A Pragmatic View of Proper Name Reference

A Pragmatic View of Proper Name Reference Peter J. Ridley King’s College London Submitted for the degree of PhD in Philosophy June 2016 1 To my friends in the King’s Philosophy Department, 2006–2016 They have inspired and influenced my thinking more than anything else 2 Proper names are poetry in the raw. Like all poetry they are untranslatable. —W.H. Auden And we must at all costs avoid over-simplification, which one might be tempted to call the occupational disease of philosophers if it were not their occupation. —J.L. Austin 3 Abstract I argue, in this thesis, that proper name reference is a wholly pragmatic phenomenon. The reference of a proper name is neither constitutive of, nor determined by, the semantic content of that name, but is determined, on an occasion of use, by pragmatic factors. The majority of views in the literature on proper name reference claim that reference is in some way determined by the semantics of the name, either because their reference simply constitutes their semantics (which generally requires a very fine-grained individuation of names), or because names have an indexical-like semantics that returns a referent given certain specific contextual parameters. I discuss and criticize these views in detail, arguing, essentially, in both cases, that there can be no determinate criteria for reference determination—a claim required by both types of semantic view. I also consider a less common view on proper name reference: that it is determined wholly by speakers’ intentions. I argue that the most plausible version of this view—a strong neo-Gricean position whereby all utterance content is determined by the communicative intentions of the speaker—is implausible in light of psychological data. In the positive part of my thesis, I develop a pragmatic view of proper name reference that is influenced primarily by the work of Charles Travis. I argue that the reference of proper names can only be satisfactorily accounted for by claiming that reference occurs not at the level of word meaning, but at the pragmatic level, on an occasion of utterance. I claim that the contextual mechanisms that determine the reference of a name on an occasion are the same kinds of thing that determine the truth-values of utterances according to Travis. Thus, names are, effectively, occasionsensitive in the way that Travis claims predicates and sentences (amongst other expressions) are. Finally, I discuss how further research might address how my pragmatic view of reference affects traditional issues in the literature on names, and the consequences of the view for the semantics of names. 4 Acknowledgments I can no other answer make but thanks, And thanks; and ever thanks; Twelfth Night T his thesis could not have been written without a great deal of guidance, support, comfort, persuasion, discussion, and sympathy from a fair number of other people, some of whom have been very deeply involved, and some of whom have been at the peripheries, but have played a part nonetheless. I have written more words than I ever imagined I could, whatever their quality, and I’m fairly sure that my primary school teachers, who were unable to get me to write more four or five words in an hour, would not believe it. I owe a huge debt of thanks to all those who helped me achieve this feat, whatever their rôle, as well as to the AHRC, who provided me with funding. I have effectively had four different supervisors for my PhD, all of whom have significantly contributed to the project in different ways and at different times. I have been working with Wilfried Meyer-Viol, my primary supervisor, for at least eight years, and he has been a constant purveyor of support, interest, ideas, and insightful commentary during that time. When I wrote my PhD research proposal, I expected my research to follow on from my MPhilSt thesis, which focussed on the treatment of names in modal logic. However, my work drifted further and further from formal considerations, to the point at which there is no formalism of my own in this thesis at all. For this, I apologize wholeheartedly to Wilfried. Of course, his capacities go well beyond the logical, and although we did not end up work together as closely as we might have, Wilfried has been an invaluable intellectual guide, and I would not have got to this stage without him. Eliot Michaelson, my 5 secondary supervisor, arrived at King’s halfway through the third year of my PhD. Having someone with a new and different perspective, and with knowledge of all the relevant literature, approach my work at this stage in its development was invaluable. Eliot was able to point out where the weaknesses in my work lay, and where my thinking had become staid. Although his efforts to make me engage with positions other than my own has strengthened my resolve in my position—albeit with a deeper appreciation for the complexity of the debate—my arguments, both positive and negative, are certainly much stronger for those efforts. Mark Textor was my secondary supervisor prior to Eliot, and he essentially started me off on the trail that would form this thesis, working closely with me, first when Wilfried was on sabbatical, and then when my work was dominated by a debate that he himself was engaged in. Mark’s ability to point to a paper or a passage in a book that is directly relevant to any topic under discussion is astounding, as is his capacity to immediately find both the most promising, and the most problematic parts of any piece of work. I was also lucky enough to have the opportunity to spend a term in Austin in my second year, working with Mark Sainsbury. Mark is not only one of the most incisive thinkers I know, but also one of the most generous. He turned around my research from a mere collection of ideas and principles, into a feasible research project that propounded a position diametrically opposed to his own. I am extremely proud of the fact that Mark seemed to think the occasion-sensitivity of proper name reference to be a less absurd idea when I left Austin than he did when I arrived. I also thank Charles Travis. I only had two or three tutorials with him, but Charles was extremely encouraging of my work. He pointed me in the right direction, and gave me the motivation to keep going. As will become extremely obvious, his written work has also had a essential impact on my thought. Although I did not work with either of them during my PhD, Shalom Lappin and Ruth Kempson have both been enduring presences in my intellectual life since I started the MPhilSt. I have learnt a huge amount from them, both academically and personally. I first became really interested in the debates concerning proper name reference during a reading group on names that consisted of just Dolf Rami, Peter Sutton, and me. The discussions that we had on those summer mornings were hugely productive, and were the genesis of this thesis. I’m indebted to Dolf, who introduced me to the literature and its authors, got me excited about the topic, and supported my development within it. Peter Sutton has been my guide and bellweather through philosophy and linguistics since we first met as new postgraduate students at King’s in 2009. I’ve had few good ideas 6 that haven’t been grounded in some way in discussions with, or suggestions from him. When he left for Düsseldorf in 2014, I was intellectually lost for some time. Both he and Dolf have read sections of the thesis, and provided detailed and invaluable comments. Other friends who have, at some point, contributed either discussion, comments, proofreading, or all three, include Josh Armstrong, Lucy Campbell, Rory O’Connell, Mike Coxhead, Alex Davies, Nils Kurbis, and Matt McKeever. Giulia Felappi, Katharine Harris, and Jen Wright, in particular, provided very detailed comments and proofreading. Many thanks for comments and conversations also go to the participants of various reading groups, seminars, conferences and workshops I have spoken at or otherwise attended during the course of my PhD, including the Advanced Research Seminar, the Language and Cognition Research Seminar, the Word Meaning group, the Seriously Summery Summer Surrogate Seminar Series, two workshops on names in Göttingen, the names workshop at King’s, the London graduate conferences, PhiLang in Łódz, and SPEs 7 & 8. Many thanks in particular are due to Robyn Carston, Alex Clarke, Louis deRosset, Delia Graff Fara, Stacie Friend, Aidan Gray, Frank Jackson, Hans Kamp, Genoveva Martí, Ora Matushansky, Adriana Pavic, Tim Pritchard, and Ede Zimmermann. I also benefitted enormously from conversations in Austin with Ray Buchanan and Josh Dever. I am hugely grateful to my family for all the support, encouragement, and love they’ve given me over the course of my PhD, and throughout my life, in particular, my parents, my sister, Kathryn, my brother-in-law, Ge, and my nephew, Harris, who was born on the day I arrived in Austin. They have put up with increasingly infrequent phone calls and visits, as well as extended visits abroad, with great understanding, and very little complaint. I couldn’t have got here without them, and I will ensure that I visit more often when I’m free of this thesis. I’d also like to thank Annie and Gary for welcoming me into their family with such warmth, and providing both understanding and encouragement. My friends, both within and without King’s, have been essential to my emotional, academic, and, on at least one occasion, physical, wellbeing. My love and thanks go to: Luke Brunning, Lucy Campbell, Rory O’Connell, Helena Drage, Ben Jeffery, Tom Parrott (my soul-flatmate), Caspar Wilson, Jen Wright, and John D. Wright, who were there at the beginning; Bobby Ford, Jazz Kilburn-Toppin, Chris Pepper, Emmi Edwards, David Shipley, Beth Tobutt and my Godson, George, and Anna Wardell, who were there even before that; Mike Campbell, Alex Davies, Dave Dewhurst, Gary Hayes, Sam Kukuthas, Clare Moriaty, Parysa Mostajir, Dolf Rami, Tom Reekie, Ceri Edmunds, Hannah Shrimpton, Mike O’Sullivan, Peter Sutton, and Mike Withey, who came along 7 next; Mike Coxhead, Ben Davies, Paul Doody, Owen Engelfield, Andrea Fassolas, Giulia Felappi, Alex Franklin, Alexander Greenberg, John Heron, Kate Tomas, Sarah Tropper, and Pasi Valtonen, who emerged alongside my PhD; Jim Evans, and Matt McKeever for fun in Texas (and thereafter); Shalom Lappin, Nils Kurbis, and Haskell, for the music; Ben Cox, for providing beer, and a fine space in which to drink it; and Alex Diner, Seb Jenner, John Marr, and Rachael Palmer, recently acquired, but thoroughly welcoming friends. Finally, I thank Katharine Harris. When I met her, I had just over a year left to finish my PhD and I had little desire to do so. She brought love and happiness back into my life, and through gentle encouragement, she renewed my determination to get to the end. I could not hope for a more supportive and understanding partner. I have often relied on her too heavily, but she has never buckled. I hope that I can support her in the same way as she nears the final throes of her thesis, and thereafter. Peter Ridley London, June 2016 8 Contents Abstract 4 Acknowledgments 5 Contents 9 I Introductory Matters 13 1 Introduction 14 2 Methodological and Philosophical Preliminaries 2.1 Reference . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 21 Proper Names . . . . . . . . . . . . . . . . . . . . . . . . . 22 Data on Reference . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2.1 Corpus Data . . . . . . . . . . . . . . . . . . . . . . . . . . 23 24 2.2.2 Data from Judgements . . . . . . . . . . . . . . . . . . . . 25 2.2.3 2.2.4 Judgements in Practice . . . . . . . . . . . . . . . . . . . . Against Expert Judgements . . . . . . . . . . . . . . . . . 28 31 2.2.5 Ordinary Speakers’ Judgements Again . . . . . . . . . . 34 2.2.6 Without Judgements . . . . . . . . . . . . . . . . . . . . . Semantics and Pragmatics . . . . . . . . . . . . . . . . . . . . . . 40 42 Occasion Sensitivity and Travis Cases . . . . . . . . . . . Occasion Sensitivity and Reference . . . . . . . . . . . . 44 51 Communication . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52 Communication and Language . . . . . . . . . . . . . . . Communication and Proper Names . . . . . . . . . . . . 53 54 2.1.1 2.2 2.3 2.3.1 2.3.2 2.4 2.4.1 2.4.2 9 II Semantic Views of Reference 56 Introduction to Semantic Views 57 3 Causal-Homophonism 3.1 The Argument from Shared Names . . . . . . . . . . . . . . . . 61 64 Evidence Against Homophony . . . . . . . . . . . . . . . Homophonist Responses . . . . . . . . . . . . . . . . . . 66 69 3.1.1 3.1.2 3.2 The Argument from Communication . . . . . . . . . . . . . . . 71 Communication and Causal Chains . . . . . . . . . . . . Causal Chains Without Communication . . . . . . . . . 73 74 3.2.3 Speaker Reference . . . . . . . . . . . . . . . . . . . . . . The Argument from Family Names . . . . . . . . . . . . . . . . 80 85 Which Family? . . . . . . . . . . . . . . . . . . . . . . . . Which Individual? . . . . . . . . . . . . . . . . . . . . . . 86 88 3.2.1 3.2.2 3.3 3.3.1 3.3.2 3.4 The Argument from Pluralism about Reference Determination: Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 Indexicalism 97 4.1 Proponents of Indexicalism . . . . . . . . . . . . . . . . . . . . . 100 4.2 4.1.1 Recanati . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100 4.1.2 4.1.3 Pelczar and Rainsbury . . . . . . . . . . . . . . . . . . . . 107 Rami . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110 Problems with Parameters . . . . . . . . . . . . . . . . . . . . . . 119 4.2.1 4.2.2 4.3 4.4 Salience . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120 Other Parameters . . . . . . . . . . . . . . . . . . . . . . . 123 Problems with Bearers . . . . . . . . . . . . . . . . . . . . . . . . 125 4.3.1 4.3.2 5 94 Many Ways to Bear, and Not Bear, Names . . . . . . . . 125 Reference Without Bearers . . . . . . . . . . . . . . . . . 129 Indexicalism: Not all Bad . . . . . . . . . . . . . . . . . . . . . . 131 Reference as Speaker Reference: Intention-Based Accounts 133 5.1 Constraint-based Intentionalism . . . . . . . . . . . . . . . . . . 135 5.1.1 5.2 Michaelson . . . . . . . . . . . . . . . . . . . . . . . . . . 135 5.1.2 Problems with Michaelson . . . . . . . . . . . . . . . . . 136 Against a Strong Neo-Griceanism . . . . . . . . . . . . . . . . . 138 5.2.1 Neale on speaker meaning . . . . . . . . . . . . . . . . . 141 5.2.2 Data on utterance formation . . . . . . . . . . . . . . . . 143 5.2.3 5.2.4 Ways Out for Neale . . . . . . . . . . . . . . . . . . . . . 150 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . 155 10 III A Pragmatic View of Reference 6 The Pragmatic View Stated 6.1 6.2 6.2.1 Reference and Truth . . . . . . . . . . . . . . . . . . . . . 163 6.2.2 6.2.3 Reference and Empty Names . . . . . . . . . . . . . . . . 165 Reference Over Time . . . . . . . . . . . . . . . . . . . . . 166 6.3 6.4 Context-Sensitivity . . . . . . . . . . . . . . . . . . . . . . . . . . 167 Semantics and Reference . . . . . . . . . . . . . . . . . . . . . . . 168 6.4.1 Necessary and Sufficient Conditions . . . . . . . . . . . . 170 6.5 Reference and Bearers . . . . . . . . . . . . . . . . . . . . . . . . 171 6.5.1 Social Practices . . . . . . . . . . . . . . . . . . . . . . . . 172 6.5.2 Nicknames and One-Off Names . . . . . . . . . . . . . . 174 6.5.3 6.5.4 A Defeasible Constraint? . . . . . . . . . . . . . . . . . . 178 Analytic Truths . . . . . . . . . . . . . . . . . . . . . . . . 180 6.6 How to Individuate Names . . . . . . . . . . . . . . . . . . . . . 182 6.7 Names and Occasion-Sensitivity . . . . . . . . . . . . . . . . . . 183 6.7.1 Occasion-Sensitivity in Thought . . . . . . . . . . . . . . 183 6.8 Names’ Contribution to Thought . . . . . . . . . . . . . . 185 6.7.3 Reference and Travis-Cases . . . . . . . . . . . . . . . . . 190 Objections to the Pragmatic View . . . . . . . . . . . . . . . . . . 192 6.8.1 6.8.2 Prediction . . . . . . . . . . . . . . . . . . . . . . . . . . . 192 Do We Need Criteria? . . . . . . . . . . . . . . . . . . . . 196 Who’s Mum? 200 7.1 ‘Mum’ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200 7.1.1 Constraints . . . . . . . . . . . . . . . . . . . . . . . . . . 201 7.2 7.3 8 158 Principles of the Pragmatic View . . . . . . . . . . . . . . . . . . 160 Names and Reference . . . . . . . . . . . . . . . . . . . . . . . . 162 6.7.2 7 157 7.1.2 7.1.3 My Mum or Yours? . . . . . . . . . . . . . . . . . . . . . . 202 No Parameters . . . . . . . . . . . . . . . . . . . . . . . . 203 7.1.4 Familial Practices . . . . . . . . . . . . . . . . . . . . . . . 204 From ‘Mom’ to ‘Tom’ . . . . . . . . . . . . . . . . . . . . . . . . . 205 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 206 Conclusions and Considerations 8.1 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207 8.1.1 8.1.2 8.2 207 Background . . . . . . . . . . . . . . . . . . . . . . . . . . 208 Semantic Views . . . . . . . . . . . . . . . . . . . . . . . . 209 8.1.3 A Pragmatic View . . . . . . . . . . . . . . . . . . . . . . 211 Semantics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 213 8.2.1 Predicativism . . . . . . . . . . . . . . . . . . . . . . . . . 214 11 Contents 8.2.2 8.3 Problems of Names . . . . . . . . . . . . . . . . . . . . . . . . . . 218 8.3.1 8.3.2 8.4 Variablism . . . . . . . . . . . . . . . . . . . . . . . . . . . 216 Rigidity . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219 Name Change and Reference Shift . . . . . . . . . . . . . 220 8.3.3 Attitudes and Beliefs, Fiction and Empty Reference . . . 221 The Final Summary . . . . . . . . . . . . . . . . . . . . . . . . . . 224 Bibliography 225 12 Part I Introductory Matters 13 Chapter 1 Introduction He telleth the number of the stars : and calleth them all by their names Psalm 147:4 U rsula le Guin’s Earthsea books describe a world in which people, animals, and other things have true names—proper names in the case of people and dragons, common names in the cases of other things—that they bear naturally, and which are hidden from most language users, the names belonging to an ancient language. Knowledge of the true name of an object enables manipulation of the object through magic. The idea that names are innate, and shed light on the true nature of things is also discussed in Plato’s Cratylus. In reality, there are no true names, proper or common. Language does not have magical properties. There are no magical ties between words and the world. The meanings of words do not effect the world, except in virtue of what speakers do with them, and they only have those meanings in virtue of how they are used. Proper names, in particular, are not borne by what bears them essentially, and they do not refer to what they do essentially. They refer to what they are used to refer to. This thesis will be concerned with elaborating on the idea of proper name reference, and arguing that it neither requires, nor constitutes, any special connection between word and object. Wittgenstein begins the Investigations (1953) with a quotation from Augustine that he takes to capture a particular view about how words connect to the world: that the meanings of words are objects for which the words 14 stand (Wittgenstein 1953, § 1).1 Proper names are sometimes thought of as being paradigmatic of the kind of view that Wittgenstein objects to: we give objects proper names, and, it appears, those names now name those objects such that using the name automatically summons the object into whatever thought is being composed. Names, so thought of, are somewhat like labels on strings attached to the objects that bear them and that they refer to (ibid., § 26). Of course, there is something to this thought: we do, in fact, give things proper names—make them bearers of those names—in large part so that we can then use those names to refer to them with relative ease. This is one use of the verb ‘to name’: to give something a name. Another use being to refer to something using a particular name, or to speak of something by its name, as in ‘she named Lucy as her successor’. However, it is, I claim, a mistake to think that simply by giving something a name one can thereby refer to it with that name. Just as for other kinds of words, the circumstances in which one uses a name, which may include the fact that one or more objects have been given that name in the past, will affect what one can do with it. Something akin to the view that all that is required for a name to refer is that it has been given as a label to a particular object has been popular at least since Kripke (1981), in the form of the view I call ‘causal-homophonism’. Opposition to the idea that proper names are simply labels of objects, that reference is an aspect of that labeling, and that, in order to refer to an object using a proper name, one just needs to utter a name that refers to or labels that object, is a major theme of this work. And, indeed, a motivating theme. To generalize these ideas, my main thesis is that it is not essential to any proper name that it refers to, or can be used to refer to, what it is used to refer to. That is, the meanings of proper names do not determine, or constitute, what they refer to. Rather, though it may be part of the meaning of names that they refer in some manner to some thing, upon a use, their reference is determined at the point of use, by the context of the occasion of use, in a way that is not specified by their meaning. One might, therefore, say that if proper names do label objects, they label them only at the time that they are being used. Of course, the fact that something has been given a name in the past, might well provide both good reason to use that name to refer to that thing, and (for the same reason) good evidence that that is what it is being used to refer to, but other factors will also play a rôle. Indeed, they must, given that many objects bear the same names. Evidence that names function in the way I have just suggested can be found through consideration 1 Travis (2006b) argues that it is not in fact Augustine’s claims, which amount to banality, that Wittgenstein objects to, but rather the misinterpretation and misapplication of such banalities (pp. 10–11). 15 of the way that they are, in fact, used in natural languages, by speakers of those languages. However, these vagaries of use are frequently overlooked or downplayed by proponents of accounts of proper name reference. My position on proper name reference attempts to address such vagaries, rather than gloss over them or idealize the use of names. As will become clear, I am heavily influenced in the development of this position by the work of J.L. Austin and the later Wittgenstein, in particular, via Charles Travis’s criticisms of the project of mainstream truth-conditional semantics (see, e.g., Travis 2006b, 2008). Travis put forward the beginnings of a view on reference that incorporates his broader views on language and semantics in relatively early papers such as his (1980) and (1981a). My view draws on these papers, but significantly builds on the picture Travis presents in them, developing it with regard to proper names in particular, and in order to address contemporary literature on reference. Travis’s general position, discussed at length in §2.3, is that the meanings of words and sentences radically under-determine what it is that is expressed— the truth-evaluable content—of utterances of those words and sentences. Knowledge of the meanings of words is in no way sufficient to interpret what speakers use them to say. Domain-general knowledge of the world— and in particular of the broad circumstances of an utterance—is required for interpretation. In this thesis, I show how this view applies to proper name reference as much as for any other phenomenon that contributes to the truth-evaluability of utterances. The failure of existing accounts of proper name reference to consider either the general problems of truth-conditional semantics, or the particular complexities of name use, is typified in the following quotation from Sainsbury (2005), in which he is specifically talking about the views of Travis: A specific version of this view is that the ordinary ‘meaning’ of expressions falls short of what is required to determine truth-conditions (even when indexicality and other well-recognized forms of contextual effect are bracketed). Identifying truth-conditions is part of what is involved in interpretation. So if the completeness of semantics is to be tested by the provision of sufficient information to permit interpretation, there may be no complete semantics. I know of no general arguments against such views. One has simply to examine the examples case by case. Yet even if semantics falls short of truth-conditions, the truth-conditional approach constitutes a useful idealization, and the kind of phenomenon that might make one think it inadequate in general have no role to play in the discussions of this book. Even if a conception of meaning closely tied to the demands of interpretation turns out to require the exercise of 16 general skills, rather than only language-specific ones, I think the issues to be discussed here will not be affected. (Sainsbury 2005, pp. 50–51) It is my view that, for the most part, the truth-conditional approach to semantics does not provide a useful semantics, but rather, results in a dismal failure to account for the relations between meaning, use, communication, content, and truth. Furthermore, I disagree that these issues do not affect the view of reference put forward by Sainsbury, or any other proponent of views about proper name reference. In this thesis, my account of proper names is initially developed through consideration of existing accounts that maintain (in general) that reference is determined by the semantics of names. These views fail, or are inadequate, for exactly the kinds of reasons just discussed. In chapter 2, I provide some methodological and philosophical background to my project. I briefly discuss what it is that I mean by proper name reference, attempting to capture a relatively intuitive notion, and observing its relation to truth. I explain that my purpose is to account for what it is that determines proper name reference: to explain what makes it such that a name, or use of a name, or a speaker using a name, refers to what it does when it does. I then go on to consider at length what it is that could count as data for the purposes of testing and comparing accounts. Assuming that those investigating name reference are primarily interested in the way that names are used to refer by language-users, or what names do in virtue of how they are generally used, I consider two possible sources of data: corpus data, and the judgements of language-users about cases. I argue that corpus data—transcripts, recordings, or other documentation of actual language use—is of little use with regard to assessing the reference of proper names, but that in order to interpret it, we must appeal to the judgements of other language-users, so we must anyway appeal to such data. I further divide the category of data from speaker judgements into that from the judgements of experts, and that from the judgements of ordinary speakers. I consider a large amount of literature on the value of each category, and conclude that any faith we should put in the judgements of experts stems simply from the fact that they are also ordinary speakers. I conclude that, if there is significant data by which to judge accounts of proper name reference, it probably comes from the judgements of ordinary speakers about cases of proper name use. §2.3 lays out my conception of the semantics/pragmatics divide, whereby semantics constitutes the meaning of words that is context-insensitive, and everything to do with content and interpretation that is context-sensitive— the way the world at large effects what we do and can do with language—is pragmatic. I explain this distinction in the context of the work of Charles Travis, 17 spending some time expounding the important features of Travis’s views of language and truth, and explaining how they will effect my view of proper name reference. In the final section of chapter 2, I discuss communication and its relevance to language, truth, and reference. I claim that communication is key to understanding why we are able to say what we do with our words, and, in particular, why we can and do refer to what we do when we use names. I posit a view of reference whereby what can be communicated with a use of a name is key to what it refers to, regardless of what a speaker intends to refer to. Part II introduces the class of views that are opposed to my conception of proper name reference. I call these ‘semantic views’, in contrast to my pragmatic view. I introduce a range of theories and accounts of proper names in the introduction to the part, sectioning off two types of semantic view that I will discuss and criticize in detail—causal-homophonism and indexicalism—both of which claim that the reference of a proper name is determined in some manner by its semantics. I also identify a third type of view which is not properly a semantic view, but includes accounts of how proper name reference is determined that are at odds with mine. These are intention-based accounts. I begin my criticism of views existing in the literature in chapter 3, attacking causal-homophonist accounts of name reference. These accounts have in common the idea that the semantics of a name is simply constituted by its reference, and so determines it in virtue of being it. In order that this view be at all plausible, given the existence of apparent shared names, it is claimed that names must be individuated in such a way that each name can only have one referent (at a time), and that any instances of apparently shared names are in fact examples of homonymy or homophony. Each name, then, has its referent, which it refers to tout court. Names gain their reference through the baptism or dubbing of objects with the name, and maintain their reference, and(/or) their identity, via causal chains linking each use back to the baptism. I attack this view through four separate arguments: that it does not account effectively for shared names; that it does not account effectively for communication using proper names; that it does not account effectively for the use of family names; and that it does not account for the apparent pluralism of proper name reference. I do not take these arguments to refute causal-homophonism definitively, but, rather, to show that it strains to account properly for the way that proper names are actually used, and must be excessively complicated to do so. In chapter 4, I turn my attention to indexicalism: the view, or family of views, that proper names are, properly speaking, a variety of indexical, or at least bear strong resemblances to indexicals in important ways. This amounts 18 to the claim that names are context-sensitive with regard to their reference, and have a semantics that specifies how they interact with context in order to refer. My general argument against indexicalism, influenced by Travis, is essentially the claim that, for any way that it is claimed the context of utterance fixes the reference of a name, it will either be too general, and so will not specify a particular way that the context must be to determine an extension—and so fail to provide truth-conditions—or it will be wrong, and the specified contextual parameters can be held constant whilst the reference of the name can be varied. I consider three particular indexical views proposed in the literature, observing particular problems with each, and then go on to discuss general issues with indexicalism. Firstly, I discuss the problem of proposing particular contextual parameters to determine reference generally, and then I consider the idea that name/bearer relations are contextual factors that can play a primary rôle in reference determination, presenting a number of problems for it. Chapter 5 addresses intention-based accounts of proper name reference. I consider two different forms of such an account: constraint-based intentionalism, which I associate in particular with Michaelson (2013); and a strong neoGricean approach to reference, which I associate most closely with Stephen Neale (2004, ms.). I argue against constraint-based intentionalism on the grounds that it fails to account properly for the vagaries of proper name reference. I then take a somewhat different approach to neo-Gricean views, proposing that the kind of strong general position that would be required to support the view that it is speakers’ referential intentions alone that determine reference is untenable given particular data from experimental psychology. In part III of the thesis, I introduce my own pragmatic view of proper name reference, already discussed above, which, I claim, is better able to deal with the vagaries of language use that presented problems for the positions I discuss in part II. I begin by presenting seven principles that provide the key points of the view, which