply substituting ‘Superman’ for ‘he’ in (18)-, although perhaps not very natural, is false even if (18) is true. Depending on one’s semantic theory of propositional attitude attributions, one will think that the syntactic and the semantic distinctions coincide or not, as it will be clear later on. The metaphysical distinction: An attribution is metaphysically de re just in case it contains a directly referential term in the scope of the attitude verb. Otherwise, it is metaphysically de dicto.8 !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! 7 I use this example rather than ‘Clark Kent is such that Lois Lane doubts whether he can fly.’ precisely to illustrate the point that in these semantically de re reports substitution of coreferential terms is allowed. Both semantically de re reports are true. Also, if one perceives these reports to be syntactically forced, one might as well consider ‘Superman is doubted by Lois Lane as to whether he can fly.’ and ‘Clark Kent is doubted by Lois Lane as to whether he can fly.’ In using reports like (18), I follow Forbes (1987, 1990). 8 I borrow this terminology from McKay and Nelson (2000), whose views on the de dicto/de re distinction are discussed below. ! '+! It is harder to see which attributions are de re and which ones are de dicto in this sense, because it is controversial what terms are directly referential. In my opinion, the notion of direct reference needs some refinement. For there is a broad understanding of ‘contribution’, according to which a name, for instance, contributes something more than just its referent when it occurs in ‘so-called’ sentences such as ‘Giorgione is so-called because of his size.’, given that the presence of the name in such sentences indirectly affects the semantic content of the utterance. Nonetheless, proper names are supposed to be paradigmatic directly referential terms. In any case, I think that the third de dicto/de re distinction, the so-called ‘metaphysical distinction’, is most relevant to epistemological discussions rather than to semantic ones, and not so much relevant either to the metaphysical discussions that I discuss here. The de dicto/de re distinction that I use, therefore, is the syntactic one, unless otherwise noticed. In order to illustrate this later point, let me now briefly consider a particular discussion about this distinction by Thomas McKay and Michael Nelson (2000). McKay and Nelson discuss the third distinction in more detail -although they present it differently. They claim that because of issues having to do with this third distinction, views that reject what they call ‘Neo-Russellianism’ face a problem: the problem of having to explain why a report such as (19) does not entail a report such as (20). (19) Sally believes that the person at Disneyland is happy. (20) The person at Disneyland is such that Sally believes that he is happy. (19), they think, may be true without (20) being true. Their explanation of this lack of entailment consists in the fact that (19) does not require Sally being in the belief relation to a singular proposition, whereas this is precisely what (20) attributes to Sally. They claim that even if Bill uniquely satisfies the description ‘the person at Disneyland’, the case may be such that Bill himself does not have the property of being believed by Sally ! '#! to be happy. I do not share their intuition: I think that if Bill is the person at Disneyland, one cannot believe that the person at Disneyland is happy without believing that Bill is happy too. It seems to me that it all comes to an epistemic distinction of believing something about someone directly or indirectly: after all, if Bill happens to be the person at Disneyland, Sally’s belief is somehow (i.e. indirectly) about him. But, as I have said, the issue seems to me to be an epistemic one that can be (and maybe should be) captured in other ways. Be that as it may, the view that I defend is anyway capable of accommodating the lack of entailment from (19) to (20), if necessary. This epistemic distinction between general beliefs such as the one attributed by (19) and singular beliefs such as the one attributed by (20) can be captured by noticing that even if in that case Bill has the property of being believed by Sally to be happy, and thus, (20) is true as well, it could have been different compatibly with the truth of (19). That is, even if no one satisfied the definite description ‘the person at Disneyland’, and so, even if Sally’s belief were about nobody and hence, (20) were not true, (19) could have been true. All of this is compatible with the fact that if both (19) is true and Bill is the person at Disneyland, then it follows that Sally is in the belief relation to the singular proposition containing Bill himself. This epistemic distinction is not so important for the project at hand because it does not affect the truth conditions of propositional attitude reports at least in the following sense: even if one argues that Bill has the property of being believed by Sally to be happy (indirectly), Bill is not relevant to the truth conditions of (19) and that is already captured by the fact that (19) attributes to Sally a belief with a descriptive content, rather than a singular content with Bill as its constituent. Notice that nothing relevant for this discussion would be different even if the case considered was even more complex, as the one that follows. Consider the case being such that Sally believes that the person at Disneyland is happy because she erroneously ! '$! believes that John is the person at Disneyland and believes (correctly, let's suppose) that John is happy. Suppose that it so happens that it is Bill and not John who happens to be the person at Disneyland but that Sally is entirely unaware of Bill. Neither would this be a case in which Sally believed that the person at Disneyland is happy, but where she did not believe of Bill, who happens to be the person at Disneyland, that he is happy. Rather, this would be a case in which she would believe both things, in addition to wrongly believing that John is the person at Disneyland and correctly believing that John is happy. Of course there is something epistemically wrong going on in a case like this: that is, Sally mistakenly thinks that the person at Disneyland is John and because of this, she ends up believing something about someone she does not even know, i.e. Bill.9 Having explained some basic principles, notions and puzzles that are central in the propositional attitude attributions semantic debate, in the following two sections I present my own account of them. I start with the metaphysics in Section 2 and then I turn to the semantics in Section 3. 4.2. A Pragmatic Non-Millian View of Propositional Attitude Ascriptions: Metaphysics In this section and the following, I propose a view on the metaphysics and semantics of propositional attitudes and their ascriptions that I call ‘The Pragmatic Non-Millian View’ (‘PNM’, for short). This account shares some features with other existing accounts, but it is different in important respects (as I make more explicit in Section 6 of this chapter). !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! 9 ! I am aware that this is at odds with a widespread view, which I do not understand. '%! Propositional attitudes, propositions and mental representations One of the central ideas of PNM is a commitment to the existence of mental representations that play a crucial role in both the metaphysics of propositional attitudes and the semantics of their reports. Propositional attitudes are mental states such as beliefs, desires, thoughts, knowledge, hopes, doubts, and so on, that are characterized by essentially involving some representational content. In fact, propositional attitudes themselves are mental representations, that is, mental items that represent, and are, therefore, evaluable as true or false, satisfied or unsatisfied, and so on, depending on their content. For instance, a belief that the Earth is flat has as its content the representation that the Earth is flat (or, in other words, it is a mental representation of the Earth being flat), and since this representation is false, so is the belief. Likewise, a desire to live in a place with warm weather is a certain mental state with the content that one lives in a place with warm weather, so that if that content is false the desire is unsatisfied. The representational contents that these propositional states have are propositions, being this the reason why this specific set of mental states obtained their name. Propositional attitudes are, thus, mental states that relate individuals holding them to propositions that are the contents of their state. Propositions are representations of possible state of affairs10. It is worth noticing that propositions, or the contents of propositional attitudes, are not (or not in general) representations of representations, but rather, as I have just said, representations of possible state of affairs, which may or may not obtain. In this way, propositional attitudes such as beliefs represent possible state of affairs; they do it via their representational contents. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! 10 Here I am talking about mental states that are propositional attitudes, which at least include some beliefs and desires. It should be acknowledged, however, that there are mental states, including beliefs and desires that are not obviously propositional, such as the ones attributed by ‘Christians believe in God.’ and ‘Children always desire presents.’ In any case, it can be argued that mental states like these beliefs and desires are propositional attitudes too. The one attributed by the first sentence mentioned before being the belief that God exists, and the one attributed by the second sentence being the desire that they receive presents. ! '&! Propositions are abstract representational entities that are both the meaning of utterances of sentences as well as the content of mental states. For instance, again, a belief that the Earth is flat consists of a relation towards the proposition that the Earth is flat, which at the same time is the semantic content expressed by an utterance of the simple sentence ‘the Earth is flat.’ Or my desire to live in a place with warm weather is a relation to the proposition that I live in a place with warm weather, which at the same time is the literal content that my utterance of ‘I live in a place with warm weather.’ has. It is a controversial issue how much structure propositions have and what kind of things they are. Some think that they are sets of possible worlds without structure or constituents, while others assume that they are language-like structured entities with individuals and properties constituting them. These two conceptions do not exhaust the list, but are two of the most dominant views (See Chapter 2). I think that we can accept the existence of propositions as abstract entities, as far as they play a theoretical role in helping us explain the world and there is no reason against accepting their existence. The real questions are: which roles these different kinds of propositions play and whether all of them are needed in a final theory of anything. What distinguishes among the different propositional attitudes, on the other hand, is the role they play in explaining behavior, even if we take those mental states to be real causal dispositional properties somehow internally represented, rather than mere patterns of behavior. Following Stalnaker (1984), I take it that beliefs, for instance, are dispositions to act in ways that would tend to satisfy one’s desires in a world in which those beliefs are true, and desires are those dispositions to act in ways that would tend to bring it about what one desires in a world in which one’s beliefs were true. However, our mental states represent what they represent not only because of the behavior that they tend to cause, but also because of the events and states that tend to cause them. Propositional attitudes are internally represented in some form or other, be that as sentences in the language of thought or as pictures, maps, or none of those but something else. Some argue that the form that those representations take are not essential to their ! '"! content in the sense that this form is not a matter of what is represented, but of how it is represented. However, and contrary to some philosophers, I think that the form of those mental representations is useful and indispensable to characterizing the contents of at least certain attitudes such as beliefs and desires. They provide a clear way to determine the content of the attitudes; their precise content, more exactly. I am aware that the postulation of mental representations may look controversial to some philosophers. However, it is also true that many philosophers accept that some sort of mental representations constitute mental states. In particular, in the area of the semantics of propositional attitude ascriptions, for instance, Salmon (1989), Soames (1987), Saul (1998), and Braun (1998) follow Perry (1979)’s distinction between belief states and contents, where a belief state is some sort of way of believing some Russellian content. Salmon, Soames, and Saul defend a pragmatic view according to which these ways of believing do not enter into the semantic contents of belief reports, while Braun defends a non-pragmatic view according to which these ways of believing do not enter into the semantic content of belief reports either. On the other hand, Perry and Crimmins (1989) proposes a view on which these ways of believing captured in terms of notions and ideas do end up being part of the semantic content of belief reports. Furthermore, Richard (1989) argues for a view according to which mental representations partly constitute some propositions, which in some contexts are good translations of what an alleged believer is said to believe in a belief attribution. In any case, as I said, all of them accept the existence and participation of some sort of mental representations in our mental states. Although they may have more concrete ideas on the nature of mental representations, I think that all of them share the minimal idea that a mental representation is a semantically evaluable mental item, and it is this minimal conception that I assume in what follows. If at some points I speak as if mental representations are something more concrete, like mental photographs, for instance, that is because I think of them this way, but my view would be essentially the same if cognitive science showed us that the specific nature of these mental representations is quite different from the way I represent it. ! ''! I explain now the example that Perry (1979) uses to motivate his distinction between belief states and belief contents, as a mere illustration of the role that these mental representations are meant to play. This does not mean that the specific characteristics of the particular cases discussed by Perry in this paper are precisely the ones that could not be explained without appealing to mental representations. But this appeal seems natural to me and I see no reason not to do it, and the reason why I, as well as the philosophers mentioned above, appeal to mental representations in explaining mental states and our reports of them is precisely because we think that without this appeal no theory of mental states will succeed. Perry (1979)’s example of the messy shopper goes somehow as follows: John Perry11 is in a supermarket and starts following a shopper that is leaving a track of sugar after him every way he goes. He follows him because he believes that that shopper is making a mess and wants to alert him so that he can stop it. As it happens, the shopper who is making a mess is Perry himself, but he does not realize it. Finally, John Perry comes to realize that it is he himself who is making the mess and stops. Independently of what Perry himself thought about this scenario, in what follows, I offer my own account of it. It seems that at the beginning John Perry has a certain specific belief. He believed that a particular shopper was making a mess; that he, the shopper, that shopper, was making a mess. Perry would have expressed his belief uttering any of the following sentences (1) That shopper is making a mess. (2) He is making a mess. In spite of the fact that John Perry is that shopper, however, Perry would not have expressed his beliefs by uttering any of the following !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! 11 Notice that here, as well as in the rest of the discussion of this example, I am talking about John Perry as the character of the story, because it is useful to do it and he himself does it when presenting the case. In so doing, however, I am not ascribing any views on the matter to him. Rather, I offer the explanation that I think is correct. ! )((! (3) I am making a mess. (4) John Perry is making a mess. It seems wrong to use (5) to capture that belief of Perry even if (6) is true. (5) John Perry believes that he himself is making a mess. (6) John Perry believes of himself that he is making a mess. (5) is just false. PNM acknowledges the distinction that there is between belief reports that are semantically semi de re, but syntactically de dicto, like (6)12, and belief reports that are semantically de dicto, but syntactically de re, like (5). The representation with which Perry represents the shopper in his belief is relevant to determine this particular belief’s content and I think it has to do with the image Perry has of that shopper, the way Perry pictures that shopper, thinks of that shopper, conceives of that shopper, or what have you; and what goes about the shopper goes about the making the mess event as well. Perry’s mental representations of the shopper and of the event of that shopper making a mess are partly caused by the shopper himself and the particular event of him making a mess. Take R to be that complex representation. That is, the complex mental representation that Perry has of the event of that particular shopper making a mess, which is constituted in part by the mental representation that he has of the shopper. The proposition, content, of the belief Perry has at the start (that is, what he believed) may be characterized as follows !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! 12 I say here that (6) is syntactically de dicto because I consider that the indexical ‘he’ in (6) is not an anaphora, depending on the reference of the proper name ‘John Perry’ that appears before, even if as it happens, in the case under consideration, both expressions corefer. Anaphora, as I understand it, is not entailed by mere coreferentiality. ‘he’ refers to the same individual than ‘John Perry’ in (6), but could have referred to someone else. ! )()! (7) <R, <making a mess, John Perry>> Following Mark Richard (1990) I use the term ‘hybrid propositions’ to refer to propositions such as (7), which are like Russellian propositions, except that they are constituted by mental representations in addition to individuals, properties and relations. Now, when Perry realizes that it is he himself who is that shopper making the mess, he seems to acquire a new belief, a belief that he did not have before. That is, a belief that he would express now by uttering (3) and (4), and that we would truly report by uttering (5). PNM accounts for the acquisition of that belief and assigns it the following content (8) <R*, <making a mess, John Perry>> which is another hybrid proposition constituted by a different mental representation than the mental representation that constituted the hybrid proposition considered above (i.e. (7)). R*, in (8) above, stands for this other mental representation. Perry acquired a different conception of the event in the supermarket, an event that consisted in a particular individual, i.e. John Perry, making a mess. This new conception identifies his new acquired belief and so, we can easily attribute those distinct beliefs to Perry by appealing, referring to, making use of, those representations. Remember that belief states are mental representations, at least in the minimal sense of mental representation that is here assumed. So to the extent that Perry acquires a new belief when he comes to realize that he himself is the shopper making the mess, he acquires a new mental representation. According to Perry’s view, he comes to believe a proposition that he already believed, but by being in a different belief state. According to the view that I am presenting here, he comes to believe a new proposition, one that is partly constituted by the acquired new mental representation. The general important point of views like PNM is that the contents of attitudes are quite fine-grained. In Stalnakerian words, to identify certain attitudes or their contents, we need ! )(*! to partition the space of possibilities much more finely than by merely identifying the possible worlds in which those attitudes are true, or identifying the individuals and properties that these propositional attitudes are about. Admitting this general point, one might think, however, that it is not mental representations that helps to determine content more finely, but rather, the things that the attitudes are about as well as their aspects, parts, guises or appearances. For things have aspects or parts, and appear under certain guises or appearances after all. Moore (1999), Soames (1987), Salmon (1989) and Saul (1998), among others, appealed to such things at some point or other in the views they defended13. Since it is clear that things have these aspects or parts, and appear under certain guises or appearances, it might seem to some to be less objectionable to use them in place of mental representations, in order to identify propositional attitudes. One immediate possible problem with this, however, is that propositional attitudes are often about things such as the city of London or the superhero Superman rather than about their parts, aspects, guises or appearances. Correspondingly, their evaluation depends on the former things and not the latter. One way of seeing this clearly is by thinking of the properties usually attributed in the content of the attitudes: being pretty, being able to fly, etc. These are properties attributable to things such as cities and superheroes, not to parts, aspects, guises or appearances they may have (or, at the very least, when attributed to something they are not attributed to the parts, aspects, guises or appearances of that thing at the same time. For, of course, in some sense of ‘guises’, there are pretty guises). For the same exact reasons, it is not an option to consider that the constituents of content are things under their appearances or guises, or even things seen as consisting of certain parts or aspects, instead of the things or their appearances and guises simpliciter. Finally, the possession of propositional attitudes is not dependent on the way things present themselves to the subjects, in the following sense: one does not hold attitudes about an object only when it is presented under a certain way, but not when !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! 13 Soames, Salmon and Saul did not appeal to these things as something that determines the contents of beliefs, though, but as ways of believing (or parts of these). ! )(+! it is presented in another different way. The fact that an object is presented or considered under a certain way does not make attitudes about it temporarily appear or disappear. If, for instance, one believes of London that it is pretty because one possesses a belief representing London in a certain specific way, he or she believes this of London however or whenever the city is presented under any way. Moreover, in any case, and independently of these reasons, I am not sure that the acceptance of entities such as aspects, guises or appearances makes one’s ontology less objectionable than accepting mental representations. I, like many philosophers, believe mental representations to be perfectly acceptable. This reliance on metal representations is in my view justified on two main grounds. First, on an inference to the best explanation: I offer an account that appeals to mental representations because I take it to be the best explanation of the mental and linguistic phenomena that I want to explain, and being able to explain all of these phenomena in a unified and quite simple way counts as a very good argument in its favor for me. Second, on the fact that, in my view, there is nothing intrinsically unacceptable on mental representations. They are not, after all, Cartesian souls or Aristotelian final causes. They are entities perfectly amenable to scientific study, the nature of which can be further studied by cognitive science. Of course, I am aware that someone who were in principle opposed to mental representations would find my theoretical proposal unconvincing. I think the burden of the proof is in their side to offer a compelling argument against the acceptability of mental representation, but I fully acknowledge that if such an argument were provided, my proposal would be compromised. So, the more specific point that PNM makes is that the appropriate way of identifying these fine-grained contents is by means of the concrete mental representations that the subjects of the attitudes have and acquired in representing the world to themselves. Mental representations are the appropriate extra determiners of the content of propositional attitudes because propositional attitudes depend on the subject that holds them in a way that someone’s kicking a ball, for instance, does not depend on the subject of the kicking. An objective event, including the kicking of a ball by someone, is ! )(#! independent of the way in which any subject represents it, even if the event depends in an important sense on a subject, like the kicking of a ball depends on the subject doing the kicking. But a propositional attitude such as a belief is not so independent, as it is itself the subject’s representation of something else. In that sense, propositional attitudes are subjective and strongly depend on the subject’s point of view. Propositional attitudes are representations of the world and subjects cannot represent the world but from a certain perspective, from their point of view. These perspectives or points of views are often, in most cases, perspectives or points of views about something, which is objective and external to the mind. Note, furthermore, that the use of mental representations to determine content does not face any of the problems indicated above as problems that other possibilities such as parts, aspects, guises or appearances face. In the typical cases, propositional attitudes are still about external things such as cities and superheroes, rather than mental representations about them, and everything predicated inside an attitude is predicated of these things, rather than of mental representations that some individual may have of them. I think that it is ultimately crucial, in fact, that those representations bear some usually causal sort of (maybe distant, indirect) relation to the things in the world they are representations of, at least in those cases we have been considering. In the shopper’s example it is pretty clear that those representations are brought into existence out of the fact that Perry is looking at that man and the making of the mess. Other cases might be much more complex. On the other hand, it is important to note that these representations are not assumed to be part of any language of thought, or expressed by any definite description in particular, although they might end up being so. By committing to those representations and their role in determining the content of attitudes, PNM is not thereby committed to any of these. As pointed out above, the more specific nature of those representations seems irrelevant to the task at issue here. ! )($! Let me insist that, as noticed before, neither beliefs nor their contents or propositions (even if hybrid) are in general about mental representations. The fact that mental representations partly constitute the propositions that are the contents of beliefs does not make those beliefs or those contents to be about mental representations at all. Rather, they are about the things that these mental representations represent instead. Let me try to clarify this by contrasting the following: the belief that Lois Lane holds that Superman can fly, on the one hand, and my belief that the mental representation that Lois Lane has of Superman is partial, on the other. So Lois Lane believes the proposition (9) <R, <being able to fly, Superman>> and I believe that her mental representation of Superman is partial. Thus, I believe the following proposition (10) <R’, <being partial, R>> -where R is Lois Lane’s mental representation of Superman and R’ is my mental representation of her mental representation being partial. Multiplicity of propositions Fine-grained propositions such as (7) and (8), for instance, repeated below, are propositions that beliefs and other propositional attitudes are related to. (7) <R, <making a mess, John Perry>> (8) <R*, <making a mess, John Perry>> Nonetheless, this does not mean that propositional attitudes cannot be directed towards other less fine-grained propositions at the same time. Propositional attitudes are relations ! )(%! to fine-grained propositions, but also to those coarse-grained propositions that the finegrained propositions determine. For instance, in the case of the shopper, Perry is also belief-related to the following Russellian proposition (which (7) and (8) determine) by being in the belief relation to the propositions (7) and (8): (11) <making a mess, John Perry> Furthermore, Perry is also belief-related to the set of possible worlds at which the utterance of (4) (repeated below) is true. (4) John Perry is making a mess. Beliefs directed to these propositions can be said to be attributed in semantically semi de re -although syntactically de dicto- reports such as (12) and (13), in which the representation of Perry that is involved in the belief ascribed is not relevant at all. (12) John Perry is such that Perry believes that he is making a mess. (13) Perry believes of John Perry that he is making a mess. The fact is that those less fine-grained propositions do not suffice to identify the beliefs that Perry has at each moment in the story, or to identify the literal main semantic content of the semantically de dicto, although syntactically de re, attributions such as (5) above, repeated below. (5) John Perry believes that he himself is making a mess. However, nothing in the conception of propositional attitudes as relations to fine-grained propositions precludes them to be related to other propositions as well. In fact, it can be said that it requires it. ! )(&! It is worth insisting that on PNM propositional attitudes are binary relations between individuals and propositions. For instance, beliefs are binary relations between believers and objects of belief -i.e. what is believed. This is different from other views, which take propositional attitudes to be ternary relations to modes of presentation, or ways of believing as well. According to these other views, propositional attitudes would be ternary relations among individuals, propositions and these modes of presentation, instead. Intuitively, after all, the predicate 'believe' seems to be a dyadic predicate that refers to a binary relation and mental representations seem to determine what individuals who have them believe, not just the way they believe whatever is that they believe. I say that this is intuitive (intending to include not only my own intuitions) for three reasons. First, I think that the initial natural thought that philosophers and linguists have is also that belief is a binary relation instead of a ternary relation, and that the reasons that lead philosophers to think that belief is, rather, a ternary relation that holds among a believer, a proposition, and a way of believing are theoretical. That is, I think that they concluded this by an inference to the best explanation, rather than the other way around. In order to see why these common initial intuitions are natural, consider the following disanalogy: we cannot think of the event of cutting something, for instance, without conceiving it as a cutting with some tool or other (as a third relata of the cutting), but we can think of believing something without conceiving it as a believing in a certain way or other (as a third relata of the believing). Second, the predicate ‘believes’ seems to behave as a binary predicate on the surface. Third, we do start thinking of meaning and mental states as mental representations (as attested by how naturally internalist theories of language and mind like Locke’s are grasped by undergraduates14). In any case bear in mind that my claim here is just that this is the natural starting point, not a mere theoretical conclusion. Notice though that those ternary relations might exist as well. The point is that they are not the relations of belief, desire, doubt, or any other usual propositional attitude. Thus, for instance, Perry might be in the relation of belief to the proposition !