can be summarized as follows: the reference of an utterance of a name is determined by the circumstances of utterance— pragmatic factors of the context—in a way that is not specified in advance by the semantics of the term. In the rest of the chapter I go into much greater detail, discussing various important aspects of the view, including reference, truth, context-sensitivity, semantics, necessary and sufficient conditions, and bearers. I then consider the way that my position on proper names relates to Travis’s account of occasion-sensitivity more generally, and finally how I can address two potential criticisms of my view: that it is unable to make predictions about reference, and that necessary and sufficient conditions are required for there to be proper name reference at all. I consider this 19 second potential objection further in chapter 7. In this chapter, I discuss at length referential uses of familial terms such as ‘mum’, and conclude that the only way to account for the range of these uses is by maintaining that their reference is fixed wholly pragmatically, and is not determined or significantly constrained by their semantics. I take this conclusion to add plausibility to the claim as it applies to proper names. In chapter 8, which concludes the thesis, I offer a summary of what has gone before, laying out what I have established. I then go on to consider some of the lose ends and topics for further research. I briefly address what the semantics of names might have to look like, in order to accommodate my view of their reference. I suggest that two existing accounts might plausibly be adapted to fit with my view. Finally, I consider how my pragmatic view of proper name reference deals with some of the classic issues and problems associated with names: rigidity, reference shift, puzzles about belief, and fiction and empty names, suggesting how my view can address these issues, and how I might need to develop it in light of them. 20 Chapter 2 Methodological and Philosophical Preliminaries That what we ordinarily say and mean may have a direct and deep control over what we can philosophically say and mean is an idea which many philosophers find oppressive. S. Cavell, ‘Must We Mean What We Say?’ 2.1 Reference W hat are philosophers and semanticists talking about when they talk about reference? It is generally accepted that there is a phe- nomenon of reference. At the very least, within ordinary language, (uses of) proper names and other types of expression are spoken about as referring to things or being about them. For example, if a speaker says, in a very ordinary type of situation, ‘Lucy is dead’, or ‘This soup is disgusting’, or ‘I met the Queen once’, they are likely to be taken to have said something about an object or entity in each case: ‘Lucy is dead’ is likely to say of some (previously) animate object, probably bearing the name ‘Lucy’, that it is dead; ‘This soup is disgusting’ is likely to say of some demonstrated or salient soup (or soup-like substance) that it is disgusting; and ‘I met the Queen once’ is likely to say of some particular, contextually-determined queen (or queen-like object) that the speaker once met them. In each utterance, it is a particular expression—‘Lucy’; ‘this soup’; ‘the Queen’—that tells us what object or 21 2.1. Reference entity the utterance is about. It appears that the truth of an utterance which contains such an expression—that is about a particular object—turns on how the particular object in question is.1 If I say ‘Lucy is dead’, and succeed in expressing something truth-evaluable, then I’ve said something true if and only if the object the utterance is about can correctly be described as being dead. It is this relation of (an utterance or use of) a word or expression connecting to some object or entity (or connecting that object or entity to what an utterance is about) that I call reference, or, more specifically, singular reference. It is also the relation that I believe other philosophers and semanticists are talking about when they talk about singular reference, even though I have couched my description of it in language with which not all other theorists would be happy. Given this account of what kind of thing we mean by ‘reference’, what is it that philosophers and semanticists are doing when they theorize about reference? One ongoing programme, at least within philosophy of language, is to explain what makes it such that various kinds of referring expression, for example, indexicals, demonstratives, proper names, and, probably, definite descriptions, refer to what they do, when they do, either by providing one unified story for all such terms, or by offering different stories for different types of term. The purpose of this thesis is to address this question as applied to proper names, and suggest an answer (of sorts) to it. 2.1.1 Proper Names Thus far, I have been talking about proper names, but have said nothing about what they are. Clearly, taking too detailed or precise a position on exactly what counts as a proper name risks assuming strong theoretical claims about them, for example about how their reference is determined, before any discussion or argument takes place. However, most examples of proper names are largely uncontroversial, and it is these that I will focus on in the course of this thesis. The most paradigmatic examples of proper names seem to be the names of people, such as ‘Edwina Currie’; ‘Lucy Campbell’; ‘Muhammad Ali’; and ‘César Chávez’; followed by the names of places, such as ‘Walthamstow’; ‘Antarctica’; ‘Kiribati’; and ‘Abuja’. The paradigmatic form for names in English, then, seems to be that of a bare referential term, (without an apparent determiner beforehand), spelt with a capital letter at the beginning. There is debate over whether some expressions that resemble (definite) descriptions are in fact proper names, particularly examples such 1 That is, of course, if the utterance is a truth-apt kind of utterance. If the utterance is apt for another kind of correctness, then that turns on how the object in question is. 22 2.2. Data on Reference as ‘The Holy Roman Empire’, the name for something neither holy, Roman, nor an empire (Kripke 1981, p. 26), or ‘Coffee Hour’, the name used in the University of Chicago Philosophy Department for a regular gathering on Friday afternoons that involves no coffee, and lasts substantially longer than an hour. In these examples, the apparent descriptive content not only is not satisfied by the referent, but seemingly plays no rôle (any longer, perhaps) in fixing the referent. As observed by Ziff (1977), however, at least in English, there are a range of uncontroversial examples of names that do not conform to the paradigm, beginning with a definite article, or otherwise diverging. For example, pub names, band names, company names, and various others: ‘The Jerusalem Tavern’; ‘The Chequers’; ‘The Be Good Tanyas’; ‘The E Street Band’; ‘Ben Folds Five’; ‘The White Company’; ‘It Takes a Nation of Millions to Hold Us Back’. In general, it appears that we give names to specific and particular things that are significant in some manner, and that we want to be able to identify and re-identify with some ease. 2.2 Data on Reference In order to address the question of what names refer to when they refer to it, which I generally express more succinctly as ‘what determines the reference of proper names?’, there need to be some data about what names or other expressions refer to, such that any proffered answer can be deemed to be adequate or not. I take it that these are questions about how names are used by speakers, or, perhaps less controversially, one or more of: what names do when they are used; what referential utterances of names do; or what speakers are doing when they utter names referentially. Whether one thinks that names have referential content inherently, given the way in which they are introduced into a language, or whether one thinks that only uses of names, or the speakers using them, are capable of reference, one must presumably think that this reference is grounded in, or exemplified by, the practice of language users. It is this practice of referring on which data is required. There are at least three potential sources for such data: linguistic corpora, the judgement of ordinary speakers, and the judgements of experts. I will argue, however, that these do not really represent three distinct sources, but that the use of corpora breaks down into appeal to the judgements of either experts or ordinary speakers, and the judgements of experts provide no more or better data than the judgements of ordinary speakers. 23 2.2. Data on Reference 2.2.1 Corpus Data A linguistic corpus is a structured collection of examples of uses of (a particular) language, collected from real-world occurrences, such as recorded conversations of different kinds, newspapers, books, and academic texts. Corpora are frequently tagged for parts-of-speech, which would enable one to search for occurrences of proper names, and observe how those names are being used in a wider linguistic context. However, whilst this might provide one with some interesting and illuminating examples of the use of names, it is unclear whether it would provide much data on proper name reference. This is because, whilst we might assume that many of the occurrences of names in a corpus refer, the corpus will rarely provide information useful to the researcher about what is being referred to. Consider the following entry in the British National Corpus, which is taken from a recording of an orchestral society committee meeting. The minutes of a previous meeting are being read: Present secra– chairman, secretary, Julia, Judith. Apologies were received from Phil, Malcolm, John and Pauline. Er minutes of the meeting held on the tenth of January were read and approved. [BNC:F7V] Numerous names are used in this snippet of dialogue. All presumably referred to something when they were used, and their use seems very typical of name use. But, whilst any account of proper name reference will presumably have a story to tell about how those names refer or referred to whom they do or did, there is very little provided by the corpus data that would lead us to prefer one account over another. Whilst it seems very likely that ‘Julia’, ‘Judith’, ‘Phil’, ‘Malcolm’, ‘John’ and ‘Pauline’ all referred to members of the committee bearing those names, we have no means of knowing who those people were, such that we could judge, on any basis, whether theoretical predictions about reference are correct. Certain corpus extracts may provide more information, for example by showing apparently unusual uses of names, or cases in which the reference of a name appears to be at issue within the dialogue, such as the following: A: Now look, I think, I think he’s got another brother called Carl B: A younger brother, yes A: Oh B: Gavin’s is older, er is, second oldest A: yeah and Thomas is the other one B: Yeah but the littlest one A: yeah, but she got off with him B: Clare he’s only about five (laugh) A: (laugh) Oh who’s Carl then? I 24 2.2. Data on Reference B: (unclear) A: I reckon it’s one of those boys from the army camp, remember (pause) I heard her talking about a Carl from the army camp (pause) yes (BNC:KBN) Whilst it may be enlightening to consider what is going on with the name ‘Carl’ in this dialogue, any conclusions we draw must be based on our own judgements about what the speakers are doing, informed by our competence as speakers of the language, or as experts in some relevant field, familiar with the kinds of thing the dialogue’s participants appear to be doing. 2.2.2 Data from Judgements Any use of corpus data looks, then, like it will reduce to use of one of the second and third potential sources of data about the reference of proper names: the judgements of ordinary speakers, and the judgements of experts. These both involve considering scenarios—which could be hypothetical or real—in which a proper name is used by a speaker, and making judgements about what has been referred to. For example, the man with the martini and Jones’s killer in Donnellan (1966) (though, of course, these involve definite descriptions rather than names); Smith and Jones and the leaf-raking in Kripke (1977); and Gödel and Schmidt in Kripke (1981). Whilst there may be many worries about the set-up and use of these kinds of cases in philosophy of language, some of which I will address below, it is unclear whether anything other than the judgements they are designed to elicit, whether they be from ordinary speakers or experts, could ever constitute data of the kind under discussion. There has been much discussion in recent years about the rôle intuitions play in philosophy, and whether they should or should not play that rôle,2 some of it on the topic of judgements about reference.3,4 But, when the object of study is language itself, regarding word meanings, or reference, or truth-conditions, for example, intuitions or judgements about the meanings of words, the reference of expressions, or the truth-values of sentences or utterances appear to be all there is to go on. This point was borne out by my consideration of the use of corpus data; when attempting to establish the reference of a use or occurrence of a referential term, there is no natural indicator or test that will reveal the referent. Instead, although what we are interested in is how names are used referentially, we must rely on the 2 For example, Cappelen (2013); De Cruz (2015); Malmgren (2011); Nagel (2012); Weinberg et al. (2010); Williamson (2007). 3 For example, Deutsch (2009); Machery et al. (2004); Mallon et al. (2009); Martí (2009). 4 Note that by using the term ‘judgement’, I am intending something broader than is usually intended by ‘intuition’, particularly with regard to the judgement of experts. I will return to this below. 25 2.2. Data on Reference judgements of the interlocutors involved in the discourse, if they are available, or of experts or speakers who are familiar with the details of the case. Even Mallon et al. (2009), who deny that intuitions are a viable source of evidence for choosing between theories of reference, state, with regard to intuitions about reference: [W]e have no idea what other considerations philosophers of language might appeal to. Thus, in the absence of concrete suggestions, we remain skeptical of the proposal to downplay the role of intuitions in choosing a theory of reference.5 (ibid., p. 343) We might draw a distinction here with other kinds of empirical investigation, for which there are other methods of data collection. If we wanted to investigate the precise dynamics of spade-use in undertakers, it might not be fruitful to collect merely the judgements of undertakers. Although they presumably do have some insight into the way they move their spades, much of their practice might be learned merely through digging, without any intellectual reflection upon it. Thus, asking the undertakers to talk about their practice may result in very little detail, or even in incorrect, data. Fortunately for the spade-dynamics enthusiast, since spade use is an observable, external phenomenon, there will be ways of recording, in great detail, the subtle threedimensional movements of a spade as it digs a grave. This is not so with reference. In observing someone utter the words ‘Lucy is dead’, there are no external measurements or observations we can make that will inform us of the reference of ‘Lucy’. We must rely on the capacity of language-users with the right kind of access to the utterance. The Judgements of Ordinary Speakers The second source of data about proper name reference that I will consider is the judgements of ordinary speakers. Specifically, the judgements of ordinary, competent speakers of the language6 about what object or entity a particular expression referred to, if it referred to anything, on an occasion of utterance. Such judgements might be collected either by asking direct questions about what was referred to, or by formulating questions about whether utterances of certain sentences would be true or false under certain circumstances. The choice of questions would, of course, require careful consideration. ‘Ordinary’ should be taken to mean that the speaker is making judgements based on 5 Mallon et al. are in fact talking specifically about the intuitions of competent speakers of the language, rather than the broader range of judgements I am discussing, but a fortiori, they support my point. 6 Since all the utterances I will dealing with will be in English, ordinary speakers should be taken to be ordinary English speakers. 26 2.2. Data on Reference the knowledge they have just in virtue of being a competent speaker, not in virtue of having theory-laden beliefs about how their language functions, held independently of their being a speaker of the language. There are at least two different approaches to the collection and use of ordinary speaker judgements. One assumes that the judgements of all ordinary speakers, perhaps within a linguistic community, will be similar enough that surveying the judgements of a very small number of speakers—perhaps just one—is sufficient to form a conclusion about what has been referred to. The other assumes that a more rigorous method of data collection is required, including larger sample sizes, if one’s conclusions are to be trusted. The latter approach, as it applies to singular reference, has not obviously ever been practiced with the aim of producing positive results in the philosophy of language. However it has been used recently in attempts to demonstrate the shortcomings of the former approach by experimental philosophers such as Machery et al. (2004) and Mallon et al. (2009). I will address experimental philosophy in detail from p. 34 onwards. The Judgements of Experts The view that the judgements of ordinary speakers about the kinds of cases discussed above is a major source, or the sole source, of data about reference is contrasted with the view that the judgements of experts, our third source of data, are a significant source. This view tends to take the form of the assertion that philosophers, and perhaps linguists, have, through their study, achieved expertise in the mechanisms and functions of language and reference. Thus, they are better placed to make judgements about the truth-values, truthconditions, or reference of linguistic expressions than are competent speakers who do not have a background in linguistic research. Such a view, about the significance of the judgement of experts, can take two forms that are interrelated. One is the view that the intuitions elicited in an expert by exposure to cases or scenarios of the kind discussed above are more reliable than those of non-expert speakers. The experts are thus better at correctly identifying the referent of a term. The second form of the view is that, because experts are well-informed about language in general, they are able to see how reference must be in order to be consistent with their well-informed picture of language in general, and so make informed judgements.7 For example, if a theorist deems some version of truth-conditional semantics to be true, that may entail that reference must be a certain way, and so the theorist will make judgements 7 On this view, the judgements being made are not, in the standard sense, intuitions, which is one reason I distinguished judgements from intuitions in n. 4, p. 25. 27 2.2. Data on Reference about cases of reference accordingly.8 I claim that the two forms of the expert judgement view are inter-related because it is hard to see how expert intuitions could be superior to pretheoretic intuitions, except by being informed by the theoretical commitments they have taken on. There is little in philosophy of language or semantics that is agreed upon across the board, so, in studying and researching in these areas, although one may develop a good knowledge of the whole field, any particular single view or theory to which one is avowed will be contentious. Two experts avowed to very different semantic theories seem unlikely to have matching intuitions about a referential case if the intuition is inconsistent with one of their semantic theories. If it is, they will likely be moved either to adapt their theory, or to renounce their intuition. It may be that developing an expertise in philosophy or linguistics improves one’s intuitions in other ways, for example by sharpening one’s capacity to spot inconsistency, or for appreciating the subtlety of a scenario. But, in the cases and scenarios that are relevant to judgements about reference, it is unclear how these skills would aid one, if not by helping one to see how the cases are affected by one’s theoretical commitments. 2.2.3 Judgements in Practice Despite the very different methodological approaches they represent, it appears that the view that ordinary speaker judgements should be the primary source of data about proper name reference, and the view that the judgements of experts should be, are often not properly discriminated. In many texts on proper names (particularly, perhaps, prior to the advent of experimental philosophy) cases bearing on reference are presented with an answer to the question of what is referred to. The cases are designed to demonstrate how a particular philosopher’s account of reference deals with a particular kind of situation, or how, even though one’s account appears to make predictions at odds with the apparent answer, the account can be adapted to allow for deviations. Presumably, the answers in these cases are those provided by the judgement of the author, but, often, it appears to be assumed that the reader will share that judgement, or at least will be easily persuaded of its correctness. This approach is found in the works by Donnellan and Kripke already cited, and also in Evans (1973), Fara (2015), Kaplan (1989b), King 8 A third view on experts is that they are better than ordinary speakers at distinguishing relevant from irrelevant cases, and thus relevant from irrelevant judgements, without explicitly consulting theory. However, this is not a view about the value of experts’ first-order judgements about reference, but rather of their judgements about which kinds of cases they are interested in. This is more obviously the purview of the expert, even if they are not good at judging who should be making the first-order judgements, nor at collecting that data. 28 2.2. Data on Reference (2014), Jeshion (2015), and Sainsbury (2005), amongst others. But if this is the case, it is unclear whether the judgements are supposed to be in accordance with those of ordinary speakers, or whether the author is presenting some form of expert opinion. Let us take, as an example, Kripke’s Gödel/Schmidt case. Kripke presents the case as a counter-example to the view that the name ‘Gödel’ is semantically equivalent to, and so, for Kripke, necessarily co-referential with, the definite description ‘the man who discovered the incompleteness of arithmetic’. Suppose that Gödel was not in fact the author of [the incompleteness] theorem. A man named ‘Schmidt’, whose body was found in Vienna under mysterious circumstances many years ago, actually did the work in question. His friend Gödel somehow got hold of the manuscript and it was thereafter attributed to Gödel. On the view in question, then, when our ordinary man uses the name ‘Gödel’, he really means to refer to Schmidt, because Schmidt is the unique person satisfying the description, ‘the man who discovered the incompleteness of arithmetic’.[. . . ] So, since the man who discovered the incompleteness of arithmetic is in fact Schmidt, we, when we talk about ‘Gödel’, are in fact always referring to Schmidt. But it seems to me that we are not. We simply are not. (Kripke 1981, pp. 83–84. My emphasis) Kripke is emphatic in his judgement about the reference of ‘Gödel’ in the case he provides, although he acknowledges it to be how things seem to him. He does not state that his expertise led him to this conclusion, although he has clearly thought at length about the issue. But neither does he assert that he expects any ordinary speakers to share his judgement, although if a significant portion of his audience or readers were to demur, it does seem likely that his conclusions would be less convincing. As it happens, it appears that many philosophers were, and continue to be, fairly convinced by Kripke’s judgements about his cases. Mallon et al. (2009) assume that, with regard to the Gödel/Schmidt case, Kripke is appealing to the judgements or intuitions of ordinary speakers. Indeed, they take the general tradition I have been discussing, of using cases and scenarios to elicit judgements about reference, to be designed to engage with the general, and perhaps universal, judgements or intuitions of ordinary speakers. They label it ‘the method of cases’: The method of cases: The correct theory of reference for a class of terms T is the theory which is best supported by the intuitions competent users of T have about the reference of members of T across actual and possible cases. (ibid., p. 338) 29 2.2. Data on Reference Given this interpretation of Kripke’s method, Mallon et al. argue that his case fails to demonstrate his conclusion by appealling to data reported in their previous paper, Machery et al. (2004). They allege that this data shows there is significant cross-cultural variation amongst ordinary English speakers in their judgements about Gödel/Schmidt-type cases, and even that intraculturally, there is limited consensus. More broadly, Mallon et al. claim that any argument that appeals to a position about reference supported by the method of cases is undermined in the same way. Deutsch (2009) challenges the conclusions of Mallon et al. by claiming that Kripke is not appealing to the judgements or intuitions of ordinary speakers in the Gödel/Schmidt case, but rather is appealing to something more like the judgement of experts: Competent speakers can get it wrong, however. Competence in a language does not bestow philosophical insight. At worst, surveys of competent speakers will record judgements about merely pragmatic implicatures, instead of semantic contents, and so their results will not qualify as evidence that different groups of people have genuinely different intuitions about what counts as a case of referring to Gödel, or an intentionally produced side effect. We saw that, in the case of Kripke’s argument against descriptivism, it need not be assumed—and is not assumed by Kripke himself—that Kripke’s judgement about the Gödel/Schmidt case is shared by all, or even most, competent speakers of English. What matters is whether the judgement is true. And whether one is justified in supposing that it is true will depend on the quality of the arguments for its truth, not on how many people intuit that it is true. (ibid., pp. 464–5) Deutsch claims both that the judgements of ordinary speakers are unreliable, and that Kripke’s argument is not effected by whether or not any number of competent speakers agree with his judgements about the case. There is, then, a very clear disagreement between Deutsch and Mallon et al. about whether Kripke, and presumably other philosophers who appear to use a similar method, are appealing to the judgements of ordinary speakers, or experts, or something other than judgements altogether. I will argue that, whether or not philosophers take themselves to be putting forward expert or argued opinion when making judgements about reference, only their status as ordinary speakers is relevant with regard to the value of their judgements. 30 2.2. Data on Reference 2.2.4 Against Expert Judgements I will press Deutsch (2009)’s response to Mallon et al. (2009) further, as it illuminates some of the major problems with the use of expert judgements. Deutsch claims that the worst possible outcome of Mallon et al.’s experimental methodology is that, in surveying ordinary speakers, they will “record judgements about merely pragmatic implicatures, instead of semantic contents”, which is problematic since he clearly thinks that “what counts as a case of referring to Gödel” is an issue of semantic content (Deutsch 2009, pp. 464– 5). Since the aim of this thesis is to argue that proper name reference is a pragmatic, rather than a semantic, phenomenon, Deutsch’s objection gives me the strong impression that his defense of Kripke rests on theory-laden ground: If the judgements of ordinary speakers were sensitive to pragmatic effects, rather than semantic content, and reference were a pragmatic effect, those judgements would be good evidence regarding reference. As I have already stated, the view that the judgements of experts are a reliable source of data on reference appears to assume that judgements consistent with a particular theory of reference or language more broadly are a good thing. But to criticize Mallon et al. in this manner is not to attack them on their own terms. To do so is to assume the expert judgements view. Deutsch’s view appears to be that there is a fact of the matter about what referential terms refer to that is independent of the judgements of users of the language, and possibly independent even of how they use the language, and that working out what this fact of the matter is requires ‘philosophical insight’ rather than linguistic competence. However, Deutsch also claims that Kripke presents arguments for his conclusions regarding the nature of proper name reference, illustrated by his claims about the Gödel/Schmidt case, rather than relying on his judgements about the case to support or motivate those conclusions. So, not only is Kripke not appealing to the intuitions of ordinary speakers, he is not making use of his own intuitions. I observed in n. 4, p. 25 and n. 7, p. 27 that a distinction can be drawn between judgements and intuitions on the grounds that judgements may be made about cases in order to accord with theoretical commitments in spite of contrary pre-theoretic intuitions. The arguments Deutsch claims Kripke uses for his conclusions (ibid., pp. 451–2) are not arguments from universally agreeable premises, but instead rely on one or more of the following: intuitions about other cases; significant theoretical assumptions about reference and names; or an assumption of the correctness of the causalhistorical account of proper name reference with which Kripke proposes to replace the descriptivist picture that the Gödel/Schmidt case is supposed to 31 2.2. Data on Reference be a counter-example to. Thus, even given Deutsch’s interpretation, Kripke is certainly relying on his judgements about the Gödel/Schmidt case in drawing conclusions about it, even if he is not merely relying on his direct intuitions about it. When I introduced the view that the judgements of experts are a significant source of data about reference, I claimed that the view could take two interrelated forms: that experts simply have better intuitions than ordinary speakers, in virtue of their expertise; and that experts make better judgements than ordinary speakers because they are able to make judgements about reference in accordance with broader theories they believe to have good standing. Deutsch’s defense of Kripke, once we have probed it, is predominantly an example of the latter form. Judgements of experts of this sort will not provide reliable evidence about reference of the kind we want if our aim is to compare rival accounts of proper name reference. If the approach is to tell us anything about reference as a genuine phenomenon of language—the phenomenon I discussed at the beginning of § 2.1, that determines what objects utterances containing names are about, such that the truth of those utterances (if they are truth-evaluable) turns on how those objects are—it relies on the accuracy and correctness of the theoretical commitments in accordance with which the judgements about reference are being made. And this accuracy and correctness must also be measured against language as it is used. It cannot be based on idealizations and abstractions which spawn nice theory but do not describe any real phenomena. Papineau (1996) argues for the kind of approach that places theoretical commitments ahead of intuitions, though not specifically with regard to language and reference. He claims that it is not grounds for dismissing an a posteriori theory that some of its consequences are at odds with the intuitions of ordinary speakers or thinkers, since the theory is not intended as merely an analysis of everyday notions (p. 130). However, it is doubtful that the kind of a posteriori theory Papineau has in mind, that does not aim to explicate the everyday notions of speakers, is the kind of theory that is of interest to semanticists and philosophers of language who are aiming to describe how speakers use language. As I discussed at the beginning of §2.1, the phenomenon of proper name reference I am interested in is the kind of thing that can be identified by questions such as ‘Who was Katharine talking about when she said “Lucy is dead”?’ or ‘Who did you mean when you said “Lucy”?’. This seems to be a very ordinary kind of phenomenon utilized very frequently by speakers in saying the things they do, and whilst speakers need not be reflective about their practice, they generally do know what they’re doing, at least to the extent that they know—to some degree—what objects 32 2.2. Data on Reference they’re talking about. Investigating the judgements of speakers about their referential practice is not an a priori activity that merely surveys people’s existing notions, it is at least an attempt at collecting empirical data. If a broader theory about semantics, pragmatics, or some other aspect of language were to entail an account of reference as I am conceiving it, it would itself need to be informed by the practice of speakers. Moreover, the account of reference entailed by such a theory would presumably, as much as possible, be in accord with the judgements of users of the language about reference. With regard to the idea that expert judgements are a good source of data about proper name reference, the justification that experts’s judgements will be made in accordance with well motivated theoretical considerations looks decidedly shaky, given the notion of reference we are interested in. It looks even less justified in light of the fact that the whole point of looking for data on reference, at least as I introduced it, was to be able to make discriminations and choices between rival accounts and theories of reference. Knowing that one account of reference fits better with a broader theory will not help choose between rival accounts if those accounts are being proposed by theorists who espouse different broader theories. As I argued above, the first form of the expert judgement view, that experts simply have better intuitions than ordinary speakers, is closely related to the second form just dismissed, because it is unclear why experts could have better intuitions except by being informed by their theories. To the extent that this is true, the first form will be subject to the same shortcomings and criticisms as the second. However, an alternative interpretation of the first form of the view bears consideration since it appears, at least superficially, to be independent of the second form. This interpretation of the view is stated by Weinberg et al. (2010, p. 333) in the following way: [P]hilosophers’ intuitions are sufficiently less susceptible to the kinds of unreliability that seems to afflict the folk intuitions studied by experimental philosophers. (Emphasis supressed) Weinberg et al. associate this view with Hales (2006), Ludwig (2007), and Williamson (2007). The statement of this interpretation in terms of reliability and unreliabilty appears to separate it from my interpretation on which socalled philosophical expertise just grants one a capacity to see how a case relates to one’s theoretical commitments. Inherent in the idea that intuitions can be more or less reliable is the idea that there is some fact of the matter independent from those intuitions. When it comes to proper name reference, this idea is rather complicated. I do not deny that, with regard to any particular occasion of use of a proper name, 33 2.2. Data on Reference there can be a fact of the matter about what was referred to (though there needn’t be), but, to the extent that there is, it is facilitated by the ways in which speakers use proper names. The kind of phenomenon I am interested in cannot be divorced from what users of the language actually do. Therefore, there cannot be facts of the matter about what names refer to such that users of the language systematically use names wrongly. Since this is primarily a matter of how names are used, it is consistent with a critique of reliance on intuitions which says that speakers, whilst perfectly capable of using names and other referential terms correctly, are bad at reflecting on this practice, and so will be unreliable judges when asked to make judgements about scenarios in which they are not themselves involved. However, whilst this critique might bring one to question the use of the judgements of ordinary speakers as a source of data on reference (as it does for Mallon et al.), it certainly should not lead one to think that the judgements of experts are a good source. The kinds of analytical and subject-specific skills that seem to be involved in philosophical expertise are not obviously skills that better attune one to the ways in which names are used, at least not more so than any other careful language-user. Weinberg et al. (2010, p. 334) suggest that philosophical training typically brings “a mastery of relevant literatures[. . . ], and even specific technical skills such as argument evaluation and construction”. They survey a range of psychological literature on expertise and roundly reject the claim that there is reason to think that philosophical expertise would be likely to improve judgements or intuitions of the kind at issue. Three hypotheses are considered with regard to how philosophical expertise might improve one’s judgement, and all three are found implausible on the basis of existing psychological literature. They are that philosophers have ‘superior conceptual schemata’; that they ‘deploy more sophisticated theories’ (the kind of hypothesis I have already discussed); and that they ‘possess a more finely-tuned set of cognitive skills’ (ibid., p. 336). In light of the various arguments and considerations I have offered here, I conclude that any sensible judgements that experts make in cases designed to elicit judgements about the reference of proper names are sensible just as a result of the experts being competent speakers of the language, and not in virtue of any further expertise they might possess. 2.2.5 Ordinary Speakers’ Judgements Again If the judgements philosophers and other reference theorists make about the cases they present are simply the judgements of competent speakers, then 34 2.2. Data on Reference we must ask whether these judgements count as a good source of data about reference, and if not, whether the judgements of competent, ordinary speakers can ever count as a good source of data. As we have already begun to see, there are significant problems with the way in which philosophers have generally used ordinary, or at least competent, language intuitions and speaker judgements about reference. Just as I have claimed Kripke does, philosophers frequently present their own intuitions—often highly informed by their theoretical interests and upbringing—as self-evident. This is particularly problematic since there is a well-known demographic problem in analytic philosophy,9 so the intuitions being recorded, taught and reproduced tend to be those of a very limited range of English (or otherwise) speakers, i.e. well-educated, affluent, white cis men from North America or Western Europe. Kripke (1981) is the prime example of this kind of use of intuitions, and, since Naming and Necessity has become so influential and widely-taught in analytic philosophy departments, those intuitions have largely become orthodoxy. Kripke’s stated judgements about, for example, the Gödel/Schmidt case, have become the received judgements, and anyone wishing to develop a contemporary descriptivist account of proper name reference must account for such cases in line with these judgements (e.g. Jackson 1998). However, as I have already alluded, Machery et al. (2004) claim to show that there is not a consensus amongst English speakers upon Kripke’s judgements. Some philosophers, being more self-aware in their use of judgements, may attempt, in various ways, to collect judgements about reference from a wider range of speakers, perhaps even non-philosophers.10 Whilst such activities may produce more reliable data about the use of referential terms than pure introspection, the results may not be as reliable or as rigorous as they could be. Linguists who collect speaker-judgements in order to investigate, for example, the acceptability or grammaticality of a sentence, aim for as rigorous a method of data-collection as possible, including large and diverse sample size, and thorough, informed consideration of the questions being asked. Of course, philosophers frequently lack the training in experimental methods and the resources necessary to carry out rigorous studies into speaker-judgements, but it certainly appears that, if speaker judgements are significant to us, such methods should be what is aimed at when making claims about what an expression refers to on an occasion. As already discussed, experimental 9 See, e.g., APA (2013); Beebee (2013); Beebee and Saul (2011); Gines (2011); Haslanger (2008); Healey (2011). 10 Although he doesn’t mention it in the text, Michaelson (2013) claims to have informally solicited judgements from a variety of speakers with regard to his sneaky reference cases (pp. 88– 93). 35 2.2. Data on Reference philosophers do aim at utilizing rigorous experimental methods in order to collect judgements from non-philosophers. With regard to reference, the findings of experimental philosophers must be preferred over data gathered less rigorously, given all that I have said. However, the results of Machery et al. (2004), who carried out experiments on groups of English-speaking Hong Kong Chinese and American students to test their referential intuitions, do not in fact give reference theorists the kind of exciting data that they would like. As mentioned in the preceding paragraph, these experimental philosophers claim that there is significant diversity of judgement about cases similar to the Gödel/Schmidt case, both within and between distinct English-speaking cultural groups. I have argued that the only serious contender for being a source of data about reference is ordinary speakers’ judgements about cases and scenarios involving referential terms. If the kind of rigorous collection of ordinary speaker judgements about reference I appear to have been championing suggests that speakers frequently do not make the same judgements about cases, then we may have a serious methodological problem. There are at least two ways to go in the face of this claim. One is to question the veracity of the claim by challenging the experimental methodology of Machery et al. (ibid.) and highlighting general problems with the collection of such data. This view would also be consistent with accepting that speakers’ judgements about referential cases differ significantly, but put this down to problems with language users’ awareness of their own practice. The second way to go is to accept the claim, and use it to argue that there is no consensus in judgement because there is in fact no consistency in the practice of reference. Something akin to this second way is indicated by the hypothetical position introduced by Mallon et al. (2009, pp. 343–47) called ‘referential pluralism’. This position essentially maintains that the judgements made by a particular speaker in response to referential cases are evidence for their general referential practice. Thus, there may be groups of English speakers for whom the reference of proper names is determined descriptively, and distinct groups for whom it is determined in the causal-historical manner outlined by Kripke (1981). As a result, large numbers of speakers may end up talking past one another. Referential pluralism, as presented by Mallon et al., is clearly unworkable, and is presented merely as a reductio of attempts to appeal to ordinary speaker judgements in spite of Machery et al.’s claims about lack of consensus. In fact, in chapter 6, I will endorse a form of pluralism concerning the mechanisms that determine the reference of names. However, the position is significantly different to that outlined by Mallon et al. Most significantly, I do not claim that referential mechanisms are relativized to different linguistic groups, 36 2.2. Data on Reference identified only by the ways their judgements turn. Rather, I claim that different mechanisms may come into play according to contextual variation, such that, on any particular occasion, different mechanisms can determine reference, but speakers do not each have their own mechanisms. As Mallon et al. (2009) indicate with their reductio of referential pluralism, it is not plausible that every speaker judgement about a hypothetical referential case is a strong indicator of that speaker’s referential practice. This eliminates the second way of going in reaction to the claims of Machery et al. (2004). What, then, of the first? This involves questioning the veracity of the claims, and observing general problems with the project. This is the route taken by Martí (2009). Martí argues that the questions asked by Machery et al. in their study were inadequate. In the study, subjects are presented with written scenarios in which there is tension between the origin of a name and the information frequently associated with it, just as in the Gödel/Schmidt case, and Kripke’s Jonah case (Kripke 1981, pp. 66-67). The subjects are then asked counterfactual meta-linguistic questions about what the name would refer to, given the situation. Martí claims that these questions do not test the subjects’ intuitions about how names are used, but rather, they test their intuitions about meta-linguistic theories of reference. That is, the questions addressed how the subjects think names work, not how they use names. Martí is appealing to the idea that speakers may not be fully aware of their own practice. She points to the fact that Mill and Russell differed radically in their claims about how (or if) names refer, but presumably did not differ in the way they used names. Martí goes so far as to claim that “the experiment does not provide any evidence at all about name use” (Martí 2009, p. 46). It should be noted that neither Machery et al. (2004) nor Mallon et al. (2009) actually claim that their results suggest that different communities of English speakers use names differently, or that reference functions differently for different speakers. Strictly speaking, they do not aim to provide any evidence about name use. Their claim is that ordinary speaker judgements are frequently used as evidence for how names are used, and that this methodology is undermined by showing that there is variation and lack of consensus in ordinary speaker judgements. That we could infer from this variation in judgements that there is variation in referential practice is the position assigned to the referential pluralist, and rejected. Martí’s argument must then be that it is possible for speaker judgements to provide evidence about name use (Martí 2009, p. 47), but that Machery et al.’s experiments fail to do so because they do not ask the right kinds of questions and do not elicit the right kinds of judgements. Although I think there is something to Martí’s criticism of Machery et al.’s 37 2.2. Data on Reference methodology, her claim that the study can tell us nothing about name use seems to go too far. I have argued that speaker’s judgements about referential cases are the only genuine source of data about reference. Whilst there will certainly be occasions on which speaker’s judgements are out of step with their actual practice, particularly if the situation in which their judgement is solicited is sub-prime, language-users are clearly, in general, aware of what they and those around them are referring to when they use names. If they weren’t, then something very strange would be going on. It may well be that Machery et al.’s scenarios were too complex, and the questions they asked too theoretical to elicit judgements closely tied to the subjects’ own nameusing practice. But, since Martí believes that Kripke’s causal-historical picture of reference is in fact correct, her claim that the experiments tell us nothing about the way the subjects use names leaves her in the uncomfortable position of claiming that the Hong Kong Chinese subjects of the experiment have statistically significantly poorer insight into their referential practice than their non-Chinese US counterparts. For her own part, Martí does not dismiss the idea of collecting judgements about reference from larger numbers of ordinary speakers, and suggests a way of asking questions about Machery et al.’s scenarios that relies more on the subjects’ referential practice than their meta-linguistic views. She suggests that the questions about the Gödel/Schmidt case be something like the following: One day, the fraud is exposed, and John exclaims: ‘Today is a sad day: we have found out that Gödel was a thief and a liar’. What do you think about John’s reaction? (Martí 2009, p. 47) Martí claims that a speaker who really used and understood names in the descriptivist manner would interpret John’s use of ‘Gödel’ as referring to the actual discoverer of the incompleteness theorem (i.e. Schmidt), and so feel that John was being extremely unfair. Whilst a causal-historical-type speaker would deem the utterance to be fair, or at least would understand John’s feelings. This general approach seems promising since it does indeed ask for a judgement that more closely involves the way the subject takes the name in question to be used. And, indeed, it appears that this question would show whether a strong form of descriptivism, according to which the reference of a name is always fixed by the same descriptive content, makes the wrong prediction. However, I don’t think that this particular question would be effective at showing either that the causal picture of reference is correct, or that more nuanced versions of descriptivism might be correct. For reasons that will become clear in chapters 3, 4, & 6, I take the view that only 38 2.2. Data on Reference uses of names, or, rather, speakers using names, refer. Furthermore, what a speaker refers to with a name is closely connected to what they are able to communicate on the occasion of use, given the way that hearers are likely to interpret them. This means that linguistic context (as well as other aspects of context) can affect reference (cf. Wettstein 1984, n. 31, p. 82). Thus, I would predict that John’s utterance ‘We have found out that Gödel was a thief and a liar’ is likely to incline an audience, in this case the experimental subjects, to take John to have referred to the theorem-usurping friend of Schmidt. This is also the prediction of the causal-historical picture, but I make it for different reasons. The prediction might also be made by a descriptivist who claimed that relevant descriptions culled from context can determine reference, or one who claimed, much like causal theorists often do (see chapter 3), that whilst pragmatic factors could affect what language users take the reference of a use of a name to be, the true reference is still determined by a description. A better form of questioning to follow Machery et al.’s Gödel/Schmidt scenario might attempt to discover how the subjects themselves would expect to use the name ‘Gödel’ in light of the scenario. Perhaps simply something like: Did Gödel discover the incompleteness of arithmetic? or Would you say that Gödel discovered the incompleteness of arithmetic? However, these questions may themselves run into problems because the name ‘Gödel’ has already been used to refer in a particular way in the description of the scenario, and this may lead the subject towards a particular judgement. There are also concerns, identified by Martí (n. 5, p. 47), about whether the scenario is to be interpreted as counterfactual, or pretend actual, and how much background knowledge the subjects have. All of these might affect the results and should be carefully considered in the design of similar experiments. There are, then, problems with the methodology of Machery et al. (2004). This might lead us to doubt the force of their claims that ordinary speaker judgements about reference accord with different theories of reference in different communities, and that, therefore, such judgements cannot be evidence for name using practice. The view of proper name reference I will put forward in chapter 6 maintains that the reference of an utterance of a name is highly context-sensitive. This means that, in order to appreciate either what will be referred to when one uses a name, or what a speaker is referring to with a name, one must be attuned to various aspects of the context. This, in 39 2.2. Data on Reference turn, means that in order to construct scenarios that accurately test speakers’ referential judgements, sufficient contextual information must be provided about the case. The precise details and quantity of context provided is likely to affect judgements. Therefore, I would argue, great care must be taken in designing rigorous studies of speakers’ referential judgements. We might also be concerned that Machery et al.’s study (two studies, in fact—they ran a pilot study with similar findings) is the only study to have investigated interor intra-cultural variation in judgements about the kinds of referential cases appealed to by philosophers of language. 2.2.6 Without Judgements It is possible that, even with the greatest care in experimental methodology, the judgements of ordinary speakers about constructed referential cases are simply hopelessly lacking in consistency. If this were the case, I would have to admit that the value of appealing to speaker judgements as data about reference, or of using referential cases and examples as a way of motivating or illustrating theoretical points, would be extremely limited. Whilst I will write this thesis assuming that there is at least some value in these activities11 it is instructive to consider the alternative. As it is, I am extremely suspicious of the conception of reference as an objective, metaphysical relation that connects names and objects in a string-like fashion. This conception is, I think, present in much of the work on names in the causal-historical tradition that followed Kripke (1981), and possibly much of the thought about proper names that preceded it. I prefer to think of referring as something that people (are able to) do with names (and other referential terms) in order to talk to other people about specific things in a dynamic and circumstance-specific way. The idea of the referent of a name, or even of a use or utterance of a name, that we can point to as its objective content becomes problematic on this way of thinking about things. It would be particularly problematic if there were no consensus available about what reasonable and competent users of the language took the reference of a (use of a) name to be. Lack of consistency of judgement amongst speakers about the kinds of referential cases discussed above would not, of course, entail that there would be eternal confusion about what is being referred to any time a name is used. However, it does call into question the plausibility of a theoretical device that can be useful to reference theorists. In cases in which there is disagreement between a speaker and their audience about what has been referred to with a name, something like a 11 Indeed, given my lack of capacity for collecting data in any kind of rigorous manner, I tend simply to provide cases that I find convincing, and that I have found others to be similarly convinced by. 40 2.2. Data on Reference partially-idealized reasonable, competent language-user might be appealed to to play the part of theoretical adjudicator (see, e.g. Cohen 1980; King 2014; Neale 2004). Deciding what is ideal or reasonable, if there is no consensus amongst real language-users, is a problem. In a situation in which there is no broad consensus on referential judgements or idealized language-use, there seem to be two things left that we can appeal to when it comes to proper name reference. One is what the speaker was trying to refer to when they used a name. In most cases, it seems that a speaker is trying to refer to something, even if they have only the haziest idea of what it is. The second is what the hearer took the speaker to have referred to. Again, this can be a very hazy notion, and need amount only to the hearer’s recognition that the speaker referred to something and their being willing to engage on that basis. The hearer can also fail to appreciate anything having been referred to, and request clarification or revision. These two aspects of reference can be thought of as what the speaker wants to refer to (we might also say ‘intends’ or ‘means to refer to’, though these terms come with more theoretical baggage), and what referential information actually gets communicated to an audience. When all goes right—and presumably much or most of the time this is the case—a speaker succeeds in communicating what they wanted to their audience. In such cases it is easy to say what has been referred to: it is what speaker and audience agree upon.12 In cases in which a speaker does not succeed in communicating about the thing they wanted to, we might generally want to ask, was this the fault of the speaker or the audience, or was it beyond the control of either? One interlocutor might be at fault because they failed to fully appreciate the context of utterance, which might include appreciating certain things about their interlocutor that they could be expected to appreciate, such as their communicative interests (see also, §2.4). However, without the notion of what it is reasonable to take the reference of a name to have been, given the context, it is difficult to motivate the idea that one party or the other is at fault. Given all this, I suggest that, if it were to turn out that there is widespread lack of consistency amongst speakers of a language in their judgements about the reference of proper names, as is suggested by Machery et al. (2004), and this lack of consistency extended beyond philosopher’s hypothetical thoughtexperiments into real life cases, then we would likely have to give up on the idea of objective reference in the case in which there is no agreement between speaker and hearer about what is being referred to. 12 Note that neither using a name referentially, nor accepting a speaker’s referential use of a name seem to entail being able to provide further information about the referent, so agreement upon reference may simply consist in deferring to one’s interlocutor. 41 2.3. Semantics and Pragmatics 2.3 Semantics and Pragmatics In this thesis, I will lean heavily on the distinction between semantics and pragmatics. In part II, I will argue against what I call semantic views of reference, and in part III, I argue in favour of a pragmatic view. The terms ‘semantics’ and ‘pragmatics’ are used in very different ways by different theorists, and whilst I recognize that at least some debate about the semantics/pragmatics divide may rest merely on terminological differences, I consider my usage to capture a significant distinction that can easily be recognized by most linguists and philosophers of language, whether or not they recognize it as the distinction between semantics and pragmatics, and whether or not they believe it to play the rôle that I claim that it does. My use of the term ‘pragmatic’ should be distinguished from a common use of that term, perhaps associable primarily now with minimal semantics (Borg 2004; Cappelen and Lepore 2005), whereby pragmatics is what goes on in linguistic practice beyond what words and sentences (literally) say or mean, and it is assumed that this is constituted or determined by truth-conditional (semantic) content that is independent of particular uses. It should be still further distinguished from the sense whereby it refers to those phenomena affected by a speaker’s intentions, and from the sense in which it is used by Neale (2004), such that pragmatics deals with the ways in which audiences interpret speakers utterances, but not with what determines what they express. I consider to be pragmatic all those features of language that go beyond the context-independent contribution that words make to an utterance, i.e. whatever is context-dependent. So, whatever the context-independent contribution that words make to an utterance is, according to my terminological dichotomy, semantic. It may be informative to consider how this dichotomy relates to, for example, the classical Kaplanian picture of indexicals. On this picture, part of the meaning of an indexical is a rule stating that certain features of the context affect the reference in a specified way, for example, that the speaker or the addressee is the referent. The reference of indexicals is, therefore, dependent on context. But, given my conception of pragmatics, it might be considered both pragmatic and semantic, because there is a context-insensitive rule that determines the reference given a specified aspect of context. Moreover, the conception of context on the Kaplanian picture is extremely limited: context is thought of just as a tuple of a few elements of immediate context, such as speaker, addressee, time, place, and elements constituting the tuple are specified by the semantic framework. In this thesis I employ a rich and varied conception of context as the circumstances of an utterance, constituted by any number of pragmatic factors and mechanisms. 42 2.3. Semantics and Pragmatics My conception of pragmatics might appear to have much in common with other conceptions, particularly the first conception, that I associated with minimal semantics, but it comes drastically apart from them when it is observed that the context-independent properties of words and sentences do not provide truth-conditions, but in fact, their truth-evaluable content varies from utterance to utterance, depending upon the occasion, and is not determined by speakers’ intentions. Such a position, most prominently propounded by Travis (e.g. 1978, 1985, 1996, 1997, 2006a), remains highly controversial. Travis’s general thesis is that predicates, as well as other types of sub-sentential expression, and therefore sentences, are occasion-sensitive— their truth-conditions depend upon pragmatic factors in such a way that they cannot be determinately encoded into a fixed lexical meaning, either as content or as Kaplanian-style character. Other philosophers and linguists have also proposed that semantics, or lexical meaning, (more or less) radically underdetermines the truth-evaluable content of utterances, for example Carston (2002, 2013); Recanati (2004, 2010); Sperber and Wilson (1986). However, these other radical contextualists differ significantly from Travis in various respects, most notably with respect to two claims made by the other contextualists: Firstly that, at least in many cases, words have a core or standing meaning that is capable of contributing truth-conditional content, but which can be modulated and altered by context to give a different content. Secondly, that there is a determinate, fully truth-conditional language of thought, or something similar, such that utterances of natural language expressions can, on particular occasions of utterance, be translated into expressions of that language that can only be satisfied by one particular way for the world to be. Travis rejects both of these claims (see §6.7, particularly with regard to the second claim). The phenomenon I have identified as semantics may fall into the scope of both compositional semantics and lexical semantics, as they are traditionally conceived. Given the Travis-influenced picture of the rôle of pragmatics mentioned in the previous paragraph, lexical semantics, dealing as it does with word meaning, is tasked not with providing truth-conditional content, but with specifying what kind of context-insensitive content words need to provide to a compositional structure (whose construction procedures are presumably the remit of compositional semantics), such that pragmatics can act upon it to give truth-evaluable content. In this thesis, I will provide very little by way of a positive account of semantics. In chapter 6, I will argue that proper name reference is a pragmatic, rather than a semantic phenomenon, and, as the thesis is primarily concerned with reference, I will largely leave the question of the semantics of names open. However, in §8.2, I will briefly consider what the semantics of names could and could not be like in order 43 2.3. Semantics and Pragmatics to be consistent with my account of their reference, suggesting avenues for further research. With regard to this thesis, the position of Travis that I am most interested in, mentioned above, is that predicates and sentences, and various other types of expression, are occasion-sensitive. That is, they are ubiquitously contextsensitive in such a way that the meanings of words and sentences—their semantics, in the sense I have characterized—radically under-determines the truth-evaluable content of (or what is said or expressed by) utterances of those words and sentences, and that content is determined by pragmatic mechanisms in a way dictated by the occasion, rather than by semantics. I will not argue at any length for the occasion sensitivity of predicates and sentences, beyond presenting the bare bones of Travis’s own arguments. Whilst my arguments for the pragmatic view of reference do not assume the correctness of Travis’s general position, and I motivate my position independently, it will benefit the reader to know where my broader philosophical and linguistic commitments lie. The rest of this section will be devoted to exposition and expository discussion of Travis’s thesis regarding occasion-sensitivity. 2.3.1 Occasion Sensitivity and Travis Cases Travis maintains that words can only be said to express anything truthevaluable (or in some way evaluable for correctness or appropriateness) on an occasion of utterance, upon which they can be correctly understood to be saying that things are a certain way. Thus, he treats utterances as the primary linguistic bearers of truth-values—as has presumably become clear, I follow him in this respect—but he is reticent to suggest anything resembling contextual parameters or criteria that fix what is expressed by an utterance, or what an utterance’s truth-value is. The certain ways that particular words or sentences can describe things as being, given certain understandings of them, are not, according to Travis, the kinds of things that can be determinately individuated in discrete ways, as it is often thought propositions can be (Travis 2008, p. 7). Nor is there anything like a determinate set of factors that determine how one ought to understand an utterance on an occasion, or how things need to be for an utterance to be true given how it is to be understood. Travis offers some insight into when utterances are true, making use of the locution ‘counts as being’. In response to the question of what sometimes makes it true to say of given leaves that they are green, and sometimes that they are not, he says: What could make the given words ‘The leaves are green’ true, other than the presumed ‘fact that the leaves are green’, is the fact that the leaves 44 2.3. Semantics and Pragmatics counted as green on the occasion of that speaking. Since what sometimes counts as green may sometimes not, that may still be something to make other [token] words ‘The leaves are green’ false, namely, that on the occasion of their speaking, those leaves (at that time) did not count as green. (Travis 1997, p. 125) This locution does not itself tell us much more about what it is about the contextual circumstances of an occasion of utterance that determine whether or not the utterance is true, since we are left to ask what might make a leaf count as green. Travis will again respond, correctly, I think, that there are no determinate kinds of general criteria to be offered that would answer such a question. However, examples Travis provides throughout his work do offer insight into the kinds of pragmatic factors that, on particular occasions, appear to be relevant to whether something counts as being so described. Travis (1996) discusses several cases in which the purposes or the interests of interlocutors appear to affect whether things count as being correctly described. In other cases it appears that there is a certain kind of charity at play in finding the ‘truth near to mind that could be so expressed’ (Travis 2008, p. 9). This positive view of truth and utterance content represents one aspect of occasion-sensitivity. It arises in parallel to, and is made more perspicuous by, another aspect, which is Travis’s negative position about what the meaning of words and sentence cannot be. This aspect, and to some extent the positive aspect, is motivated, at least in part, by consideration of examples of discourse, known as Travis cases, that demonstrate the complexity and volatility of truthvalue judgements, and the pervasive rôle of context in such judgements. Travis cases, and other arguments for occasion-sensitivity, can be found throughout Travis’s work in various forms.13 I will give three examples of Travis cases, before going on to address what it is that they purport to show, and, in doing so, present a characterization of them. A well known case from Travis (1996) involves whether a ball is round: A game of squash is in progress, and two groups of people are watching from the viewing gallery. One of the groups contains a viewer who has never come across squash before. He asks one of his group what shape the ball is, since it is moving around the court too fast to see, and she replies ‘The ball is round’, and, in so doing, speaks a truth. The other group viewing the game are engineers interested in squash ball dynamics, who are monitoring the shape of the ball during play. Just as the ball hits the wall, one of the engineers asks what shape the ball is, and her colleague replies ‘The ball 13 Travis (1985, pp. 199–203); Travis (1989, pp. 18–20) Travis (1996, p. 97); Travis (1997, pp. 111– 113); Travis (2000, pp. 3–4); Travis (2006a, pp. 152–153); Travis (2006b, pp. 21–24); and Travis (2008, p. 2). 45 2.3. Semantics and Pragmatics is round’, coincidentally, at the very same time as the member of the other group. However, this time the reply is false, because at that moment, the ball has been squashed into a very non-round shape. It is plausible to suppose the meaning of the sentence ‘The ball is round’ is not different in its two utterances—in both cases the same ball is being referred to, and there seems to be no variation in the meaning of the predicate being applied to it. There is not even a variation in the standards of roundness being applied to the ball. But despite the constancy of the meanings of the utterances, they cannot have the same truth-conditions, since they have different truth-values, despite being spoken at the same time with reference to the same object. Another example of a Travis case comes from Travis (1985). Here, Travis presents two separate but parallel stories, that are equivalent to the two separate groups of interlocutors in the squash ball case. Story I: Smith is quite proud of the results of the rigorous diet he has followed. He has lost easily 15 kg. Stepping on the scales one morning, he notes with satisfaction that they register a thick hair or two below 80 kilos. At the office, he proudly announces, ‘I now weigh 80 kilos’ But the tiresome Melvin replies, ‘What! In that heavy tweed suit? Not very likely’ and, pulling a bathroom scale out of his bottom desk drawer and pushing Smith on to it, notes with satisfaction, ‘Look. 83 and a bit.’ (For good measure, let us suppose Smith not yet to have taken off his overcoat, so that the scale actually reads 86.) Of course, we would say, what Melvin has demonstrated does not count against what Smith said. (Contrasting) Story II: Smith, dressed in the last way, is about to step into a crowded elevator. ‘Wait a minute.’, someone says, ‘This elevator is really very delicate. We can only take 80 more kilos.’ ‘Coincidentally, that’s exactly what I weigh.’, replies Smith. In he steps, and down they plummet. So it appears that what Smith said this time is false. (ibid., pp. 199–200) In this case, nothing changes about the way Smith is, but the situations he finds himself in shift the truth of whether or not he weighs 80kg. A third example, in which Travis contrasts different ways of understanding a situation, rather than comparing different utterances, comes from Travis (2011): Returning from the market, Sid crosses the white rug, heading for the kitchen. As he does so, the bottom of his recycled-paper bag breaks, and the kidneys he has bought for the mixed grill fall in their butcher paper to the rug. Is there, now, red meat on the white rug? One question is whether kidneys count as meat. Though one would not offer them to vegetarians, their usual rubric (at any rate, in Sid’s parts) is offal. Meat, on one understanding of the term, is flesh. And flesh, as one typically 46 2.3. Semantics and Pragmatics would understand that, is muscle. Thinking in this way, one should say that, however unenviable the condition of the rug, it does not, at any rate, have meat on it. On the other hand, what Sid dropped is, in fact, the meat course for Sunday brunch. So meat can be understood to be something which those kidneys are. Then there is the matter of being on the rug. The meat is still wrapped in its butcher paper. So, on one way of understanding things, it is (so far) only the paper that is on the rug. On the other hand, if Sid, now missing the kidneys, asks where they are, telling him, ‘On the rug’, may be saying no less than where they are. (Travis 2011, pp. 243–244) Travis’s Target(s) What Travis aims to show in investigating Travis cases is expressed, in simple terms, in the following quotation: . . . it would seem that the meanings of sentences are not to be explained in terms of truth-conditions, sentences bearing the former but not the latter, and mutatis mutandis for sentence parts. (Travis 1985, p. 188) The alledged gap between meaning and truth-conditions is the key issue. But exactly what this amounts to is unclear, since ‘meaning’ is used, both in ordinary language and by philosophers and linguists, in a variety of ways. It seems that the best way to understand what Travis means by ‘meaning’ is that which is held constant between uses of sentences and sub-sentential expressions. Davies (2014) has this to say about Travis on meaning: Meaning here is defined so that it is an open question what the relation between meaning and satisfaction and truth and reference may be— something that could be investigated by seeing what else stays constant across a word’s uses when its meaning, in this sense, is held constant. This is something upon which Travis (1981b, p. 3, pp. 9-10) lays a particularly heavy emphasis. (Davies 2014, p. 502) Davies discusses the target of Travis cases in the context of arguing that many attempts at addressing Travis’s work on language are off target. He claims that many of Travis’s alleged opponents take Travis to be attacking the compositionality of sentence meaning, i.e. that the meaning of a sentence is a function of the meanings of its parts. In fact, Travis leaves such a compositionality principle untouched, and instead is attacking positions that hold that meanings (either sentential or sub-sentential) determine (in a systematic way) the content of utterances, or, put in other ways, that meanings are either contents or characters,14 or that meanings are truth-conditions. This 14 In the Kaplanian senses: content being a complete (though not necessarily wholly determinate), truth-evaluable proposition or thought (at least for assertoric utterances), and character 47 2.3. Semantics and Pragmatics confusion stems from an ambiguity between two semantic projects that are both thought of as truth-conditional. There is a ‘compositional project of identifying properties that words may have which combine to form the truthconditions of the sentences the words can be used to construct [. . . and there is] a reductive project of showing that the meanings of linguistic expressions are either contents or characters’ (Davies 2014, p. 501). Compositional projects may look as if they’re providing truth-conditions for sentences by providing biconditionals of the form pS is true iff Pq, but, according to Travis, they only do so in a problematic way if what occurs on the right-hand-side is determinate, or should be, according to the project. If the right-hand-side of a putative biconditional does not aim to be determinate, then it does not provide truth-conditions in the sense we are interested in.15 Projects of the reductive persuasion—what I called ‘lexical semantics’ above—may often not explicitly provide biconditionals with a determinate right-hand-side, but rather, suppose that such conditions could, in principle, be provided, i.e. they suggest that, given a sentence, or a sentence and a determinate set of contextual parameters, there is a determinate way the world must be to make the sentence true. Meanings as Contents Davies separates out two forms of reductive lexical semantics. One aims to show that meanings are contents, and the other aims to show that they are characters. The two projects require somewhat different responses. Most Travis cases appear to target the view that meanings are contents, and are used as part of arguments with the following form: 1. The meaning of an expression is its content only if: the meaning changes if and only if the content changes. 2. The content of an expression can change even though the meaning does not change. 3. The meaning of an expression is not its content (from 1., 2., and modus tollens). (ibid., p. 506) Travis cases are intended to demonstrate premise 2. They do this by providing convincing examples in which multiple utterances of a sentence vary in being a rule of use modelled as a relation between words and contents: an n-place function (or, possibly, relation) which, given n determinate contextual parameters as arguments, will return a content. 15 This is not to say that all compositional projects are wholly innocent in truth-conditional terms. 48 2.3. Semantics and Pragmatics the truth-evaluable content they express—usually evidenced by varying truth-values in different situations—whilst their meaning does not appear to vary. Consider Travis’s example, borrowed from Grice, of a table covered in butter (Travis 1991, pp. 67–69). The sentence ‘The table is covered in butter’ can, without any variation of meaning (given a certain understanding of ‘meaning’—see the following paragraph for discussion), be satisfied in at least two different ways: the table could be spread with a layer of butter, or there could be a layer of foil-covered butter pats on the table top. One can imagine a scenario in which an utterance of ‘Is the table covered in butter?’ can be answered affirmatively if and only if a particular one of these satisfactionconditions is met. Imagine a disreputable butter sculptor who cheats at her work by covering non-butter items, rather than sculpting directly from blocks of butter. She might leave her assistant to smother a table whilst she works on something more complicated. If she says to the assistant ‘Is the table covered in butter?’, the assistant cannot correctly answer ‘Yes’ if the otherwise bare table is scattered liberally with packets of Anchor. Judgements both about the truth-value of an utterance in a particular situation, and about whether or not the meaning of the words has shifted, is left to the intuitions of speakers of the language in question. Such speakers should be expected to be reasonable judges of the truth-values of utterances in their language about fairly mundane occurrences. If they were not, it would seem that there was a serious problem with the language. Of course, there may be disagreement over particular cases, but the idea behind an effective Travis case is that the relevant judgements are plausible to most languageusers.16 Intuitions about whether the meanings of the words of the sentence in question shift across the utterances are less essentially reliable, especially since the word ‘meaning’ itself is used in so many different ways, and we have not settled on anything approaching a technical definition. However, there is good reason to accept that something fundamental to the semantics of the sentence and its sub-sentential expressions is being held constant across the relevant utterances—if we were to posit that words such as ‘green’ were ambiguous between subtly distinct meanings, each of which carried its own, unchanging truth-conditions, then the apparent ease with which new Travis cases are generated would suggest that there might be indefinitely many of these semantic entries for many, if not all, words. This result seems to be highly undesirable for a variety of reasons, most notably because it is very hard to see how humans could ever learn a language with indefinitely many semantic entries for each sub-sentential expression. 16 See § 2.2 for discussion of the rôle of judgements and intuitions in philosophy of language. 49 2.3. Semantics and Pragmatics Meanings as Characters As Davies (2014) observes, the project of showing that meanings are contents cannot, in reality, be wholly separate from that of showing that they are characters, since at least some lexical items are undeniably indexical in nature, and, in light of Travis cases like those just discussed that show that meanings cannot be contents, theorists may be inclined to claim that they are characters. This is a natural thought—if Travis has shown that the truth-conditions of many more expressions than we previously thought are context-sensitive, then we should adapt our treatment of the meanings of these expressions so that, together with context, they provide determinate truth-conditions. Kaplan’s (1989b) treatment of indexicals does just this, so seems a likely choice. It is effective against the type of Travis cases already considered as long as the contextual factors that seem to be relevant in shifting truthconditions in any particular case are included in the parameters taken as arguments by the relevant characters. Including parameters in this way may be problematic since it requires identifying the right contextual factors with the right degree of specificity without being ad hoc. However, a bigger problem is the potential generation of further Travis cases that turn on contextual factors not considered by one’s parameters. Travis does not often explicitly address the meaning-is-character project in his work (cf. Travis 1978, pp. 412–413), but Davies provides a schema for systematically developing an argument against the project, along the same lines as his schema for arguments against meanings as contents: 1. Assumption: The meaning of an expression e is a character only if there is a function F that provides a model of e such that: a) F takes the parameters of the character as arguments and returns the contents of e as values. b) F stays constant if and only if the meaning of e stays constant. 2. Assumption: For any expression e, let F be a function from a set of parameters to the content of e that is a model for a character proposed to be the meaning of e. The meaning of e and the parameters in the argument of F can be held constant whilst the content of e is varied by altering contextual factors that are not amongst the parameters in the argument of F. 3. Conclusion: The meaning of an expression e cannot be modelled as a function from parameters to contents, and thus cannot be a character. (Davies 2014, p. 9, substantially revised) 50 2.3. Semantics and Pragmatics The idea of this argument form is that Travis cases can be found showing that no semantic analysis can be provided for an expression such that it meets conditions a) and b) for being a character. Such cases would require taking a proposed analysis, which provided determinate parameters, and holding those parameters constant whilst varying the content of the expression by shifting contextual factors not included amongst the parameters. If those factors are added to the parameters, we then repeat the process by finding other contextual factors and showing that the content can be varied by varying those whilst holding the parameters fixed. Of course, it cannot be shown a priori that this can be done for every proposed character. This is not a general argument that shows that every set of parameters fails. Given enough imagination and dedication, the process of proposing a character for an expression, and showing that it fails, can run and run. However, the plausibility of the characters will presumably decrease as they become ever more complex, and the plausibility of the hypothesis that no character is satisfactory will increase as each is shown to fail. 2.3.2 Occasion Sensitivity and Reference Given the preceding elucidation of Travis’s position, one might reasonably wonder how occasion sensitivity will be relevant to the reference of proper names. As I have already mentioned, my major thesis about proper name reference is that it is a pragmatic, rather than a semantic phenomenon, and this position was initially, at least partially, motivated by Travis’s position on the occasion sensitivity of other types of expression. As I have discussed, occasion sensitivity (for predicates) is, or has the result that, the contribution that is made by predicates to the truth or falsity of an utterance is radically under-determined by their semantic content, where semantic content is taken to be something determinate and constant between uses. Being able to tell whether or not a predicate is being correctly or appropriately applied on an occasion is something that requires domaingeneral knowledge about the ways the world is on the occasion, not merely knowledge of the semantics of the predicate. Thus, what is said with an utterance—the content of an utterance—is not dependent upon the semantic meaning of the utterance. That the semantics of predicates and sentences under-determine utterance content might lead one to think that semantics per se does not aim at providing determinate utterance content. Reference, however, is a phenomenon that has an impact on the truth-evaluable content of an utterance—in order to analyze what is said by an utterance containing a proper name, and whether the utterance (or what is said) is true, we must 51 2.4. Communication know (in some sense) to what the name refers (at least, in most circumstances). Indeed, a name can be said to refer to an object only if the truth of the utterance containing it turns on the way that object is (again, in most circumstances). Given these two claims—that semantics is not utterance content-oriented, and that reference is utterance content-contributing—there is reason to think that reference is not a semantic phenomenon. If reference belongs in the domain of truth-evaluable content, and truth-evaluable content is a pragmatic phenomenon, reference is pragmatic. Of course, an argument of this sort is not going to convince anyone who is not already inclined to take what Travis calls ‘the pragmatic view’ (Travis 1997, p. 111-3), but, again, it will be helpful to the reader to appreciate my own initial motivations for developing the pragmatic view of proper name reference. In chapters 3–6, I will try to motivate that view independently of the broader Travisian framework. 2.4 Communication Whether or not, and how, communication is relevant to what words mean, whether and how it is relevant to when sentences, utterances, or thoughts are true, and whether and how it is relevant to what proper names refer to, are all controversial questions in the philosophy of language. I will not attempt to explore them in great depth here, but since I make certain assumptions about the rôle of communication, it will do well at least to stake my ground. It seems to be a minimal constraint upon a theory of natural language and meaning that it does not exclude the possibility of communication, even if it does not explain it. Moreover, since communication is, at the very least, something that humans use language for a great deal—to my mind, its primary use17 —it must be considered a boon if a theory of language and meaning makes successful communication plausible. Indeed, we should look very unfavourably upon a theory that makes communication implausible.18 Therefore, an account of proper name reference should not make communication impossible, and, if it has nothing to say with regard to communication, we might reasonably expect 17 Chomsky has, famously, doubted such a claim. He observes: Language is used in many different ways. Language can be used to transmit information, but it also serves many other purposes: to establish relations among people, to express or clarify thought, for play, for creative mental activity, to gain understanding, and so on. In my opinion, there is no reason to accord privileged status to one or the other of these modes. (Chomsky 1979) However, it is unclear that any of his examples are genuine challenges to the significance of communication. All of the examples Chomsky provides seem to require information transfer, at least at some stage, and I will claim below that communication amounts to nothing more than information transfer. 18 Thanks to Peter Sutton for highlighting these points to me. 52 2.4. Communication to be able to see how it makes communication possible, and why it is that it does not need to address communication. In this section, I will highlight why it is that I think that communication is relevant to language and meaning, and how it relates to proper name reference. First, a word about what I take communication to be. The most plausible, minimal account of communication that I am aware of is that in the vein of Shannon-influenced models,19 whereby communication is simply the transfer of information. Information here need not be in the form of thoughts or propositions, in the sense of fully determinate, truth-evaluable entities, but could simply be an ‘indicator of probable states of affairs’ (Sutton 2013, p. 126), though not necessarily correct indicators. This idea of communication and information is compatible with the thought that words on their own communicate relatively little, but that uses of words in context, together with the application of various kinds of awareness about the circumstances of those uses, allows for a great deal more information to be transferred. Speakers and audiences are able to make use both of the information that might be carried by words independently of any particular circumstance of utterance—what I characterized in the previous section as their (semantic) meaning—but also the information that they can carry upon particular occasions of utterance, being aware that those words, having been spoken by the speaker, who has particular purposes, provide reasons to believe certain things. All this is of relatively little importance for the bulk of what I will have to say in this thesis, except to make the point that what I take to be communicated by a referential use of a name is just that information (which need not be true, or correct, or intended) that is transferred to the audience given a speaker’s use of the name. Communication and Language 2.4.1 Language is used by people to do certain things. It is a tool that enables people to do those things, and it can be used to do those things because language-users are such that they can make it do them. Language is able to represent, or carry information about, things as being a certain way because it can provide reasons to suppose that things are that way: [W]hat one thus says as to how things are, so when one would have spoken truly, is determined by what it is reasonable to expect of that particular describing—reasonable to hold one thus responsible for—given the circumstances of its giving; what expectations would, in those circumstances, reasonably be aroused. (Travis 2006b, p. 31) 19 See Sutton (2013) for arguments in favour of such models. 53 2.4. Communication What it is that we use our words to do, therefore, is not a question (merely) of what we intend them to do, but of what we actually do do with them. And that is inherently connected with, and constrained by, the effect they can have within particular linguistic communities, broadly interpreted, given the practices of those communities. And that brings in what is communicated: the information carried by our words, giving other language-users reasons to think that things are a particular way. It will not do to think that only one’s own purposes or intentions are relevant to what one’s utterance expresses, because that utterance will carry different information under different circumstances, depending, at least, upon the practices of the communities in which it is uttered, and carrying that information is what enables, or constitutes, what an utterance expresses. Moreover, what is key to the idea of an utterance carrying information at all is that that information be recognizable. There is no sense in a signal—some words—carrying information, if the information is not, in principle, recognizable by someone who is able to track what it would be reasonable to use those words for. But recognizing what has been said with an utterance, or what information it carries about the way things might be, is the basis of communication. Thus, if I want to express a thought with an utterance, that utterance must make that thought recognizable—communicable—in the particular circumstances of utterance, that is, recognizable by a reasonable member of the relevant linguistic community, who will have particular expectations, given those circumstances. So, what one communicates with an utterance is directly relevant to what one expresses with an utterance. And what one can communicate is determined by the circumstances, and a reasonable audience’s expectations in those circumstances. 2.4.2 Communication and Proper Names Given all this, then, when a proper name is used to refer, what it refers to must be affected by what it can communicate—what, all things being equal, a member of the relevant linguistic community—which might be specified in a way specific to a particular interaction—would take them to refer to with the name, given the circumstances (which must include the linguistic context). Of course, an audience must be in the right kind of position to appreciate the circumstances in the relevant way. If they are not concentrating, or for some other reason fail to understand what the speaker has said, despite the speaker acting exactly as appropriate in the circumstances, then the audience is not in a position to recognize what they should. However, it is also clear that particular facts about the relationship between the speaker and their audience, or particular facts about the audience that the speaker is in a position to 54 2.4. Communication appreciate, can affect the circumstances in such a way that it will not do merely to substitute a generic rational speaker for the audience when describing the significance of what should be understood by a referential utterance of a name, and how that effects its reference. These kinds of considerations might influence the way in which it is necessary to specify the relevant community. Cohen (1980), Neale (2004), and King (2014) all appeal to the idea that some kind of idealized rational language-user is what is needed to judge (or is in someway relevant to) what a speaker has referred to with a use of a name or other referring term. What would that person take the speaker to have referred to in these particular circumstances? In order to capture the fact that speakers frequently use names or other words in a way that makes use of shared knowledge or past mutual experiences that they may have with the audience, we might try to grant this kind of knowledge to the idealized language-user too: Neale (2004) proposes a ‘a rational, reasonably well-informed interpreter in [the audience]’s shoes’ (p. 80). However, it is hard to see how it would be possible to capture all that would be necessary in an expression such as ‘in the audience’s shoes’ without simply meaning ‘has all the relevant knowledge and awareness required to get the right answer’. I agree entirely that it is important to appeal to the notion of what would be communicated by a use of a proper name, if all were to go well, or something similar, but I do not think that appeal either directly to the actual audience, or to an idealized audience is necessary, given the right understanding of communication and occasion of utterance, though, as I have suggested, we might appeal to a relevant linguistic community. If the circumstances of an occasion of utterance are understood as including particulars of the speaker and hearer that are recognized by or recognizable to each other, or if the relevance of a specified community is sensitive to these issues, then there is a way (or perhaps a gradation of ways) that an utterance of a proper name should be understood, given those circumstances. This idea already presupposes that the interlocutors are rational in the right way, and that they are engaged with the circumstances in the right way. 55 Part II Semantic Views of Reference 56 Introduction to Semantic Views T reatments of proper names in the philosophical literature tend to characterize either their semantics, or their referential mechanisms— if indeed the treatment deems them to be referential—or both. Most accounts of proper name reference in the last fifty or more years state that it is determined by semantics. Thus their characterizations of referential mechanisms can be considered as dealing in meta-semantics. Such accounts hold that there is something in the semantic contribution a particular name makes to a sentence or utterance containing it that either wholly or partially determines, fixes, or specifies to what that name, or use of the name, refers. They frequently, therefore, provide a story both about the semantics of names, and about their referential mechanisms. It is characteristic of these semantic accounts that they attempt to specify, in some way, necessary and sufficient conditions for proper name reference, or at least imply that such conditions are, in principle, available. I will call views that some semantic account of proper name reference is correct semantic views. There are five broad groups of accounts of proper names in the contemporary literature. Not all are mutually incompatible, and each is subject to substantial internal variation. Some provide characterizations of both the semantics and the referential mechanisms of names, and some provide characterizations of only one. The five groups of accounts are causal-homophonism; indexicalism; classical descriptivism; predicativism; and variablism. Causal-homophonism20 is a group of accounts of both the semantics and meta-semantics of names. It holds that names refer to their referents tout court 20 See, for example, Corazza (2004); Jeshion (2015); Justice (2001); Kaplan (1990); Kripke (1981); Mercier (1998); Sainsbury (2005, 2015); Salmon (2005); Soames (2002). 