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! 14 Notice that here I am talking only about what is natural and intuitive. I am not defending an internalist conception. The view I am offering in this dissertation is externalist. ! )("! (11) <making a mess, John Perry> and he might be said to be in that relation under a certain mode of presentation R. But the relation of belief is still merely between Perry and (11) and hence binary. This is so even if there exists this other ternary relation that holds among Perry, (11) and R, and has the binary relation of belief that holds between Perry and (11) as a component. We could call this relation ‘belief*’, rather than ‘belief’. In general, for any propositional attitude, there is a ternary relation that has that attitude as a constituent. These are not the relations attributed in the standard propositional attitude attributions, although perhaps they might be attributed by more complex attributions of one of the following forms: (14) A Vs that S under mode of presentation M. (15) A Vs that S in the way W. However, this is not a case in which it is clear whether the phrases ‘under mode of presentation M’ or ‘in the way W’ introduce a third relatum for a ternary relation, or rather, are adverbial modifiers instead (See Ludlow (1995) and Schiffer (1996)’s discussion about this). Another related, but different, issue is the semantic-pragmatic issue of what proposition is the one that gets literally expressed by what propositional attitude report and how. Since, as I have just said, propositional attitudes are distinguished by the role they play in explaining behavior, it is plausible to think that there are as many attitudes as needed to explain all behavior. Also, it is natural and reasonable to think that their propositional content is as fine-grained as needed to distinguish as many propositional attitudes as needed to explain all behavior. After all, we are aiming at the best possible explanation of all the relevant data. ! )('! This is ultimately why propositional attitudes seem to be relations not only to sets of possible worlds, but also to structured propositions such as Russellian propositions and, further, to hybrid propositions of the kind I advocate. The first step, i.e. to accept something like Russellian propositions, seems to be needed at least to distinguish between the behavior typically caused by the belief that Hesperus is Phosphorus, for instance, and the behavior typically caused by a simple mathematical belief such as the belief that 2+2=4. The final step, i.e. to accept something even more fine-grained such as the hybrid propositions, seems to be needed to distinguish among the behavior typically caused by propositional attitudes in puzzle cases such as the ones discussed at the very start, including linguistic behavior such as assenting to or affirming to certain utterances of sentences15. Let me now take stock. In this section I have discussed the metaphysics of propositional attitudes, propositions and mental representations. In the next section I present what PNM claims about propositional attitude attributions and their semantics. 4.3. A Pragmatic Non-Millian View of Propositional Attitude Ascriptions: Semantics Very briefly and basically, the proposal behind PNM is that propositional attitude attributions of the form: (1) A Vs that S semantically express that the individual named A stands in the relation expressed by Vs to the proposition referred to by the that-clause that S. That S refers to that hybrid proposition composed out of the referent of that and the main proposition expressed by S. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! 15 I am aware that there are Possible Worlds views such as David Lewis (1979) and Robert Stalnaker (2008) that try to explain all the relevant data by only appealing to sets of possible worlds, which I think are wrong but do not even consider here. ! ))(! The complementizer ‘that’, in these that-clauses, works as a demonstrative that refers to the mental representations which are made salient in the context with the help of the linguistic Fregean sense expressed by S. For example, the report ‘John believes that Mary is sleeping.’ semantically expresses that John stands in the binary belief relation to the proposition referred to by ‘that Mary is sleeping’. ‘that Mary is sleeping’ refers to the hybrid proposition that might be characterized as <R, <sleeping, Mary> -where R is the mental representation which is made salient in the context by the linguistic Fregean sense associated with the words ‘Mary is sleeping.’. This hybrid proposition is composed out of the referent of ‘that’ and the main proposition expressed by ‘Mary is sleeping.’. ‘That’ demonstratively refers to R in that context16. In more detail and using another example, consider the following belief ascription. (2) Pierre believes that London is pretty. According to the view that I am proposing, belief ascriptions like (2) semantically express both a descriptive proposition and a singular proposition, as it is generally the case for utterances of sentences which contain singular terms (as I explain in Chapter 3). That is, (2) expresses a descriptive proposition such as (3) and a singular proposition such as (4) below. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! 16 Jeffrey C. King (2007) has quite recently been discussing a problem called ‘the problem of the unity of propositions’, which he traces back to Russell. Some might think that PNM would have that problem. In particular, one might think that the problem arises specially in that PNM defends that the ‘that’-clauses of propositional attitude reports refer to propositions which are composed out of the propositions expressed by the embedded sentences and the mental representations referred to by the complementizers ‘that’. However, I do not think that PNM faces any such unity problem. In general, I think I follow the Tractatus in thinking that both things (including properties, relations, and mental representations), on the one hand, and linguistic expressions, on the other, are bound together by their very nature. That is, individuals have properties and bear relations to other individuals, partly because it is in their very nature to do such things. Individuals come with properties and relations, and vice versa. Likewise, proper names combine with predicates, whereas, for instance, a concatenation of proper names is not grammatically possible, partly because it is in the very nature of proper names and predicates to combine in these ways together. I think that it is useful to think about things and linguistic expressions as pieces of LEGO, which fit together by their very nature. No extra element, that is, no glue, is needed. ! )))! (3) The proposition that the individual named ‘Pierre’ believes that the individual named ‘London’ is pretty. (4) <believing, <Pierre, <R, <being pretty, London>>>> -where R stands for a mental representationHowever, (2) is true if and only if Pierre bears the binary relation of belief towards the hybrid proposition <R, <being pretty, London>>. Let me now insist on five main ideas of this view. First, remember that the proper name ‘Pierre’ refers to the individual Pierre and expresses the linguistic sense expressed by the description ‘the individual named ‘Pierre’’ (and proper names are individuated as explained in Chapter 3). Second, the predicate ‘believes’ refers to the binary relation of belief that holds between believers and structured propositions. Third, the embedded sentence ‘London is pretty.’ expresses two propositions as well, 1) a descriptive proposition such as the proposition that the individual named ‘London’ is pretty, and 2) the singular proposition <being pretty, London>. Fourth, the expression ‘that’ in the ‘that’-clause ‘that London is pretty’ functions as a demonstrative referring to those representations made salient in each context partly because of the linguistic sense expressed by the embedded sentence in the ‘that’-clause. The expression ‘that’ in the ‘that’-clause ‘that London is pretty’, when embedded in the belief ascription (2), for instance, refers to a representation that Pierre has of London, i.e, R, a psychological Fregean sense, as one might call it. Finally, the ‘that’-clause ‘that London is pretty’ behaves as a singular term that refers to a proposition, the singular more fine-grained proposition <R, <being pretty, London>>. Thus, again, if it is true that Pierre believes that London is pretty, then Pierre bears the belief relation towards the proposition <R, <being pretty, London>> -where R stands for a mental representation that Pierre has of London, and which is partly determined by or associated with the name ‘London’. Note, however, that if it is true that Pierre believes that London is pretty, then Pierre also bears the belief relation towards the more coarsegrained proposition <being pretty, London> and if it is true that Pierre believes that ! ))*! London is pretty, then Pierre may not bear the belief relation towards the another finegrained proposition <R*, <being pretty, London>> -where R* stands for another mental representation that Pierre may have of London, and which may be partly determined by another name or token; R* ! R. This way the theory makes the following predictions about the ascriptions below, for instance. (2) Pierre believes that London is pretty. (5) Pierre creu que Londres és bonica. (6) London is such that Pierre believes it to be pretty. (7) Pierre believes of London that it is pretty. (5) may be false even if (2) is true, and vice versa. But (6) and (7) cannot be false if (2) or (5) are true. (2) or (5) do not follow from a report like (6) or (7), or the fact that Pierre is belief-related towards the proposition <being pretty, London>. The account I propose appeals to Fregean senses. However, I propose that the Fregean notion of sense should be split into two distinct notions: linguistic senses and psychological senses (This is a proposal that I have learned other philosophers have made too, such as Recanati). Each of these notions plays one of the roles that are typically associated with Fregean senses. Linguistic senses are something like the semantic contents of definite descriptions conventionally associated with linguistic expressions, though not synonymous with them. Psychological senses are mental representations, which can also be associated with linguistic expressions. Both linguistic and psychological senses may be thought of as modes of presentation. Psychological senses are mental representations that often come into existence by causal interaction with the things they are about in the world. As it is clear from our discussion in previous chapters, the appeal to Fregean senses does not prevent singular terms such as proper names and indexicals from being rigid designators. To be a rigid designator is to be a designator that refers to the same thing at ! ))+! every possible world at which that thing exists, and to nothing at those worlds at which it does not, and singular terms do this in addition to expressing a Fregean sense. Moreover, in simple sentences Direct Reference holds for referential expressions (understanding Direct Reference as the thesis that the only contribution of any singular term to the proposition mainly expressed by the simple sentences in which it occurs is its referent). Taking as non-simple sentences cases such as ‘so-called’ sentences or sentences involving quotation, in which there is an expression whose semantic content depends on the referential expressions occurring in them, so that the later may be said to also contribute to that. Pragmatic explanation Finally, in the remainder of this section I explain what the role of pragmatic explanations is within PNM. One of the classical objections to Fregean theories of propositional attitude reports (see, for instance, Crimmins (2002)) is that they can only explain the ones that are semantically de dicto, but that there are others that are intuitively semantically de re, and they are not just a few. For instance, even about the Superman story one finds intuitively semantically de re attributions made, or that could be made, by the use of sentences that are considered to be semantically de dicto. In contexts where it is clear to all participants in the conversation that Lois Lane does not realize that the man called ‘Superman’ is the same man that is called ‘Clark Kent’, or that she does not know that those two names are coreferential, and it is common knowledge among participants that all of them take this to be clear and hence this fact about Lois Lane is presupposed, a participant (even myself) could utter (8) to successfully convey the intuitively semantically de re true attribution that Lois Lane desires of the man called both ‘Clark Kent’ and ‘Superman’ that he marry her, ! ))#! independently of the mental representation that more finely determines the content of her desire. (8) Lois Lane desires that Clark Kent marry her. Contextualist views of propositional attitude attributions (Richard (1990), Crimmins and Perry (1989), Ludlow and Larson (1993)) aim at capturing both kinds of intuitions (i.e. semantically de re and de dicto) by making the literal content of those attributions vary according to context in a way that corresponds to these intuitions. My view is that at least in some cases, syntactically de re or semantically semi de re attributions get to be communicated by attributions syntactically and semantically de dicto merely via pragmatics. In particular, sometimes when this happens it is due to Gricean conversational implicatures. The literal content of attributions does not vary as much as the contextualists would have it, because the mental representation that is made salient in a given context is made salient by the help of the property (i.e. linguistic sense) expressed by the terms used in the attribution, even if this is made contextually. Those terms restrict the possibilities of reference. Let me now explain what Gricean conversational implicatures are. Conversations are rational activities governed by certain maxims. Sometimes we utter sentences to convey propositions that are different from the ones literally expressed by them. Gricean conversational implicatures are those propositions that are pragmatically conveyed in a context due to the violation of some maxim of conversation. They clearly depend on the context of utterance. Different utterances of one and the same sentence may convey different conversational implicatures. These implicatures have two properties that identify them: they are calculable and cancelable. For an implicature to be calculable is for it to follow via an accessible (to the participants) inference from the literal content of the sentence uttered and the violation of some maxim of conversation. For an implicature to be cancelable is for it to be such that the conjunction of its negation and the literal content expressed by the sentence uttered is not contradictory. If a proposition p is conversationally implicated by an utterance of a sentence S whose literal content is q, in a ! ))$! context C, q and not-p should not be a contradiction. That is, it should be possible to express the negation of p after uttering S without contradicting oneself. Among the maxims of conversation, there is, for instance, the maxim of quality (do not say what you believe to be false) and the maxim of relation (be relevant). A clear example of a conversational implicature is the one conveyed by an utterance of (9) made by a waiter in a restaurant intending to mean what an utterance of (10) literally expresses. (9) The ham sandwich left without paying. (10) The person who recently had a ham sandwich in that restaurant left without paying. The literal contents of utterances of (10) and (9) are different, but the literal content of an utterance of (10) is conversationally implicated by the utterance of (9) somehow as follows: among other things the waiter literally said that a sandwich left the restaurant. But this cannot be. Sandwiches do not have any such capacity and hence, what the waiter literally said is obviously false. It is clear that the waiter knows this. So, since the waiter is cooperative and realizes that by uttering (9) he is violating one of the maxims of conversation –i.e. the maxim of quality (do not say what is false), he must have meant something else. What the waiter meant is what an utterance of (10) literally expresses. This is the way in which the content of an utterance of (10) can be calculated in that context. Right after uttering (9), one does not contradict oneself by adding something such as (11) But no one had a ham sandwich. In other words, an utterance of (9) in conjunction with an utterance of (11), or some negation of (10), does not express a contradiction. Therefore, the content of an utterance of (10) is cancelable as well. It is important to note that one does not need to process the derivation of the implicature in an explicit way. On the contrary, this is often done tacitly. ! ))%! Let me return now to the Pierre/London case and the particular French translation (13) of (12) below. (12) Pierre believes that London is not pretty. (13) Pierre croit que Londres n’est pas jolie. If an utterance of (12) were true, but an utterance of (13) were false, as it seems to be according to the case that Kripke introduced, an utterance of (13) could be used to convey a true conversational implicature such as the syntactically de re or semantically semi de re French belief report (14), which is translated into English as (15), by violating one of the maxims of quality (do not say what is false), in contexts where it is clear that what is uttered is a translation of an attribution in another language. (14) Pierre croit de Londres que c’est n’est pas jolie. (15) Pierre believes of London that it is not pretty. That is to say, contexts where it is clear that, as it is usual in the practice of translating between languages, something may have got lost in the translation -such as the precise way in which Pierre would have expressed his belief and, thereby, the precise way in which he believes of the city of London that it is not pretty. Another case in which this could occur would also be one in which what matters in the conversation is to get across what city is the one of which Pierre believes that it is pretty, and nothing more, to someone who does not speak English, but French, and has never heard of the English name ‘London’. Indeed, many other cases like these could be described. Consider again one of our previous examples. As I said, sometimes one (including myself) may use a false report like (8) to convey what a report like (16) or (17), instead of (8), semantically expresses; that is to say, to report that Lois Lane is desire-related towards the proposition (18) without caring about the mental representation that better – more precisely- determines Lois Lane’s desire. ! ))&! (8) Lois Lane desires that Clark Kent marry her. (16) Clark Kent is such that Lois Lane desires that he marry her. (17) Lois Lane desires of Clark Kent that he marry her. (18) <marry, <Superman, Lois Lane>> We utter (8) instead of (16) or (17) due to reasons of economy or convenience -as it happens when certain expressions are more available and ready, or better suited, to be used than others. In uttering (8), we may manage to convey this other information in some contexts and we do it, but pragmatically. We do it by means of Gricean mechanisms: by conversational implicatures. In uttering (8), one (including myself) may manage to convey a conversational implicature similar to (19) or even the simpler (20) below. (19) and (20), like any other conversational implicature, is calculable and cancelable. (19) Lois Lane desires of Superman that he marry her and what the details of her desire other than that are (such as the mental representation involved in what she desires, or the precise proposition that she desires) is not relevant for the purposes of what I intend to communicate at that moment by uttering (8). (20) Lois Lane desires of Superman that he marry her. Notice, though, that I am not here defending that the conversational implicature conveyed by an utterance of (8) needs to be exactly the literal content expressed by an utterance of (19) or (20). Rather, I am defending that it must be something similar. That is, what I am saying is that in uttering (8), one may convey a conversational implicature that does not commit the speaker to the fact that the exact propositional attitude reported is the exact one that the subject to which it is attributed holds, or, in other words, that does not commit the speaker to her report being accurate. ! ))"! Notice also that there are cases in which one (including myself) may utter (8) without implicating anything, and even if knowing that no implicature will be understood by the hearers, even if one thinks that one is thereby asserting something literally false. These are cases in which one asserts something literally false in order to communicate something true, realizing that the false information transmitted is not important at all in those cases. In short, what I am arguing is that accuracy is often not the most important goal in a conversation and that this fact is what accounts for the intuitively semantically de re attributions. Let me finish this section by discussing a potential worry that one might have about the comparison of cases like the ham sandwich explained above and the intuitively semantically de re reports. The worry is that whereas the former cases are clear cases in which the Gricean explanation applies, the latter cases are not. My reply to this worry is as follows: first, I acknowledge that the former cases are paradigmatic cases of the phenomenon that the Gricean explanation explains, but that the latter cases are not. This is precisely the reason why these are the cases that are used in presenting that explanation. However, this does not mean that all the cases to which the Gricean explanation applies are paradigmatic as well. Second, notice that the common intuitions are that a report such as (8) is false, rather than true, and that this is widely acknowledged, even by philosophers who contrary to intuitions have defended that (8) is true (philosophers such as Salmon, Saul, Soames, and Braun). Thus, it is not surprising that if these reports, which are thought to be clearly false by competent speakers, are still used by them, it is, as PNM claims, to communicate something other than its exact or entire literal content. Third, whereas the intuitions that reports such as (8) are false are intuitions that arise out of explicit consideration of the truth-value of these reports by competent speakers, and include pretheoretical intuitions, I think that the observation that we, speakers of a language, often use syntactically de dicto reports to communicate what a semantically de re report would literally express does not arise out of explicit consideration of the truth-value of these reports, but out of observing what reports we use ! ))'! in what circumstances. PNM does not deny this observation. Quite the contrary, PNM fully endorses it. Moreover, I think that if there are any intuitions of these reports being de re, these are not pretheoretical, unlike the intuitions that indicate that reports like these are (or at least are used as) semantically de dicto indeed. Finally, I may only add that I often observe reactions in the speakers of the kind that PNM, but not its opponents, predicts. I mean linguistic reactions that would indicate and confirm that it is a phenomenon like the one PNM describes that is going on. 4.4. Dealing with Traditional Problems Having presented PNM, I now defend some of its virtues. I mainly show how the view deals with some traditional problems about propositional attitude reports. One of these traditional problems is that of the apparent failure of the principle of substitutivity salva veritate of coreferential expressions (i.e. the principle that substituting one expression in a sentence for another that has the same reference cannot change its truth-value). For example, consider the following pair of sentences: (1) Lois Lane knows that Superman is Superman. (2) Lois Lane knows that Clark Kent is Superman. Assuming that the fictional story of Superman is not fictional, utterances of (1) and (2) seem to have different truth-values –i.e. (1) is true while (2) is false. However, if the referents of ‘Superman’ and ‘Clark Kent’ in (1) and (2) are the same, utterances of (1) and (2) should share truth-value according to the principle of substitutivity salva veritate of coreferential terms. PNM successfully deals with this problem. Although the stated principle of substitutivity salva veritate of coreferential terms fails in cases like (1) and (2), PNM provides a simple ! )*(! explanation of its failure that is compatible with a plausible modified version of the principle. According to PNM, the ‘that’ in the ‘that’-clauses of (1) and (2) does not have the same referent. The ‘that’ in (1) refers to a mental representation partly made salient by the Fregean sense associated with the name ‘Superman’, while that in (2) refers to a mental representation partly made salient in the context by the Fregean sense associated with the name ‘Clark Kent’ as well. These representations are different according to the story and this is why utterances of (1) and (2) differ in content and truth-value even though ‘Superman’ and ‘Clark Kent’ are coreferential terms. This is compatible with the following plausible version of the principle of substitutivity salva veritate of coreferential terms: the principle that substituting one expression in a sentence for another that has the same reference cannot change its truth-value, unless there is any other variation in reference in that sentence. To illustrate this version of the principle, consider the ‘so-called’ sentences discussed by Quine: (3) Giorgione is so-called because of his size (4) Barbarelli is so-called because of his size Since the name ‘Giorgione’, in addition to being a proper name, means big George, and hence, means something relevant to sizes, but ‘Barbarelli’ does not, (3) is true and (4) false. This is so, even if Giorgione and Barbarelli are one and the same person: the italian painter Giorgio Barbarelli da Castelfranco. Sentences with the ‘so-called’ expression are a clear case of opacity. The so-called case is a case in which substitution of coreferential names does not preserve truth. However, as in the case of propositional attitude attributions, in these cases, we also have contextual reference being made which easily explains the failure of the simple and unmodified substitutivity principle. In particular, we have a clear context-dependent expression ‘so’ whose referent varies in context depending on the proper name used to refer to the person ! )*)! that these sentences are about. In fact, ‘so’ refers to the proper name used. Thus, notice that whereas these sentences show that the simple version of the substitutivity principle fails, they are compatible with the proposed modified version of the principle. Let me explain now how PNM fares with respect to the puzzles advanced by Kripke: the Pierre/London puzzle and the Peter/Paderewsky puzzle. PNM can handle those puzzles too. For it makes the right predictions about the following reports. (5) Pierre believes that London is pretty. (6) Pierre croit que Londres est jolie. (7) Pierre believes that Londres is pretty. (8) Pierre believes that London is not pretty. (9) Pierre does not believe that London is pretty. (10) Peter believes that Paderewsky is musically talented. (11) Peter believes that Paderewsky is not musically talented. (12) Peter does not believe that Paderewsky is musically talented. According to PNM, in the story of Pierre, (5) is false while (6)-(9) are all true, and in the story of Peter, (10)-(12) can all be true due to the presence of the different modes of presentation involved in that story. That is, the ‘that’ in (10) may refer to a different mental representation than the ‘that’ in (11) and (12), corresponding to the different mental representations that Peter has of the Polish politician and musician. Theories appealing to mental representations have to face additional problems having to do with whose mental representation gets included in the content of these reports, whether the mental representation of the reporter or of the individuals to whom propositional attitudes were ascribed. PNM has it that the reference is to mental representations of the later and may not always be description-like. The only definite descriptions needed to understand these reports are those conventionally associated with the linguistic expressions used and thus, the knowledge of which is already required by ! )**! the linguistic competence. Distinguishing between linguistic and psychological senses dissolves any possible conflict between the subjectivity of the modes of presentation and the objectivity of language. It might be useful to consider this other case which I am told is due to Erin Eaker: Omar is an Iraqui who is deceived by the Iraqui government to believe that Saddam’s double is the real Saddam. One might think that in that case an utterance of the following report is true: (13) Omar believes that Saddam’s double is the real Saddam. In addition, one might think that this poses a problem for PNM because this is a case in which Omar would never assent to the following sentence: (14) Saddam’s double is the real Saddam. However, I think this case does not pose any problem for PNM and that it can be dealt with as follows. First, notice that ‘Saddam’s double’ is, in fact, the definite description ‘the double of Saddam’ and, hence, according to PNM and theories that accept Russell’s theory of definite descriptions (or a sophisticated version of it), (13) is ambiguous between a syntactically de re and a syntactically de dicto interpretations that arise due to the interaction of ‘believes’ and the determiner ‘the’. The de re reading of (13) is (15) There is a unique double of Saddam and Omar believes that it is Saddam. This reading is true. The de dicto reading of (13) is (16) Omar believes that there is a unique double of Saddam who is Saddam. This reading is false. Second, notice that according to the description of the case, Omar does have a representation of Saddam’s double, even if Omar ignores that he is merely a ! )*+! double. This is precisely the mental representation that Omar has of the individual in question that would be contributed to the proposition which is truly said to be believed by him. In addition to solving the traditional problems discussed above, PNM has other advantages. In the first place, the view does not become implausible when indexicals get involved. Consider, for instance, (17) Peter believes that he is not musically talented. (where ‘he’ refers to Paderewsky) (17) is false, when uttered in a context where it is the mental representation that Peter has of Paderewsky as the musician which is salient, but true, when uttered in a context where the representation which is made salient is that one that Peter has of Paderewsky as the Polish politician. In the second place, PNM does not have any problem in accounting for attributions to more than one person. Consider, as an example, (18) Many people believe that London is pretty. In those cases it might be that either the ‘that’ might not get a referent because the context makes it clear that there is no particular mental representation that is salient, or it may refer to some general enough representation that many people share by having tokens of the same type. Notice that the first possibility would not be a problem for PNM precisely because according to PNM, the proposition referred to by the ‘that’-clause is always a proposition composed out of the mental representation contextually referred to by the complementizer ‘that’ and the proposition semantically expressed by the embedded sentence. In a case (if any) in which the ‘that’ does not refer to any mental representation, perhaps because there is none that is made salient in the context, there is a proposition available to be referred to by the ‘that’-clause in question. Indeed, in a case like this, it would be natural that this is the proposition so referred and said to be ! )*#! believed. Notice too that this case would not be a case in which the fact that an expression does not refer affects the reference and truth-value of the more complex expressions in which it occurs. If this were so, it would be a problem. However, it is not. However, according to PNM, there is reference to some general mental representationtype in cases of attribution of an attitude to many people. This must be the case, for consider: (19) Many people believe that Superman can fly, whereas many do not believe that Clark Kent can fly. Also, consider: (20) Every student thinks that she is the best student in the class. A report like (20) also requires PNM to acknowledge reference to some general mental representation-type. It is clear that a report like (20) does not make any reference to the specific characteristic mental representation involved in the thought that each student is said to have. Notice that there are propositional attitude reports in which the expression that makes a difference is not a singular term, but a common noun. These have also been extensively discussed in the literature. Consider, for instance: (21) Pablo believes that lawyers are selfish. (22) Pablo believes that attorneys are selfish. In my view, these cases are less clear. Unlike the cases in which distinct corefering proper names are involved, these cases seem to involve some sort of deficient linguistic competence with the words. To know the meaning of ‘lawyer’ and ‘attorney’ requires to know that they are synonymous. In any case, PNM could treat these other cases as well. ! )*$! Finally, notice that PNM does not require us to abandon the plausibility or simplicity of predicates such as ‘believe’ being binary, or referring to binary relations, or that the meaning of them does not change according to context, nor has it to appeal to unarticulated constituents. PNM does not appeal to ambiguities either. 