57 2.4. Communication in virtue of causal chains (or trees) of uses from the name to the referent, and the semantics of a name is just constituted by its reference. Names are individuated according to reference or baptismal-origin such that no name can have more than one referent at a time (within one name-using community). Indexicalism21 is primarily a position on the semantics of proper names, however, that semantic position entails a particular kind of meta-semantics. Indexicalism holds that names behave somewhat like indexicals, in that they do not refer tout court, but have a Kaplanian character (Kaplan 1989b) which specifies that certain determinate contextual parameters determine the reference of a use of a name in a particular way. Classical Descriptivism22 can be either an account just of how the reference of a name is determined, or an account both of its reference and its semantics. It holds that names (or name-using practices) are associated in some way with a definite description, or a cluster or collection of descriptions, the satisfier of which is the referent of the name. Predicativism23 is a group of accounts just of the semantics of names. It holds that proper names have the same basic semantic type as predicates, rather than being basically referring terms. (In the jargon of traditional formal semantics, they are terms of type he, ti rather than type e.) When names are used to refer, they are proceeded by a determiner, almost always unpronounced in English. On many views the determiner is a definite article, forming a definite description with the name predicate, on other views it is a demonstrative such as ‘that’, forming a complex demonstrative. Variablism24 is also a group of accounts of name semantics. It holds that names have the semantic form of variables, as has been proposed for certain occurrences of pronouns, such that referential uses of names behave like free variables. As stated, of these five groups of account, only the first two—causalhomophonism, and indexicalism—are essentially committed to positions on reference determination for proper names, and both of these positions fall within the semantic view: causal-homophonism takes reference to be constitutive of a name’s semantics; and indexicalism, although it takes the reference of a name to be partially determined by contextual factors, states that the manner in which this occurs is stipulated by the semantics of the name. Predicativism is not, in and of itself, committed to any particular view on how 21 See, for example, Burks (1951); Cohen (1980); Pelczar (2001); Pelczar and Rainsbury (1998); Rami (2014, 2015); Recanati (1993); Tiedke (2011); Zimmermann and Lerner (1991). 22 See, for example, Ahmed (2007); Jackson (1998); Searle (1958). 23 See, for example, Bach (1987, 2002); Burge (1973); Elbourne (2005); Fara (2015); Geurts (1997); Gray (2014, forthcoming); Katz (2001); Matushansky (2008); Sawyer (2010). 24 See, for example, Cumming (2008); Dever (1998). 58 2.4. Communication the reference of names is determined, since this would depend on how the reference of the complex referring terms claimed to explain referential uses of names is determined. This holds similarly for variablism: if referential uses of names are free variables, then the way the variable assignment is determined will determine the reference of the name.25 Classical descriptivism is a broad church, and so whether or not it falls within the semantic view will depend both on how names are taken to interact with their reference-fixing descriptions, and on what the semantics of those descriptions is taken to be. I will not discuss classical descriptivism at any length in this thesis, as it is has not played any major part in the literature on names in the last thirty years.26 There is a comparison to be drawn between the two types of account constituting the semantic view of proper name reference, causal-homophonism and indexicalism, and the two main positions opposing the (pragmatic explanation of the) occasion sensitivity of predicates, discussed in Travis (1997) and Davies (2014). Homophonist views about names are analogous to ambiguity views about predicates, which attempt to account for Travis cases (and polysemy) by claiming that each predicate can refer to multiple distinct properties, e.g. ❣r❡❡♥1 , . . . , ❣r❡❡♥367 , . . . , ❣r❡❡♥n (Bosch 2007). Indexicalist views about names (and other ‘moderate’ contextualist views about name reference, such as might be added to predicativist or variablist accounts) are analogous to contextualist views about predicates that suggest that predicates as well as indexicals have complex Kaplanian characters (or similar parameterbased structures) such that, given enough contextual parameters, the right content will be forthcoming for every utterance (Hansen 2011; Rothschild and Segal 2009; Szabo 2001). In the following two chapters I discuss these two versions of the semantic view, and offer arguments against each. I begin with causal-homophonism, which I believe to be the more egregious of the two, since it treats names, for the most part, as terms which refer essentially to particular objects, and maintains that they are completely context-insensitive. However, it is also the harder view to argue against because it shares fewer of my background assumptions about what constitutes an adequate and plausible account of proper name reference. There is another group of views about proper name reference that are not, or need not be, semantic views, but I include discussion of all of them 25 In fact, Dever (1998) does have a view on how reference is determined, though this is secondary to his view about the semantics of names. 26 A significant anomaly to this claim is Jackson (1998). However, this paper did not spark any particular resurgence of descriptivism, and I understand Jackson himself subsequently abandoned the view. There may also be psychological evidence that suggests that classical descriptivism is mistaken. See Garcia Ramirez (2011) for a survey. 59 2.4. Communication amongst the genuinely semantic views because they also do not treat names as context-sensitive in the way that the pragmatic view I propose in part III does. These are intention-based or intentionalist views that maintain that a speaker’s intention to refer to a particular object with a use of a name determines the reference of the name, or contributes to that determination in a significant way. I discuss two types of intention-based view in chapter 5. Firstly, constraint-based intentionalism, which could be considered a semantic view, and maintains that the referent of a proper name upon a particular use is whatever the speaker intends it to be, but within a constrained domain of objects, for example, bearers of the name used. Intentions to refer to other objects do not produce reference. Secondly, a variety of neo-Gricean view that maintains that the reference of a use of a name is determined by a particular kind of structured intention to communicate about a particular object. In this case it is the intention itself that is constrained—one can only intend to refer to certain objects with each name in particular situations. I will argue against the first kind of view on the basis that it fails to result in a convincing account of name reference. However, I will argue against the second kind of view much more generally, by claiming that the kind of general view that would be required to accommodate such a view of reference is incompatible with certain empirical psychological data. 60 Chapter 3 Causal-Homophonism ‘We name things and then we can talk about them: can refer to them in talk.’—As if what we did next were given with the mere act of naming. As if there were only one thing called ‘talking about things’ L. Wittgenstein, Philosophical Investigations T wo common views on proper names after Kripke (1981) have been that, (i) since they have no descriptive meaning which determines their reference, the semantic content or contribution of a name just is its referent (this is a consequence of the assumption that semantics is truthconditional); and (ii) a name has a constant semantic content, i.e. it has the same meaning (its referent) independently of any particular use. These two views entail homophonism (or the ambiguity thesis), which is the view that proper names are properly individuated in such a way that they have only one (semantic) referent (at a time: cf. Sainsbury 2015). This means that, despite the apparent ubiquity of multiple objects having the same name, strictly speaking, these ‘shared’ names are actually multiple homophonous names on a par with ‘bank’/‘bank’. Thus, John Smith the Tadcaster brewer, and John Smith the late Labour Party leader did not share a name, but had distinct names that are spelt the same and sound the same. This view is made somewhat more plausible by Kaplan (1990), who distinguishes between generic names, which are something like name templates—the kinds of things found in books of baby names—and have no referential content, and specific names (called ‘common-currency names’ by Kaplan), which are the 61 things we actually use to refer to objects, and which cannot be shared. Kaplan has the following to say about the distinction, upon introducing it: There is the generic name “David”, and then there is my name “David”, there is David Lewis’ name “David”, and so on. These are all three distinct words. The latter two [being specific names] have—and here I speak carefully—a semantic function: They name someone. The first, the generic name doesn’t name anyone (doesn’t name anyone, perhaps it names or is an unnatural kind). Furthermore, it doesn’t pretend to name anyone (as certain empty [specific] names do). (Kaplan 1990, p. 111) Thus, the two John Smiths have something in common: their specific names, though distinct, both stand in the same relation to one particular generic name. The relationship between generic and specific names is somewhat unclear. It appears to be something like instantiation, though Sainsbury (2015) refers to it as exemplification (p. 196). Kaplan compares the relation to that of speciesindividual (Kaplan 1990, p.208). Causal-homophonists combine the homophonist individuation of names with various elaborations of the causal picture of reference put forward by Kripke (1981). Homophonists developed Kripke’s picture into a theory of reference and/or names, maintaining that the tout court reference of a name is determined and maintained by a causal chain connecting uses of the name back to an originating baptismal act that named the referent and brought the name into existence. Uses of a name are uses of a particular specific name, with its particular referent, in virtue of the causal connection that that use has to previous uses, tracing it back to the particular baptism. The causal connections are usually explained in terms of speakers intending to use the name in the same way, or with the same reference, as some previous speaker’s use. I will offer four arguments against the causal-homophonist picture in §§3.1– 3.4. They are as follows: The argument from shared names (§3.1): Homophonism explicitly endorses the claim that there are no shared names, merely multiple homophonous/homonymous names. This is implausible given the ordinary language conception of a name. Whilst defenders of the view claim that this argument has no teeth, I believe that there is something to it, and causalhomophonism’s counter-intuitive claims should at least count as a mark against it. The argument from communication (§3.2): Causal-homophonists and their ilk assume that there are reference-fixing mechanisms (i.e. causal chains) at play in our use of proper names that are independent of any awareness particular speakers and hearers have of them. This means that, in order 62 to successfully communicate using names, there must generally be other mechanisms in play that facilitate awareness of what’s being referred to, particularly where shared names, etc. are at play. The homophonist does not deny this, but dismisses them as presemantics, not actually involved in truth-conditional reference determination, and liable to come apart from the metaphysical, truth-conditional-level reference. I argue that this is seriously detrimental to an account of name reference because it divorces reference from communication in a way that ignores how speakers actually use names. The argument from family names (§3.3): Family names (or surnames), present a problem for causal-homophonists, both as they are used to refer to families, and, in particular, as they are used to refer to individual bearers of family names. As far as I’m aware, Sainsbury (2015) presents the only significant discussion of family names amongst homophonists, and his discussion is very brief. Family names appear, almost essentially, to be shared amongst members of a family. Sainsbury embraces this intuition, claiming that specific family names refer to families as the specific names of individuals refer to their bearers, and that it is these specific family names that combine with individuals’ specific first names to create the two-part names such as ‘Katharine Harris’ typical in Western, English-speaking cultures, as well various others. This presents two problems for Sainsbury: firstly, how to individuate families in the right way, particularly nuclear families; and, secondly, how to account for the very common practice of using family names to refer to individual people. I consider various ways in which homophonists might alter Sainsbury’s account in order to account for this practice. I conclude that each either undermines causal-homophonism by introducing contextual elements to reference determination, or is implausible because it cannot account for the familial nature of family names. The argument for pluralism about reference determination (§ 3.4): The causal chain view of reference determination is extremely inflexible. As my other arguments will have revealed, in order to account for the full range of proper name reference it either needs to make various implausible assumptions, like the existence of inadvertent baptism, accidental attachment to casual chains, and the speaker/semantic reference distinction,1 or it must take the view that in some cases there is simply a fact of the matter about what is being referred to, and it’s not what people think. None of these are knock-down arguments against causal-homophonism, nor, together, do they offer an indefeasible refutation. However, they do indicate that the position is not nearly as inviting or elegant as its proponents have suggested. Indeed, in order to account for the vagaries of the use 1 This is a general problem for all semantic accounts of reference. 63 3.1. The Argument from Shared Names of proper names, it must accept significant complications that weaken the strength of its simplicity. Alternatively, causal-homophonists can maintain their position’s simplicity by refusing to accommodate the complications, and simply drawing a strong distinction between what names truly refer to, for the purposes of contributions to truth-conditional sentence content, and what people use them to achieve. In reality, homophonists often appear to lurk somewhere between these two positions, accepting certain complications to accommodate features of use, but only to a certain extent, so that significant gaps still emerge between what the theory says about reference and how it appears that speakers actually use names. All this leaves causal-homophonism in a rather poor position, such that an account of proper name reference that could more easily deal with the data, whilst being able to do everything that causal-homophonism can, would be eminently preferably. 3.1 The Argument from Shared Names It appears that many objects, particularly people and places, are named in the same way—there are many people called ‘Lucy’ and ‘John Smith’, and multiple places called ‘Newton’ and ‘Springfield’. Indeed, it is reasonably difficult to think of real objects with names that are known to be unique. If we did come across one, the chances are that it would be only coincidentally unique, and there would be nothing to stop us ending the uniqueness by simply granting something else the name. Even in domains where attempts are made to name things uniquely, such as various scientific fields, or pedigree animal or racehorse naming,2 the names are only unique within that domain: whilst astronomers might not name more than one nebula ‘M 78’ (assuming this is a name), nothing about that would prevent a pharmaceutical researcher also naming a test subject ‘M 78’. And though rules laid down by horse racing authorities might preclude me naming a new racehorse ‘Shergar’, I’m quite free (in the UK at least) to so name my child. Moreover, Travis (1981a) observes that in a case in which a name really only has one bearer (imagine an unusual name given to a child which, coincidentally no one else has ever thought of—‘Almost Chantilly’, perhaps, or ‘Nils Kürbis’), the name would not, when used, always refer to her (pp. 26–27): Consider a talent show-style competition in which participants are knocked out one by one. The participants each wear a name badge, this being the only way that the judges and competition 2 In fact, it appears that in the UK and US, the names of racehorses can be reused, as long as any previous bearers are no longer registered with the relevant authority and are either dead or above a certain age, and the name is not on the ‘International and Domestic protected lists’, which seem to include the names of all successful and famous racehorses, whether or not they are still alive. 64 3.1. The Argument from Shared Names organizers can identify them. When the judges rule against a candidate, they ask the security team to remove them bodily. Suppose Nils Kürbis—who tells me that he believes himself to be the only bearer of his name in the world3 —is taking part in this competition, but through some administrative error his name badge has been mixed up with another contestant’s, Herbert Melon, but neither has had the opportunity to point out the error. At some stage in the proceedings, Herbert Melon, the competitor wearing Nils’s name badge is eliminated, and one of the judge calls out ‘Remove Nils Kürbis!’, meaning to refer to Herbert, and the security team remove him, identifying him by his badge. In such a context it is clear that it is Herbert that has been referred to with the name ‘Nils Kürbis’, even though someone else is the only bearer of that name in the world. However, as discussed above, according to causal-homophonism, proper names cannot be shared. This is because, if the meaning of a name is just its reference, then either most names have multiple meanings, and are ambiguous between them, or each name can in fact only have one referent, and there are multiple, homophonous names. That there are multiple, homophonous names ‘Lucy’, rather than one, multiply ambiguous name comes down to how fine-grained one deems names to be: If one Lucy’s name is to another Lucy’s name as ‘bank’ the financial institution is to ‘bank’ the river’s edge, which seems to be the case if they have wholly different, unrelated meanings (one’s meaning is constituted by its reference to one person, and the other’s to a different person), then it is reasonable to claim that they are distinct, yet homophonous, lexical items, just as is claimed with ‘bank’ and ‘bank’. So, although there may be some sense in which there is one ambiguous name, there is a significant sense in which there are multiple names—one for each Lucy. This is the sense which implies that there are no shared names.4 The position that each proper name can only denote one thing seems to have the consequence that utterances such as ‘I share a first name with Lucy Campbell’ on the lips of speakers named ‘Lucy’ are actually (literally) false. Yet such sentences are frequently held to be true, and I believe speakers of English have a strong (pre-theoretic) intuition that they can be true simpliciter, and not merely true when interpreted as a loose way of speaking, or as paraphrases for some other sentences. Rami (2014, pp. 123) observes that proper names ‘are social artifacts and their nature essentially depends on how we think about them and how we use them.’ It is implausible, he claims, that speakers are massively and systematically ignorant of the correct individuation of names. 3 His name, when styled in English, would be ‘Neil Pumpkin’ after all. idea that ‘Lucy’ is to non-co-referential ‘Lucy’ as ‘bank’ the name of a financial institution is to ‘bank’ the name of a sloping side or sloping sided thing seems itself quite wrong. See Katz (2001) for discussion and arguments to this end. 4 The 65 3.1. The Argument from Shared Names Therefore, I take it that it is true that I share a name with the many other Peters in the world, and a theory entailing that this, and all similar utterances, are (always) false must be wrong. It is important to note that many of the philosophers positing the thesis that there are no shared names are not merely suggesting that names can be adequately modelled as non-logical constant terms for formal semantical purposes, they are saying something about the real metaphysics of the actual lexical items we use to talk about things. Similarly, the sense in which I claim that names can be shared is the sense that is relevant to semantic inquiry, since, as I will argue, it is the only sense generally recognized by speakers of English. The homophonist’s notion of a name is an artificial construct left over from an idealized view of natural language, with little to do with how names are actually used. Furthermore, there is no explanatory payoff to positing the existence of names that cannot be shared, given the other problems I will raise for causal-homophonism in subsequent sections. 3.1.1 Evidence Against Homophony Generic Names One might imagine that, on Kaplan’s account, the intuitions about sharing names can be accounted for—there is an ambiguity in ‘name’ over generic and specific names. But, as is pointed out in Rami (2014, pp. 123–4) it will not work to say that it is generic names that are shared, because generic names are not the sorts of things that can be ‘had’ or ‘borne’, at least not according to the notion of a generic name employed by Kaplan (1990) and Sainsbury (2015). Sainsbury explicitly states that ‘it’s only specific names, and not generic names, that have bearers’, and that generic names have no significant semantics at all (p. 196). Kaplan, as already stated, sees generic names as species to individual specific names, and thinks that if they have any semantics, it is to refer to kinds. Even if generic names are like species, it would not follow that they were things that were had by bearers. Indeed, if it were claimed that every object that bore a specific name also bore the generic name that that specific name instantiated, then each object would bear two homophonous names, but in radically different ways, since the names would be radically different kinds of word. This would create an ambiguity not just in the word ‘name’, but also in ‘bear’ or ‘have’ as they relate objects and names. Rami observes (p. 124) that, if generic names have bearers derivatively upon the specific names that instantiate them having bearers, there is nothing to prevent the claim that generic names also derivatively have referents, since, for the homophonist, a specific name’s bearer is just its unique referent. But there being referential 66 3.1. The Argument from Shared Names names with multiple bearers is precisely what the distinction between generic and specific names was supposed to avoid. It seems then, at least according to the standard conception of generic names, that assertions that multiple objects bear the same generic name must be false, just as those asserting the same of specific names are. The closest true sentences must say something like ‘My first name is an instance of the same generic name as Peter Winch’s first name’, and this is not what anyone (except the theoretically encumbered) believe they are asserting when they say something about shared names. Whilst it may be possible for people to be wrong about what it is they are expressing with their words from time to time, it is implausible that all English speakers are systematically mistaken about the content of utterances which appear to assert sameness of name between two distinct objects. Similarly, no English speaker, to my knowledge, believes there is a sense in which Peters have distinct names that is accounted for by positing specific names. This is a different kind of case from simply having false beliefs about certain ways in which language works, because it constitutes a systematic mistake about what it is that we say with our words—their content. Quantitative vs Qualitative Identity In many cases where the expression pthe same xq is used, it is possible to distinguish both a qualitative and a quantitative reading. For instance, ‘Peter and I have the same netbook’ could mean either ‘Peter and I have the same model of netbook’, or ‘Peter and I share a netbook’. It has been suggested that this distinction could be applied in the case of ‘the same name’, thus justifying the generic/specific dichotomy, and providing a sense in which Peters do not have the same name (Justice 2001). However, this does not survive scrutiny. The sentence ‘Peter and I have the same name’ can be reformulated as ‘Peter and I share a name’, salve veritate, and this is not so in cases where only qualitative identity is meant; sharing usually entails numerical identity of what is shared. For example, sharing an address, sharing a bed, sharing a packet of Rolos, sharing a great grandfather. In all these cases, what is shared is considered to be a single thing. There are some cases of sharing which do not entail simple numerical identity, mostly in cases of things that it’s not obvious how to count. For example, intentional emotions can be shared—‘We share a hatred for Rick Trainor’—as can dispositions and desires. However, it is clear is that these are not simple cases of qualitative identity either. There also appears to be prosodic evidence that it is not merely qualitative 67 3.1. The Argument from Shared Names identity meant by ‘the same name’5 : In English prosody, expressions of the two identities are often marked differently to get a different focus. For example, ‘We don’t have the same netbook’ denies quantitative, though possibly not qualitative identity. Compare this with ‘We don’t have the same netbook’, which seems to deny quantitative and qualitative identity (although this will depend to some degree on the context). Inserting ‘name’ for ‘netbook’, the first marking—‘We don’t have the same name’—seems to be infelicitous, or at least does not seem to receive a quantitative reading: Imagine a pedant responding to an ambiguous statement about the qualitative identity of netbooks by saying ‘They’re not the same netbook’, intending to assert numerical distinctness. Although they might be annoying, we would probably understand what they were getting at. But if the same pedant were to respond in that way to a statement about name identity—‘But you don’t have the same name’—we would think they’d either said something very odd, or that they were making a pedantic point about some qualitative difference, like the spelling of names. These two pieces of evidence strongly suggest that the ordinary language notion of having the same name is not merely a qualitative one. Loose Talk Another possibility for the homophonist is that people are speaking loosely, somehow, when they say that two Peters share a name. But loose talk is generally recognized as such, and no one does think that this is loose talk. There is simply nothing to suggest that ordinary talk of sharing names is loose talk. Kaplan himself admits that his view is highly counterintuitive, citing the incredulity a presentation of the view incited amongst distinguished philosophers of language (Kaplan 1990, pp. 110–11), and such incredulity would not arise if same-name talk was widely regarded as being loose or ambiguous in some way. Moreover, as I have claimed above, the best paraphrase available to the homophonist is wholly unconvincing. A further argument against the homophonist thesis is made by Cohen (1980, p. 143), who observes that the multiple reference of proper names is very unlike genuine homophony, since names seem to be intrinsically versatile, whereas homophony, such as ‘bank’/‘bank’, ‘bear’/‘bear’, or ‘fluke’/‘fluke’, is a comparatively rare phenomenon, or at the least, it is frequently coincidental. It is simply wrong to suggest that the reuse of names is an inconvenience arising from the impreciseness of natural language, as Frege (1892) does, or that occurrences of shared names are deviant from the norm. The very opposite seems to be true: in many cases people very intentionally reuse 5 Many thanks to Peter Sutton for pointing this out to me. 68 3.1. The Argument from Shared Names existing names, indeed in many cultures it would be unusual or even deviant to introduce an entirely new name for a person or place. 3.1.2 Homophonist Responses Those inclined towards the homophonist picture of names are frequently disposed to react dismissively to arguments from the existence of shared names.6 They are inclined to see such arguments from ordinary language use as akin to arguments from folk-intuitions against scientific discoveries. For example, arguing that names must be shared, because sentences such as ‘there are many people who share the name “Lucy Campbell”’ are widely deemed to be true, may seem analogous to arguing against heliocentrism on the grounds that sentences such as ‘the sun moves across the sky’ are frequently deemed to be true. Ordinary folk, it is argued, do not have the same insight into the nature of the words and language they use as theorists who are in the business of analyzing those words and language, and presenting theories to best explain how they work. Whilst I do not claim that ordinary language arguments about shared names are definitive against homophonism, I do not think the analogy between its claims with regard to name individuation, and the claims of empirical sciences hold up. When epistemologists attempt to define knowledge, it seems that they are under a certain obligation to account for many, if not all, English speakers’ uses of the term ‘knowledge’, at least of a particular kind. There is an idea, at least on some understandings of the project, that speakers are in possession of a concept knowledge, and the epistemologist is aiming to provide an analysis of that concept. If their analysis fails to account for what look like good cases in which the speakers’ concept applies, then it is inadequate. Whilst it might be an analysis of some epistemological or psychological phenomenon, it is not knowledge. It may be that there is a distinction to draw between the use of ordinary language intuitions in such a case, and claims about proper name individuation. Homophonists will claim that, in providing an account of the individuation of names, they are not in the business of providing an analysis of any ordinary language concept proper name. Rather, they are observing how names are used, or, more likely, what names are purported to do, and attempting to provide an analysis of how names must be in order to do that, within a broader semantic framework. Thus, homophonists may claim that they are under no obligation to respect the ‘folk-intuitions’ of speakers about how their language works. Speakers qua speakers have no more insight into 6 In person at least. Published responses to such arguments are harder to come by. 69 3.1. The Argument from Shared Names the true nature of the words they use than thinkers do into the true nature of the brain’s workings. This account of what the homophonist is doing in denying that names can be shared is misguided, however. As mentioned above, homophonists are saying something about the metaphysics of the actual proper names we use to talk about things. They are not introducing a technical term, ‘proper name’, that fits into a model-theoretic account of language. If they were to do that, although the account might not entail the falsity of utterances of English, we would be left with the question of whether or not the technical term described anything actually present in natural language generally, or English in particular. In general, renaissance astronomers and contemporary neuroscientists purport to be describing the world as it is, rather than creating self-consistent models. But unlike them, the homophonist, when making claims about the individuation of names, does not observe how names are used, or consult with speakers who use names. Homophonists come to the conclusion that speakers are wrong in their conception of names because they (the homophonists) begin with a very particular, and theoretically-encumbered, view of how proper names function. They take them to refer to objects, which is plausible enough, but they take that reference to be a substantial metaphysical relation between a word and the object that holds independently of any particular use. They also assume, as mentioned at the beginning of chapter 3, a truth-conditional conception of semantics, and that the only semantic contribution a name makes to the truth-conditions of a sentence containing it is its reference. It is these assumptions, when taken together, that lead to the counter-intuitive homophonist individuation of names: If a name refers tout court, it must refer independently of any context of utterance, and if it is to contribute a reference to the truth-conditions of a sentence such that a determinate truth-value could be returned, it must have only one referent. The assumptions that lead to the homophonist individuation of names are not, however, empirically grounded in the way that the claims of renaissance astronomers or contemporary neuroscientists are generally assumed to be. The claim that proper names refer tout court has little empirical or intuitive sway: names never occur independently of a use in a context, so there can be little reason beyond a historical predilection for assuming that word meanings are not context-sensitive for assuming that names refer independently of use. And indeed, many contemporary scholars deny that names refer tout court, even though they have access to all the same data as the homophonists.7 Similarly, 7 See, for example, Burge (1973); Cohen (1980); Cumming (2008); Dever (1998); Elbourne (2005); Fara (2015); Gray (forthcoming); Katz (2001); Matushansky (2008); Michaelson (2013); 70 3.2. The Argument from Communication truth-conditional approaches to semantics, although widespread, owe more to Frege’s place at the conception of formal semantics than they do to empirical observation, and dissent from the applicability of truth-conditional semantics to natural language is common. Although there is some initial intuitive pull to the idea that knowing the meaning of an expression is to know under what conditions the sentence in which it occurs is true, consideration of some examples of natural language usage, for instance, Travis cases (see §2.3), demonstrates that things are more complex. The assumptions that lead homophonists to their position on name individuation are a result of theoretical rather than empirical considerations. So it appears that the situation in which they claim that the beliefs of ordinary speakers about the nature of names is rather more like that of the epistemologist who claims to have a better idea of what knowledge is than speakers who employ the notion daily, and less like the astronomer who, through celestial observations, realizes that the Earth orbits the Sun rather than the other way around, as had previously been assumed. Sainsbury (2015) is explicit that causal-homophonism, and the notion of semantic reference it creates, are theoretical, and suit theoretical purposes (p. 209). However, this leaves the position open to the claim that its theoretical posits fail to capture anything significant about natural language, particularly if an alternative account of proper names is available that adequately captures the referential use of names in natural language without burdensome theoretical notions that entail the falsehood of ordinary kinds of utterance. 3.2 The Argument from Communication Travis (1981a) argues that semantic conventions that stipulate or limit what a referring term can refer to are impotent in the face of pragmatic cues for reference or for communication (ibid., pp. 24–26). Donnellan cases (Donnellan 1966) provide examples of this. Consider Donnellan’s famous man with a martini: a speaker can succeed in using the expression ‘the man with the martini’ to refer to someone who is not drinking a martini, even though it is generally thought that the semantics of definite descriptions stipulate that it can only be satisfied by an object fulfilling the description, given the right context. For example, if there is a man present drinking water from a martini glass, and no other perceptually salient men are drinking from martini glasses, then he is likely to be the referent of ‘the man with the martini’, in the sense of the utterance being about that man and best taken to be so, the truth of the utterance depending upon him (on reasonable accounts of the truth of Pelczar and Rainsbury (1998); Rami (2014); Recanati (1993); Sawyer (2010); Travis (1981a). 71 3.2. The Argument from Communication utterances). This can be the case even if there are other men nearby who are covertly drinking martinis. This can also be the case if the speaker is aware that their referent is only drinking water, and that there are covert martini drinkers around. For example, the speaker may use the expression in a tongue-in-cheek manner, yet still succeed in referring, or simply be aware that the expression is the best way to refer, given her audience’s perspective. One plausible analysis of such cases is that it is what a use of a name can reasonably be taken to make recognizable in the particular circumstances of utterance that is important to what gets referred to, rather than how the world might be, independently of the aims and interests of interlocutors. And what words are best understood by a reasonable audience to have made recognizable as what is being said with them is essentially what those words are communicating under the particular circumstance. The homophonist account of proper names does a very good job of explaining how proper names could refer just in virtue of their semantic content, given that many names appear to be multiply referential. However, it goes little way towards explaining how such names are used to successfully communicate about single objects, since the causal chains which individuate homophonic names are often not cognitively accessible to users of those names. For homophonists, the mechanisms that allow hearers to appreciate what a speaker’s use of a name refers to are entirely separate from the mechanisms that determine the name’s reference. Sainsbury (2015) accuses Pelczar and Rainsbury (1998) of failing to recognize the distinction between the facts that determine the bearer of an utterance of a name (which for him constitutes its reference), and the evidence an interpreter might have for it (p. 197). Homophonists are well aware that language-users must often make use of contextual information in order to work out what a name refers to on an occasion of use, at the very least in order to disambiguate homophony, but they refer to such information, and its processing, as pre-semantics (Kaplan 1989b; Perry 1998)—the kind of non-semantic cues that allow language-users to appreciate which words comprise a phonological string, prior to parsing their semantics. They maintain that this use of contextual information does not contribute to what is strictly-speaking referred to, i.e., to the truth-conditions of the sentence or utterance: Each name-using practice involves exactly one baptism; baptisms metaphysically individuate practices, and thus fix the referent, if any, of a practice, though when we wish to know to which practice a given use of a name belongs, or what the referent of a practice is, it is rare that we can reach an answer by first identifying the baptism. Normally our evidence is associated information, even though this is evidence only, and does not 72 3.2. The Argument from Communication make a practice the practice it is. (Sainsbury 2005, p. 106) 3.2.1 Communication and Causal Chains The causal-homophonist account of reference is not, however, completely silent on communication. The causal theory of reference, as developed from the picture presented by Kripke (1981), notably by Kaplan (1990) and Sainsbury (2005), states that the reference of a name is transmitted causally by the deference of users to prior uses. This principle extends beyond the reference of names-as-uttered to singular thought: object-oriented thoughts that are about a particular object directly, not in virtue of any general, descriptive content. Being tied into a causal chain of reference enables one to have singular thoughts about the referent. Thus, although causal chains are not generally cognitively accessible, a speaker can communicate a singular thought to their interlocutor using a name, just in virtue of admitting them to the chain. According to Sainsbury (ibid.) this process can convey knowledge of the referent of a name to a hearer, even though they may have no knowledge of the referent other than that it is the referent of the name: When we come to master the use of a name with a bearer, we thereby come to know (I hold) who or what the bearer is, and more generally we satisfy any reasonable demand for discriminating knowledge of the bearer. (ibid., p. 97) Mastery of the use of a name here just amounts to ‘having a resolution to use the name “in the way it is used” by the existing users’ (ibid., p. 113), although it presumably also requires that one not also be systematically confused about the use of which one is a master. In claiming that resolving to use a name in the way that others use it grants one discriminating knowledge of its bearer, Sainsbury is making the strong claim that simply employing a name that is causally connected to an object grants one not only the capacity to have singular thoughts about that object, but also the capacity to single that object out from all other objects, even though one has no other capacities, such as perceptual or descriptive capacities, to do so. This is in contrast to Evans (1982), who claims that ‘a capacity to distinguish [an] object of his judgement from all other things’ is required in order to have singular thoughts about that object (p. 89), but he denies that ‘the use of a proper name [. . . is] an autonomous way of satisfying the requirement that one have discriminating knowledge of the objects of one’s thoughts’ (p. 403). 73 3.2. The Argument from Communication According to Sainsbury’s causal-homophonist account of reference transmission,8 the use of a name transfers the capacity to have singular thought about its bearer from speaker to hearer, and so allows the communication of the singular thought expressed by the speaker. The sentence/utterance and the thought are about the referent of the name because they are causally connected to it. The fact that a hearer may in fact have plenty of other descriptive or perceptual information about the referent that is not connected to the causal chain is irrelevant. As observed by Evans (1973), this model of communication using names appears to work very well in a limited range of cases: those in which a hearer is entirely unacquainted with a referent, or in which a speaker truly does begin using a name deferentially—what Evans calls ‘mouthpiece cases’, and Rami (2016) calls ‘parasitic uses’. For example: Lucy joins her friends Peter and Alex at the lunch table whilst Peter and Alex are already involved in a conversation about someone called ‘Rory’. Lucy does not know Rory, and has not heard of him before. Lucy can learn things about Rory from what Peter and Alex say, and she can even join in the conversation about him by saying things like ‘Rory sounds like he needs some help’. So Peter and Alex can communicate with Lucy about Rory using the name ‘Rory’, even though she had no prior knowledge of him, and she is able to talk about him and to refer to him successfully using the name. If Lucy were asked ‘Who’s Rory?’, she might plausibly reply—in certain contexts—that Rory is the person about whom Alex and Peter are talking, and state some relevant information about him she has picked up from the conversation. 3.2.2 Causal Chains Without Communication The causal-homophonist explanation of communication using names works well in cases such as Lucy’s because it explains how Lucy is able to refer to Rory just in virtue of having picked up his name from Peter and Alex, and her deference to their reference. However, as already indicated, this is a very particular kind of communicative interaction. We can well imagine that if Lucy were to approach Jen, who was not involved in the conversation with Peter and Alex, and who does not know Rory, and say, without providing any context, ‘Rory sounds like he needs some help’, she would fail to communicate anything about Rory—at least on a reasonable understanding of the kind of uptake required for communication—because Jen simply does not know who ‘Rory’ is, and it seems very unlikely that Jen would thereby be able to have 8 Sainsbury does not, in fact, commit himself to the homophony thesis about the individuation of names in his (2005); he applies the Kripkean causal theory of reference to naming practices, rather than directly to names themselves. Sainsbury (2015) explicitly adopts homophonism and outlines a causal theory of names, drawing on Kaplan (1990). 74 3.2. The Argument from Communication a singular thought about Rory. Jen might well respond ‘What’re you talking about? Who’s Rory? Leave me alone!’. Furthermore, if Jen were to instead respond ‘Rory sounds like he needs to pull himself together’, or even ‘He sounds like he needs to pull himself together’, quite as a joke, it is not clear that Jen would have said anything about the particular person Alex and Peter were talking about, even though she appeared to use the name in deference to Lucy’s use of the name, or used ‘he’ anaphorically upon it. ‘My friend Katharine’ It is notable that when a speaker wants to communicate about a person (or other named object) that their interlocutor is unfamiliar with, or unfamiliar with by name, in many contexts it is normal and—plausibly—necessary, to form a description with the name of the person, or insert some small piece of descriptive information about them. For example, if Helena wants to tell Clare about Katharine’s love of Bowie, but she knows that Clare doesn’t know Katharine, and isn’t familiar with talk about her using her name, then she will not simply begin ‘Katharine loves Bowie’, but will say something like ‘My friend Katharine loves Bowie’ or ‘I have a friend from school, Katharine, who loves Bowie’.9 Not only does this seem like a polite convention on Helena’s part, it seems that, at least in many contexts, Clare would be wholly confused by Helena’s utterance, and unable to interpret it, if it were just to contain the name without the descriptive content. She will reject the attempt at reference, and request clarification by asking something like ‘Who’s Katharine?’: it appears that Helena is unable to use the name to communicate referentially. Indeed, it appears that, in English, in many contexts, if a name is used bare, without additional descriptive content, there is a presupposition that the audience is already familiar with the use of that name to refer to the speaker’s intended referent, and will be able to suitably appreciate the reference. Adding some kind of minimal descriptive content (and it is clear that it can be extremely minimal) to the use of the name seems to mitigate this constraint on communication, although the question of whether singular thought will be elicited in a hearer unfamiliar with the referent by such a referential expression bears further consideration. Regardless of one’s view on this question of singular thought, it seems clear that the causal-homophonist cannot account for this quirk of communication using proper names. According to the causal theory, Helena’s use of the name ‘Katharine’ should be enough to induct Clare into the relevant naming 9 Providing philosophical examples is one obvious counter-example to this practice: I am not actually trying to refer to particular people in my examples, and I can indicate this brute introduction of characters by using names without any accompanying description. 75 3.2. The Argument from Communication practice and communicate to her the referential thought being expressed. The homophonist must claim that, had Helena just said ‘Katharine loves Bowie’, she would have referred to Katharine, and either Clare does know who Katharine is in virtue of Helena’s reference, but seeks further information about her, or Clare is stubbornly refusing to appreciate the reference proffered to her, demanding richer information. Both these options are problematic. The first option is problematic because, although it might appear that Helena referred and Clare was able to pick up on her reference by saying ‘Who’s Katharine?’, in fact, nothing more was communicated than that Helena was trying to talk about someone using the name ‘Katharine’. She might well have simply said ‘“Katharine”?’. It is clear that, on any understanding of ‘knowing who’ that is relevant to her, Clare does not know who Katharine is in this context. She would actively disavow such knowledge, and might not even ascent to Katharine being whomever Helena referred to with the name. The second option is implausible because Clare’s situation of ignorance seems eminently reasonable: she is not being obtuse, or difficult, or in any way trying to be unco-operative. Deference and Context Further problems for the view of communication by reference put forward by the causal-homophonist can be brought out with a similar example to that of Lucy, involving picking up names from the newspaper, and then using those names in discussing the news, apparently referentially. This is a fairly common scenario. Suppose Mike reads a newspaper article that contains the sentence ‘Police are searching for a third man, Brian Blessed, in connection with the robbery’, that being the only time that the name ‘Brian Blessed’ occurs in the article. Mike can (under certain circumstances) say ‘If I were Brian Blessed, I’d have skipped the country by now’, and successfully refer to the person mentioned in the article. The causal-homophonist can explain this reference by claiming that Mike’s deference to the use of the name in the article ties him in to the chain of referential uses stemming back to the person themself. But now consider two supplementary scenarios: In the first, the person Mike addresses when he speaks, Luke, has not read the article, and is unfamiliar with the story; in the second, the author of the newspaper article has made a mistake, and the suspect the police are searching for is actually called ‘Ryan Jones’. In the first situation, it seems very likely that Luke, not knowing the content of the article Mike was reading, would take him to be talking about the well-known actor called ‘Brian Blessed’, rather than the suspect mentioned 76 3.2. The Argument from Communication in the article. Mike has then failed to communicate about the person he meant to, and has, in virtue of failure to appreciate the context of their utterance, unintentionally communicated about someone quite different, but with a same-sounding name. The causal chain connecting Mike’s use of the name to its bearer has failed in its task of facilitating a singular thought about the bearer in Luke, and Luke has a singular thought about someone else entirely. But it seems entirely reasonable that Luke took Mike to be referring to the actor Brian Blessed: Mike used the name out-of-the-blue, without ensuring that Luke was aware of the content of the article, and Luke assumed—apparently very naturally—that Mike was trying to communicate in a way perspicuous to him, i.e. by using ‘Brian Blessed’ in a way he was familiar with. According to the homophonists, Luke has failed to appreciate which name Mike was using because of a breakdown at the pre-semantic level. Consider the second situation, in which the author of the newspaper article made a mistake, and the suspect the police are looking for is not in fact named ‘Brian Blessed’, but ‘Ryan Jones’. Suppose the author intended to reproduce the correct information provided by the police, but that, as she was composing her copy, someone mentioned the actor Brian Blessed and, in a moment of aberration, she wrote that name rather than that of the suspect. Suppose also, this time, that both Mike and Luke have read the article containing the anomalous name. When Mike says ‘If I were Brian Blessed, I’d have skipped the country by now’, if Luke, taking Mike to be talking about the suspect mentioned in the article, replies ‘If I were him, I’d be living it up in the South Pacific’, it seems that they are having a conversation about the suspect, and are referring to him. This is in spite of the fact that the name printed in the paper is causally connected, in some manner, to Brian Blessed, the actor. The homophonist is not obviously committed to the claim that it is Brian Blessed referred to by either the author of the article or Luke and Mike, since the author did not intend to use the name ‘Brian Blessed’ in the way it is normally used—this generally being a requirement of being a link in a referential causal chain, e.g. for Sainsbury (2005)—but it is not clear how the homophonist could happily explain Mike’s use of ‘Brian Blessed’ apparently referring to Ryan Jones. They might claim that an inadvertent baptism had taken place, and that the author of the article had created a new name ‘Brian Blessed’, referring to the suspect, or that such a name is created at some point, given Mike’s use of the name. Such a claim is implausible. It is entirely opaque when or how such an act could have occurred: The author did not intend to use the name ‘Brian Blessed’ to refer to the suspect. Though Mike did, it does not seem in keeping with the tenets of homophonism to assert that one use of a name, which the speaker also intended to be in keeping with previous uses, could 77 3.2. The Argument from Communication create a new name and baptize an object with it. Shared Names and Causal Chains Evans (1973) provides an example designed to undermine the causal theory of reference, in which a quiz show contestant asserts, when asked to name a capital city, ‘Kingston is the Capital of Jamaica’. Evans has the strong intuition that the contestant has ‘said something strictly and literally true’, even though, in his example, it turns out that contestant had learned this piece of (apparent) information from a racist who actually meant to refer to Kingston-upon-Thames (p. 194). The causal theory of reference predicts that, since the contestant picked up the name ‘Kingston’ from someone who was intending to use it as part of a causal chain attached to the settlement now in southwest London, the contestant’s use is also a part of that causal chain, and she too refers to Kingston-upon-Thames, rather than Kingston, Jamaica. I agree, of course, with Evans that the contestant says something true in the context of the quiz, and certainly answers the question correctly, although I do not accept the notion of something being strictly and literally true: there might be contexts in which the contestant will end up saying something false with the same words. With just the bare bones of Kripke’s causal picture of reference, the causal theorist does not have much to counter Evans’s intuition about the Kingston case, but by introducing the homophonist individuation of names, they have a little more to say. According to the causal-homophonist, the quiz show contestant says something literally false, because she is not in fact using the name of the Jamaican capital at all, although it sounds like it, but is using the name of a quite distinct town. They might allow that, in quiz contexts, what is literally said is less important than a favourable interpretation of what might have been said, but the homophonist is still committed to the idea that the contestant literally spoke of Kingston-upon-Thames. Whilst this response does not get the homophonist the most intuitive result, it does provide an explanation of why the intuition arises (the traps of homophony), and why the intuition is incorrect. However, examples can be constructed which highlight the fact that the homophonist response is unsatisfactory. Consider Sam, who is told by Sarah, ‘Lucy is upstairs’. Sarah is talking about Lucy Goldstein, who she believes to be the only person upstairs. In fact, Sarah is wrong: only Lucy Campbell is upstairs. Sam knows neither Lucy. He goes upstairs, bumps into Lucy Campbell and says ‘You must be Lucy’, intending to use the name as Sarah used it. Lucy replies ‘Yes, I am. Nice to meet you’. Does Sam now know Lucy Campbell’s name? Intuitively, he does: she just agreed that she 78 3.2. The Argument from Communication was Lucy, which, under normal introductory circumstances, entails that her name is ‘Lucy’.10 So Sam knows that Lucy’s name is ‘Lucy’. But according to the most obvious reading of the causal-homophonist thesis, he does not.11 When he said ‘You must be Lucy’, he uttered Lucy Goldstein’s name, not Lucy Campbell’s, because his use was tied, via his intention to reproduce Sarah’s name use, to the causal history of the former Lucy’s name. However, it is clear that Sam and Lucy’s interaction was in fact perfectly felicitous, and Sam learnt Lucy’s name. Of course, things would be very different if Sam had an appointment with Lucy Goldstein, and Sarah sent Sam upstairs for his appointment. Then, it is clear that Sam is looking for a much more particular Lucy, and there is a case of mistaken identity between Sam and Lucy Campbell. As it is, although there is a mistake, it is solely Sarah’s. And while it is pure chance, arising because of Sarah’s mistake, that Sam got Lucy’s name right, he did so nevertheless. Sainsbury (2005) and (2015) both seem to allow that something like the case just described can arise, but it is unclear whether either can properly account for it. Sainsbury (2005), does not assume the homophonist individuation of names, but rather relies on name-using practices, which are brought into existence, and individuated, by baptisms, such that for each name-using practice there is a unique baptism (p. 106). Moreover, Sainsbury claims that, as a practice’s baptism fixes its reference, a practice has its reference essentially (ibid., p. 113). If a name appears to change its reference, then a new practice has arisen, which must have been begun by a new baptism, even if that baptism is inadvertent (p. 111). In Sainsbury (2015), name-using practices have been replaced by names individuated in the homophonist manner. Like name-using practices, names are individuated by a unique baptism that brings them into being,12 but unlike the practices of (2005), the names of (2015) can change their referent without the occurrence of a new baptism, inadvertent or otherwise. Sainsbury (2005) must treat the above case as a failed initiation into the practice of using ‘Lucy’ to refer to Lucy Goldstein: Sarah tries to initiate Sam into the practice, but, because his very first attempt at using the name as part of the practice goes awry, Sam is simply not initiated into the ‘Lucy Goldstein’ practice (ibid., pp. 113–4), although perhaps he is, quite coincidentally, inducted into the practice of using ‘Lucy’ to refer to Lucy Campbell, as a result of 10 Though certainly not all circumstances. to Clayton Littlejohn for highlighting this kind of epistemic problem for the homophonist. 12 Note that in neither case must the baptism have an object, empty names being a primary motivation for Sainsbury’s work on proper names. 11 Thanks 79 3.2. The Argument from Communication his interaction with her. This explanation is inadequate, however.13 In order to account for an initiation into a name-using practice taking, or, conversely, not taking, Sainsbury appeals to a speaker’s speaker reference—the object of a speaker’s referential intention, discussed further in §3.2.3—coinciding, or not, with the referent of the name-using practice in question—the semantic reference. So, when Sam says to Lucy Campbell ‘You must be Lucy’, in order for him to fail in his use of the practice into which Sarah tried to initiate him, the object of his referential intention must not be Lucy Goldstein. In fact, it should presumably be Lucy Campbell, if he is going to come to know her name as a result. But it is far from clear that Lucy Campbell is the speaker referent of his utterance of ‘Lucy’. He is asking her whether she is Lucy, and attempting to use Lucy in the way Sarah did. If Lucy Campbell’s name were not ‘Lucy’, then she might have said ‘No. I’m not. Lucy just went down the fire escape.’, or some such, and then it seems unlikely that we would have the intuition that she was his speaker referent, and it would be clear that Lucy Goldstein was both the speaker and semantic referent (assuming we allow for such notions at all). The notion of speaker reference is also essential to the account of names in Sainsbury (2015). Sainsbury uses speaker reference to explain the continued reference of a name to its bearer, and the possibility of a name changing its reference. Unlike his (2005) account, he allows that a speaker can properly acquire use of a particular specific name, but be using it to speaker refer to something other than its bearer, at least for a while. The example he gives is of a child who has misunderstood their parents’ introduction of the name ‘Whiskers’ for a kitten, and wrongly applies it to a puppy until corrected (Sainsbury 2015, p. 206). It seems likely that Sainsbury would use this same idea in the case of Sam and Lucy. Although Sam acquires Lucy Goldstein’s name from Sarah, upon meeting Lucy Campbell, he speaker refers to her with the name. However, as just discussed, it is not at all clear that Lucy Campbell is his speaker referent. Moreover, this would still leave Sam with the wrong belief. 3.2.3 Speaker Reference Kripke The notion of speaker reference, as a distinct phenomenon to semantic reference, plays a rôle not just in Sainsbury’s accounts of proper name 13 Since the explanation is mine rather than Sainsbury’s, the inadequacy is presumably mine. However, it is hard to see how else Sainsbury would approach the problem using the framework he lays out in his (2005). 