4.5. Other Related Phenomena PNM is a proposal about propositional attitude attributions that has the virtue of constituting a part of a general, simple, elegant and unified account of many phenomena in language –which I called above ‘The Demonstrative Discourse View’ (‘DD’, for short). DD covers phenomena such as proper names and indexicals, indirect speech, quotation, ‘so-called’ sentences, the informativeness and cognitive value of identity statements and simple sentences, and the meaningfulness of statements with empty names and fictional discourse in general. This general account explains these phenomena, in part, by appealing to the linguistic mechanism of demonstration and, thus, construes them as parts of a demonstrative discourse. It is for this reason important to present PNM together with the other accounts for the other specific phenomena. I now present how DD accounts for the rest of those phenomena case by case, except for the case of fiction that I discuss more deeply in the next two chapters. Identity statements Consider the following pair of statements: (1) Hesperus is Hesperus (2) Hesperus is Phosphorus ! )*%! My view is that although utterances of (1) and (2) have the same literal main content, they both semantically express the proposition characterized by (3). (3) <=, <Venus, Venus>> The informativeness of an utterance of (2) is explained by the additional descriptive proposition it expresses. That is, in addition to the proposition characterized by (3), an utterance of (1) also expresses the proposition indicated by (4), whereas an utterance of (2) additionally expresses the proposition indicated by (5). (4) That the individual called ‘Hesperus’ is the individual called ‘Hesperus’. (5) That the individual called ‘Hesperus’ is the individual called ‘Phosphorus’. So the proposition indicated by (5) is the informative content of an utterance of (2). On the other hand, an utterance of (2) can also pragmatically convey a proposition that (5) helps determine such as the one indicated by (6) in certain contexts. (6) That the morning star is the evening star. Those contexts in which participants of the conversation associate the names ‘Hesperus’ and ‘Phosphorus’ with the definite descriptions ‘the morning star’ and ‘the evening star’, respectively. Thus, the proposition characterized by (6) will be informative content pragmatically conveyed by an utterance of (2) in those contexts too. ! )*&! Simple sentences The case of other simple sentences that seem to differ in their informativeness and cognitive value as well, which Jennifer Saul (1997) brought to our attention, is similar to the case of identity statements just explained. Consider the variation in informativeness of the following groups of sentences: (7) Clark Kent goes into a phone booth and Clark Kent goes out (8) Clark Kent goes into a phone booth and Superman goes out or (9) Superman is much sexier than Superman (10) Superman is much sexier than Clark Kent In my view, utterances of (7) and (8) semantically express the same main singular proposition, i.e. something like the proposition characterized by (11), and thereby, share truth value, but semantically express distinct additional descriptive propositions as well, i.e. something like the proposition mainly expressed by utterances of (12) and (13), respectively. (11) <and, <to go into, <Superman, phone booth>>, <to go out, <Superman, phone booth>>> (12) The individual named ‘Clark Kent’ goes into a phone booth and the individual named ‘Clark Kent’ goes out. (13) The individual named ‘Clark Kent’ goes into a phone booth and the individual named ‘Superman’ goes out. ! )*"! It is my view that utterances of (7) and (8) differ in their cognitive value and informativeness because of their additional content: the propositions indicated by (12) and (13), respectively again. Now. There might be something extra to account for in these last sorts of cases and that is that some competent speakers of English who well know the identity17 of the reporter and superhero might, in spite of this, very firmly judge utterances of pairs of sentences such as (7) and (8), and (9) and (10), to differ in truth value. Since these intuitions would go against the semantic view defended, they should be explained away. This is a point that Jennifer Saul made in bringing out these cases to our attention. DD explains them away by appealing to very simple Gricean mechanisms here as well. Competent speakers of English who are well connoisseurs of the identity judge utterances of these pairs to differ in truth-value in contexts where they convey different conversational implicatures. The conversational maxim that gets violated in these cases is the maxim that says be as informative as you can. The Gricean explanation of the intuitions goes as follows: competent speakers of English who know the identity of Superman would judge an utterance of (7) to be false while an utterance of (8) to be true, even if they express the same main semantic content. This is because they mistakenly take the proposition, which is only pragmatically and conversationally implicated by utterances of them to be their main semantic content instead. The additional proposition that these utterances semantically express (i.e. (12) and (13), respectively) actually helps determine these conversational implicatures conveyed by them, which are, respectively, something like (14) Clark Kent goes dressed as a reporter into the phone booth to afterwards go out dressed as a reporter as well. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! 17 Of course, someone who is unaware of the relevant identities can perfectly well mistake the semantic content and truth-value of utterances of sentences like these as well as identity statements, even if she is competent in English. But these other cases are explainable just because of that: this person just ignores the coreferentiality of the names. ! )*'! (15) Clark Kent goes dressed as a reporter into the phone booth and afterwards he goes out dressed as a superhero. Utterances of (12) and (13) help determine the main semantic content expressed by utterances of (14) and (15), respectively, precisely because people associate with the names ‘Clark Kent’ and ‘Superman’ the different ways of dressing even if they know that Superman and Clark Kent are one and the same individual. The calculation of the implicature goes -participants in the conversation reason- as follows: the speaker could have used either of the two coreferential names in each of these places in the sentences and it is common ground that with these two different names there are associated different modes of presentation of their referent. So in using ‘Clark Kent’ rather than ‘Superman’, or vice versa, she must have meant something relevant about the mode of presentation of the individual referred to by these names. Otherwise, she would be violating the maxim be as informative as you can by not being as informative as she could in uttering (7) instead of (8) –for she would have been more informative in uttering (8) instead-, or she would be violating the maxim do not say what you believe to be false by suggesting something false about the referent’s appearance in the situation that the speaker is talking about when she utters (8) instead of (7). Importantly, however, this pragmatic explanation suffices to make the competent speakers accommodate their intuitions about the truth-values of simple sentences. This does not seem to be so with respect to attempts to do something analogous in cases of propositional attitude attributions18. Jennifer Saul argued that since the right way to go to account for the intuitions in cases of simple sentences is by appealing to pragmatics, it could be that the right way to go to account for the intuitions in propositional attitude reports is pragmatic as well, rather than semantic. But as I just said, there is an important difference that distinguishes both types of cases and justifies their different treatment. After all, it makes sense that modes of presentation affect the content and truth-value of !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! 18 I am referring to the pragmatic theories of propositional attitude reports offered by Nathan Salmon (1989) and Scott Soames (1987), which try to do just that. ! )+(! propositional attitude reports, whereas they do not affect the content and truth-values of simple sentences. Indirect speech Cases of indirect speech or reports have the following form: (16) A said that S where A stands for the name of the subject who is being reported and S for the sentence embedded in the ‘that’-clause. Indirect reports semantically express that the individual named A stands in the binary relation expressed by Vs to the proposition referred to by the ‘that’-clause that S. DD’s proposal for indirect reports is very close to the one for propositional attitude reports, that is to say, PNM. The only difference is that the hybrid propositions reported as being said are a bit different: they are composed out of the main proposition expressed by the sentence embedded in the ‘that’-clause and the additional descriptive and metalinguistic proposition that it also expresses, or at least some of the elements of the latter. For example, consider: (17) Lois Lane said that Superman would come. (18) Lois Lane said that Clark Kent would come. Utterances of these reports express different main propositions and hence, differ in truthvalue. The propositions that Lois Lane said according to these reports are hybrid propositions, that is, propositions that get individuated in part by the distinct metalinguistic descriptions associated with the names ‘Clark Kent’ and ‘Superman’. ! )+)! What the report reports is that Lois Lane said something more fine-grained than the proposition that the individual Clark Kent has the property of coming to a particular place. The apparent exceptions to that explanation are treated analogously as apparent exceptions in the case of propositional attitude reports, which has already been explained in full detail. Quotation There are various cases of quotation. There are cases of quotation such as (19) and (20), and cases closer to the case of indirect speech such as (21). (19) ‘Boston’ contains six letters (20) ‘Superman’ refers to Clark Kent (21) Lois Lane said ‘Superman will come’ In addition, there are cases that are even closer to the case of indirect speech, like mixed quotations, which are indeed very commonly used such as the following: (22) Lois Lane said that ‘Superman’ will come My view is that as in the other three cases of opacity, in quotation there is also reference being contextually made that changes the content of the statements and hence, affects their truth-value, and which depends on the particular expressions used to make such statements. DD adopts some version of the Davidsonian demonstrative theory of quotation. Versions of this theory have been defended by Cappelen & Lepore (1997), Davidson (1979), García-Carpintero (1994, 2004, forthcoming), and Predelli (2008), among others. ! )+*! According to García-Carpintero’s version of it, quotation marks are the contextdependent expressions, which may refer to different things in different statements depending on which properties of the words quoted are salient in the context. Thus, according to such a proposal, quotation marks behave as a demonstrative. Treating quotation marks as context-dependent expressions which behave as demonstratives is analogous to treating the expression ‘so’ in the ‘so-called’ sentences and treating the ‘that’ in the ‘that’-clauses in propositional attitude and indirect reports as contextdependent expressions which also behave as demonstratives. Treating all these cases this way would provide an elegant unified view of opacity in general. Let me now summarize the contents of this section. In this section I have shown how DD can account for a wide variety of phenomena: identity statements, simple sentences, ‘socalled’ sentences, indirect speech and quotation. In the next section I discuss how DD relates to other extant views in the literature and how it can answer some objections. 4.6. Other Views In this chapter I have made a new proposal on the metaphysics and semantics of propositional attitude reports which, nonetheless, builds on several ideas from different proposals already made in the literature, most notably Davidson (1968), Frege (1892, 1918), García-Carpintero (1994, 1998, 2000, 2004, forthcoming), Grice (1967), Kripke (1972), Pietrosky (1996), Richard (1990)), Russell (1905, 1910), Salmon (1986, 1989) and Soames (1987). PNM follows several traditional insights on different phenomena: first, it follows the Davidsonian insight on the phenomena of propositional attitude reports and quotation by considering propositional attitude reports and quotation as mainly involving some sort of demonstration. Secondly, it follows the Fregean insight on the semantics of singular terms and the metaphysics and epistemology of modes of presentation by accepting some role for Fregean senses in the semantics. More particularly, PNM is influenced by the ! )++! concrete Davidsonian and Fregean theories of quotation and singular terms that GarcíaCarpintero defends, on the one hand, and of propositional attitude reports that Pietrosky defends, on the other. Thirdly, PNM follows Russellian insights on definite descriptions and the nature of propositions by accepting some version of Russell’s analysis of descriptions, and the view that Russellian propositions are the main propositions that utterances of sentences in general express. Finally, PNM follows Grice’s insight on the nature of conversations and pragmatic explanations involving conversational implicatures to explain away certain intuitions. Having noticed these affinities, it would be useful to put PNM in the broader context of the views defended on the semantics of propositional attitude attributions in general. There are five main types of views: the kind of traditional Fregean approaches (Forbes (1987, 1990), Frege (1892, 1918), Kaplan (1969), Pietrosky (1996)), the traditional Russellian approaches (Braun (1998), Russell (1905, 1910)), the Neo-Russellian views that treat propositional attitude reports as a highly contextual matter (Ludlow & Larson (1993), Perry & Crimmins (1989), Richard (1990)), the naïve Millian views that try to explain away those counterintuitive intuitions that they face by appealing to pragmatics (Salmon (1986, 1989), Soames (1987), Saul (1998)) and finally, the Possible World views (Lewis (1979), Stalnaker (2008)) Let me now very briefly explain some of these theories. At one first extreme of the spectrum, Nathan Salmon and Scott Soames defend a pragmatic naïve Millian and Russellian view of propositional attitude ascriptions, according to which, reports such as (1) and (2) semantically express the same proposition and hence, have the same truthvalue, in part because their embedded sentences express the same proposition as well. (1) Hammurabi believes that Hesperus is Hesperus (2) Hammurabi believes that Hesperus is Phosphorus According to this view, the only semantic function of proper names such as ‘Hesperus’ and ‘Phosphorus’ is to stand for their bearers. However, utterances of (1) and (2) ! )+#! pragmatically convey different propositions and it is this fact that explains that according to common but mistaken intuitions, they differ in truth-value. David Braun also defends a naïve Millian and Russellian view of propositional attitude ascriptions, but disagrees with Salmon and Soames’ pragmatic explanation of our common intuitions about the truth-values of the ascriptions. Like Salmon and Soames, Braun follows Perry’s distinction between belief states and belief contents, according to which one may believe a proposition –i.e. belief content- by being in different mental states –i.e. belief states. According to Braun’s view, however, in having our common intuitions what we do is to commit very simple mistakes, as simple as the ones that students commit in an introductory Logic course. On the other hand, there are the Contextualist views such as that of Perry and Crimmins, and the interpreted logical form theories of Mark Richard and Ludlow and Larson. Perry and Crimmins defended that belief is a ternary rather than binary relation among a believer, a proposition, and a way of believing under which the believer believes the proposition in question. Belief reports have an unarticulated constituent that refers to the particular way of believing in each context. This way their view can explain that in some contexts reports such as (1) and (2) differ in their semantic content and hence, truthvalue. Mark Richard’s view is that beliefs are binary relations between a believer and what he calls a ‘Russellian Annotated Matrix’ (or ‘RAM’, in short). RAMs are something like tuples of pairs of linguistic expressions, or mental representations, and their Russellian interpretations. RAMs are also the contents of sentences. According to Richard’s contextualist view, it is predicates such as ‘believes’ that are context-sensitive, expressing different relations in different contexts. Basically, the idea is that, for instance, an utterance of (1) is true relative to some context c and a world of evaluation w just in case Hammurabi has a belief in w that is properly represented (relative to the standards in effect in c) by the sentence ‘Hesperus is Hesperus’. What counts as properly represented highly depends on the context of utterance of the report. ! )+$! Larson and Ludlow proposed the view that propositional attitudes are binary relations between individuals and Interpreted Logical Forms (ILF). ILFs are structured syntactic trees, the nodes of which have Russellian interpretations assigned. Finally, at the other extreme of the spectrum, there are the non-contextualist Fregean views of Pietroski and Forbes. Pietroski’s view is that propositional attitudes are binary relations between individuals and Fregean thoughts –namely, propositions constituted by Fregean senses. The idea is that, although the embedded sentences in propositional attitude reports have their usual senses and references, the ‘that’ in the ‘that’-clauses of propositional attitude reports refers to an ILF, constituted by linguistic items and Fregean senses, and the ‘that’-clause itself ends up referring to the Fregean thought. For instance, a report like (1), according to Pietroski’s view, would be true if and only if Hammurabi bore the belief relation towards the Fregean thought expressed by the embedded sentence ‘Hesperus is Hesperus’ and referred to by the ‘that’-clause ‘that Hesperus is Hesperus’. Forbes’ view is a more traditional Fregean view according to which expressions in the context of propositional attitudes refer to their customary senses, rather than to their customary references. Forbes goes further than Frege, however. His view follows Kaplan’s proposal of quantification into the context of a propositional attitude and makes some unusual locutions of English such as (3) central in expressing the content of ascriptions in those cases in which one does not know the specific sense that fills the proposition which is the object of the attitude in question. (3) Phosphorus is something which Hammurabi believes to be Hesperus The analysis of an utterance of (3) would be something along the following lines (4) "# (# is a way of thinking of Phosphorus & B (Hammurabi, # ˆ ¬is Hesperus¬) (5) Phosphorus is such that for Hammurabi’s so-labelled way of thinking of it #, B (Hammurabi, # ˆ ¬is Hesperus¬). ! )+%! (where # stands for psychological senses, ˆ stands for the combination of senses, and we use ¬¬ around expressions to refer to their senses) PNM lies near Pietroski’s and Forbes’ views, mostly for being Fregean in that expressions express some kind of sense. However, like Salmon and Soames’ views, PNM also applies Russell’s conception of propositions and Gricean pragmatic explanations in the case of propositional attitude attributions (even if to defend quite a different view). In addition, PNM agrees with Richard on the hybrid propositions involving mental representations that propositional attitudes are directed to. The attempt to solve problems by recurring to pragmatic mechanisms has not proved easy, though. As I have explained above, Salmon and Soames’ pragmatic explanations of common intuitions have not convinced many. Thus, the Fregean or contextualist views seem to be better positioned, since they count with strong intuitions on their side. However, (Neo-) Millian and Russellians not only appeal to strong arguments against Descriptivism, but also to some of other intuitions that favor their view over a (Neo-) Fregean. This allows me to bring into focus one of the original aspects of PNM. The (Neo-) Millian and Russellian views recurred to pragmatics to explain intuitions away. However, to my knowledge, no one has thought that it is the (Neo-) Fregean views the ones that should do it, and successfully so. PNM, the Neo-Fregean and Russellian view that I defend in this chapter, does precisely this. PNM appeals to pragmatics, though unlike Millianism, in a successful way. Before concluding this brief literature review, let me notice that, as I indicated above, among the views on the semantics and metaphysics of propositional attitude reports there are different Possible Worlds views as well, such as the views of Stalnaker and Lewis. However, these views have specific problems to solve due to their conception of the nature of the propositions that are the contents of propositional attitudes and utterances of sentences in general. I do not discuss this family of views in this dissertation. ! )+&! 4.7. Objections and Replies Let us now see how PNM could answer some objections. A classical objection to a Davidsonian view on the semantics of propositional attitude ascriptions, like PNM, consists in complaining that since the view treats the complementizer ‘that’ as a referring expression that refers in a similar fashion as a demonstrative, it treats it exactly like a demonstrative when it does not seem to be one. There are two rationale for this complaint: first, unlike a demonstrative, the complementizer cannot refer to anything in the context. According to PNM, the complementizer ‘that’ always refers to something that is determined by the sense of the embedded sentence. Second, unlike a demonstrative, the reference of the complementizer ‘that’ cannot be anaphorically referred back later in the discourse. For instance, demonstratives like ‘that’ occur in subject or object positions in sentences and can be anaphorically referred back later in the discourse as in the following examples. (1) That does not work. It is broken. (2) Superman can fly. Lois Lane believes that, and it is true. In utterances of (1) the demonstrative ‘that’ that occurs in subject position refers to something that is anaphorically referred back again later by the pronoun ‘it’. In utterances of (2) the demonstrative ‘that’ occurs in object position, referring to whatever it is that Lois Lane is reported to believe and anaphorically referred back again later by the pronoun ‘it’. Thus, part of the objection advanced is to notice that there are no cases like these in which the thing that is referred to by the complementizer ‘that’ in a propositional attitude sentence is later anaphorically referred back in the discourse. Now, I agree that all of these differences exist, but I do not see that this constitutes any evidence against the idea that the complementizer ‘that’ refers to something in a way that is close to the way that demonstratives refer. For this is not to say that the complementizer ‘that’ behaves exactly like, or that is identical to, a demonstrative. ! )+"! According to PNM, the complementizer ‘that’ is a referring expression that refers to things that are made contextually salient by the sense expressed by the embedded sentence in the ‘that’-clause, but the complementizer is not the same expression as the demonstrative ‘that’. In particular, PNM agrees that unlike the demonstrative, the complementizer cannot refer to anything and that its reference cannot be anaphorically referred back later in the discourse. The reason for the latter being simply that unlike the demonstrative, the complementizer ‘that’ does not occupy the subject or object positions in the sentences in which it occurs, but instead is only part of expressions that do occupy those positions, namely the ‘that’-clauses which also refer. Consider the following belief report: (3) Lois Lane believes that Superman is sexy, and it is true. The pronoun ‘it’ in an utterance of (3) anaphorically refers back to the reference of the entire ‘that-clause, which occupies the object position in (3) and is, according to PNM, a singular term whose reference is partly determined by the reference of the complementizer ‘that’. In short, one may claim that an expression behaves like another only in certain respects, without claiming that they are exactly alike. This is what PNM does19. Another classical objection to Davidsonian approaches is that not all ‘that’-clauses seem to create opacity. ‘That’-clauses under the scope of ‘It is true’, ‘It is necessary’, ‘It is possible’, or ‘It is a fact’ do not seem to do it. Thus, the objection goes, the complementizer ‘that’ does not seem to play any role or have any effect on the opacity of propositional attitude reports and other constructions. Quite the contrary, it seems to be a syncategorematic expression –i.e. an expression that does not have any meaning. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! 19 Other close alternative hypotheses occur to me. One is the hypothesis that ‘that’-clauses behave like complex demonstratives such as ‘that book’. According to this other hypothesis, the ‘that’ in a ‘that’-clause, like the ‘that’ in a complex demonstrative, would not refer to anything but would contribute to determine the reference of the ‘that’-clause. A second hypothesis is the hypothesis that the complementizers ‘that’ do not refer to anything, but the semantic rule that composes the meanings of the ‘that’-clauses from the meanings of their complementizers ‘that’ and their embedded sentences does. I prefer the hypothesis adopted by PNM in the main text. ! )+'! I do not see how this constitutes any evidence against PNM either. An expression may have reference in certain contexts, but not in others. An indexical such as ‘he’ may lack reference in some contexts due to the absence of certain conditions. The complementizer ‘that’ lacks reference in some contexts also due to the absence of certain conditions, which is this case have to do with the absence of an expression that induces the opacity. Another example, although proper names are paradigmatic cases of referential expressions, there are some that do not refer. In short, the complementizer ‘that’ may function as a referential expression when it follows a propositional attitude verb or a verb like ‘explains’ or ‘says’, even if it does not refer to anything when it follows other expressions such as ‘true’ or ‘necessary’. Notice that this per se does not entail that there is no unified explanation of propositional attitude attributions and other constructions involving ‘that’-clauses such as the ones that predicate truth and the ones that predicate modalities. According to the view that I am proposing, after all, all ‘that’-clauses could be said to refer to the proposition composed out of the proposition expressed by the embedded sentence and the reference of the complementizer ‘that’, if there is any. Let me consider a third classical objection now: opacity seems to arise under the scope of propositional attitude verbs even when they are not followed by ‘that’-clauses but by different clauses such as ‘whether’-clauses. But it seems even more implausible that complementizers such as ‘whether’ are or behave as a demonstrative, for in cases like these, the complementizers do not even look like one. Consider the following example: (4) Lois Lane doubts whether Clark Kent is Superman. I do not think this objection is any more successful than the previous ones. One can claim that the thesis that the complementizer ‘that’ behaves like a demonstrative does not have much to do with the similarity in syntactic or phonetic form of that specific complementizer and the demonstrative ‘that’; in the same way that there is no such similarity between quotation marks and the expression ‘so’, on the one hand, and the ! )#(! demonstrative ‘that’, on the other. Therefore, there is no added issue in defending the same semantic thesis for other complementizers like ‘whether’. After all, a similar thesis would be defended for quotation marks in the case of quotation and, as pointed out above, quotation marks are unlike any demonstrative too in appearance. Before ending this chapter, I would like to answer a couple of more general worries that one might have about PNM. First, one might think that semantics should not be something dependable on the results of other disciplines such as cognitive science, but PNM clearly is. PNM posits mental representations, which cognitive science might end up discovering that do not exist. Thus, cognitive science could prove PNM wrong and this should not be the case: semantics should be independent in this respect. First of all, I do not think that semantics should be independent in this way of the possible discoveries of other disciplines. Quite the contrary, I regard semantic theories in general as dependable on cognitive science, as well as on many other disciplines such as linguistics, biology and physics. In the particular case of a semantic theory of propositional attitude attributions, many will agree that propositional attitude attributions make some kind of reference to propositional attitudes themselves. In fact, our reasons to posit propositional attitudes and appeal to them in our explanation are on a par with our reasons to posit mental representations as their constituents as well (propositional attitudes are mental representations after all). This is so even if cognitive science could prove our hypotheses to be mistaken. We hypothesize the existence of propositional attitudes in order to explain behavior. But if positing propositional attitudes is a reasonable thing to do to explain behavior, so is it to posit some kind or other of mental representation that constitutes them and that does that same job. If this is so, positing constituent mental representations should not be more risky than it is to posit propositional attitudes. Cognitive science might show the hypothesis about the existence of constituent mental representations to be wrong, as much as it might show the hypothesis about the existence of propositional attitudes to be wrong. In any case, however, I think that mental representations are well accepted by current cognitive science. Although there may be attempts to explain behavior, propositional attitudes or ! )#)! propositional attitude reports without appealing to constituent mental representations as parts of the content of the attitudes, such as Ludlow and Larson’s considered above, it is an open or controversial question whether they succeed and whether they are finally preferable. In the case of Ludlow and Larson’s account, it does so by individuating words in a quite odd and I think ad hoc way. According to Ludlow and Larson’s account, different pronunciations of ‘Harvard’ or ‘Paderewsky’ correspond to different proper names, for instance. Finally, one might think that even if there were properties associated to linguistic expressions besides their reference or extension, like DD has it, these would not be part of their semantics, but part of their metasemantics. However, it seems to me that the inclusion of these properties into the semantics is quite justified. The reasons are the following: first, these properties are conventionally associated with the linguistic expressions. Second, they are part of what a competent speaker knows just by being competent of a language. Third, they are part of what gets semantically expressed if only additionally. Finally, they play a role in determining the main propositions that utterances of sentences semantically express, which is what determines the utterances’ truth conditions. Let me conclude these remarks by quoting a text by Perry (1988) in which I think he expresses something similar to what I have tried to express here in his reply to Wettstein (1986): Wettstein (…) accepts that cognitive significance is not the semanticist’s worry. Is this reasonable? What the semanticist should worry about, depends on what the semanticist is trying to do. A logician who is trying to come up with a semantic account of some logic that will enable the construction of a completeness proof should be relieved of worrying about cognitive significance. But this is not the sort of project Wettstein has in mind. By “semantic theory” he means a theory of linguistic meaning for natural languages. Wettstein is interested in how demonstratives, descriptions and proper names work in ! )#*! English, not how variables, iota operators, and individual constants work in languages for quantification theory. If this is the semanticist’s interest, then he has picked a subject matter that, whether he likes it or not, ties in with a number of other subject matters, including the study of cognition. After all, our main interest in language is the way its use can communicate beliefs, inspire action, and have other effects on what we think and do. In these uses of language, it is aspects of the meaning of the language used that are crucial. A theory of linguistic meaning should help provide us with an understanding of the properties sentences have that lead us to produce them under different circumstances, and react as we do to their utterance by others.” (pp. 3-4) In this chapter, I have presented and defended, as part of DD, what I think is the best theory on the metaphysics and semantics of propositional attitudes and their ascriptions, and other related phenomena. In the next two chapters, I focus on the issue of fiction and fictional discourse and explain what DD claims about them. ! )#+! 5. Fiction and Its Contingencies This chapter and the next one are devoted to fiction and fictional discourse. In chapter 6 I present and defend an original proposal about these issues that is part of DD (that is, the view that I am offering in this thesis). I call this part of DD ‘Demonstrated Fictional Names’ (‘DFN’ for short). Before getting to that, however, in this chapter I explain some of the main problems that a view on fiction has to deal with, and some of the main views in the literature. Among these views, I discuss David Lewis and Gregory Currie’s descriptivist view in more detail, in order to show what I take to be some of its major flaws, and to distinguish it from DFN, the descriptivist view that I propose. I also consider Frederick Kroon’s pragmatic pretense-based view and his arguments against the ellipsis strategy, because this is a strategy that DFN makes use of. 5.1. Problems In this section I explain some of the main problems to be solved by any philosophical theory on fiction. The list is not intended to be exhaustive, but I think the central issues are covered. The aboutness problem Ordinarily, when we talk about someone we use her name or some appropriate indexical. Proper names and indexicals refer to individuals. When we use them, we normally say something meaningful, and their referents become part of the content of our statements and mental states. On the other hand, also to create and talk about fiction we normally use linguistic expressions of the kind we ordinarily use to refer to individuals. We use referring terms such as proper names and indexicals. However, when dealing with fiction ! "##! (at least when we engage in fiction as a storyteller or as someone who participates in the corresponding make-believe) we do not seem to use them to refer, and they do not seem to refer either. In spite of this, the language we use obtains meaning, and we create and communicate content this way. Moreover, aside from the language we use, we also seem to engage in meaningful mental states. When dealing with fiction, we usually use our imagination to picture the fiction's content, for instance. However, it seems that there are no more individuals that we are imagining than there are that we are talking about. One of the central problems in theorizing about fiction is precisely this. Even if meaningful, in some important respect, fiction seems to be about nothing, except for relations or properties. That is, in addition to properties or relations, there does not seem to be anything fiction is about. I call this 'the problem of aboutness'1. The problem of untrue but meaningful utterances of sentences In certain situations, we seem to produce meaningful but untrue utterances of sentences. We seem to do it, when we create fiction or when we mistake fiction for reality. In the first case, in storytelling, we use sentences such as (1) to tell a fictional story. (1) E.T. is an extraterrestrial. Utterances produced in storytelling such as an utterance of (1) are meaningful and determine the content of the fictional story. An utterance of (1) seems to say of the individual referred to by the name 'E.T.' that he or she has the property of being an extraterrestrial. However, 'E.T.' is a fictional name that does not seem to refer to anyone. Hence, there does not seem to be an individual referred to by that name to whom the !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! 1 It is worth mentioning that in telling some fictions there are some proper names that are used, which either are the same proper names that are also used outside storytelling, or homonymous to proper names so used, such as ‘Napoleon’ or ‘Russia’ in Tolstoy’s War and Peace. It might seem that these names do refer and that the aboutness problem does not arise with respect to them. I consider these cases later in the chapter. ! "#$! property of being an extraterrestrial could be ascribed. That seems to mean that there is no individual that the utterances are about, and this seems to entail that these utterances cannot be true2. In the other situation, the case of utterances made out of mistaking fiction for reality seems to be like the case of those utterances made in storytelling, except for the distinct attitude of the utterers in both occasions. In storytelling, we seem to utter sentences like (1) without the intention of saying anything true about the world (we only pretend to do it), while not so in this last case. Unlike what we do in situations of storytelling, in cases in which we mistake fiction for reality, we seem to ordinarily utter those sentences, with no other intention than to say something that is true. Let me introduce a qualification to what has just been said. While it seems true that uttering sentences such as (1) in fictional storytelling is done without the intention of saying anything true about the world, uttering more complex sentences or telling whole fictional stories might not be. For even when a fiction presents us with very unrealistic situations, the storyteller might be telling us about such real things as love, hate, fidelity, envy, and the human condition, or even such other mundane things as the mean character of a neighbor –and all this with the intention to do so. In any case, these uses of sentences such as (1) present the following problem: the problem of accounting for their apparent meaningfulness in spite of the fact that there seem to be no individuals that the utterances are about and that they are untrue. I call this problem 'the problem of untrue but meaningful utterances of sentences'. This problem is closely related to the aboutness problem discussed before. The difference is that the aboutness problem is a more general and metaphysical problem (because it mainly concerns what in the world the fiction is about), whereas the present problem is more specific and semantic (because it mainly concerns the meaning and truth-value of certain !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! 2 Here, and in what follows, I am taking that the intuitive datum is that these fictional proper names and indexicals do not seem to refer. I think that this is the intuitive starting point even for those philosophers who end up defending the view that these fictional proper names and indexicals refer, such as Millians like Nathan Salmon and Meinongians like Graham Priest. ! "#%! linguistic expressions). The problem of describing fictional stories So far we have been considering uses of sentences within storytelling, and uses of these same sentences as a result of ignorance. Sometimes, though, we use (what look like) this same kind of sentences to describe existing fictional stories as such, rather than to tell or create them. When we do so, we manage to state truths about the story. We seem to utter sentences such as (1) meaning that what they say is true according to a specific fiction – that is, meaning the same as when we utter sentences such as (2). (2) According to the fiction E.T. The Extraterrestrial, E.T. is an extraterrestrial. The problem of accounting for the fact that we sometimes use sentences such as (1) and manage to state truths like (2) is the problem I call ‘the problem of describing fictional stories’. This is the problem of determining the linguistic mechanisms that we exploit when we do this. On the other hand, it might also seem that when uttering sentences like (1) in order to describe an existing fiction, we are not doing the same thing as if we were to utter sentences like (2). We are doing something else, at least in part. Utterances of (1) and (2) do not have the same effects: (1) may be employed within certain games of make-believe, within which (2) may not. There is, then, the need for an explanation of what it is that we do when we utter these sentences, what the true content is that we manage to communicate in doing so, and how we manage to do it. ! "#&! The problem of truth in fiction Another problem is that of finding out what it is that determines what is true in a fiction. Things that have been explicitly stated in storytelling are usually truths in the fiction told. In addition, there are also certain truths in fiction that are not explicitly stated in the telling of the story. Consequences of what is explicitly stated together with background assumptions normally are truths in the fiction as well. For instance, it is not explicitly stated in the Sherlock Holmes Stories that Sherlock Holmes has two hands and two feet, but this is clearly true according to it. It follows from both what is explicitly stated and background assumptions. Notice that the solution to this problem is not that straightforward as it may seem at first sight. Consider, for instance, those fictions where the fictional teller is a liar or tells the story with irony (or with some other insincere or non-literal intentions). In these cases what is explicitly stated by the narrator is not fictionally true. The film The Usual Suspects, where the narrator lies, is an example. The problem of comparing characters from different fictions Sometimes we make comparisons between characters from different fictions. For instance, we say things such as (3) E.T. is more homesick than Alf. (4) Alf is sassier than E.T. However, unlike the cases that we have been considering so far, cases like (3) and (4) do not seem to be true in any particular fiction. Thus, it is not clear how we should interpret this kind of statements. It is clear that utterances of (5) and (6) below do not capture what we mean by utterances of (3) and (4). ! "#'! (5) According to the fiction E.T. The Extraterrestrial, E.T. is more homesick than Alf. (6) According to the fiction Alf, Alf is sassier than E.T. An alternative explanation is therefore needed. Things are somewhat more complex than this. For we also utter sentences to compare alleged fictional individuals with real individuals. The carryover problem There are fictions that are written as continuations of stories told in other fictions about the same characters, even by different authors. In these cases, fictional truths according to the first fiction might be contradicted in the second, attributing different traits or histories to their characters. There is a problem, then, in discerning the real properties and histories that these fictional characters have when this happens. If so, what properties or history the characters fictionally have might end up being relative to each of the fictions told. Otherwise, if what is stated by the first fiction is not contradicted by the second, then it seems to carry over in the latter and be one of the nonrelative fictional truths that are not made explicit in the second fiction. To illustrate this problem, consider the case of El Quijote. After Cervantes wrote the first part of El Quijote, and before he wrote its second part, another Spanish writer, Avellaneda, wrote a continuation to the first part. Cervantes’ second part included replies in the mouths of the characters to this other continuation of El Quijote written by Avellaneda. In fact, in Cervantes’ second part, Quixote himself denies having done the things that Avellaneda’s story reported that he did. ! "#(! The problem of thinking about fiction from the outside In addition to speaking and thinking about the content of fictional stories and fictional individuals represented in these stories, we also say things about the author of the fiction and her creations. Thus, we utter sentences such as (7) Spielberg created E.T. (8) E.T. is a fictional character. Utterances such as these seem to be true, but do not seem to be about the content of the story and hence not true according to it. Instead, they seem to be about the fiction as a created work of art – that is, about the fiction “from the outside”, as we may call it. Thus, their truth cannot be explained by the prefixing of operators such as ‘in the fiction’ or ‘according to the fiction’. According to the fiction E.T. The Extra-Terrestrial, E.T. is an extraterrestrial rather than a fictional character and Spielberg plays no role in bringing it to the world. Utterances of sentences such as (9) and (10) are, thus, false. (9) According to the fiction E.T. The Extra-Terrestrial, Spielberg created E.T. (10) According to the fiction E.T. The Extra-Terrestrial, E.T. is a fictional character. The problem of real world individuals There are fictions that seem to be closer to reality in certain ways than others. Thus, actual London seems to bear a strong relation to the city where Holmes fictionally lives, and the real world Napoleon seems to be a character in Tolstoy's War and Peace. It may seem that real world entities, other than real relations and properties, come to be part of those fictions and are referred to in them when using what at the very least are homonymous proper names. Theories of fiction need to make a decision about whether these appearances are taken at face value or not. ! "$)! The problem of attitudes towards fiction We seem to bear attitudes towards fictional characters and stories. Thus, for instance, we seem to pity E.T. and Anna Karenina, admire Superman or Sherlock Holmes, fear Freddy Krugger, and so on and so forth (or perhaps, to fear E.T. and Sherlock Holmes, pity Superman and Freddy Krugger and admire Anna Karenina). But this is difficult to explain if there are no individuals towards which we bear these attitudes. The non-existence of fictional individuals does not seem to be the only obstacle that an account of these apparent attitudes encounters, however. There seems to be another condition to be met for us to be able to hold these attitudes towards them. As Kendall Walton (1978) indicates, fear, to take an example, seems to require the belief that the thing that one fears exists and is dangerous. So despite of all the usual agitation and physical reactions that people feel when watching a horror movie, they do not really believe any of the characters in the movie to be dangerous and hence, they do not seem to fear them after all. This is precisely why, in fact, they never rush to leave the theater. On the other hand, if this is right, then one needs to give an account of what other kind of attitudes people hold towards fiction if any, or what other kind of things they hold these attitudes towards if any. After all, we seem to hold some attitude or other towards something or other which is “close” to the one that we would be wrongly judging to hold. We do say such things as (11), (12) and (13) and regard them as true. Or at least, we seem to manage to say something true when uttering them. (11) I admire Sherlock Holmes. (12) My friend pities E.T. (13) I am afraid of Freddy Krugger. In order to explain the apparent truth of these utterances, it does not seem that we can prefix the 'according to the fiction' operator to them, since we do not seem to be describing any of these fictions (i.e. Sherlock Holmes’, E.T.’s or Freddy Krugger’s) when ! "$"! we utter them. In any case, we say something about our real attitudes towards something and the theory should explain what. And, further, we do not seem to be part of any of these fictions either. Notice, however, that it may seem that even if we did believe Freddy Krugger to be dangerous, we would not be able to hold the attitude of fear towards anything by that name if there was nothing that it named. Children believe that E.T. exists and that he is defenseless, and they seem to pity E.T. But what do they pity, if E.T. does not exist? The problem of true negative existentials Despite the fact that we say such things as (14)-(17) when we want to distinguish fiction from reality we sometimes also use negative existentials such as (18). (14) E.T. is an extraterrestrial. (15) Mary pities E.T. (16) Alf is sassier than E.T. (17) Spielberg created E.T. (18) E.T. does not exist. On the one hand, negative existentials such as (18) seem to work as any other negative attributive sentence by which one refers to something and then denies to it the possession of an attribute. For instance, utterances of (19)-(21) make reference to something, i.e. a particular table, the species of elephants, and the utterer respectively, and then deny that the thing in question has a certain property, which in these cases are the properties of being round, of flying and of existence, respectively. (19) This table is not round. ! "$*! (20) Elephants do not fly. (21) I do not exist. So an utterance of (18) seems to be referring to someone or something with the use of the name 'E.T.' and denying that this individual has the property of existence. On the other hand, it is precisely because that statement is true that there seems to be nothing to which to attribute that property. But then it is difficult to tell how this statement could be meaningful and true. Again, the matter does not seem to be accountable by prefixing the 'according to the fiction' operator, because it is the opposite of what these negations state that the corresponding fictions state. On the contrary, negative existentials are true according to reality. Before moving on to the next problem, let me make the following clarification. As stated in a previous footnote, I take as the shared intuitive starting point the idea that fictional proper names such as ‘E.T.’ do not seem to refer. Similarly, I take as the shared intuitive starting point the idea that negative existentials like ‘E.T. does not exist.’ seem to be true. I take the views that proper names refer to either abstract entities or to beings that do not exist as theoretical views (which I consider below). Analogously, I regard as theoretical the view that negative existentials like ‘E.T. does not exist.’ are usually a way of saying that a certain fictional character, E.T. in this case, is not non-fictional, not flesh-andblood, or not spatiotemporal. These theoretical views are not taken into account here because in this section I am presenting the problems that arise from what many of us believe to be pretheoretical intuitions. But, of course, these theoretical views might well provide the solutions to these problems. On the other hand, one might think that what many of us take as pretheoretical intuitions shared by philosophers and non-philosophers are not such, and that we are wrong in this respect. There is no much to say about this disagreement really. But let me just notice that in any case, it suffices for having the problem of true negative existentials, discussed in ! "$+! the literature, that there is a single seemingly true utterance of a negative existential that uses the predicate ‘exists’ at face-value. It seems to many of us that such an utterance obviously exists, even if it is one uttered by some philosophers rather than by nonphilosophers. Indeed, even if it were true that non-philosophers never make this kind of utterance, nothing prevents them from doing it. Even if it were true that non-philosophers take this kind of utterance to be false, they could be wrong about this. Moreover, this is wholly compatible with accepting that non-philosophers often use negative existentials to pragmatically convey many other different propositions, like the ones suggested above. Let me provide a piece of evidence in favor of my view that also non-philosophers think that E.T., Superman, and the like, do not exist. Recently, I found this picture in Facebook, which is supposed to be a joke. The (perhaps, bad) joke of an atheist, obviously. ,-./01!$2"3!,4516778! 97:;! My point in including this example here is that this joke is based on the fact that people in general believe that Spiderman does not exist. In fact, there were people complaining about this joke in Facebook. They would claim that it was “obviously ridiculous” to believe that Spiderman existed, whereas it was not ridiculous to believe that God exists. I must confess, however, that I have some evidence to the contrary. This year I was told that a student of mine rebelled against what I said in one of my lectures by writing that she did believe that Superman existed in a washroom of the University. ! "$#! 5.2. Some Main Views In this section I discuss some major views that try to solve the problems that we have just presented: Meinongianism, different versions of Creationism, and Frederick Kroon (2004, 2009)’s recent Pretense-based theory. I reserve a more lengthy discussion of David Lewis and Gregory Currie’s descriptivist view for section three. Meinongianism One of the most well known views about fiction is Meinongianism. Its characteristic thesis is that there are things that do not exist. Amongst these things that are deprived of existence there are the fictional creatures whose stories fictions tell. There are E.T., Alf, Sherlock Holmes, Anna Karenina, and the like. These creatures are and do what the corresponding fictions say that they are and do, except that they do not exist. Thus, according to this view, there is an extraterrestrial, called 'E.T.' by his human friends, who gets lost on Earth and wants to go back home. In spite of all of this, that is, in spite of E.T.'s being something, his acting in certain ways and his feeling certain things, E.T. does not enjoy existence. Meinongianism offers a straightforward way out to many if not to all of the problems described before. Consider first the problem of aboutness. According to Meinongianism, there are things fictions are about. It is only that these things do not exist. Second, about the problem of untrue but meaningful utterances of sentences, Meinongianism can easily deny that these utterances are untrue and explain their meaning in the same way that meaning is explained in any ordinary utterance with no empty names. That is, if it also adopts the semantic thesis that fictional names refer to these nonexistent creatures, then, on this view, utterances of sentences such as (1) will be plainly true and meaningful, since 'E.T.' would refer to the non-existent being E.T., and this being would have the property that (1) attributes to him, the property of being an ! "$$! extraterrestrial3. (1) E.T. is an extraterrestrial. Third, if Meinongianism is correct, the problem of describing fictional stories indicated above dissolves. For one would be able to describe fictional stories both just by uttering true sentences such as (1) and by uttering sentences explicitly prefixed with the ‘according to the fiction’ operator. On the other hand, notice that in the special case of the property of existence, we do not utter such things as ‘E.T. exists’, we rather say the opposite. So the fact that, according to Meinongianism, E.T. does not exist, is compatible with the fact that according to the fiction E.T. The Extra-Terrestrial, E.T. exists, since we would never express this latter truth without (understanding or meaning it as including) the ‘according to the fiction’ operator at the beginning of the sentence. Fourth, according to Meinongianism, there is no problem in comparing different fictional creatures from different fictions: doing so is just like comparing different existent individuals. What it takes for Alf to be sassier than E.T. is exactly what it takes for Berlusconi to be sassier than Sarkozy. Fifth, for this same reason there is no problem in saying things about fiction from the outside: according to Meinongianism, there is a thing, E.T., that Spielberg created. So there is no problem in explaining how (2) can be true. (2) Spielberg created E.T. Sixth, Meinongianism can either accept that real world entities are sometimes referred to !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! 