80 3.2. The Argument from Communication reference, but in many semantic views of reference, both in order to account for discrepancies in intuitions about reference from the predictions of theories, and, as in Sainsbury’s causal accounts, in order to ground semantic reference. The speaker reference/semantic reference distinction first arises in Kripke (1977), which is, in part, a response to Donnellan (1966). As already mentioned, Donnellan argues that definite descriptions can be used to refer to objects, even if the referent fails to fulfil the relevant description. Kripke’s paper introduces the distinction between speaker’s reference and semantic reference in order to account for this alleged phenomenon. The distinction is claimed to be a special case of Grice’s distinction between speaker meaning and semantic meaning (Kripke 1977, p. 263). Kripke claims that referential expressions such as definite descriptions, proper names, and indexicals have a semantic reference, which is what they refer to in virtue of their meaning, and determines their truth-conditional content. Thus, the expression ‘the King of Spain’ can only semantically refer to a person who is uniquely the King of Spain; the name ‘Felipe VI’ can semantically refer only to its bearer, which will be determined causally; and the indexical ‘yesterday’ uttered by me at the time of writing will semantically refer only to the 16th February, 2016. Kripke claims that the meanings he appeals to are determined by conventions in a speakers idiolect, which will usually be closely connected to a common language (ibid., n. 20), and that the particular convention that determines the semantic reference of a referring expression is a general intention to refer to a particular object with the expression (ibid., p. 264). Speaker reference, on the other hand, is determined, by a speaker’s specific intention to refer to an object on an occasion, together with a belief that the object is the semantic referent of the term being used. A speaker can, therefore, speaker refer to an object using an expression not conventionally used for such a referent so long as she has a specific referential intention to so refer, and believes it is semantically appropriate. However, in such a case the semantic reference will not be the speaker referent, and so, in spite of what the speaker might succeed in communicating about the speaker referent, the literal truth-value of the utterance will depend upon the semantic referent. Sainsbury Sainsbury appeals to speaker reference in both his (2005) and (2015), but does not specify what he means by it in either case, so I think it is largely safe to assume that his conception is similar to Kripke’s: a speaker, in using a proper name, speaker refers to whatever object they specifically intend to 81 3.2. The Argument from Communication refer to using the name (where a referential intention may be a specific sort of thing). However, we might credit Sainsbury with dropping Kripke’s condition that a speaker must believe that the object of their referential intention is the semantic referent of the name they’re using. Such a condition is in tension with, for example, certain kinds of accidental baptisms. Diverging further from Kripke, Sainsbury (2015) goes on to define semantic reference as a conventional matter, supervening on a community’s use of a proper name (or other expression), that is, on individual members’ speaker reference: The “semantic reference” of a name, as used in a community, is its conventionalized, stabilized or normalized speaker-reference in the community. “London” refers to London among many speakers who live in England (and elsewhere) because it’s a conventional or stabilized or normal fact about these speakers that they use the specific name “London” (I hope you know which specific name of the genus “London” I have in mind) only if they intend thereby to refer to London. The notion of semantic reference is a theoretical one, and one that needs to be constructed to suit theoretical purposes. In the present case, we need a conception of semantic reference that will supervene on use and help explain features of usage (for example agreement, disagreement, correction). Basing semantic reference on speaker-reference is the most straightforward, and perhaps the only, way to achieve this. Speaker-reference can be theoretically described without any theoretical commitment to semantic reference, so the supervenience has a reductive character. (ibid., p. 209) This is in contrast to Kripke, who bases the conventions of semantic reference on speakers’ general intentions to use a name in a particular way within their idiolect. The view that there is a distinction to be drawn between semantic reference and speaker reference is very much the orthodoxy amongst philosophers of language working on reference. With regard to proper names, it is fairly essential to any semantic view of reference because such views are characterized by their insistence that proper name reference is determined by (or constitutes) the semantics of the name: it is the semantic reference being determined. Any apparent deviation from the semantic reference must be accounted for by the notion of speaker reference, and, indeed, in the case of the causal theory of Sainsbury and others, a notion of reference prior to semantic reference is posited to ground the causal chains of semantic reference. Speaker Reference and Communication Speaker reference, and its distinction from semantic reference, are relevant to the argument from communication against causal-homophonism because 82 3.2. The Argument from Communication speaker reference is often appealed to in cases in which a proper name is used to communicate about an object that is not its bearer, and thus, according to homophonism, is not its semantic referent. I have two main objections to the use of speaker reference by homophonists. The first is that examples are available in which a name is used, but the object communicated about—and what might intuitively be thought of as the referent of the use of a name—is neither what a homophonist would consider to be the semantic referent, nor the speaker referent, if we consider that to be just the object of the speaker’s referential intentions. The second objection is that the homophonists’ use of the speaker reference/semantic reference distinction provides a picture of reference that is overly complicated since it posits an unnecessary ambiguity in the notion of reference: If one notion of reference were able to account for everything we might want to say about reference, the distinction would be redundant. My first objection is that there are cases of apparent reference which are explained by neither speaker nor semantic reference, and fall into neither category. Consider the following example: David Cameron has been attacked by a vengeful citizen, and is critically injured. In the newsroom of a broadcaster, political correspondents and journalists await further news. A man runs in and shouts ‘Cameron’s dead! Cameron’s dead!’. Naturally, those present take the speaker to be telling them that David Cameron, Prime Minister of the UK, has succumbed to his injuries, and, given the context, it seems very natural to claim that the speaker communicated that to his audience, and referred to David Cameron. However, suppose that the speaker, oblivious to the developing political situation, was actually trying to convey information about American actor Cameron Diaz, of whom he is a big fan. We can certainly agree that the speaker was trying to refer and communicate about Diaz. But, because his audience was primed to hear news of David Cameron, and, furthermore, they expected the name ‘Cameron’ to be used to refer to him, it is implausible that the speaker did refer to or communicate about Diaz: given the context in which he spoke, he did not have good reason to use the name he did without elaboration, and the name he used did not provide his audience with good reason for taking him to be talking about Cameron Diaz.14 However, the 14 It is worth noting that the locution ‘was referring to’, employing the imperfective aspect, can be used to report what a speaker meant to refer to, even if the speaker is at fault in causing a misunderstanding. However, it is not clear that this imperfective locution entails success: Just as ‘I was running home’ does not entail ‘I ran home’, ‘She was referring to Lucy’ might not entail ‘She referred to Lucy’. It might not even entail partaking in an activity that might lead to success. Particularly in first-person reporting (‘I was referring to’), such locutions may simply be reports of a speaker’s intentions, or what they believed they intended, even if they were not in a position to fulfil those intentions or supposed intentions. This point is raised by Travis (1980, p. 142; 1981a, p. 16). The same point seems to hold for the locution ‘was talking about’. 83 3.2. The Argument from Communication causal-homophonist has nothing to say about the speaker’s relationship to David Cameron, except that his audience wrongly took him to be referring to him, and this places the mistake on the audience and not the speaker. Cameron Diaz is both the semantic and speaker referent of the speaker’s utterance of ‘Cameron’ because he was using her name ‘Cameron’, causally linked to her baptism, and he intended to refer to her. My second objection questions the need for positing a notion of semantic reference in addition to a notion of reference supposed to explain what happens in communication. As already discussed, Sainsbury (2015) maintains that semantic reference supervenes on speaker reference, and invokes speaker reference in explaining how names maintain or change their bearers and therefore (semantic) reference. Without speakers speaker referring with their uses of names, therefore, there would be no semantic reference: there would be no specific names with their referential semantic content. But if it is accepted that people can use names referentially, without those names being connected to their referents in virtue of their semantics, and, in doing so, at least sometimes communicate successfully (presumably when their intentions are perspicuous), then it is not obvious what rôle semantic reference fulfils, since it certainly does not provide correct truth-conditions in all cases. It is my view that the perceived need for proper names (and other referring expressions) to have a reference in virtue of their semantics, particularly a reference tout court, as homophonists contend, is symptomatic of the view that sentences (rather than utterances) (perhaps together with a formally represented context to allow for indexicals and tense) have a literal meaning that is truth-conditional, and thus are literally true or false simpliciter. Such a view is particularly apparent in Kripke (1977).15 Kripke reveals his rather conservative conception of semantics and pragmatics in a dismissal of Donnellan’s apparent suggestion that the use of language is not merely epiphenomenal: It is not “uses”, in some pragmatic sense, but senses of a sentence which can be analyzed. If the sentence in not (syntactically or) semantically ambiguous, it has only one analysis; to say that it has two distinct analyses is to attribute a syntactic or semantic ambiguity to it. (ibid., p. 262) Whilst Kripke does not deny that speakers can use their words to do things that are not mandated by their semantics, he maintains that this is a pragmatic issue that must be accounted for separately. Kripke chastises Donnellan for positing unnecessary semantic ambiguities (ibid., p. 268), but does not consider that he is himself positing an ambiguity around reference that would 15 Which is, admittedly, nearly forty years old. 84 3.3. The Argument from Family Names not be required if all reference were accounted for at the pragmatic level (as it will be in chapter 6). Travis (1981a) argues that there is a genuine natural language phenomenon of reference, which he calls reference simpliciter, but that neither speaker reference, nor semantic reference, capture it adequately. Reference simpliciter is the kind of phenomenon I introduced in §2.1, and discussed in §2.4: it is supposed to accord with judgements about what an utterance is about, and what its truth turns on, in virtue of particular uses of proper names, both of which are connected to what is recognizable, given the use of the name, in the particular circumstances of utterance. Both Kripke and Sainsbury seem to assume that semantic reference is the kind of thing that is required to capture intuitions about reference in natural language, but they realize that it is not enough on its own. Neither considers the possibility that they could do away with semantic reference entirely. Whilst something like the pragmatic phenomenon of speaker reference that Kripke describes, and to which Sainsbury presumably defers, might get closer to reference simpliciter, because of its reliance on speaker intentions, it clearly will not account for instances such as the ‘Cameron’s dead!’ case. I present a more detailed critique of intention-based accounts of reference in chapter 5. 3.3 The Argument from Family Names The homophonist view is usually stated in terms of ‘shared’ first names, or full names, but less attention has been paid to family names.16 Whilst it may not be totally absurd to think that two Lucys have distinct names ‘Lucy’, and two Lucy Goldsteins have distinct names ‘Lucy Goldstein’, it is much less plausible that a grandmother and granddaughter have distinct family names ‘Goldstein’, since the whole point of family names seems to be that they are shared between members of the same family. Sainsbury (2015) offers a brief discussion of family names (p. 197, particularly n. 2). He claims that family names name families in the same way that the names of individual people name individuals, i.e., one (specific) name refers to each family, and, if there are multiple families apparently named ‘Smith’, there are multiple, homophonous (specific) names ‘Smith’. So, while members of individual families all share a family name that forms part of their own unique individual name, this is not the case between members of distinct homophonously-named families. 16 Many different customs and practices exist with regard to family names. In this section I will consider only how family names are generally used in Western, English-speaking communities, though it may apply more broadly. 85 3.3. The Argument from Family Names 3.3.1 Which Family? Sainsbury’s view of family names presents an immediate problem if he wishes to account for their use to refer to nuclear families, or other limited family groups. Families are not discrete entities in the way that people are. Families are, on the blood-linear understanding of family that goes hand-in-hand with the sharing of patriarchal family names, iterative.17 This is because the relation of ‘belonging to the same family’ seems to minimally apply symmetrically to parents and their children, and to be transitive. The problem, then, is this: according to Sainsbury, family names refer to families. He doesn’t specify what he takes a family to be, but it is clear that ‘The Goldsteins’ can refer to a nuclear family, that is, just one or two generations who are directly related. Sainsbury claims that each name ‘Goldstein’ will refer to just one family (at a time), just as the names of individuals do. If more than one family appears to be named ‘Goldstein’, there is more than one homophonous name. However, families cannot be neatly individuated into nuclear families. Nuclear families overlap. If two parents and two children are one nuclear family named ‘Goldstein’, then one of those parents together with their siblings and parents are also a distinct nuclear family named ‘Goldstein’. And it cannot be that those families bear distinct homophonic names, because both families considered together are also a family named ‘Goldstein’, and the younger nuclear family are only called ‘Goldstein’ because they are a part of a larger family stretching back many generations, all of which is called ‘Goldstein’, and every individual member of which is a Goldstein. The point of homophonism is to simplify the name/reference relation such that each name can only refer to one object. But it seems that in the case of families, one name can be used to refer to a whole host of differently-composed families, some of which overlap, and some of which do not. Consider two nuclear families living near each other, each of which contains one of two siblings whose family name is ‘Yu’. Both families are referred to in the area as ‘The Yus’ when the context doesn’t require further disambiguation, and, according to Sainsbury’s view that family names refer to families, a bare name ‘Yu’ refers to each family, regardless of any rôle the definite article might play. This is a natural case for homophonists to explain using their account of causal name individuation. And yet it is clear that the families are both called ‘Yu’ because they contain members of a larger family called ‘Yu’, and there are times when ‘Yu’ can be used to refer to both families together, and their common near-ancestors. If the homophonist were to account for every 17 And, of course, this is probably a very bad way of thinking of families, for all sorts of reasons. Not least that, in many contemporary societies, members of families increasingly do not share names with one another, and are not formed along patriarchal bloodlines. 86 3.3. The Argument from Family Names different referential use by positing a distinct name, we would be dealing with an implausible proliferation of explanatorily-connected homophonic names. A potential response to this problem for the homophonist view18 is observing that families are vague entities, and so this kind of extensional variation is to be expected. That an object has vague identity conditions is not a problem for reference. Places (and indeed people) are often such that their names can refer to lesser or greater geographical (spatial) areas on different occasions: ‘Los Angeles’ can refer to just the City of Los Angeles, or to the County of Los Angeles, or to the metropolitan area, and to gradations more or less of each at their boundaries. ‘Mark Sainsbury’ can refer to different physical quantities of the same man, withstanding loss of hair, or even limbs, and the gain of hair or weight. But the problem for families is not simply that their constitution is vague. Vagueness for a property or term is generally thought of as a phenomenon whereby there are points which definitely count as falling within the extension, and points which definitely don’t count, and the points in between are less clear, such that any particular, determinate boundary would be arbitrary. For example, in a classic sorites case, whilst n grains is definitely a heap, and n − 1000 grains definitely isn’t, there is no stage at which the removal of a single grain can definitely be said to change the heap into a non-heap, such that a boundary may be drawn. But this is not the case for families as we are thinking of them.19 Consider a family tree. Suppose one man at the top is definitely a member of the Goldstein family. His son is definitely a member of that family, because of the way family related-ness was defined. But for the same reason, the son’s son is also definitely a member. This continues, such that a Goldstein at the bottom of the tree, directly descended (down the male line), is also definitely a member of the same Goldstein family. But, unlike in a sorites sequence, we have not reached a false conclusion. It is perfectly plausible that these relatives are all members of the same family. The whole lot of Goldsteins is legitimately a single family called ‘Goldstein’, but so, as mentioned, are many smaller groupings of those people, which need not all even overlap with each other. The problem is not simply that there’s no non-arbitrary place to draw a line which limits a family, it is that there is nowhere along a patri-linear family tree at which a person cannot legitimately be counted as a member of the family, and that many smaller groupings along the tree also legitimately count as families. This is not an issue of vagueness in the way that the case of drawing the boundary of, for example, the green part of a colour line is, because, in 18 Suggested 19 Although, to me by Mark Sainsbury. of course, relatedness, and being in the same family, generally are vague in this way. 87 3.3. The Argument from Family Names that case, you eventually stray into definitively not-green. Sainsbury, then, must say more either about what kind of family each family name refers to, without invoking arbitrary or ad hoc divisions between families, or he must specify how and why one name can refer to differently-composed families in different contexts, perhaps by specifying the nature of the vagueness at play. 3.3.2 Which Individual? Family names present a further problem for homophonists. If not as they refer to families, then as they are used to refer to individuals who bear the family names. It is very common in a whole host of domains to refer to people by just their family name, for example, using just ‘Frege’ to refer to the philosopher and mathematician Gottlob Frege. In such cases it seems implausible to claim that a specific name unique to the referent is being uttered, since family names are shared by members of the same family. Indeed, Sainsbury is explicit that the full names of individuals contain names referring to their family: ‘John Smith’s name is a combination of his given specific name “John” and the specific name of his family, “Smith”’ (Sainsbury 2015, p. 197, n. 2). He also suggests that the family name contained in an individual’s name is not granted to the individual by baptism: he says of George (H.W.) Bush that his ‘dubbing involved just “George”. The family name “Bush” was applicable merely in virtue of his parents having that name, not in virtue of a dubbing’ (p. 197).20 Thus Sainsbury, and homophonists generally, cannot obviously explain how ‘Frege’ can refer to Gottlob Frege and not to his mother, Auguste Frege.21 Family Names as Abbreviations One solution would be to claim, apparently contra Sainsbury, that whenever one uses just a family name to refer to an individual, one is, in some sense, using an abbreviation of their full name, so the name has the semantic value of the person’s full name. Clearly, a speaker need not be aware of the full name of every person they refer to using a family name, particularly given that the position presumably entails that the unabbreviated version of a name contains middle-names (etc.) as well as first names, so even ‘Lucy Goldstein’ 20 Whilst it is plausibly true of Bush that he has the family name he does just because it was his parents’ family name, the point no longer generalizes throughout the English-speaking world. In the UK, when a birth is registered, the child may be given any family name the parent or parents wish. So, even if a child receives the family name of one of its parents, and even if the parents do not deliberate about this, there is a definite sense in which both a child’s first names and its family name must be decided upon, and the child is thus dubbed with the whole name. 21 This point is raised by Travis (1981a), who makes the point regarding Frege and his uncle Dieter. 88 3.3. The Argument from Family Names is an abbreviation if the bearer’s full name is ‘Lucy Susan Levy Goldstein’. Note with respect to this that Frege’s full name was ‘Friedrich Ludwig Gottlob Frege’. This need not be a problem, however: one generally does not need to know what a word is an abbreviation of—or even that it is an abbreviation— in order to use it meaningfully. For example, many English speakers are, I imagine, unaware that the word ‘Gestapo’ is (or was) an acronym for ‘Geheime Staats-Polizei’, or that ‘Haribo’ is an acronym for ‘Hans Riegel, Bonn’. But that does not prevent them from using these words to refer to the Nazi secret police, and the German confectionary company, respectively. However, in the case of these names, there are not multiple, different long-form names, all of which have the same acronym, i.e. ‘Haribo’ isn’t also an acronym for ‘Hannah Riegel, Bonn’.22 Something must determine that one utterance of ‘Frege’ is an abbreviation of ‘Friedrich Ludwig Gottlob Frege’, but another is an abbreviation of ‘Auguste Wilhelmine Sophie Frege’, when the speaker is unaware of either abbreviation. I will consider two possible answers to the question of what might determine which full name ‘Frege’ abbreviates on a use. One view would be that the use of ‘Frege’ is causally connected to the full name of one family member, and it is that full name that it is an abbreviation of. This is clearly an overly complicated picture. The homophonist already thinks that the full name is causally related to its referent (or to a baptismal event)—this is what determines its reference—and it is hard to see why any causal chain relating the abbreviation to the full name would not essentially be an extension of the referential chain. Thus, we can simply say that a particular use of ‘Frege’ refers to Gottlob rather than Auguste in virtue of its causal connection to him (or to other uses of his name, or an abbreviation of it, to refer to him). At this point, there are two more possible routes to take: one is to maintain that there is one family name ‘Frege’, shared by Gottlob and Auguste; and the other is to claim that in fact, there are two distinct individual names ‘Frege’ that abbreviate the two full names of Gottlob and Auguste. On the first route, that there is one family name ‘Frege’, we now have a causal chain picture of reference without homophony. The very same name can refer to different individuals (though only within one family so far) in virtue of a causal connection, but only on an occasion of use—the family name cannot refer to an individual person tout court. So, within a homophonist 22 Of course, many acronyms, initialisms, and other abbreviations do abbreviate more than one expression (or, perhaps, there are many homonymous abbreviations), but these generally differ from family names in that they are not related to each other: it is by chance that ‘am’ can mean both ‘ante meridiem’ and ‘amplitude modulation’, but it is not by chance that ‘Frege’ can refer both to Gottlob and his mother. Nor is it by chance that ‘Sellars’ can refer to both Wilfrid Sellars, and the unrelated Peter Sellars. 89 3.3. The Argument from Family Names picture, we’ve found a way of referring quite properly with a single name to multiple objects, within a restricted domain. But, if that were accepted, it seems that there would be little reason to maintain the homophonist individuation of names at all. Why should multiple referents of one name ‘Frege’ via causal connection be sanctioned, but not multiple referents of one name ‘Friedrich Ludwig Gottlob Frege’ by the same mechanism? I suggest that there would be no good reason to block such a move, and we are left with a position upon which proper name reference occurs only on an occasion of utterance, and, on each occasion of utterance, the reference is determined by a causal connection of the utterance to the referent, or a baptism of the object. Something like this is the indexicalist view of Pelczar and Rainsbury (1998). So this approach to family names appears to undermine homophonism, if not the causal picture of reference. Multiple Specific Family Names The second route, that there are two distinct individual names ‘Frege’, looks more likely to preserve homophonism, and seems likely to be the most coherent option for the homophonist. However, it requires us to abandon the view that the members of a family share a family name. On this route, since ‘Friedrich Ludwig Gottlob Frege’ is a specific name unique to Gottlob Frege, the abbreviation ‘Frege’ that refers to him is also unique to him, and is effectively just another specific name (in virtue of being an abbreviation of a specific name). Thus, any use of the name ‘Frege’ just is part of the causal chain linking its long-form version to Gottlob. However, there is no particular connection between Gottlob’s specific name ‘Frege’, Augusta’s specific name ‘Frege’, and the specific family name ‘Frege’, other than that they are ‘instantiations’ of the same generic name. Moreover, it is not clear what the relation of the individuals is to the specific family name. Certainly, Gottlob and Auguste both belong to the Frege family, which bears that specific name, but they do not bear it; it does not form part of their full names, although it could be that they were given their specific names ‘Frege’ because they instantiate the same generic name as their specific family name. This route has the problem that it seems that one can refer to an individual using just their family name—or, according to the view in hand, a name that is distinct from, yet homophonous with, their family name—without ever having been connected to the causal chain associated with (or forming) their specific name. For example, suppose a speaker, Frau Großbart, sees a young Gottlob Frege with his mother. Frau Großbart knows Auguste’s name, but not Gottlob’s. She could say something like ‘Ah! Young Herr Frege 90 3.3. The Argument from Family Names is out for a walk with his mother’, and successfully refer to Gottlob, even though she inferred his name from his mother’s specific name (or perhaps from the family’s specific name, itself inferred from Auguste’s) and has never been inducted into the causal chain of ‘Friedrich Ludwig Gottlob Frege’.23 A response that the causal-homophonist may be tempted to make is to claim that Frau Großbart was in fact able to tap into the correct causal chain, simply by brute force. Even though she was not explicitly inducted into the chain, in virtue of knowing that it was very likely that a child’s family name would instantiate the same generic family name as its mother’s (perhaps more than likely in nineteenth-century Mecklenburg-Schwerin), she was able to infer what the child’s specific family name would be, and thus she was able to use that with the intention of using it in the same way as previous speakers. This response to the problem of causally unconnected uses of a specific name is not especially convincing. The point seems to rely upon the idea that Frau Großbart is inferring that there is a specific name ‘Frege’, which is an abbreviation of a longer specific name, and with which she was previously unfamiliar, and then, crucially, she intends to use that name as her community does. Sainsbury is clear that such an intention is key to the acquisition of a specific name (2015, pp. 205–6).24 Both parts of this idea seem implausible as a description of the speaker’s actual psychology. Whilst the homophonist may be able to argue that Frau Großbart’s assumption that Auguste and Gottlob share a family name is in fact an inference about different specific names, it does not seem plausible that her—presumably object-oriented—intention to use that family name to refer to Gottlob can be reinterpreted as an intention to use a specific name in the way that it has been previously used. Problems for the idea that each person’s family name is a distinct specific name unique to them also arise from conjunctive referential uses of joint family names such as ‘Lucy and Michaela Goldstein’. According to the view 23 Evans (1973) raises essentially the same problem for the causal theory of reference more generally, with his example of Wagera Indians and their alleged practice of naming children after relatives in a systematic manner, according to the order of their birth (p. 195). Evans suggests that Kripke, whose presentation of the causal theory (or picture) he is criticizing, will respond that the ‘denotation of a name in a community is [. . . ] to be found by tracing a causal chain of reference preserving links back to some item’ (ibid., p. 195), even though ‘a knowledgeable speaker may excogitate a name and use it to denote some item which bears it without any causal connexion whatever with the use by others of that name’. This response, which does not presuppose the homophonist individuation of names, looks like it will entail a very much reduced version of the causal picture than that suggested by the homophonist. If participation in causal chains is not the only means by which a speaker can refer using a name, and it may not even be the primary means, then the causal theorist is making a rather meagre claim about one means of referring. It is also not clear what could be forming the causal chain of uses of a name if most of the uses are not causally connected to one another. Evans’s response also does not address the issue for the homophonist that a speaker presumably needs to be using the specific name of the intended referent in order to refer to them, it will not do to use their mother’s specific name. 24 As he is for the earlier incarnation of that process, initiation into a name-using practice (2005, pp. 113–6). 91 3.3. The Argument from Family Names in hand, ‘Goldstein’ can be any one of three (relevant) distinct specific names (or an abbreviation of them), but it can’t be more than one at once, because the semantic value of each is just its referent, i.e. Lucy, Michaela, or the Family Goldstein. In order to account for these kind of constructions, and, indeed, as we have seen, the use of family names generally, the homophonist must complicate their account of proper name reference significantly, adding new levels of name-type and offering ad hoc explanations. For instance, to account for ‘Lucy and Michaela Goldstein’-type cases, homophonists might say that ‘Goldstein’, as it appears in the phrase, is just part of the name ‘Michaela Goldstein’, which appears in long(ish)-form, and ‘Lucy’ appears as an abbreviation of ‘Lucy Goldstein’, the ‘Goldstein’ part of her name not appearing. However, this would fail to capture the apparent meaning of such expressions. There is a strong impression that ‘Goldstein’ is functioning to complete both names, and is able to do so because it is Lucy and Michaela’s shared family name. Alternatively, the homophonist might claim that there is some kind of ellipsis at work. However, it is very unclear how this would work. The phrase would presumably be elliptical for ‘Lucy Goldstein and Michaela Goldstein’, but since the second ‘Goldstein’ is just part of the longer specific name ‘Michaela Goldstein’, whose semantic contribution is just its referent, it’s unclear how it could license the elision of part of the name ‘Lucy Goldstein’. The semantic contribution of that name is just its referent, and, there is no semantic connection whatsoever between the two names. The elision (or appearance of it) might be explained as a form of zeugma or semantic syllepsis,25 but it bears few, if any, of the characteristic marks of these figures of speech. Moreover, any claim that there is ellipsis or syllepsis at play would also fail to explain why ‘John and John Smith’ would not be used to refer to two unrelated John Smiths. This is because, if the specific names which exemplify or instantiate the generic name ‘Goldstein’ in both Lucy and Michaela’s names are distinct, and the specific name referring to their family is also distinct, there is nothing in their semantics or meta-semantics that connects their names to the exclusion of the names of members of any other family called ‘Goldstein’—all merely have specific names that instantiate the same generic name. A third approach, that goes someway towards responding to the problems of the first two, is to claim that there is some kind of pragmatic rule that indicates how to form conjunctive names of relatives who have homophonic family names in virtue of being related. The rule might state that one can put 25 As employed by Flanders and Swann in the song ‘Madeira, M’Dear’: “as he hastened to put out the cat, the wine, his cigar, and the lamps”; “she made no reply, up her mind, and a dash for the door”. 