3 There are different theories that are Meinongian. What all of them have in common is their central claim that there are things that do not exist. One of the things in which they differ is in what properties they attribute to non-existent objects. Usually, a distinction is made between existence-entailing properties and non-existence-entailing properties. I am here considering what I think is the simpler version of Meinongianism, which I think would have more advantages if ultimately tenable. According to this version, fictional beings have all the properties that fictions say they have, except for existence and a few other existence-entailing properties. ! "$%! in fictions, or claim that there are things such as the Sherlock Holmes Stories' London or War and Peace's Napoleon that are similar in important respects to real world entities but that do not exist and are not these real world entities after all. Seventh, if there are fictional creatures, then there are things towards which we can bear attitudes such as fear, admiration, pity, and belief. Hence, there is no difficulty in explaining the apparent truth of utterances of sentences such as (3) with respect to this, because there is E.T. and so Mary can hold such an attitude. (3) Mary believes in E.T. However, there is still a question that remains: whether things that only are but do not exist can be dangerous, admirable or pitiful, or even be believed to be dangerous, admirable, or pitiful when one knows that they do not exist. Hence, Meinongianism has to say something more in order to explain these other apparent attitudes and truths, such as the ones stated in the following utterances (4) Mary pities E.T. (5) I fear Freddy Krugger. According to some Meingnonians, there are certain properties that are existenceentailing. What this means is that they are properties that can only be had by things that exist. Thus, fictional creatures such as E.T., Sherlock Holmes and Anna Karenina possess only those properties that do not require existence, but not the ones that do. It could be then that such properties as the properties of being dangerous, admirable or pitiful are amongst the existence-entailing ones. But if this is so, Meinongianism also has to provide us with an explanation of the apparent attitudes and truths mentioned above. Otherwise, that is, if these properties are not existence-entailing and can be had by fictional creatures, Meinongianism has no trouble in accounting for this, although it might have some trouble in explaining our seemingly inconsistent behavior in relation to them. For we do not flee from the theater when we see some of Freddy Kruggers atrocities, and we ! "$&! do not try to go and help E.T. when we see him in his most desperate moments. On the contrary, in fact, we do nothing that could signal a real fear or pity about it. Meinongianism needs to say something more about the determination of fictional truth and the problem of carryover as well. Even if there were fictional creatures, their being, characteristics and story would be determined not only by what the fiction explicitly tells, but also by what it follows from what it explicitly tells and its background assumptions. So the problem remains as to how these or similar things determine what truth in fiction is. The same goes for the problem of carryover. The fact that there are fictional creatures does not seem to shed any light on the issue of what effects on the fictional truth the existence of a follow-up fiction has and how these effects work. Finally, a considerable problem that does disappear if Meinongianism is adopted is the problem of true negative existentials with proper names. This is so because if fictional creatures have some kind of being and we actually refer to them when using the names introduced by fictions, true negative existentials such as are as easily explained as any other ordinary negative attributive utterance. (5) E.T. does not exist. In uttering a sentence such as (5) one is referring to an individual, E.T. in this case, and denying its possession of the property of existence. According to such a view, there is no problem with that precisely because there are things such as E.T. that do not exist. Thus, utterances of sentences such as (5) can simply be true. Meinongianism, thus, seems a very appealing view on the face of it. However, it is not ultimately appealing in spite of all of these advantages. For the core Meinongian thesis that there are things that do not exist seems incomprehensible, and it is because of this that has been widely rejected. It is hard to understand what being without existence might be. A theory that claimed that there is a single ghost which is the thing being referred to by all fictional names and had all the properties that fictions tell their characters have, ! "$'! etc., would also be able to explain all these data, but would be implausible and therefore widely rejected for similar reasons. Besides, we reject Meinongianism partly because we believe that we can explain all these data in a much more plausible and understandable manner. Of course, I do not claim to have refuted Meinongianism. Rather, I am simply joining the generalized attitude in the field, which in addition I think is the commonsense attitude held since the beginnings of this discussion. Creationism Another important view in the literature is the view that fictional characters exist but are abstract objects rather than concrete. These abstract objects are created by the authors of fictions at the same time and in the same way they create the fiction. On this view, i.e. Creationism, such characters as E.T. and Sherlock Holmes exist but are not concrete. Creationism is a metaphysical view that can be complemented with different semantic theses. I will focus here on two views that result from complementing Creationism with two different semantic theses about fictional names: the Ambiguity view suggested by Saul Kripke (1973) and the Non-ambiguity view defended by Nathan Salmon (1998). My own view, which I mainly defend in Chapter 6, is also a version of Creationism, which nevertheless departs from both Kripke’s and Salmon’s views with respect to their semantic theses. The ambiguity view Kripke holds the view that fictional names such as 'E.T.' or 'Sherlock Holmes' are ambiguous. They are both used as pretended names of pretended fictional creatures and as names of those abstract fictional characters that are created at the same time and in the same way that the fiction is. When used as pretended names, they are empty, because there are no fictional creatures to name. ! "$(! More specifically, according to the Ambiguity view, there actually are two homonymous names. The name introduced in telling the story is prior to the other name that starts its existence out of the former. Once the fiction exists and the non-referring name has already been used in pretense, the name to refer to the abstract fictional character created to talk about the fiction as such, from the outside, is then also created. The practice seems to be to use a homonymous name of the pretended name within the fiction to do that. According to this view, utterances of sentences such as (6) and (7) are thus ambiguous. (6) E.T. is an extraterrestrial. (7) Alf is sassier than E.T. There are several ways in which utterances of sentences such as (6) may be intended and interpreted: first, they may be meant to talk about the merely pretended fictional creatures of fictional stories and to say something within the pretense, in which case they would be fictionally true4 at least in the case of (6)5. If so, fictional names are pretended to be used as if they were names that referred. Second, the utterances may be meant to talk about these unreal creatures too, but to say something about reality, in which case they would be meaningless. For fictional names would then be used as names that refer, when they do not. Third, they may be meant to talk about the real and existing abstract fictional characters and say something about reality, in which case they would be false. In this case, fictional names would be used as names of the existing abstract fictional characters and would refer to them, but utterances would be false because they would say that these existing abstract fictional characters have properties that they do not have. Finally, the utterances may be meant to talk about these abstract entities, but to say something about the story by using a special interpretation of the predicates involved, so that 'is an extraterrestrial', for instance, means that the fictional character corresponds to a fictional !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! 4 Notice that fictional truth is just truth in some fiction, or according to some fiction. It is not an especial kind of truth. 5 Thus, this should then be explained, and their content too. For the view also needs to say what the content of the fiction pretended is and how it is determined. ! "%)! creature that is actually an extraterrestrial, in which case they would be false, because there is no such fictional creature6. Let us now see how Kripke’s view can deal with the problem of true negative existentials. True negative existentials are meant as talk about the fictional creatures whose existence is merely pretended. Hence, the names used in true negative existentials such as the following are the non-referring names used by the authors of the fictions in pretending to refer to fictional creatures. (8) E.T. does not exist. The problem is that they are uttered to say about reality rather than fiction that these unreal fictional creatures do not exist. On the one hand, it would follow from the view that these utterances do not say anything, since there is no referent of the name used in them. But, on the other, they are meaningful and true. However, according to Kripke, true singular negative existentials such as (8) are, in fact, a special kind of speech act and express something such as the content literally expressed by an utterance of (9) There is no true proposition that E.T. exists. In general, the account would be that statements of the form (10) ! does not exist !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! 6 One might want to add another possible interpretation to this list along the lines of the later. On this additional interpretation, the predicate ‘is an extraterrestrial’ is interpreted as a shorthand for ‘is an extraterrestrial character in the fiction’. Notice though that, even if this interpretation (if plausible) would be literally true, it would amount to the interpretation that results from prefixing the ‘according to the fiction’ operator plus claiming that the fictional name ‘E.T.’ refers to an existing but abstract character. Thus, (6) would be true according to this interpretation, but as an effect of the prefixation. In fact, I discuss in more detail views that make these sorts of claims in what follows. ! "%"! express a proposition of the form (11) There is no true proposition that ! exists and the proposition expressed will be true for one of two reasons: either because there is a proposition, but it is not true, or because there is no proposition. If ! is a definite description such as 'the king of France', the statement would be true for the former reason, but if ! is a fictional name, the negative existential would be true because of the latter. In fact, analogous interpretations are available whenever negation is used, as in the case of 'E.T. is not homesick' for example, which could be interpreted as expressing that there is no true proposition that E.T. is homesick. Further, the view would be extended to other non-fictional empty names such as 'Vulcan'. A fatal problem that the Ambiguity view has is that it leaves content of fictional discourse unexplained. It passes the problem from non-referring names to more complex expressions containing them. For if 'E.T.' does not mean anything or has no content, neither do complex expressions containing that name such as 'that E.T. exists', 'the proposition that E.T. exists', or 'According to the film E.T. The Extraterrestrial, E.T. is an extraterrestrial'. In spite of these problems, there is something intuitively right about the idea that talk about E.T. or Sherlock Holmes as fictional characters from outside the fiction starts once the fiction is finished. It does not seem to be that the authors of fictions start using these fictional names to refer to any existing fictional character at all. This is why one may find the ambiguity thesis attractive. The non-ambiguity view However, Nathan Salmon (1998) does not seem to find ambiguity appealing. On the contrary, he holds the view that fictional names are not ambiguous. According to his ! "%*! view, since the very moment that they get meaning, all of their relevant uses name abstract fictional characters. According to the Non-ambiguity view, the authors of fictions create fictional characters when telling their fictions. These created fictional characters –like the stories told- are abstract artifacts in reality, but according to the fictions they are different things. According to many fictions, in fact, they are not abstract even if they are so in reality. On the other hand, fictional names are also introduced by authors in storytelling in pretending to refer to fictional individuals whose stories they are telling. In spite of the fact that the fictional characters already exist, fictional names at the beginning do not refer to them. Authors do not intend to refer to them in introducing these fictional names either. They only pretend to use them to refer to fictional individuals. Initially, in fact, there are no uses of names (that is, these expressions used at the beginning are not, contrary to appearances, names)7. It is only later that the merely pretended use of the name within the fiction becomes a real use of the name in reality to name the created abstract artifact that is the fictional character. This abstract artifact is, on Salmon’s view, though, the same thing that the name was already pretended to refer to in the fiction, only that the name now really names it. Let us now consider an example in order to better understand what this view is. According to this view, in telling us the story E.T. The Extraterrestrial, Steven Spielberg created E.T., a fictional character, which is an existent man-made abstract artifact. When Steven Spielberg introduces the name, he merely pretends to use it to refer to a fictional alien, but he does neither actually do so nor intends to do so. This individual that, according to E.T. The Extraterrestrial, is an alien is in fact the abstract fictional character E.T. that Spielberg has created, even though within the story is not abstract at all, but !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! 7 Notice that the claim that these expressions used are not names is different from the claim that the uses of these names are non-referring even if they are names. To illustrate the difference contrast the view that fictional names are empty names, that is names that do not refer, with the view that these expressions even if homonymous to proper names are not names after all. The difference resides in the acceptance of the possibility of names that are used as names but as not referring at the same time, on one side of the debate, and in the identification of proper names with expressions used (with the intention) to refer, in the other. ! "%+! rather a non-abstract alien. So, at first, there is an abstract fictional character, but there is no use of the name ‘E.T.’ at all. There is no alien in reality either, only an abstract artifact that according to the story is an alien, but it is not. When, afterwards, the name ‘E.T’ starts to be used, it refers to the fictional character that already existed without a name but that was the fictional referent of the name since the very beginning. Therefore, the name ‘E.T’ is not ambiguous because once it gets a use, it is to refer to this abstract fictional character that Steven Spielberg created in creating the story E.T. The Extraterrestrial. Contrarily to what Kripke said, the name is never non-referring because it never gets a use in which the name is non-referring. Strictly speaking, it (i.e. the expression used) is never used as a name before, and therefore, according to this view, is not a name. This view has some important disadvantages. One of them is that the view is committed either to an implausible semantics of certain fictional sentences or to an implausible metaphysics of fictional characters. For instance, it follows from this view, initial utterances of (12) are meaningless, since the name ‘Sherlock Holmes’ does not have a meaning. (12) Sherlock Holmes smokes a pipe. In fact, it is not even used as a name at all, but only pretended to be so. Once ‘Sherlock Holmes’ is used as a name of the abstract fictional character Sherlock Holmes, utterances of (12) are made to predicate something true about it. However, there seem to be two options on such view: first, utterances of (12) are asserting that an abstract entity has the property of being someone who smokes a pipe, and for an abstract fictional character to have a property such as being someone who smokes a pipe is for it to represent a fictional individual who fictionally possesses that property. But there is no independent motivation for holding this metaphysical thesis. On the contrary, one would have thought that smoking a pipe is the same thing for every entity, be it abstract or concrete. The only motivation that I see for denying this is to ad hoc solve this problem. Second, utterances of (12) instead claim of the abstract entity Sherlock Holmes that it represents a fictional individual who fictionally possesses the property of being someone who smokes a pipe, ! "%#! in which case the meaning assigned to the predicate ‘smokes a pipe’ in these contexts is far-fetched. Let us now go back to the problem of true negative singular existentials and consider again the following example. (13) E.T. does not exist. According to Salmon’s view, contrary to our intuitions, utterances of (13) are false, because according to it, utterances of (13) claims that the abstract fictional character E.T., which exists, does not exist. However, by uttering negative existentials such as (13), we manage to pragmatically communicate something true, which in the case of (13) would be something like that the fictional alien represented by the abstract fictional character E.T. does not exist, which is true. Our intuitions would be due to mistaking pragmatics for semantics and would thus be explained away by this pragmatic explanation. Even if the Non-ambiguity view defended by Salmon offers possible solutions to many of the problems discussed above, the theory seems prima facie implausible in that, as explained above, runs against our initial intuitions. It would be nice to have a theory without these counterintuitive added complexities. On the other hand, it is worth noting that Millianism is preserved by a theory like Salmon’s and, therefore, to the extend that Millianism is an attractive view on the semantics of proper names, this is an important good reason in favor of such a view. However, such a motivation vanishes for anyone who is not a Millian, like myself. Kroon’s pragmatic pretense-based theory DFN, the view that I propose and defend in Chapter 6, is a version of Creationism, but one according to which fictional names do not refer. This is why I think it is worth ! "%$! considering here a family of views that agree with DFN on the thesis that fictional names do not refer. To that effect, I focus on the recent proposal of Frederick Kroon (2004, 2009), which belongs to a broad family of views, according to which sentences containing empty names do not even semantically express any proposition, but we use them, and pretend them to be true, in participating in games of make-believe. This is in a nutshell the proposal of Gareth Evans (1982)8 and Kendall Walton (1990). I first discuss Kroon’s positive view and then I turn to his criticism to what he calls ‘the Ellipsis Strategy’. This is also especially relevant because DFN makes use of the ellipsis strategy as well, and I deem it important to defend this strategy from Kroon’s arguments. As I argue below, Kroon’s arguments are not successful. According to Kroon, by means of pretense, speakers who utter negative existentials such as (14) manage to assert something informative, even if these negative existentials are otherwise contradictory. (14) E.T. does not exist. In general, when speakers utter contradictory sentences such as (14), they pretend that they refer to something and then they use this pretense to finally assert that there is no such individual. According to this view, what speakers assert by uttering such sentences is not what these sentences would literally mean. It is not semantically, but pragmatically, expressed. For instance, in uttering (14) speakers pretend that they refer to some individual that exists by their use of the name ‘E.T.’, and use the resulting interpretative tension to assert, in part, something as follows: (15) Outside of the pretense that there is a unique actual individual called ‘E.T.’ whom I !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! 8 Notice that even if Kroon and Evans defend descriptivist accounts of the semantics of proper names, the pretense-account of fiction that they both defend is not descriptivist. That is, neither Kroon nor Evans account for the content of utterances of sentences containing fictional names in terms of a semantically expressed descriptive proposition. ! "%%! have seen or heard described as a homesick alien who wants to go back home, etc., there is no such individual. We can therefore see how Kroon solves the problem of true negative singular existentials. He gives the same kind of explanation for other sentences containing empty names, as well as informative identity statements. Let me very briefly mention several worries that I have about this account. First, in my opinion, this proposal does not make sufficiently clear how these contents, that according to it we communicate in uttering sentences with empty names, are pragmatically conveyed. Kroon rejects a Gricean explanation of them. Second, the account explains all content communicated by sentences with empty names as pragmatically conveyed. That is, we never use these sentences to communicate their literal content. I find this suspicious for at least the following two reasons: one reason is that it strikes me as odd that no utterances of sentences in which empty names occur, which are nevertheless heavily used, have literal content. This would mean that we use sentences without meaning a lot, while understanding each other without problems nonetheless. The other reason is that it is true that pragmatics plays a major role in communication, but it also seems true to me that pragmatics crucially rests on semantics and the literal contents of the sentences uttered. The Gricean explanation of conversational implicatures is a paradigmatic case of a plausible pragmatic explanation, but it crucially depends on the semantics in the sense that it starts from the semantic content literally expressed by utterances of sentences. Given that according to pretense-views like Kroon’s, utterances of sentences containing empty names do not semantically express any proposition, this kind of pragmatic explanations for the other propositions (which according to it are pragmatically conveyed) is not available to them. Third, the pragmatic content that according to Kroon’s proposal gets communicated explicitly talks about pretense, when it does not seem that people uttering or hearing these sentences are so conscious about the pretense that might be involved in them, and at the very least, do not seem to be saying anything about it. However, notice that this does not mean that they are not conscious that they are talking about fiction. Quite the contrary, I think that they are fully aware of ! "%&! that. Moreover, notice that neither this means that I think that they are never pretending when discussing fiction. Rather, I think they often are. Where I disagree with a pretenseview like Kroon’s is in that I think it implausible that speakers are conscious of their pretense every time they engage in any pretense, and further, that they even communicate about the pretense they engage every time they use sentences containing empty names. Fourth, the proposal does not make a distinction between the fictional case and other cases that involve non-fictional empty names such as ‘Vulcan’, in which it is not even clear that there is ever any pretense involved. Having outlined and briefly discussed Kroon’s pragmatic proposal, let me now focus on an argument that Kroon advances for his view -an argument that in fact has a much wider interest and motivation. Kroon argues that the typical Russellian treatment of negative existentials that contain definite descriptions does not work. To illustrate his point he asks us to consider the following complex existential (16) The golden mountain –you know, the elusive mountain that Dr. Zak wrote about in his book on famous journeys of exploration- does not in fact exist. According to the typical Russellian treatment, this is understood with negation taking a wide scope over the description. However, against this widespread treatment, Kroon interprets (16) with negation taking a narrow scope instead. Kroon argues as follows. (17), interpreted in its turn as (18), cannot be (16)’s interpretation because (16), but not (17) or (18), is false in some worlds in which there is no golden mountain and Dr. Zak did not write any book about journeys of exploration. (17) The entity that is at once the golden mountain and the elusive mountain that Dr. Zak wrote about in his book on famous journeys of exploration does not exist. (18) It is not the case that there is a unique entity that is at once the golden mountain and the elusive mountain that Dr. Zak wrote about in his book on famous journeys of ! "%'! exploration. Kroon further argues that (19) cannot be an interpretation of (16) either, because it requires both the existence and the non-existence of the golden mountain. (19) The golden mountain does not exist, and (the golden mountain = the elusive mountain that Dr. Zak wrote about in his book on famous journeys of exploration) To see this notice that on the typical Russellian treatment, the first conjunct of (19) would be interpreted as (19’) and the second conjunct, as (19’’). (19’) It is not the case that there is a unique golden mountain. (19’’) There is a unique golden mountain and there is a unique elusive mountain that Dr. Zak wrote about in his book on famous journeys of exploration and they are the same thing. Thus, on the typical Russellian treatment, (19’) asserts the non-existence of the golden mountain while (19’’) asserts the existence of the golden mountain. Therefore, on the typical Russellian interpretation, (19) cannot be true, while (16) can. Kroon considers other initially possible interpretations of (16), but all of them turn out to be as inconsistent as (19). According to Kroon, the use of definite descriptions in negative existentials “typically feels as if it brings with it a commitment to there being an object of a certain sort, even as the speaker denies that there is such an object.” (2009, pp. 374-375). He claims to be able to account for this by appealing to pretense. Like in the cases of true negative existentials involving empty names, Kroon proposes to explain this as follows: utterances of (16) semantically express something inconsistent, such as (16’) (interpreted with negation taking a narrow scope over the existential ! "%(! quantifier), but (by means of pretense) manage to pragmatically express and assert something which is not, such as (20). (16’) There is a unique golden mountain that does not exist. (20) Dr. Zak used the description ‘the golden mountain’ in his book on famous journeys of exploration and no one has ever discovered such a mountain. It is not the case that there is a unique golden mountain. Against what is commonly accepted, Kroon claims that this is generally the case: in general, true negative existentials involving definite descriptions should be interpreted with negation taking a narrow scope under the existential quantifier. However, in my opinion, Kroon’s interpretations should be resisted. It seems that the examples that he discusses to make his point admit of other readings compatible with the usual Russellian treatment of them, according to which negation takes a wide scope over the existential quantifier. These are cases in which something like ‘alleged’ is elided as part of the definite descriptions involved. Thus, for instance, I would treat (16), repeated here as (21), as (22), interpreted in its turn as (23). (21) The golden mountain –you know, the elusive mountain that Dr. Zak wrote about in his book on famous journeys of exploration- does not in fact exist. (22) The alleged golden mountain –you know, the alleged elusive mountain that Dr. Zak wrote about in his book on famous journeys of exploration- does not in fact exist. (23) It is not the case that the entity that is both the alleged golden mountain and the alleged elusive mountain that Dr. Zak wrote about in his book on famous journeys of exploration exists. Further, taking out the predicate ‘exists’ from the end of (23) in order to avoid ! "&)! redundancy, we get the following simpler formulations (24) It is not the case that there is a unique entity that is both the alleged golden mountain and the alleged elusive mountain that Dr. Zak wrote about in his book on famous journeys of exploration. or (25) It is not the case that there is a unique entity that is both the golden mountain and the elusive mountain that Dr. Zak wrote about in his book on famous journeys of exploration as it was alleged. Unlike (17) and (18), but like (21), (22)-(25) would be false in the world considered before and would not be inconsistent like (19). (22)-(25) would be true only in those possible worlds in which at least someone thought of or pretended that an alleged golden mountain of a certain sort existed; in particular, an alleged golden mountain that had allegedly also been an elusive mountain that Dr. Zak wrote about in his book on famous journeys of exploration. Further, it would also account for the pretense phenomenology that, as Kroon points out, accompanies these statements. Thus, they would be good interpretations of the complex existential (21). It might be worth making explicit here what the constraints on an admissible interpretation of (21) are. First, an admissible interpretation should capture the intuitive meaning and truth-value of the target sentence. Second, an admissible interpretation should capture its modal truth-conditions -that is to say, it should capture the truth-value that the sentence obtains at each possible world. And this is precisely the constraint that the interpretations initially considered by Kroon fail to satisfy and the reason why he correctly rejects them. Third, an admissible interpretation should be compositionally obtainable given the expressions that explicitly constitute the sentence. That is to say, the interpretation cannot be one that we would have thought to be the literal interpretation of a too different sentence. Forth, an admissible interpretation should not be ad hoc. That is, ! "&"! it should not arise in virtue of an especial linguistic phenomenon that only obtains in the cases that one is trying to explain. Finally, one might add as a fifth constraint the capacity to explain related phenomena like, for instance, the phenomenology that accompanies the phenomena to be explained. All the interpretations that I am offering meet all of these constraints, including the fourth. We do use expressions such as ‘alleged’ or ‘allegedly’ in other contexts with similar communicative purposes, and the Russellian account of the interaction of definite descriptions with other operators is not exclusive to negative existentials. Consider a real written utterance of a sentence that illustrates my claim that we do seem to use elided expressions such as ‘alleged’ or ‘allegedly’ in different contexts. The example illustrates that this phenomenon is indeed usual and not restricted to philosophical or fictional discourse. The following is a headline that appeared in a local Catalan newspaper: (26) The English kid who had a son when he was 13 discovers that he is not the father. If there is no elided expression in (26), its utterance is obviously false. However, this utterance could be the utterance of a longer sentence than (26). The idea would then be that some expressions would have been elided in (26). If this were the case, there would be a clear reason why these expressions would have been elided. (26) would be more useful to convey in a short economic journalist fashion what true utterances of the longer sentences (27) and (28) express. (27) The English kid who we all thought had a son when he was 13 discovers that he is not the father. (28) The English kid who allegedly had a son when he was 13 discovers that he is not the father. ! "&*! Yet here is another real example of the same phenomenon that I recently found: (29) After several analyses of her remains, the parents of a girl who died in 1990 four days after being born confirm that these do not correspond to their daughter. This last example is taken from a report entitled ‘Stolen children’ from a Spanish magazine called ‘Interviu’. In my view, this is also a clear example of a written utterance of a sentence that has some of its parts elided. The corresponding longer sentence, with no expressions elided, would be the following: (30) After several analyses of what were supposed to be her remains, the parents of a girl who allegedly died in 1990 four days after being born confirm that these do not correspond to their daughter. After all, this is clearly what competent speakers understand when reading these sentences and what the reporters intended to communicate by writing them. The explanation that I am offering here, which appeals to the existence of an elided expression such as ‘alleged’, is a version of what Kroon calls ‘the Ellipsis strategy’, an strategy that DFN (the view that I defend in Chapter 6) adopts. Kroon criticizes this strategy, arguing that it does not provide correct interpretations of examples like (21), which I considered above. But I think it does, as I have shown above and I show in what follows. Let us now see this with the following example that Kroon also asks us to consider. (31) The “splendid and immaculate” hotel you booked me into was neither immaculate nor splendid. I propose to interpret this as (32), which, in turn, is interpreted as (33). ! "&+! (32) The allegedly splendid and immaculate hotel you booked me into was neither immaculate nor splendid. (33) There is a unique allegedly splendid and immaculate hotel you booked me into that was neither immaculate nor splendid. It should be noticed that the specific interpretations of the ellipsis strategy that Kroon objects to are different to the ones I propose. In the case of (31), for instance, Kroon objects to the following. (34) The hotel you booked me into, which you described as ‘splendid and immaculate’, was neither immaculate nor splendid. According to Kroon, there is no syntactic or semantic evidence in favor of such a strategy, but there is much evidence against it. His objections are the following. He first adduces the fact that there is no explicit occurrence of a phrase like ‘described as’ and then objects that it is implausible to think that the expression ‘splendid and immaculate’ is being mentioned rather than used, even if it appears within quotation marks, because the speaker could have used finger quotes, a certain tone of voice, or even the recognition by her audience of her expectation to be understood as speaking from another’s perspective when uttering this expression, rather than used these quotation marks, and she would have managed to assert the same. In my opinion these objections are not good. For, first, one often utters sentences such as (34), (32) or (33), instead of (31), making the elided phrase explicit to seemingly make the same point. Secondly, the fact that a sentence such as (31) can be uttered without any kind of quotation marks, or even without any tone of voice, does not show that the expression that appears within quotation marks in (31) is not being somehow mentioned or used as part of a demonstration of something, and that the claim that one makes by uttering (31) is not made by precisely these means. In fact, finger quotes, a tone of voice, or even speakers’ intentions might perfectly play the role that written quotation marks ! "&#! play in paradigmatic cases of quotation. Kroon also objects that the readings proposed by the ellipsis strategy that he considers, in which it is explicitly said that the hotel has been described as ‘splendid and immaculate’, miss the irony that exists in utterances of sentences such as (31). According to Kroon, this irony is even more clearly present in sentences such as (30). (35) The “splendid and immaculate” hotel you booked me into –you know, the one where maids vacuum the floors every hour and the guests are mostly Saudi princes- was neither splendid nor immaculate. Quite the opposite, I think that Kroon is wrong in this respect as well: strong irony or mimicking may be present even if the point is made more explicitly by explicitly uttering phrases such as ‘that you described as’ like in (36) or phrases such as ‘allegedly’ like in (37). (36) The hotel you booked me into which you described as ‘splendid and immaculate’ – you know, the one that according to a description like yours would have maids vacuum the floors every hour and guests would be mostly Saudi princes- was neither splendid nor immaculate. (37) The allegedly splendid and immaculate hotel you booked me into –you know, the one where allegedly maids vacuum the floors every hour and the guests are mostly Saudi princes- was neither splendid nor immaculate. That is, irony is something extra to what is said: some claims, even (36) and (37), might be made with irony or without it. This is not to deny that the irony works better if certain things are not made explicit, of course. This might be precisely why, some tend to utter sentences like (36) or (35) rather than sentences like (34), (32), (33), (36) or (37). The same goes for the other utterances that Kroon considers when making this objection, ! "&$! such as (38) and (39), which according to the ellipsis strategies would be interpreted as (40), or as (41), and as (42), or as (43), respectively. (38) Do you have any other “splendid and immaculate” hotels to show me? (39) You know the “splendid and immaculate” hotel you booked me into? Amazingly enough, and contrary to all the logic I know, it was neither immaculate nor splendid. (40) Do you have any other hotels that you would describe as ‘splendid and immaculate’ to show me? (41) Do you have any other allegedly splendid and immaculate hotels to show me? (42) You know the hotel you booked me into which you described as ‘splendid and immaculate’? Amazingly enough, and contrary to all the logic I know, it was neither immaculate nor splendid. (43) You know the allegedly splendid and immaculate hotel you booked me into? Amazingly enough, and contrary to all the logic I know, it was neither immaculate nor splendid. Finally, Kroon objects that the ellipsis strategy would render false the otherwise intuitively true Russellian interpretation of negative existentials such as (44) that arise out of a restrictive reading of the embedded relative clause ‘which is “splendid and immaculate”’ of sentences such as (31). (44) The “splendid and immaculate” hotel that you booked me into does not exist –the rooms are a disgrace. Here he does not say it, but I guess that Kroon takes the true reading to be (45) and that he takes the Russellian interpretation on the ellipsis strategy, though, to be (46) which is ! "&%! false, given that he did book her into a hotel which he described as ‘splendid and immaculate’. (45) It is not the case that there is a unique hotel you booked me into, which was “splendid and immaculate” –the rooms are a disgrace. (46) It is not the case that there is a unique hotel you booked me into that you described as ‘splendid and immaculate’ –the rooms are a disgrace. According to my version of the ellipsis strategy, (44)’s Russellian interpretation would turn out to be as (47) instead, which in turn Kroon would take to be interpreted as the false (48). (47) It is not the case that there is a unique allegedly splendid and immaculate hotel you booked me into –the rooms are a disgrace. (48) It is not the case that there is a unique hotel you booked me into which was allegedly splendid and immaculate –the rooms are a disgrace. However, the ellipsis strategies are not forced to take these false readings of the Russellian interpretation of (44). Contrarily, the right readings seem respectively to be instead (49) and (50), which are true, rather than false as objected. (49) It is not the case that there is a unique hotel you booked me into which was splendid and immaculate as you described it to be –the rooms are a disgrace. (50) It is not the case that there is a unique hotel you booked me into which was splendid and immaculate as it was allegedly to be –the rooms are a disgrace. There is nothing that prevents these true readings of (44) according to the ellipsis strategies and so Kroon’s final objection does not work either. ! "&&! Let me remind you here that Kroon is arguing that the strategy is unable to provide right interpretations of the sentences considered if it follows the traditional Russellian account of the definite descriptions and their interactions with other operators such as negation. Against this, I am showing that such an ellipsis strategy can, and does, provide such correct interpretations, which do meet the usual constraints placed in these discussions, which I explicitly indicated above. Kroon then considers negative existentials whose truth does not involve any object at all, such as the complex and long (51) uttered with sarcasm. (51) The fabulous object that Jones discovered –the golden mountain, no less, as is so clearly shown (so Jones keeps reminding us) by the golden streaks on the last slide- is, sad to say, neither golden nor a mountain. Look closely enough, and you’ll see that it does not even exist. Jones’s fabulous object is just a trick of the light –there is nothing there. According to Kroon, the ellipsis strategy that he considers cannot explain (51) either. Unlike in the case of the hotel, in that other case there is not even a thing to be described. However, note that even if the version that Kroon considers does not work, my version of the ellipsis strategy does work in that case too, and so contrarily to what Kroon assumes, there is an ellipsis strategy that works. According to this version, (51) is taken to be an abbreviation of (52) The alleged(ly) fabulous object that Jones allegedly discovered –the alleged(ly) golden mountain, no less, as is allegedly so clearly shown (so Jones keeps reminding us) by the golden streaks on the last slide- is, sad to say, neither golden nor a mountain. Look closely enough, and you’ll see that it does not even exist. Jones’s alleged(ly) fabulous object is just a trick of the light –there is nothing there. Therefore, my proposed version of the ellipsis strategy gives a simpler and unified account of all cases, and at the same time agrees with the widely hold thought that ! "&'! Russell was right on negative existentials involving descriptions. Summing up, I have discussed Kroon’s Pragmatic Pretense-based account of sentences with empty names, and his arguments against the typical Russellian treatment of negative existentials involving definite descriptions and the ellipsis strategy that I favor. If I am right, Kroon’s arguments do not succeed in showing these to be wrong. In the present section I discussed four major views on fiction: Meinongianism, two versions of Creationism (the Ambiguity view and the Non-ambiguity view), and the Pragmatic Pretense-based account. In the next section, I discuss yet another theory: a descriptivist view that is close in certain respects, but that is importantly different, to the descriptivist view that I defend in Chapter 6. 5.3. David Lewis and Gregory Currie’s Descriptivism In this section, I discuss what I call ‘The Lewis-Currie view’, which is the descriptivist account of fiction defended by David Lewis (1983) and Gregory Currie (2003). As I said, I consider this descriptivist view because I propose a different descriptivist view in Chapter 6, DFN. It is therefore important for my purposes to explain why I think that the Lewis-Currie view is not good enough while DFN is. In doing this, I argue that the fact that this descriptivist view of fiction does not work does not entail that no descriptivism about fiction succeeds. An interesting issue that arises within the topic of fiction is the issue of how to account for the intuitive contingencies of fictional characters. For at least some of the things that occur to fictional characters within a story are supposed to happen only contingently, and this might initially appear to conflict with other assumptions that we make about the modal properties of fictional characters. I discuss the Lewis-Currie view in the light of this problem because I think that the explanation of fictional contingencies that this view provides reveals what I take to be a good reason to reject it. Moreover, in addition to ! "&(! arguing that the view is mistaken, I offer an sketch of what I take to be the explanation of these contingencies. In short, I argue that fictional contingencies, such as the one that in Tolstoy’s Anna Karenina, Anna Karenina might not have fallen for Vronsky, pose a serious problem to the Lewis-Currie view, because the view cannot account for the fact that in Tolstoy’s Anna Karenina, it is Anna Karenina herself who contingently falls for Vronsky. That is, it cannot account for the fact that in Tolstoy’s Anna Karenina, Anna Karenina falls for Vronsky in the actual world but she fails to fall for him in some other possible world. Let me start now by introducing what I take to be essential to the Lewis-Currie view. The Lewis-Currie view is characterized by the following two theses: Russellian Descriptivism of fictional names: fictional names, unlike ordinary proper names, are, or are used by the author of the fiction, as possibly satisfiable definite descriptions; except for inconsistent fictions perhaps (and by ‘possibly satisfiable definite descriptions’ I just mean definite descriptions that are satisfied by some individual at some possible world). For instance, the name ‘Sherlock Holmes’ is, or is used as, the abbreviated possibly satisfiable definite description that could be extracted out from what Conan Doyle writes in the Sherlock Holmes Stories: i.e., something like ‘The man who is called ‘Sherlock Holmes’ and is a detective that smokes a pipe and lives in Baker Street at London and whose friend is Watson, and so on’ Possibility view of fiction: fictional stories that are not (essentially) inconsistent (or close consistent versions of them) are possibly true if not actually true. For instance, according to this view, Conan Doyle’s Sherlock Holmes does not actually exist because it is merely fictional, but could have existed. Thus, Sherlock Holmes is possible in the sense that there is some non-fictional individual who is Sherlock Holmes at some possible world other than the actual. Hence, the Sherlock Holmes stories are not ! "')! actual non-fictional stories, but are possible non-fictional stories as well. Notice that the Possibility view of fiction is in fact a consequence of the Descriptivism of fictional names, for according to the latter the bearer of the name will be any individual that satisfies the definite description that the name abbreviates, and such an individual is possible, even if not actual, according to this view. And this is so, thanks to the fact that the definite descriptions that the fictional names abbreviate are, again, according to the view, possibly satisfiable in the sense that they are satisfied at some possible world other than the actual. Notice also that in spite of this, the Lewis-Currie view agrees with Kripke’s point that fictional names used in storytelling do not (and cannot) actually refer (and the descriptions they would abbreviate, according to the Lewis-Currie view, do not actually denote either), because they are used by the author of the fiction with the intention of telling a fiction and without the intention to refer. Let us now consider the problem of fictional contingencies that the Lewis-Currie view faces. The problem of fictional contingencies Following Currie, consider utterances of the following pair of sentences: (1) Necessarily, someone who did not fall for Vronsky would not be Anna Karenina. (2) Someone who necessarily fell for Vronsky would not be Anna Karenina. Prior to careful reflection, utterances of (1) and (2) might seem to show an initially apparent inconsistency between them. However, at the same time, each of them seems intuitively true. On the one hand, utterances of (1) seem true because arguably anything ! "'"! that a fictional story tells about its characters is essential to them. Tolstoy’s story about Anna Karenina tells us, among other things, that Anna Karenina falls for Vronsky. Hence, unlike what happens to non-fictional people like you and me, and due to its fictionality, it is a constitutive feature of Anna Karenina that she falls for Vronsky. Thus, it is necessary that she does. On the other hand, utterances of (2) seem true because Tolstoy’s story is not a story in which Anna Karenina cannot but fall for Vronsky, but a story in which Anna Karenina falls for Vronsky only contingently. Thus, anyone who necessarily fell for Vronsky, that is to say, who fell for Vronsky non-contingently, would not be Anna Karenina. Notice that the initially apparent incompatibility between utterances of (1) and (2) cannot be explained in terms of the distinction between truth in fiction and truth simpliciter, or any other similar distinction. For both, utterances of (1) and of (2), seem to be true in one and the same reading, or at least, none of them is true in the fiction. Rather, they are about the fictional character Anna Karenina as such, as a fictional character. They specify some of Anna’s necessary qualities. Note too that appeal to the ambiguity in scope due to the interaction between modalities and definite descriptions in (1) and (2) does not help either. For the problem is that we are dealing with fiction and fictional names. Hence, there are no actual individuals that actually are these fictional characters, only actual individuals that satisfy the definite descriptions in question but at some possible world other than the actual. Thus, we can explain the consistency of utterances of (3) and (4) by noticing the distinction in scope of the occurrences of the definite description ‘the Queen of England’ in (3) and (4), and explain that utterances of (4) can be true compatibly with the truth of utterances of (3), because there is an individual –i.e. the Queen of England- who can exist in another possible world and fail to be the Queen of England in it. (3) Necessarily, the Queen of England is queen. (4) The Queen of England may not have been queen. ! "'*! As I said, unlike in the case of fiction, this is possible precisely because there is in fact an individual who is the Queen of England in the actual world, whereas there is no such individual denoted by the definite description that the fictional name ‘Anna Karenina’ allegedly abbreviates. The descriptivist way out of the problem The Lewis-Currie view would accept the truth of both claims, (1) and (2), and would explain it as follows: Anna Karenina possibly exists. That is to say, even if -as we all agree- Anna Karenina does not actually exist, there is some other possible world where she does. For to be Anna Karenina is simply to play the Anna Karenina-role, and to play the Anna Karenina-role merely amounts to satisfying the general possibly satisfiable definite description that the name ‘Anna Karenina’ abbreviates. According to this view, what one does when telling a fiction is to tell a story, which although not actual, is possible. It is to qualitatively describe part of some possible worlds other than the actual. It is to explain some ways the actual world might have been but is not. Thus, the view is that Anna Karenina could have existed and fallen for Vronsky, even if in fact this never occurred and will never do in actuality. That Anna Karenina falls for Vronsky is as possible as my turning off my laptop in a moment. What would explain the truth of utterances of (1), according to the Lewis-Currie view, is the fact that there is no possible world where someone plays the role of Anna Karenina but does not fall for Vronsky. This is so precisely because part of what it means to play this role is to fall for Vronsky. Thus, it is true in every world that anyone who plays the Anna Karenina-role in that world falls for Vronsky. Nevertheless, according to the Lewis-Currie view, utterances of (2) would be true as well because for every person who plays the Anna-role in some possible world, there is at least one more world where that same person does not fall for Vronsky, i.e. a world where she does not play the role of Anna Karenina. This would be so because it is a ! "'+! modal fact (or at least we are assuming it to be so, given the appearances and the lack of a reason to the contrary) that people do not necessarily fall in love. That is, those who fall in love for someone could have not done it. Hence, the existence of these other possible worlds is what would explain the contingency of the falling for Vronsky by Anna Karenina. This is then the solution offered by the Lewis-Currie view to the problem of fictional contingencies. However, I think that this solution is unsatisfactory. The explanation does not explain what it has to explain, that is, the fact that in the fiction, Anna Karenina has the property of falling for Vronsky but only contingently so. This amounts to the fact that Anna Karenina herself must have the property in every story-world –i.e. a world where the Anna Karenina-role is satisfied. But at the same time she (Anna Karenina and no one else) must fail to have the property of contingently falling for Vronsky, while being Anna Karenina at some world, which must be possible with respect to the story-world. But it is Anna Karenina herself who must have the contingent property at one world and lack it at another. This is what contingency means. Otherwise, it is not true that Anna Karenina falls for Vronsky in a contingent way, but that someone else does. The problem is that the only way for the Lewis-Currie view to try to explain this contingency is by appealing to possible worlds in which there is an individual who does not fall for Vronsky and hence is not Anna Karenina, but exists in yet another possible world where she does fall for Vronsky, occupies the Anna Karenina role and hence is Anna Karenina there. This individual has the property of falling for Vronsky contingently because she has it in some worlds but lacks it at others. But she is not Anna Karenina in all of these worlds. According to this account, there is no world in which Anna Karenina lacks the property of falling for Vronsky. Therefore, this account cannot explain her having this property merely contingently. In other words, there is no single individual which, at the same world, both is Anna Karenina and lacks the property of falling for Vronsky. But being Anna Karenina requires precisely that, even if Anna Karenina is a fictional individual. ! "'#! It is worth noticing that the problem just discussed is not just the Humphrey Objection famously raised by Kripke to David Lewis’ Counterpart Theory -even if it might be thought to be a variant of it, one that is used against a different, though importantly related, theory. The Humphrey Objection is the objection that modal truths about actual things cannot be explained or be about other distinct possible things. For instance, the modal truth that Humphrey could have won the election cannot be explained by the existence of a counterpart of Humphrey winning the election in another possible world. It is Humphrey himself, and not one of his counterparts, who is relevant for explaining possibilities about Humphrey. Both the Humphrey Objection and the objection discussed above are raised against attempts to explain modal properties in terms of something other than the subject of those properties. An important difference, however, is that in the case of fictional contingencies, there is no actual individual who actually is Anna Karenina. The problem of fictional contingencies lies in what it takes to be Anna Karenina according to the Lewis-Currie view: to be Anna Karenina in some world is to satisfy at it the definite description that the fictional name ‘Anna Karenina’ abbreviates. In other words, the problem arises from Russellian Descriptivism of fictional names rather than Counterpart theory. Unlike in the case of Humphrey and other non-fictional people, in the currently discussed fictional case there is no actual individual who actually satisfies the description in question to start with. What is more, Counterpart theory is not indispensable to the Lewis-Currie view. Hence, for all these reasons, even those who do not like Lewis’ Counterpart theory partly because of the Humphrey Objection might find Russellian Descriptivism about fictional names appealing in spite of the present objection, be it a variant of Kripke’s Humphrey Objection or not. After all, what the Lewis-Currie view claims is that Anna Karenina is anyone who satisfies the possibly satisfiable definite description in some non-actual possible world. I see no way in which the Lewis-Currie view can handle this problem. However, I examine some possible replies in what follows. ! "'$! One might find Currie’s explanation of fictional contingencies plausible merely due to a natural tendency to forget what the Lewis-Currie view tells us being Anna Karenina consists in. As a result, one could come to have the following thought: that this person who does not fall for Vronsky in some world in which she does not occupy the Anna Karenina-role is, nevertheless, Anna Karenina also in such a world due to the fact that she is Anna Karenina in one of the worlds of the story, where she does occupy the Anna Karenina-role and does fall for Vronsky. But to evaluate the view under this impression is to misunderstand it, as I explain below. If that other person were to be Anna Karenina in any sense also in this other world where she does not fall for Vronsky, utterances of (1) would not be true. It would not be a necessary condition for being Anna Karenina to fall for Vronsky, for there would be some possible worlds where Anna Karenina would not fall for him. These would precisely be the worlds where someone who occupies the Anna-role in one of the storyworlds exists and does not fall for Vronsky. As I argued above, however, there is no such sense for the case of being Anna Karenina. To think of that person, let’s say Jane, as being Anna Karenina also in that other world where she does not fall for Vronsky only because she does occupy the Anna Karenina-role at some world, it is to mistake what being Anna Karenina is, on such a view, for what being Jane (or, in fact, any other real person) is. Currie (2003) explains this as follows: Now consider Jane, a respectable inhabitant of the actual world. In the actual world she does not fall for Vronsky; in fact she never meets him. But, given what I have said just now, it may well be the case that Jane in some other world does fall for Vronsky; in that other world, Jane occupies the Anna-role. Does that make Jane, in this world, Anna Karenina? No. Being Anna is, according to me, something that happens to you in some worlds and not in others. It happens to you in worlds where you occupy the Anna role. In any world in which Jane occupies that role she is Anna. But that does not make her Anna in this world. Being Anna is not at all like being Jane. The person who is Jane in one world is Jane in all worlds. Being Jane is a matter of being a certain individual; being Anna, on the other hand, is a matter of occupying a certain role. Moving up a semantic step we can ! "'%! say that “Jane” is a proper name of an individual, whereas “Anna”, where it is the proper name of anything, is the proper name of a function from worlds to individuals. Of course when Tolstoy says that Anna did this or that, we are not from the point of view of our imaginative engagement with the work, to understand this as meaning that a role did this or that. This is because it is part of the fiction that “Anna” is the name of a person. But “Anna”, as used by Tolstoy, is not in fact the name of a person, nor does it purport to be. Names are expressions used in order to pick out individuals, and Tolstoy does not use “Anna” in order to do this, nor does he expect us to believe that he is. “Anna”, as used by Tolstoy, is not a name. (p. 141) On the other hand, one might also contemplate the possibility of the fictional characters enjoying of a certain autonomy with respect to their stories, in such a way that one could say that Tolstoy’s Anna Karenina could have had a different end, for instance. The idea would be that the characters would be well defined since the very beginning of the fiction -which would open possibilities for their fate other than the ones that the author chose. Considering this, one might think that the contingency of the properties of the characters could be reduced to the contingency of the writing process itself. Anna Karenina, for instance, might not have fallen for Vronsky precisely because Tolstoy might not have written that she did. However, one could see that this explanation does not work by considering the fact that one can write a fiction where characters have certain properties necessarily, and notwithstanding this, the contingency of the writing process remains; the author could have written a different story or this story a bit differently. Yet another (and perhaps better) alternative reply9 to the objection raised above could be just to bite the bullet and consider that Anna Karenina might fall for Vronsky merely contingently even if this is explained by what someone does in some world where she is not Anna Karenina; that is, by appealing to someone who is Anna Karenina in some world, but does not fall for Vronsky in some other world where she is not Anna Karenina. That is, even if the contingency of the falling for Vronsky by Anna Karenina is explained by what happens to those persons that are Anna Karenina in some world in which they !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! 9 From personal communication, I know Manuel García-Carpintero favors the Lewis-Currie view and I think that he likes this sort of reply. ! "'&! fall for Vronsky at other possible worlds in which they do not fall for Vronsky and hence are not Anna Karenina. In short, in my view, the contingency would not be entirely explained by Anna Karenina. I think this way out of the problem is ad hoc, and wrong as well. The conclusions that I think we should draw from all of this go farther than the mere conclusion that the explanation of fictional contingencies offered by the Lewis-Currie view fails. The problem of fictional contingencies and the failure of this explanation indicate a deeper problem. It shows that Russellian Descriptivism about fictional names is mistaken and perhaps so is the Possibility view of fiction, at least, for the case of consistent fictions (or consistent versions of them) told by the use of singular terms such as proper names. In short, the problem is that the Lewis-Currie view cannot explain the truth of utterances of pairs like (1) and (2). For, in particular, it cannot explain any fictional contingency of any fictional character. In any case, let me clarify here that I do not pretend to have refuted the view. Rather, I think that I just have provided a reason, which I take to be quite strong, for those sympathetic to my intuitions to reject the view. Other possible descriptivist ways out If the Lewis-Currie view claims that being Anna Karenina amounts to satisfying the possibly satisfiable definite description which has as a part the description of this woman as falling for Vronsky, it will not succeed in explaining that Anna Karenina falls for Vronsky only contingently. For the simple reason that any woman who would be Anna Karenina at all would be so only in some worlds, and precisely in those worlds where she falls for Vronsky. However, one might think, even against what Currie seems to insist, that there are two ways of being Anna Karenina: one of them, the one that we already contemplated and the one that Currie tells us about; the other, the one that the Lewis-Currie view would like to have in addition: being someone who at some story-world satisfies the description that ! "''! ‘Anna Karenina’ is, even if she does not satisfy it at some other possible worlds. In this sense anyone who met the description at some possible world, would be also Anna Karenina at all the other worlds where she existed, even if she did not satisfy the description in them. This last sense does not seem to be compatible with Russellian Descriptivism of fictional names. But for the sake of the argument, let’s assume for a moment that it is. Thus, there would be two ways of understanding the relevant pair of claims. According to the interpretation corresponding to the first sense of being Anna Karenina, utterances of (1) would be true but utterances of (2), false. And according to the interpretation corresponding to the second sense, while utterances of (2) would be true, utterances of (1) would be false. In none of these two interpretations, does one get that both claims are true. Intuitively at least, however, they seem to be true under one and the same interpretation. Both claims are about the features that characterize a fictional character, Anna Karenina. One of these features is to be someone who falls for Vronsky; another, to be someone who falls for Vronsky in a contingent way. One might think, though, that the intuitive truth of these two claims may be very well accounted for by considering a different interpretation of them in each case. However, there is no independent reason to interpret them this differently. This does not seem to be why we think that they are both true. This way out of the problem of fictional contingencies seems, therefore, ad hoc. In any case, there is no way for the Lewis-Currie view to obtain what the view really needs. That is, that Anna Karenina, one and the same thing, has the property of falling for Vronsky in the actual world of the story, but lacks it at another possible world. For it is a condition on being Anna Karenina that she does so contingently. This is what having a contingent property amounts to. Note that the independent reason to argue for the legitimacy of using two different interpretations cannot be that ‘Anna Karenina’ can be used both as a possibly satisfiable definite description and as an ordinary proper name, and that while it is used as a possibly satisfiable definite description in the case of (1), it is used as an ordinary proper name in the case of (2). For, according to the Lewis-Currie view, only within the fiction, ‘Anna Karenina’ is or comes to be used as an ordinary ! "'(! proper name. We cannot use the proper names that are used in these other possible worlds. For these referring proper names are only possible, not actual (or at least, they are not actually used as proper names that refer). Non-descriptivist possible worlds views of fiction One might think that perhaps the Possibility view of fiction can be defended independently of Russellian Descriptivism of fictional names and that in this way the problem of fictional contingencies could be handled. I see two starting options: one might defend Meinongianism (already considered before in this chapter) and say that fictional characters actually have being in some special mysterious way, and that fictional names are like ordinary proper names that refer to them. Or one might defend the view that fictional characters are abstract objects, which actually exist and to which the fictional names refer. Within this last option I see two further ones: one might say that these abstract objects are only contingently so, so that in other worlds these same objects exist but are concrete instead of abstract in these worlds (the existence of these contingently non-concrete objects is defended by Bernard Linsky and Edward N. Zalta (1994, 1996) not with respect to fictional characters but with respect to mere possible objects –i.e. possibilia). Or one might defend the view that these abstract objects, like any other abstract objects, are necessarily abstract (like the Millian Creationist view that Nathan Salmon defends and we also considered before in this same chapter), in which case, unless we reject other initially intuitive assumptions like the ones that we discussed above (see the previous discussion of this view), they only can do what their fictions tell that they do in worlds that are impossible, for there are things that only concrete objects can do, such as to smoke a pipe, for instance. Thus, if these abstract objects are to do these things, it can only occur in impossible worlds rather than possible ones. ! "()! Now. On the one hand, as I already said above in this chapter, the first option, Meinongianism, seems to me wholly mysterious and hence, no plausible at all. On the other hand, the only option left which explains fictions in terms of possibilities is the option that sees fictional characters as contingently non-concrete objects and, hence, consists in the seemingly implausible claim that some actual abstract objects can be concrete and some actual concrete objects can be abstract, when initially one might have thought abstract objects to be necessarily so10. In view of the alternatives to Russellian Descriptivism about fictional names, I think that we can safely conclude that the Possibility view of fiction needs more argument – provided that I am right that these are the only possibilities, that there is an alternative explanation of all of these data which does not include the Possibility view of fiction, and that this explanation is at least as good as any of the ones that includes it. I think that there is such an alternative explanation and I offer it in the following chapter. It must be acknowledged, though, that this is not a conclusive refutation of the Possibility view of fiction or of the Russellian Descriptivism of fictional names. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! 10 Timothy Williamson (1998) defends a similar metaphysical view but that does not include the claim that these contingently non-concrete individuals are abstract. In Williamson (1998)’s footnote 17, he makes this point explicit: Linsky and Zalta, whose approach is in some ways very similar to the present one (although they present it within the framework of actualism and take BF as primitive) waver on this point. They hold, like the present approach, that there are contingently nonspatiotemporal objects. At Linsky and Zalta (1994: 446) they define ‘concrete’ as ‘spatiotemporal’ and ‘abstract’ as ‘not concrete’; thus some abstract objects could have been concrete. At Linsky and Zalta (1996: 293) they say ‘abstract objects are not concrete at any world’. The latter approach is preferable. The inconsistency may be largely terminological. (p. 271) According to Williamson, contingently non-concrete individuals, though, do have some sort of logical existence. Non-concrete individuals are non-spatiotemporal individuals. Among these contingently non-concrete individuals, there are the dead people, the not-yet-born people, among others. And by appealing to their logical existence Williamson accounts for the Barcan Formula and propositions semantically expressed by sentences about the past and about the future. This may well be a tenable view, but at least for the case of fictional individuals, I prefer the view that I propose in Chapter 6, which I think is metaphysically and semantically more plausible. In any case, it seems to me that this sort of logical existence would also require some explanation. ! "("! Impossible worlds I think that the best way to solve the problem of fictional contingencies is simply to abandon the Possibility view of fiction and with it Russellian Descriptivism of fictional names, for the reasons just provided. In my view, story-worlds are not all possible worlds even if they are ontologically the same kind of thing: that is, something like sets of sentences or propositions. The difference between story-worlds and possible worlds would just be that only the latter represent possibilities with respect to the actual world. The fictional contingencies of fictional characters should be explained by appealing to those worlds which would be possible but only with respect to the world of the story and not with respect to the actual world. With this view in mind, we can now turn back to (1) and (2) (repeated below as (5) and (6), respectively) and see how they can be reconciled. (5) Necessarily, someone who did not fall for Vronsky would not be Anna Karenina. (6) Someone who necessarily fell for Vronsky would not be Anna Karenina. On the one hand, utterances of (5) are true because there is no world of the story in which Anna Karenina does not fall for Vronsky. This is so because it is the fiction of Anna Karenina that Anna Karenina falls for Vronsky. On the other hand, utterances of (6) are true because according to the fiction, there are some possible worlds that are possible with respect to the story-worlds in which Anna Karenina does not fall for Vronsky. This is so because according to the fiction, Anna Karenina falls for Vronsky merely contingently. This raises no problem and does not fall within the Possibility view of fiction because the story-world, if it is a possible world at all, is one that is impossible. With this I simply ! "(*! mean that it is a world that is not possible with respect to the actual world. In sum, I have argued in this section that there is a problem with fictional contingent properties that the Lewis-Currie view about fiction cannot solve. I have also argued that other alternative Possibility views on fiction do not seem plausible. Finally, I have provided some positive suggestions to develop in order to explain fiction and solve the problem posed by fictional contingencies. The telling of the same story in other worlds Before closing this chapter, and moving to my positive view on fiction, I would like to propose a modification to Lewis’ (1983) analysis of truth in fiction, which fits well with the Lewis-Currie view of fiction. In my view, Lewis’ analysis is wrong as it is, but could be easily modified. I think that at least some who would like to defend the Lewis-Currie view together with the Lewis’ analysis of truth in fiction (like García-Carpintero (2007) and Richard Hanley (2004)) would agree with me about this. Lewis’ analysis of truth in fiction is either of the following two. He does not commit to one over the other and I do not discuss them here further. ANALYSIS 1: A sentence of the form “In fiction f, !” is non-vacuously true iff some world where f is told as known fact and ! is true differs less from our actual world, on balance, than does any world where f is told as known fact and ! is not true. It is vacuously true iff there are no possible worlds where f is told as known fact. ANALYSIS 2: A sentence of the form “In fiction f, !” is non-vacuously true iff, whenever w is one of the collective belief worlds of the community of origin of f, then some world where f is told as known fact and ! is true differs less from our actual world, on balance, than does any world where f is told as known fact and ! is not true. It is vacuously true iff there are no possible worlds where f is told as known fact. ! "(+! A fiction generates truths in fiction not only from what is explicitly said in the fiction but also from the fiction’s background. The difference between these two alternative analyses precisely resides on what the relevant background of a fiction is taken to be. Analysis 1 takes the relevant background of the fiction to be just the actual world, whereas Analysis 2 takes it to be any world from the set of possible worlds that are collectively believed to be actual by the community of its origin. Let me insist, though, that my aim here is not to discuss any of these two analyses but to argue for an amendment that is necessary for either of them to be true. According to this possible worlds view, fictions such as The Holmes Stories or Anna Karenina are general stories that are satisfied in some possible worlds and not in others (including the actual one). Suppose that w1 is a world where The Holmes Stories are true. In this world, there must be someone, who must be called ‘Watson’, telling the stories about a famous detective, friend of his, called ‘Sherlock Holmes’. If the theory were right, the story told by Conan Doyle at the actual world must be the same as the one that this man called ‘Watson’ tells about his friend Sherlock Holmes at w1. But it is not. If the view is correct, the story that Conan Doyle tells is a general one about no person in particular. However, the story that Watson tells at this other world w1 is a singular story about a particular man, namely, his friend Sherlock Holmes. Moreover, in another possible world w2, another person different from w1’s Watson, who is also called ‘Watson’, also tells the story of one of his friends, Sherlock Holmes, who is a well-known detective, and smokes a pipe, and so on. w2’s Watson, at w2, tells a singular story about his friend w2’s Sherlock Holmes. This is neither the same general story Doyle is supposed to have told in the actual world, nor the other singular story that w1’s Watson tells about w1’s Sherlock Holmes. One is a general story about no particular person; another is a singular story about a particular person, namely w1’s Sherlock Holmes; and yet, the other is an even different singular story about a different particular person, namely w2’s Sherlock Holmes. ! "(#! To help see this more clearly, imagine that I tell you a general story by uttering, and doing it literally: (7) There is a man, whose name is ‘Bill’, who has had an affair with Monica Lewinsky. Consider also the story that I tell you by uttering (8) and referring to Bill Clinton in using the name ‘Bill’. (8) Bill has had an affair with Monica Lewinsky Finally, consider the story that I tell you by uttering (8) as well, but this time referring with my use of ‘Bill’ to Bill Gates instead. We have here the telling of three different stories. Likewise, the three cases of storytelling considered in relation to the example of The Holmes Stories above are cases of the telling of three different stories. Within the LewisCurrie view, it could only be said that the singular stories told at the two non-actual worlds entail the general one told by Doyle at the actual world. But they are not one and the same story. One might reply that every time that a singular story is told, it is the case that the general story that would be entailed by it is also told by the same act. But this is wrong. I think that this thought must be due to cases like the following, in which it does look as if the corresponding general story is also told even if in fact it is not. Imagine that you are having dinner with your wife and your parents-in-law. During the dinner you are telling your wife a real life story about some of your close friends. As it happens, your motherin-law misses the initial part of the story that you are telling, and while you keep on telling it to your wife, she asks about it to her husband, and he tells her that you are telling your wife the story of a guy whose name is, say, ‘John’, who had this accident and so on. In this imaginary dining situation, you tell your wife a singular story about your friend John and your father-in-law manages to mention the story you are telling by using ! "($! some general expressions such as ‘a guy’ or ‘this guy’ in such a way that enables your mother-in-law to identify the story and catch it up. But by doing this, your father-in-law is by no means telling a general story. In fact, his use of ‘John’ is wholly singular due to its adequate causal link to your previous use. My point is that a situation like this is not really an example, as one might think, in which your father-in-law tells the same story that you tell your wife, which is singular, but in a general way. Neither would it be an example of a situation in which the telling of a general story may be the telling of the same story as a singular one. In fact, no one but you is telling a story. Singular and general stories are distinct. This shows that Lewis’ analysis, and any analysis that relies on the possible telling of the same story in some possible world where the story is true, would have to be amended. For Lewis’ analysis does require that the story that an author of a fiction tells is also told in at least one possible world but as known fact. As I have argued, if the story that Doyle tells us is general like this view has it, then no Watson can tell as known fact the same story that Doyle tells as fiction by the same act of story-telling11 in which singular terms are used. According to this view, Conan Doyle’s story, like any other fictional story, is a general story about no one in particular, while the stories told by some Watson about some Sherlock Holmes in the possible worlds in question by performing the same Lewisian act of story-telling are singular stories. One could amend David Lewis’ analysis by saying instead that what is told as known fact is one of the singular stories that entail the general story (i.e. the fiction) told by the author of the fiction, in the case of fictions that are told by using singular terms, and the fiction itself in the case of fictions that are told by using only general terms. This amendment would leave Lewis’ analyses somehow as follows: !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! 11 Understood in Lewis (1983)’ terms: the act of story-telling is an act of uttering the string of words used to tell the fiction. See pp. 264-266. ! "(%! ANALYSIS 1 (amended): A sentence of the form “In fiction f, !” is non-vacuously true iff some world where either f or a singular story that entails f is told as known fact by performing the same act of story-telling that the author of the fiction performed and ! is true differs less from our actual world, on balance, than does any world where either f or a singular story that entails f is told as known fact by performing the same act of story-telling that the author of the fiction performed and ! is not true. It is vacuously true iff there are no possible worlds where f is told as known fact. ANALYSIS 2 (amended): A sentence of the form “In fiction f, !” is non-vacuously true iff, whenever w is one of the collective belief worlds of the community of origin of f, then some world where either f or a singular story that entails f is told as known fact by performing the same act of story-telling that the author of the fiction performed and ! is true differs less from our actual world, on balance, than does any world where either f or a singular story that entails f is told as known fact by performing the same act of story-telling that the author of the fiction performed and ! is not true. It is vacuously true iff there are no possible worlds where f is told as known fact. I am offering this as an amendment, because I think that there might well be philosophers who would like to defend one of these analyses (or some other version of them), while agreeing with me in the distinction that I am highlighting between singular and general stories. However, I realize that some might also think that yet another possibility that I have not considered so far might be to insist that even the stories told by the possible Watsons at some possible worlds are not singular, and that this might have been what Lewis had in mind all along. If so, the modifications to the analyses that I am suggesting would not be considered as amendments after all by some (and probably, Lewis included). However, I think that would be a mistake. In this chapter I have surveyed most of the problems and some of the main proposals about fiction that one finds in the literature, and I have discussed the accounts of Frederick Kroon, on the one hand, and David Lewis and Gregory Currie, on the other, in more detail. In particular, I have defended the ellipsis strategy (that I make use of in the ! "(&! next chapter) from Kroon’s recent objections to it. In addition, I have criticized Lewis and Currie’s general and possibilist descriptivist view, offering an amendment of Lewis’ analyses of truth in fiction at the end. In the next chapter, I offer my own account of these issues: DFN. ! "('! 6. Demonstrated Fictional Names We use language to create, enjoy and discuss fictions. We often speak about fictions, their characters and stories. But what it is that we really talk about and how this fictional discourse works is far from obvious. In this chapter I propose a theory about the semantics of fictional discourse that follows the particular thesis about the meaning of proper names, and singular terms in general, that I defended in Chapter 3. This proposal partly consists in an account of descriptive phrases in general, such as 'the Catalan writer Jaume Cabré' or 'our son John' as well. In particular, although on this proposal fictional names do not refer to anything at all, descriptive phrases in which they sometimes occur (such as ‘the fictional character Sherlock Holmes’) do have denotation and we manage to make true claims about the fiction by using them. The proposal presented here is, in part, that we manage to do this by using fictional names as demonstrata within the descriptive phrases in which they occur. Fictional names are sometimes used as demonstrata that help contribute the denotation of the descriptive phrases in which they appear. I call this proposal ‘Demonstrated Fictional Names’, ‘DFN’ for short. As explained in Chapter 5, one of the things that we do by means of the fictional discourse is to talk “about” fictional individuals, stories and worlds. We do this when we create fictions by telling fictional stories. Afterwards we talk “about” these stories and the individuals appearing in them. For this, both authors and readers, usually use such sentences as 'Anna Karenina fell for Vronsky.', 'Sherlock Holmes lived in London.' or 'Sherlock Holmes is smarter than Poirot.'. Another thing we do is to discuss fictions as literary critics or from the outside. That is, we talk about fictions as such. We use sentences such as 'Tolstoy wrote Anna Karenina.', 'Sherlock Holmes is a fictional character.' or 'Santa Claus does not exist.'. The main difficulty that fictional discourse presents us with, as theorists, is to explain the fact that we seem to be talking about things that do not exist. Moreover, we seem to do it ! "##! by using the same kind of devices that we use to speak about existing things. For instance, we use singular terms such as proper names and indexicals. But in spite of the fact that proper names are clear devices of reference and normally used to pick out individuals, fictional names like ‘Anna Karenina’, ‘Sherlock Holmes’ and ‘Santa Claus’ do not seem to have anything to refer to. After all, most of us, pretheoretically at least, believe that neither Anna Karenina, nor Sherlock Holmes, nor Santa Claus exist. In this chapter, I offer a theory that explains the semantics of at least main problematic bits of fictional discourse. As indicated above, part of this theory is based on the thesis about the semantics of singular terms that I defended in Chapter 3. This is the thesis that in addition to referring, as most singular terms do, they also express a token-reflexive rule. For our present purposes, it is sufficient to say that according to this view for any token M of a proper name N, M semantically expresses the meaning of a definite description such as the one that results from writing ‘the individual called’ followed by M –where M is the token actually used of the name. Whereas the view I propose is a descriptivist one, it is relevantly different from the descriptivist Lewis-Currie view, discussed in the previous chapter. I proceed as follows. First, I briefly bring back our attention to some of the ideas that I defended in Chapter 3 and that are most relevant to DFN. I also introduce in this first section a conceptual distinction that DFN makes use of. In Section 2, I present DFN by explaining how it accounts for the different uses of the different expressions related to fictional discourse. I also explain in this section how DFN solves the problems presented at the beginning of Chapter 5. Next, in Section 3, I go back to the Lewis-Currie view and compare it with DFN. Finally, in Section 4, I consider some objections and replies. 6.1. The Semantic and Metaphysical Basis of DFN In this section, I outline the general view on meaning on which DFN rests and explain an important plausible metaphysical assumption that it makes. ! $%%! Proper names (singular terms) According to DFN, proper names (and singular terms in general) semantically express the content of a definite description in addition to referring to an individual –in the case they do. Both properties and individuals are part of their meaning. In the case of proper names, this definite description would be something like ‘the individual called’ followed by M, which would be the token of a name N that has been used. Here it is important to remember that I share David Kaplan (1990)’s view on the individuation of names in certain important respects –though not all- and by ‘a token M of a name N’ I mean any physical event M produced by an utterance of the name N, where N is not ambiguous. Like Kaplan’s, my view is that proper names are not ambiguous. ‘David’ is not the type of a single proper name whose tokens refer to different individuals, David Kaplan and David Lewis among many others. Rather, David Kaplan and David Lewis bear different proper names which nevertheless share properties such as sound, shape or spelling. Thus, my view is that tokens of the names of David Kaplan and of David Lewis are tokens of different names. Unlike Kaplan’s, though, I see no reason to reject the type-token distinction and with it the idea that the tokens of these different names are also tokens of the even more general name or expression-type ‘David’ -that is, in addition to being tokens of different and less general names-types with that same form. The idea is that the same way that ‘A’ is a token of the letter-type ‘a’ but also of the capital-letter-type, italic-letter-type and bold-letter-type, or even of the more specific capital-italic-bold ‘a’ letter-type, this token of the name of David Kaplan ‘David’ is a token of different types: the generic-name-‘David’-type, the name-of-David Kaplan‘David’-type, the name-starting-with-a-‘D’-type, or even the English-proper-name‘David’-type1. I think of proper names as created types and since types are abstract, that proper names, like fictional characters or artworks, are created abstract objects. I think that the type-token distinction is natural and useful. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! 1 Notice that although David Kaplan rejects the type-token distinction here, he admits of the existence of generic names. ! $%"! In short, by the description obtained by writing M after ‘the individual called’ I am referring to the token of the name N that has been produced as a token of that specific and unambiguous name N –that is, something that is produced with specific concrete referential intentions and thereby individuated by them. This thing has certain referential properties essentially and could not have had different referents from the one that it actually has. A token of a name, in that sense, does not exist at other possible worlds with totally different referential properties. Notice however that the thesis is not that proper names, or singular terms in general, are synonymous with definite descriptions. They are not. Not even fictional names are abbreviations of definite descriptions or used as if they were such abbreviations. Complex linguistic expressions As explained at the end of Chapter 3, not only simple linguistic expressions perform more than one semantic function. Utterances of sentences express more than one proposition. The same way that tokens of simple linguistic expressions express a property in addition to referring to their referent or having an extension, utterances of sentences also express some descriptive propositions in addition to the main proposition expressed. So, for instance, an utterance of (1) additionally expresses a proposition that would be mainly expressed by an utterance of a descriptive sentence such as (2), in addition to expressing its main content, which is a singular proposition that is characterized by (3). (1) Frida is cute. (2) The individual called ‘Frida’ is cute. (3) <cuteness, Frida> ! $%$! The additional proposition is a mode of presentation of the main proposition and it is part of what one says in uttering the sentence and part of what is understood in hearing it by competent speakers of the language (See Chapter 3 for more details). It is worth emphasizing here that the descriptive proposition additionally expressed by an utterance of (1), and mainly expressed by an utterance of (2), is a singular proposition. This is because it is constituted by the token of the name ‘Frida’ that has been used in uttering (1), and this is an individual. Notice that this is so, even if the definite description ‘the individual called ‘Frida’’ in (2) gets a Russellian interpretation and, thus, Frida is not a constituent part of this proposition. Although the descriptive proposition additionally expressed by an utterance of (1) is singular, there is no regress. For an utterance of (2) is only considered in this theoretical explanation in order to characterize which proposition an utterance of (1) additionally expresses -that is, the proposition that an utterance of (2) mainly expresses. But in the context on which we are focusing no utterance of (2) takes place, only an utterance of (1) does. If an utterance of (2) were to be performed in addition to an utterance of (1), then the former would also mainly express a proposition -the one that an utterance of (1) additionally expresses- and would additionally express yet another descriptive proposition which would be mainly expressed in its turn by an utterance of a sentence containing a definite description such as ‘the individual called ‘‘Frida’’’, and so on and so forth. But this is not the case at issue. Conceptual distinction: fictional characters vs fictional personas In telling and thereby creating fictions, authors also seem to create fictional characters. Here I assume that fictional characters, like fictions, exist and are abstract entities. These ! $%&! are not the referents of fictional names, though. As said above, fictional names are empty and do not refer at all. Thus, for instance, in writing Anna Karenina, Tolstoy created the fictional character Anna Karenina by telling the merely pretended story of an existing woman called ‘Anna Karenina’. It is not the case that a concrete woman was created by Tolstoy. It is also not the case that the abstract entity that was indeed brought into existence by his actions is, or experiences, what the woman Anna Karenina is told to be, and to experience, in Tolstoy’s story. We need to distinguish clearly between the abstract fictional characters and the alleged fictional individuals that they allegedly represent. None of them are the bearers of the fictional names, and only the former exist. There are no fictional individuals. Philosophers such as Nathan Salmon (1998) and Peter van Inwagen (1977) also make this distinction. However, unlike DFN, their view is that fictional names refer to these abstract objects, the fictional characters, and that by saying things such as that Sherlock Holmes is a detective we are not ascribing the property of being a detective to the abstract object to which, according to their view, the name ‘Sherlock Holmes’ refers. Notice, however, that confusions arise due to the fact that the expression ‘the fictional character’ is sometimes used to talk allegedly about the alleged fictional individuals that fictional characters allegedly represent instead, and in that sense is ambiguous2. With this semantic and metaphysical basis in place, I can go on to present DFN. In the next section I consider how DFN applies to the different characteristics or uses of our fictional discourse, and indicate which of the problems discussed in Chapter 5 are thereby solved. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! 2 See Section 2 of Chapter 5 for a discussion of views that claim that fictional names do refer to fictional characters, and my reasons against them. ! $%'! 6.2. Demonstrated Fictional Names One of the features of intentionality is that it is possible to have representations that are about nothing at all; representations that allegedly represent things that do not exist, if one wishes. According to DFN, this is the case of empty names in general, but more vividly, of fictional names and fiction. According to DFN, fictional names, as already said, are empty proper names: they do not refer, they are not abbreviations of any kind of definite description, and they are not used as if they were abbreviated definite descriptions either. In telling fictions such as Tolstoy’s Anna Karenina, fictional names are used as if, in pretence, they were nonempty rigid proper names that actually referred. However, as also explained above, fictional names, as any other proper name, additionally express the meaning of a specific definite description and this is what explains their meaning in spite of their lack of referent. So, for instance, the name ‘Sherlock Holmes’ is an empty name. It does not refer to any individual, since Sherlock Holmes does not exist. In spite of this, ‘Sherlock Holmes’ expresses the content of the essentially denotationless definite description ‘the individual called ‘Sherlock Holmes’’, and provides a meaning to utterances in which it occurs. Thus, part of what one expresses when uttering a sentence containing this proper name is this descriptive content: that there is one and only one individual named ‘Sherlock Holmes’. In telling Sherlock Holmes stories, Conan Doyle used the name ‘Sherlock Holmes’ as if, in pretence, it were a non-empty ordinary proper name that actually referred to a successful English detective. In explaining what an author does in telling and thereby creating a fiction there are two very close but relevantly different possibilities. According to one of the pictures, the storyteller uses fictional names and pretends to be referring to someone by this use. He pretends to refer but he does so by actually using proper names that do not refer. Fictional names are real proper names and are used as such. This is possible even if they are not ! $%(! used to refer to anyone. According to the other, the storyteller not only pretends to refer to someone but also pretends to be using proper names without actually using proper names, or at least not using them as such. In this picture, the expression 'Sherlock Holmes' may be used in at least two ways. It could be used semantically as a proper name, or pragmatically as a definite description, in which case it is not used as a proper name at all. The later would be the use the author makes of the expression in storytelling while pretending to be using that expression as a proper name. It is this pragmatic use of the expression that determines the fiction’s content. DFN agrees with the first, not the second, picture. According to DFN, the author does use proper names that he introduces and in so doing, determines future uses of those names quite in the same way than non-fictional names get introduced and determine their posterior uses after a baptism. The author’s intentions in using fictional names as fictional, with the intention of not referring, determine that other uses of these same names will not refer either. By using fictional names in her storytelling, the author pretends these fictional names to be different, pretends them to be names of certain individuals –individuals that do not actually exist. In posterior uses of these names, the tokens produced are adequately linked to the author’s first use as much as posterior uses of ordinary proper names are this way linked to the original use. In this, DFN greatly differs from the descriptivist view of David Lewis and Gregory Currie, which adopts the alternative picture. I turn back to that point in another section below. In fictional discourse, there are several uses to consider. I present DFN by explaining how it treats them. Positive statements in fiction Among the sentences with empty names, there are sentences with fictional names such as (1), the utterances of which, like the utterances of the other sentences with empty names, are meaningful but untrue. ! $%)! (1) Anna Karenina suffered. For there is no one referred to by ‘Anna Karenina’. Only those who ignore the relevant facts about the name would make sincere use of these sentences. Storytellers utter these sentences to tell a fictional story which gets through just because of the additional descriptive meaning that they express, in spite of their lacking their main meaning. Notice that in this way DFN solves the Problem of Untrue but Meaningful Utterances. Apart from these, however, there are things that we say about the story told in the fiction that are true or false according to the fiction. As it is widely assumed, to say things that are true or false according to a particular fiction, we use an operator which is specific to the case of fiction. For example, we express the truth that Anna Karenina suffered according to the fictional story that Tolstoy created, Anna Karenina, by uttering something like (2), or some variant of it constructed with other similar phrases such as ‘in the fiction’. (2) According to Tolstoy’s Anna Karenina, Anna Karenina suffered. Utterances of (2) are true, whereas utterances of (1) are truth-valueless. Sometimes, though, we express what utterances of (2) express without making the ‘in the fiction’ operator explicit, that is, by uttering just (1). But it is clear that in these cases, the relevant expressions ‘in the fiction’ or ‘according to the fiction’ are tacitly meant. Part of what DFN claims is that what the fiction operator does is to change the interpretation of the embedded sentences. It does so by making the proposition that normally these sentences only additionally express be their main content. Thus, an utterance of (2) mainly expresses the same as an utterance of (3) which, accepting a variant of Russell’s theory of descriptions, which DFN does, is quite the same as saying what an utterance of (4) says. ! $%*! (3) According to Tolstoy’s Anna Karenina, the individual called ‘Anna Karenina’ suffered. (4) According to Tolstoy’s Anna Karenina, there is a unique individual who is called ‘Anna Karenina’ and suffered. Utterances of (3) and (4) are true and explain the content and truth of utterances of (2). Again, it is important to bear in mind that it is the token of the fictional name used in an utterance of (2) that it is mentioned in utterances of (3) and (4). This token has at least some of its referential properties essentially. Hence, utterances of (3) and (4) involve the meaning of essentially denotationless definite descriptions rather than possibly satisfiable ones. Notice that in this way DFN solves the Problem of Describing Fictional Stories. Notice that, as discussed in Chapter 5, what is true in or according to the fiction is not only what is explicitly said in it. For instance, utterances of (3) and (4) would be true even if nowhere in Tolstoy’s Anna Karenina was explicitly said that there was an (unique) individual called ‘Anna Karenina’. One might find odd that, as it follows from the fact that utterances of (3) and (4) are true, Anna Karenina is about the name ‘Anna Karenina’. However, I do not find this odd. Anna Karenina is to a small extent about that name, as it is also about anything else that forms part of the story. This is compatible with the fact that Anna Karenina is centrally about a woman who suffers, but not centrally about a name. That is, although the novel is partly about the name to a minimal extent, the name is not the central topic of the novel. What one does in uttering sentences such as (2) or (3) after the fiction is created is not totally different from what the author of the fiction did in creating it. In uttering (2) or (3) once the fiction is already created, one is describing the fiction by saying in part –though not in these words, of course- that according to it, there is a unique individual named by any token of the name 'Anna Karenina', introduced by Tolstoy, that is adequately related ! $%+! to the first token used by him. Notice, though, that to say this, that there is a unique individual named 'Anna Karenina', is not to name any individual, but to describe a certain fact, namely, that there is an individual who already has that name. By uttering (2) or (3) one is saying that this is so according to a certain fiction. When I say that there is a person near me called 'John’, I am not naming anyone, but just describing a situation in which there is someone whose name happens to be 'John' and he is near me. On the other hand, anyone who uses the fictional name to describe the fiction in utterances such as (2) or (3), engages in the same pretense that the author started. The difference between what the author does and what the others do is that only the author created the fiction, while the others are merely describing it afterwards. This is precisely why in later describing the fiction one prefixes 'according to the fiction' (even if only tacitly) to one's descriptions. The real difference between storytelling and posterior uses in which the fiction is described is that in the first case one is creating without any possibility of committing a mistake. This possibility does clearly exist in the second case. ‘Sherlock Holmes’ does not seem to be used to refer at all. However, an expression is no less of a name just because it is not used to refer. And of course, as Currie discusses, the fact that an expression is a name in a fiction does not mean that it is a name in reality either. What makes Conan Doyle’s use of the expression 'Sherlock Holmes' be the use of a proper name is its shape and one of its semantic functions. Conan Doyle uses it as a proper name even if he has no intention to refer with its use, and knowing that he will not refer at all in this way. However, using an empty proper name such as 'Sherlock Holmes' knowing well that it does not refer is not absurd, because 'Sherlock Holmes', as any other proper name, partly expresses a property (the meaning of a metalinguistic definite description), and it is by the expression of such a property that this empty proper name helps determine the content of the fiction. That is, in using an empty proper name, something is done in spite of its being empty: one expresses a property. What makes this property determine the content of the fiction is not any special process, but rather, the fact that in using a proper name one usually expresses it and that nothing more is expressed due to the emptiness of the name. ! $%#! What prevents these definite descriptions, whose meaning gets expressed by fictional names, from being possibly satisfied is that the proper name that appears in them is used to refer to a token used of this same name. Since the name is produced with the intention of creating fiction and not of referring, this token could not have been used to name any individual. According to what was explained at the end of Chapter 3 and at the beginning of this same chapter, then, no token of the same name could have been used to refer. Of course, there can be homonymous names that are homonymous tokens of the types of different names. These are different names, nonetheless. Consider, for instance, ‘London’ (Ontario) and ‘London’ (England). It is precisely this last feature that accounts for the specificity of fictional stories told by the use of singular terms such as Tolstoy’s Anna Karenina and many others. In spite of their being descriptive, they are specific this way. The current proposal is that fictional stories, although not singular in the sense of containing individuals as constituents (other than names mentioned in storytelling), can nevertheless be specific (rather than wholly general) in virtue of the use of singular terms (as singular terms rather than as descriptions)3. Thus, according to the general view that I am arguing for, there are three grades ranging between generality and particularity that may be expressed through language: the generality expressed by the use of possibly satisfiable definite descriptions, the singularity expressed by the use of non-empty singular terms, and the specificity that lies in between and that can be expressed by the use of empty names which indirectly express the meaning of essentially denotationless definite descriptions. DFN entails that most fictional stories and the alleged fictional persona they allegedly represent are not even possible in an important sense: there is no possible thing satisfying these descriptions, since there is no individual bearing the names used in the fictions given that these names are not originally used to refer when first introduced in storytelling. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! 3 ! I am here and elsewhere understanding ‘story’ as something individuated by content. $"%! Notice that the case of non-fictional empty names such as ‘Vulcan’ is different in this respect. Vulcan does not exist, but might have existed. The introduction of the name ‘Vulcan’ to name a hypothesized planet involves some sort of generality that fictional names such as ‘Sherlock Holmes’ do not. ‘Vulcan’ was introduced to name an alleged planet between Mercury and the Sun with the clear intention to refer to that planet. So in some possible world, ‘Vulcan’ names a planet and, hence, Vulcan exists. This does not entail that mere possibilia exist. The planet that the name ‘Vulcan’ names in some other possible world might well be an actual thing such as another planet, Venus perhaps. The key difference between fictional names and other empty names such as Vulcan is that whereas the latter are introduced with an intention to refer, the former are not. Now. As I said before, in addition to these sentences, there are also statements we use to talk about fiction as fiction. DFN treats them as follows. Statements about fiction as fiction Consider, for instance: (5) Sherlock Holmes is a fictional character. (6) Conan Doyle created Sherlock Holmes. These statements do not involve a prefixing of the ‘in the fiction’ operator like in the cases considered above. Consider: (7) In the fiction, Sherlock Holmes is a fictional character (8) In the fiction, Conan Doyle created Sherlock Holmes ! $""! (7) and (8) are clearly false and not what utterances of (5) and (6) express, which seem to be true, or at least have a true reading. Unless, of course, there is a fiction according to which Sherlock Holmes is a fictional character and Conan Doyle created it. A movie based Conan Doyle’s life could be such a fiction. Part of what DFN proposes is to extend the recognition of the existence and prefixing of the fiction operator ‘in the fiction’ or ‘according to the fiction’ to other expressions such as ‘the fictional character’ and ‘the fictional persona’, so that, according to DFN, the contents of utterances of (5) and (6) are, respectively, something like the contents of utterances of (9) The fictional character Sherlock Holmes is a fictional character. (10) Conan Doyle created the fictional character Sherlock Holmes. Let us call these phrases that include the word ‘fictional’, or its cognates, ‘F-expressions’. The thesis here is that an F-expression makes it salient that the fictional name in these statements is neither used to refer to the bearer of the name –it does not have one because it is fictional-, nor used to talk directly about itself. Rather, an F-expression makes it salient that the fictional name is merely used as a demonstratum to help pick out an individual satisfying the whole complex description composed of the F-expression and the fictional name, if there is one. In those cases in which the F-expression that is prefixed to the fictional name is ‘the fictional character’, such as in (9) and (10) above, the F-expression makes it salient that the fictional name is merely used as a demonstratum to identify and finally denote the fictional character in question with the whole complex definite description. But in some other cases, cases in which it is the expression ‘fictional persona’, ‘fictional detective’, or ‘fictional alien’, for instance, that is prefixed instead, the name is also used as a demonstratum, but nothing gets denoted because there is nothing that satisfies the description in question. ! $"$! An F-expression makes it clear that the statement is about fiction. The idea is to interpret utterances of (9) and (10) somehow analogously as we interpreted utterances of (2) and (3) above. In (5), (6), (9) and (10), the expression ‘the fictional character’ is prefixed and the name ‘Sherlock Holmes’ works as a demonstratum rather than ordinarily used or explicitly mentioned, with the sole purpose of identifying the particular fictional character that the statement is about. Again, contrary to van Inwagen and Salmon’s view, ‘Sherlock Holmes’ does not refer to that fictional character though; the way we get to talk about, and hence refer to, the fictional character is by using the whole complex expression ‘the fictional character Sherlock Holmes’ that denotes it. It is worth stressing that DFN does not claim that the prefixed F-expression must always be either ‘according to the fiction’ or ‘the fictional character’. Consider the following dialogue, for instance. A: ‘Who is Sherlock Holmes?’ B: ‘Sherlock Holmes is a fictional character.’ First, notice that DFN does not require interpreting this exchange as follows. A: 'Who is the fictional character Sherlock Holmes?' B: 'The fictional character Sherlock Holmes is a fictional character.' In any case, it is clear that this cannot be the interpretation of the dialogue. Otherwise, the exchange would no longer make sense. DFN would be wrong if this were to be its interpretation. But it is not. Rather, DFN would interpret the dialogue somehow as follows: A: ‘Who is the individual Sherlock Holmes?’ B: ‘The individual Sherlock Holmes is a fictional character.’ ! $"&! Or even perhaps as follows: A: ‘Who is the individual called ‘Sherlock Holmes’?’ B: ‘The individual allegedly called ‘Sherlock Holmes’ is a fictional character.’ That is, in uttering each initial sentence, both participants in the conversation would be saying the same as what they would have said if they had uttered the more explicit two last pair of sentences. In the first pair of sentences, the expression ‘the individual’, ‘the individual called’, or ‘the individual allegedly called’, is elided. However, as it is clear in my accepting the second alternative interpretation above, I do not think that the semantic contribution of 'Sherlock Holmes' to A's question must be the same as the semantic contribution of 'Sherlock Holmes' to B's answer. I think B’s answer could even be interpreted as saying that the fictional character Sherlock Holmes is a fictional character, even if A’s question does not say anything about a fictional character. One might wonder what makes one F-expression rather than another to be prefixed to a fictional name. According to DFN, it is the intention of the individual who utters the sentences, together with the context of utterance, which determines this. Language use is highly restricted by economical-practical factors that make speakers normally use simpler, or different, expressions than the ones that would express what they mean in a more explicit and strict way. In these contexts of fictional discourse is relatively clear what speakers mean when they use these sentences, so that there is no need for them to utter the longer but explicit sentence that expresses what they mean. In addition, in these contexts, pretense is involved and so, speakers often exploit this pretense when uttering these sentences as well, leaving certain expressions elided. Thus, something that determines which F-expression is the one used in each case is the fact that the speaker knows that she is speaking about a fiction, when she does, and that she thinks that her hearer knows this, when she does. Another criterion to determine what sentence is the one that expresses what a speaker literally means in uttering some other close but shorter sentence is the following: that the expression elided is an F-expression or a related expression that includes ‘alleged’ or ‘hypothesized’, or one of their cognates. That is, all ! $"'! the expressions elided have to do with some practice of pretense or supposition. Finally, yet another criterion would be that the longer sentences that DFN proposes are the ones that the speakers would have uttered if they had no economical-practical restrictions, or if there was a need to clarify what they literally meant. F-expressions and other quotation-like phenomena As already noticed, according to DFN, this use of fictional names is similar to the use that we make of names or other expressions in propositional attitude attributions, indirect reports, quotation and the ‘so-called’ statements. The behavior of fictional names would fit a more general phenomenon in natural language –a fact that gives DFN some additional support. In other linguistic contexts, we also use expressions as objects that we demonstrate in order to make reference to other things saliently related to those expressions. Those things related to the relevant expressions may be mental representations (in the case of propositional attitude reports), expressions types or tokens (in the case of quotation and the ‘so-called’ statements) and fictional characters. How F-expressions work According to DFN, the descriptive phrase ‘the fictional character Sherlock Holmes’ contributes both the property of being a fictional character and the individual that exemplifies that property and is made salient by the fictional name 'Sherlock Holmes'4. This descriptive phrase expresses the same content that is expressed by the following expression !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! 4 The main semantic content ascribed to these descriptive phrases is similar to the semantic content ascribed by Soames' (2002) theory of descriptive names to what he calls 'descriptive names'. ! $"(! (11) The x such that x is a fictional character and x = y. (relative to an assignment of the fictional character made salient by the name 'Sherlock Holmes' to y) The main content expressed by utterances of (12) and (13) is the descriptive but singular proposition expressed also by (14). (12) Sherlock Holmes is a fictional character. (13) The fictional character Sherlock Holmes is a fictional character. (14) The x such that x is a fictional character and x = y is a fictional character. (relative to an assignment of the fictional character made salient by the name 'Sherlock Holmes' to y) Fictional names can work as demonstrata to help contribute the individual this way because there are unique fictional characters that bear a relation to them, that are made salient by them and satisfy the meaning of the description in question. Notice that DFN thereby solves the Problem of Thinking about Fiction from the Outside. These uses of complex expressions which combine proper names and descriptions not only exist in the fictional discourse. These are the cases of ordinary descriptive phrases such as 'the Catalan writer Jaume Cabré' or 'our son John'. So my proposal is that, in general, for any descriptive phrase containing a proper name like these, DN, the name N is used as a demonstratum to pick out the individual satisfying the descriptive phrase DN in question: that is, satisfying the descriptive part D and bearing the contextually relevant relation to the name N. This contextually salient relation is in many cases just the one of being referred to by the name. ‘The Catalan writer Jaume Cabré’ is one of these cases. ! $")! A nice example that suggests that these complex expressions are used as I am proposing can be found in the novel El hombre sentimental (The Sensitive Man) by the Spanish writer Javier Marías (an English translation of the quote follows it below): Vivían, me contaba ese hombre al que no conozco y cuyo nombre ya ni siquiera recuerdo (pero empezaba por N, Noriega, o Navarro, o Noguer), en una de esas casas de Barcelona que allí llamamos torres, de dos o tres pisos y que se encuentran sobre todo en la parte alta. Berta, un día cualquiera (…) se había caído por las escaleras. (…) Berta se recuperó en seguida y recuperó sus quehaceres normales, pero al noveno día de la caída, que era tan sólo dos antes de que aquel hombre, Noriega, se sentara a escribirme la carta, no amaneció. (…) El marido, Navarro, no daba más explicaciones, como si las causas médicas ya no le importaran, o no debieran importarme a mí.(...) 'Hoy la he enterrado', decía en singular, como si lo hubiera hecho a solas y con sus propias manos y Berta fuera un animal doméstico. (...) Con aquellas palabras concluía la carta de Noguer, aunque añadía un post scriptum en el que me preguntaba si deseaba recuperar esos libros míos que Berta cargaba cuando cayó por las escaleras; y, meticulosamente, en una cuartilla aparte me incluía la relación, una cincuentena de títulos (…). Así, sin mención de los autores, aparecían los títulos en la lista adjunta de Noriega. (…) Y aunque no he retenido en la memoria el nombre exacto de Noguer, en cambio se me han quedado grabados fragmentos de su carta (…) No cabe duda de que Noriega sería un marido desatento. (p. 75 and ff) That man whose name I do not even remember (although it began with N, Noriega, or Navarro, or Noguer) reminded me that they used to live in one of those houses of Barcelona that are there called towers, of two or three stories and located mainly in the upper part of the city. (...) A day like any other, Berta fell down the staircase. She recovered soon and restarted her normal activities, but on the ninth day from the fall, which were only two before that man, Noriega, sat down to write me the letter, she did not wake up. (…). Her husband, Navarro, did not give further explanations, as if the medical causes were no longer of interest for him, or as if they should not interest me. (…) 'I have buried her today' he said in singular, as if he had done it by himself and with his own hands and Berta were a pet. (…) With those words the letter from Noguer finished, although he added a post scriptum in which he asked me if I wanted to recover those books of mine that Berta was bearing when she fell down the staircase; and, very ! $"*! scrupulously, he included a list in a different sheet, a list of about fifty titles. (…) That way, without mention of the authors, the titles appeared in Noriega's attached list. (…) And although I do not keep in my memory the exact name of Noguer, I do remember some passages from his letter (...). It is beyond doubt that Noriega was not an attentive husband. This text, I think, is an example in which it seems clear that the proper names used are not used to refer because they are used to end up talking about someone who does not bear them as names. My hypothesis is that in this text, expressions such as ‘the husband’ or ‘the man’ are elided right before some of the uses of the names that occur in it, and the descriptive phrases that result from making these elided expressions explicit work as DFN claims these kind of phrases work. Now. It is usual to object that, according to this account, an utterance of a sentence such as (12) becomes redundant, trivial or analytic when it does not seem to be so. But, in fact, the amount of redundancy or triviality that DFN attributes to utterances of sentences such as (12) actually corresponds to the redundancy or triviality expressed in uses of (12). Usually, sentences such as (12) are precisely uttered to make the redundant and trivial point that something that is a fictional character is a fictional character by someone who perfectly knows the fictionality and hence non-existence of Sherlock Holmes. However, these utterances are not analytic. They state the unique existence of an individual, an abstract object in that case, and hence, if true, it is not just in virtue of their meaning -the object could fail to exist after all. Consider, for instance, the difference between (13) and ‘All fictional characters are fictional characters.’ Even if quite trivial, these utterances prove informative and cognitively valuable to those who previously ignore what they express. In fact, uttering them seems to be a good way for us to make precisely this kind of point in a simple way. On the other hand, perhaps the implicit phrase that is prefixed in these cases is just ‘the character’ instead of ‘the fictional character’, just as ‘according to the story’ instead of ‘according to the fiction’ or ‘according to the fictional story’ could be the one prefixed in some other cases. DFN is quite flexible about this. ! $"+! One might think that the following pair of sentences shows that the problem here discussed remains and so, that DFN is mistaken after all. For one might argue that utterances of (13’) are trivial while utterances of (12’) are not, and so that an analysis of utterances of (12’) in terms of utterances of (13’) is mistaken. (12') If Sherlock Holmes exists, then Sherlock Holmes is a fictional character. (13') If the fictional character Sherlock Holmes exists, then the fictional character Sherlock Holmes is a fictional character. However, again, notice that DFN is not claiming that ‘the fictional character’ is to be the prefixed expression in every case. That is, DFN does not require that every utterance of (12’) means exactly what every utterance of (13’) means. On the contrary, according to DFN, (12’) might well be used to mean that if the individual Sherlock Holmes, or the individual called ‘Sherlock Holmes’, exists, then the individual Sherlock Holmes, or the individual called ‘Sherlock Holmes’, is a fictional character, which is not trivial at all. Abundance of F-expressions Finally, the involvement of F-expressions like 'the fictional character' or 'the fictional persona' in the fictional discourse is quite abundant and flexible. Utterances of other sentences such as, for instance, (15), might also involve the implicit use of other Fexpressions. (15) Sherlock Holmes is smarter than Poirot. Utterances of (15) perhaps express something like what utterances of (16) express. (16) The fictional detective Sherlock Holmes is fictionally smarter than the fictional detective Poirot. ! $"#! All of these F-expressions have in common that they use the expression ‘fiction’, or one of its cognates, to make the point or advertisement that one is talking about fiction. Alternatively, perhaps utterances of (15) may also express what utterances of the following sentence express: (15*) According to The Sherlock Holmes Stories and the Agatha Christie novels, Sherlock Holmes is smarter than Poirot. If this were the case, true utterances of (15*) may be explained as among those truths in fiction that are entailed by what is explicitly said in the stories instead of by being explicitly said. It is clear that Poirot is not a character in The Sherlock Holmes Stories and Sherlock Holmes is not a character in the Agatha Christie novels either. However, according to the former fiction, Sherlock Holmes is smart to a certain extent, and according to the latter, Poirot
You may notice some images loading slow across the Open Collections website. Thank you for your patience as we rebuild the cache to make images load faster.
UBC Theses and Dissertations
Reference through demonstration: singular terms, propositional attitudes and fiction Celestino Fernández, Gemma 2012
Notice for Google Chrome users:
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.
- 24-ubc_2012_fall_celestino_fernandez_gemma.pdf [ 1.95MB ]
- JSON: 24-1.0073292.json
- JSON-LD: 24-1.0073292-ld.json
- RDF/XML (Pretty): 24-1.0073292-rdf.xml
- RDF/JSON: 24-1.0073292-rdf.json
- Turtle: 24-1.0073292-turtle.txt
- N-Triples: 24-1.0073292-rdf-ntriples.txt
- Original Record: 24-1.0073292-source.json
- Full Text