92 3.3. The Argument from Family Names together the names of the two individuals and delete the first individual’s family name to create a plural name for both individuals that does not bear a rigorous semantic analysis of its component parts. Whilst such an explanation addresses the problem, it is ad hoc, side-stepping the issue of what the semantic status is of the occurrence of ‘Goldstein’ in the construction. However, the analysis effectively provides an additional way that family names can function to the already complicated list comprising an over-arching generic name; a specific name for each family; a specific name for every member of every family bearing the name; and, potentially, both a generic version of just the family-referring names to explain the connection between those specific names, and one for the specific names of each particular family. As we have seen throughout the discussion of causal-homophonism, accommodation of very ordinary quirks in the use and implementation of proper names requires severe complications of the homophonist account of reference. Contextual Determination of Family Name Reference A second possible answer to the question of what might determine what full name a family name abbreviates, when used to refer to an individual, is context. This view would amount to something like the following: The specific family name ‘Frege’, when used bare, can abbreviate the names of any of several members of the family who bears it, and which of those family member’s names it abbreviates depends on the context in which it is used. For example, if we are in a philosophy seminar on logicism, ‘Frege’ will probably abbreviate and thereby refer to Gottlob, but if, in 1870, we were in the school led by Auguste Frege, ‘Frege’ might well refer to her. This view is very odd. This does not sound like the way a homophonist should conceive of abbreviations. It also seems very likely to collapse into a slightly less odd view, and one suggested to me by Mark Sainsbury:26 Bare ‘Frege’, as used to refer to an individual, is not an abbreviation of that individual’s full name, but rather, has its reference determined contextually from amongst the members of the Frege family. On this view, again, it is clear that ‘Frege’ does not refer to any individual Frege tout court. This might be thought of as some kind of deferred reference or metonymy (cf. Nunberg 1993). However, the use of family names to refer to individuals seems different to other examples of metonymy, since, although Sainsbury maintains that family names refer to families, generally in English we cannot use bare family names to refer to a family, but rather, have to form definite article phrases in which the family name appears to occur 26 The suggestion was made somewhat ‘on-the-fly’, and should not be taken as reflective of Sainsbury’s considered view. 93 3.4. The Argument from Pluralism about Reference Determination: Conclusion predicatively: ‘The Freges’, ‘The Frege Family’, or even ‘The Family Frege’. Bare uses of family names are assumed to refer to individuals whose full names contain the family name: ‘Frege’ can refer to Gottlob or Auguste, but not, as it stands, their family. This seems to put pressure on this phenomenon being an example of deferred reference or metonymy, since those are assumed to rely on a term which could refer to one object referring to another in virtue of some relation holding between the objects. If Sainsbury’s micro-contextualist picture is not an example of deferred reference, then we just have the idea that a family name picks out a family, and when used bare, is contextually determined to refer to one member of that family. So, once again, it appears that we’re forced into a kind of localized use-theoretic contextualism, for family names only. But if we have this picture for family names, nothing is to stop us, in principle, from adopting the same kind of view of first names: when I use ‘John’ referentially, my reference is selected from amongst all Johns by the same kind of contextual factors that selected from amongst the Freges. And if we are granted this for first names, the whole idea of the homophonist individuation of names, and even the causal theory of reference, gives way to a contextualist, particularist account of proper name reference. A further problem for the position arises from the fact, mentioned above, that the family name ‘Frege’ either refers to a vast swathe of distantly related Freges, or there is some contextual restriction, such that a particular use can refer just to a subset of that swathe. If it’s the former, then the contextual mechanisms that allow bare ‘Frege’ to pick out Gottlob are doing a very large amount of work. If it’s the latter, then context is entering the picture twice: once to restrict the particular (e.g.) nuclear Frege family, and then to pick an individual from amongst them. In either case, it looks like the causal picture of reference is doing very little work in getting us to the individual referent of ‘Frege’. 3.4 The Argument from Pluralism about Reference Determination: Conclusion In §§3.2 & 3.3 I have offered arguments to the effect that causal-homophonism, as an account of proper name reference determination, is too inflexible to capture the vagaries of proper names use. The account must make a variety of implausible or overly-complicated modifications to the original picture in order to accommodate these vagaries, perhaps the most striking of which is the homophonist individuation of names, criticized at length in §3.1. As I discussed at the beginning of §3.2.1, the causal picture of reference is well suited to accounting for deferential mouthpiece or parasitic cases, but, 94 3.4. The Argument from Pluralism about Reference Determination: Conclusion although it may be true that most referential uses of a name refer to a bearer of that name, and most users will have learnt the name from another user, many uses do not fit neatly into the deferential model. Rami (2016) makes the point that, according to the causal pictures of Kripke (1981) and Sainsbury (2005), with each utterance of a name, a speaker intends to use the name to refer to whatever the name refers to (even if, after an initial induction into a name-using practice, a speaker no longer remembers the specific use that their first use was parasitic upon), but in many cases this is implausible. When a speaker is introduced to a person and says ‘Hello, Dave’, they presumably intend just to refer to the person. Similarly if a speaker knows the referent of a name very well: When I talk about London, referring to the capital of the UK, where I’ve lived for ten years, I intend to refer to the city,27 and have no (de dicto) intention to refer to whatever is the bearer of the name. I also observed, in §3.2.2, several types of case in which it appears that context trumps a causal connection in fixing the reference of a use of a name, most notably either because there is insufficient information for the parasitic take up of a name, so no communication, or the wrong communication takes place, or because a mistake has been made with regard to causal chains, but the context is sufficient to rectify it. In such cases, it seems that the causalhomophonist is likely to claim that the causal chain tying the particular specific name used to the baptism of its referent is still playing a rôle, and that the name still refers, in some manner, to its bearer, regardless of context, or what is communicated. In response to this claim, I ask: why, in the face of a whole variety of different ways and situations in which names appear to be used, insist that only one mechanism is at play in determining reference proper? I recognize, of course, that in many, if not most, cases, we use the name we do in a particular situation because the thing we want to refer to bears that name. Naming practices, which do seem to proliferate in part by being passed from speaker to speaker, are very important to the way we use names, and seem to be primarily what separate proper names from referring terms like pronouns and demonstratives. But homophonists do not make it clear why such practices should be the only factor in determining reference, particularly when, in order to account for shared names—an apparently ubiquitous and universal phenomenon—they have to posit a homophonist individuation of names, which I criticized at length in §3.1. Moreover, inserting name-using practices, and naming events into an account of reference determination threatens the universality of an account of proper name reference by tying it contingent and messy social conventions. Rami (2016) argues for a pluralist conception of the determination of the 27 Though, of course, I do not think my intention is relevant to my reference. 95 3.4. The Argument from Pluralism about Reference Determination: Conclusion name-bearer relation, and thus, according to him, of the determination of proper name reference. He claims that causal theories of reference, such as those of Sainsbury (2005, 2015), make too great an idealization about proper name use when they put baptisms at the heart of their theories. Although Sainsbury (2005) acknowledges that a name-using practice can appear to emerge and develop over time, such as in the case of unintentional nickname giving (p. 111) in which an initial use of a common noun to describe someone initiates a practice of using that term as a proper name, he claims that the initial use (or the initial referential use) counts as a baptism. He maintains that whether or not a novel referential use of a term counts as a baptism will depend on whether or not the use initiates a genuine name-using practice. But Rami observes that it is implausible to claim that such practices come into being with a single use in this way. He claims that it is much more plausible to see these kinds of practice as emerging over time, as multiple speakers use them. Moreover, Rami claims that this is how most name-using practices in fact begin, rather than with the paradigmatic baptismal speech act of ‘I hereby name this NN’, and that even such an act would fail to attach a name to an object, even if all the conditions were apt (the speaker had the relevant authority, etc.), if a practice of using the name for the object were not established.28 Although I think that Rami’s pluralism does not go far enough in abandoning the speaker reference/semantic reference distinction (referred to by Rami as pragmatically- and semantically-licensed reference), I think he is exactly right in observing the importance of not trying to make the many different ways in which names are used fit into a single, idealized model. It is similarly important, however, not to simply adapt one’s model by trying to fit in a disjunction of the many different social factors relevant to naming and name use. I will discuss the problems of this type of approach, typical of indexical accounts of proper name reference, in the next chapter. 28 Rami weakens his position somewhat by apparently claiming that legal acts of naming (presumably whatever process is involved in granting a name such that it’s recognized by the state) are made by paradigmatic baptismal speech acts. At least in the UK, and I imagine in other European countries, parents decide to give their child a particular name through a process of deliberation, and using sentences such as ‘Let’s call her Mary’, which Rami explicitly denies is an act of naming, and when they’ve decided, that name becomes the child’s legal name by being written on a birth certificate. 96 Chapter 4 Indexicalism ‘No physical theory of local hidden variables can ever reproduce all the predictions of quantum mechanics’ Bell’s Theorem I ndexicalism is a family of views about the semantics of proper names that suggest that names are, in some respects, similar to indexicals such as ‘I’, ‘you’, ‘here’, and ‘today’ that can only refer upon occasions of use, since their reference is sensitive to aspects of the context of use. Some indexicalist views propose that names are in fact a sub-class of indexicals, as a syntactic-semantic category, whilst others simply suggest that names have certain key features in common with indexicals. As I use the term, I consider indexical views to make at least the following claims about names: (i) Names refer only upon an occasion of utterance; (ii) The reference of (a use of) a name depends on the context of use; (iii) The semantics of a name contains an appeal to context or to non-linguistic factors as norms of their use or reference. Claims (i) and (ii) separate indexicalism from causal-homophonism, which maintains that names are context-insensitive, and refer tout court: whether they are being used or not. Since indexicalism regards names as contextsensitive expressions, it is, in my view, to be greatly preferred to homophonism. However, most indexicalist accounts of names still aim at providing a semantic 97 view of reference. This is recognized by claim (iii), which maintains that the way in which context is involved in the use of names is governed, at least in some minimal way, by their semantics. As it stands, this claim is very weak, primarily in order to include the heterodox, but self-avowedly indexicalist, account of Rami (2014), discussed in §4.1.3. Most other indexicalist views could be captured with a stronger alternative to claim (iii): (iii*) The semantics of a name specifies that some aspect(s) of context or nonlinguistic factor(s) determine(s) the reference of (a use of) the name. (iii*) makes names look much more like traditional conceptions of indexicals, whereby the way that context affects their reference is specified by their semantics. It may still be that the exact details of how context is involved are under-specified by the semantics, as we will see in §§4.1.1 & 4.1.2, but nevertheless, indexicalist accounts, especially those embracing (iii*) have in common that they are attempting to provide truth-conditions for proper names, or attempting to show how they could, in principle, be provided. The manner in which indexicalist views of proper names go about expressing their particular picture of the relation between names and context varies. Some indexicalists, such as Cohen (1980), do not attempt to provide an explicit or formal account of the semantics of names, merely discussing some contextual factors that they claim are relevant to reference determination. Such accounts may not make any connection between indexicals and names, but still fall into the category of indexicalism according to my broad definition. Other indexicalists, such as those I will discuss in detail below, are much more explicit about what the semantics of names looks like, often providing a formal or semi-formal interpretation. Such interpretations frequently involve specifying two semantic levels within the meaning of a name. Recanati (1993) does this by stating that names have, as part of their meaning, a specification that they just contribute their reference to the truth-conditions of an utterance containing them, and they also have a mode of presentation of that reference, which specifies how the reference is determined. Pelczar and Rainsbury (1998) and Rami (2014) both avail themselves of a two-level semantics provided by a Kaplanian-style logical framework, as is frequently applied to the analysis of other indexicals (Kaplan 1989a,b). This framework distinguishes between the character and the content of a word. The content provides the extension or reference of the word (relative to a world and time of evaluation), and the character determines the content, relative to specific aspects of the context. Character is usually represented or modelled as a function from context (which is represented as a tuple of contextual parameters such as speaker, time, world of context, place, and, potentially, 98 any other aspects of context deemed to be relevant to the determination of the content of a particular expression), and content is represented or modelled as a function from world and time of evaluation to an extension. In the simplest cases, in which the world and time of evaluation do not alter the extension of a term, as is the case for names, given the standard analysis of their rigidity, or in which the world of evaluation is the same as the world of context, the content of a word effectively just is its extension, and its character is just a function from context to extension, that is, a specification of the—potentially context-sensitive—truth-conditions of the term. As mentioned in the introduction to part II, there is a parallel between the homophonist and indexicalist approaches to proper names, and two approaches to the meaning of predicates and other general terms and expressions that are discussed in relation to the work of Travis in §2.3.1. The first approach— the parallel of homophonism—involves treating the meaning of expressions as contents or extensions, which Travis argues against by showing that the meaning of an expression can be held constant whilst the truth-evaluation varies, indicating a variation in extension. The second approach, parallel to indexicalism, involves treating the meanings of expression as character-like, specifying particular contextual parameters that effect the extension of the expression in a determinate and systematic way. The Travisian response to this approach, as discussed in Davies (2014) and §2.3.1, is to demonstrate that any proposed character for a term is either too general, and so does not specify a particular way that the context must be to determine an extension—and so the character does not provide any truth-conditions at all—or it is wrong, and the specified parameters can be held constant whilst the extension or content can be varied. My response to the indexicalist approach to names will be similar to the Travisian response to the second approach to general expressions. Travis (1980) also provides such an approach for referential terms—in particular, what is essentially an indexicalist account of the reference of descriptions. In §4.1, I will discuss three prominent indexicalist accounts in the literature, and discuss problems with them, the foremost of which, in each case, will be that they fail to specify any determinate contextual conditions on the determination of proper name reference. They are not only thereby too general, but they are incomplete: they do not provide an explanatory, truth-conditional theory of proper names, although they give the strong impression of being engaged in the project of doing so. In §4.2 I will offer a general discussion of the problem with providing determinate contextual parameters for reference, and argue that the indexicalist accounts I discuss fail even to provide general constraints on name reference. I then discuss, in §4.3, a particular parameter common to 99 4.1. Proponents of Indexicalism all three accounts—that of name/bearer relations—and argue that it could be neither a determining factor, nor a constraint, on name reference. Finally, in §4.4, I suggest a particular interpretation of indexicalism upon which it could avoid many of my criticisms. 4.1 Proponents of Indexicalism The first indexicalist approach to proper names seems to be Burks (1951), who explicitly relates names to indexicals. Cohen (1980) could also be considered an early proponent. Although he does not make an explicit connection to indexicals, or provide any kind of formal consideration of the semantics, he does mention various contextual factors that he maintains effect the determination of reference. Cohen is an influence on Recanati (1993), who puts forward probably the first indexical account that contains a reasonably detailed (semi-)formal indexical treatment of the semantics of names in English.1 Pelczar and Rainsbury (1998) provide an account that it similar to Recanati’s, though making use of a slightly different formalism and slightly different contextual factors. These two versions of indexicalism are probably the most cited and discussed in the literature on names, so I provide extended discussion of them in this section of the thesis. I also discuss Rami (2014) at length, since it represents a version of indexicalism that is both thoroughly informed by previous versions, and interestingly divergent from them.2 Tiedke (2011) also provides a formal indexicalist account of proper names, but I will not discuss it here since it is similar to Pelczar and Rainsbury (1998) in its use of contextual factors and its assumptions about them, only differing significantly in being designed to accommodate fictional names. 4.1.1 Recanati Indexicals Recanati (1993) does not claim that names are indexicals, but leaves it an open question of little significance whether they are a category of indexical, or merely have many close similarities to indexicals. Recanati defines indexicals thus: An indexical expression t in an utterance S(t) indicates that: 1 Zimmermann and Lerner (1991) provide a slightly earlier indexical account, but it’s in German, so unfortunately beyond the reach of my analysis. Rami (2014) suggests that it is quite similar to Recanati’s account, so subject to similar criticisms. 2 I have Rami (ibid.)’s detailed survey of indexicalism, and my exposure to the author’s general painstaking approach to literature review to thank for the breadth of this chapter. 100 4.1. Proponents of Indexicalism There is an object x which is F (= linguistic mode of presentation), such that the utterance is true if and only if x satisfies S ( ). The mode of presentation which is one aspect of the meaning of t makes a certain object contextually identifiable, and the feature REF, which is another aspect of the meaning of t, presents the utterance as true if and only if this object has the property expressed by the predicate in the sentence. For example, ‘You are G’ indicates that there is a person to whom the utterance is addressed (linguistic mode of presentation), such that the utterance is true if and only if this person is G. (Recanati 1993, p. 140) Thus, Recanati does not treat indexicals explicitly in the Kaplanian manner, but there are strong similarities: The feature REF entails that indexicals are referential terms whose only semantic contribution is their reference, just as, for Kaplan, the content of an indexical is just its reference. Similarly, the linguistic mode of presentation that determines reference contextually on each occasion of utterance is analogous to a Kaplanian character, mapping contextual parameters to contents. Names with No Meaning Recanati’s discussion of proper names begins with consideration of the view, held in some sense by the causal-homophonists, and certainly associated with direct reference theorists in the vein of Mill and Kripke, that names have no meaning in the manner of other kinds of expression—such as might specify or determine a referent—but only refer to their referent (ibid., p. 136). However, according the machinery of direct reference Recanati has specified in his (1993) prior to his discussion of names, such a position is not possible since he has defined directly referential expressions as conveying the feature REF within their meanings. He now considers that the meaning of every name might consist solely of the feature REF, producing 1*: 1* The meaning of a proper name is nothing over and beyond the feature REF. By virtue of its meaning, a proper name NN indicates that there is an entity x such that an utterance S(NN) is true iff x satisfies S( ), but it does not present this entity in a particular way. Hence a proper name has no ‘meaning’ in the sense of a (linguistic) mode of presentation of the reference. (ibid., p. 137) Recanati states that a strong reading of 1*, whereby all proper names are equivalent and indicate only that the speaker has a referential intention, without providing any clues as to what the speaker intends to refer to, is 101 4.1. Proponents of Indexicalism unacceptable as an account of the meaning of names. This is because he claims that names very obviously do provide a clue as to what they refer to: an utterance of ‘Cicero was bald’ is significantly different to ‘Aristotle was bald’, because, unlike what Recanati calls ‘pseudo-names’, such as ‘thingumybob’ and ‘what’s-their-name’, there are conventions associating those names with particular objects. Adding Bearers to Meanings Recanati moves on towards his eventual indexical account via a position reflecting a weaker reading of 1*, 1**, which establishes his view of nameusing conventions: 1** By virtue of its meaning, a proper name NN indicates only that there is an entity x such that an utterance S(NN) is true iff x satisfies S( ). NN also indicates which entity y is such that y = x, but this indication is not part of the meaning of the name: it is conveyed by the name by virtue of an extralinguistic convention, namely the convention which associates NN with its bearer. (Recanati 1993, p. 138) 1** still states that there is nothing more to the meaning of a name than the feature REF, but it also states that the name indicates—independently of its meaning—which entity it refers to, and it does this in virtue of an extra-linguistic, i.e. social, convention associating the name with its bearer. The idea that the conventions that associate names with their bearers are not linguistic conventions encoded in semantics, but, rather, social conventions that form part of the context of an utterance of a name is the most significant feature of Recanati’s indexical view of names, and is essentially what separates it from homophonist views.3 This is also the most significant point that will be carried over from indexicalism into my pragmatic view. Recanati argues that this idea is reasonable because being competent in a language clearly does not require one to know all the name bearers, given a broad understanding of what constitutes a natural language. Whereas, one presumably does need to know both the meanings of a great number of other words, and how to use a proper name, once one has been familiarized with at least one nameusing practice. Indeed, it seems plausible (if unlikely) that a speaker could be completely competent in a language without being aware of the bearers of any proper names. Once Recanati has used 1** to establish the idea of name-bearing conventions being social, rather than linguistic, phenomena, he rejects it on 3 As Recanati discusses on pp. 143–152. 102 4.1. Proponents of Indexicalism the grounds that it does not place those conventions close enough into the meaning of names. He claims that the assignment of referents to (uses of) names in virtue of these conventions is part of what defines the category of names: having one or more bearers is essential to being a proper name. This connection to name-bearing conventions must be encoded in the meanings of names. Thus, Recanati reaches his indexical account of names: A proper name NN indicates not only that there is an entity x such that an utterance S(NN) is true iff <x> satisfies S( ) [feature REF], it also indicates—simply by virtue of the fact that it is a proper name—that x is the bearer of the name NN, i.e. that there is a social convention associating x with the name NN. (Recanati 1993, p. 139) The indication that there is a social convention associating the name with its bearer is the mode of presentation of the name. This is a general, linguistic convention that holds for all names: For each proper name there exists in principle a social convention linking that name to a definite individual, called its bearer. This individual is the referent of the name. (ibid., p. 139) Unlike this linguistic convention, the social conventions it refers to are specific to each name, and, as Recanati highlights in a parenthetical note, there is sometimes more than one convention associated with each name ‘as when a name has more than one bearer’ (ibid.). Despite Recanati’s apparent under-appreciation of the ubiquity of shared names, he clearly intends that his position should be able to account for them. However, the way in which his mode of presentation goes about selecting a unique referent is in conflict with this intention. It states that for each name there is a bearer, which is the referent of the name. But, given that Recanati accepts that some names have more than one name-bearing convention associated with them, it is unclear at this point how just one bearer can be selected on an occasion of use. He says that ‘[t]he reference of NN is the entity which is called NN in the context of utterance’. He goes on to suggest that, although there may be multiple bearers of a name generally, on an occasion of utterance, the context will provide just one bearer, which is reminiscent of a Kaplanian-style logic of indexicals, in which a context is a tuple of parameters providing a single speaker, time, world, place, etc.: What happens in this sort of situation is that there are two distinct nameconventions associating the same name type, say ‘Gareth Evans’, with particular persons (the regretted British philosopher and an Australasian politician). These conventions are not appealed to in the same contexts of 103 4.1. Proponents of Indexicalism utterance, and this is why the reference may vary from one token to the next. The reference of a particular token of ‘Gareth Evans’ is the person who is related to the name type ‘Gareth Evans’ by a name-convention operative in the context of utterance of this token. (Recanati 1993, p. 141) This conception of bearer-in-a-context, determined by the ‘name-convention operative in the context’ allows for a very simple indexical character of names, but leaves much under-specified. Problems for Recanati One significant problem is that, at least according to an intuitive notion of context of utterance, there can be more than one name-convention operative, and whilst, in such contexts, speakers may generally go to lengths to disambiguate different referential uses of the same name, for example, by supplying additional descriptive information, at least according Recanati’s account as I have provided it, there is no room for this information to affect the reference of the name. Consider an utterance of ‘Gareth Evans is dead, unlike the other Gareth Evans’. There will be circumstances in which this is a perfectly comprehensible utterance, and a hearer will understand what each token utterance of ‘Gareth Evans’ refers to. In this case, ‘the other’ is providing evidence that a different name-convention is being invoked. But this needn’t be the case. Consider ‘Lucy went to the shops whilst Lucy listened to the radio on her phone’. Nothing in the content of such an utterance obviously entails that there is more than one referent of ‘Lucy’, since the activities ascribed are simultaneously compatible, and there is nothing to indicate a shift in context, in the intuitive sense of ‘context’, but there seems to be a presupposition that the repetition of a name within such a short period (rather than introducing a name and then using an anaphoric pronoun) is indicative of the uses of the name having different referents.4 In order to account for such circumstances, in which a single name is clearly used to refer to different objects, apparently within the same context, Recanati must either allow that contexts, in his technical sense of ‘context’ can shift very quickly, in which case we might justifiably ask what determines these context shifts, or that further pragmatic factors can affect the reference of a use of a name. In either case, it appears that further factors are at work, and that the indexical account of names, as Recanati presents it, is under-specified; it does not give us anything approaching a full story of the determination of proper name reference. 4 This quirk of the use of names in English—which may well also occur in other languages—is particularly complicated, and is hard to characterize in any systematic manner. Indeed, it appears to defy any such characterization. 104 4.1. Proponents of Indexicalism In fact, Recanati recognizes that the mode of presentation of an indexical, which specifies the relation that must hold in context between the indexical and the referent (the bearer-in-context relation for names), may, in some contexts, not identify a single object because more than one object stands in the correct relation to the expression in the context. In a footnote, he says: [T]he linguistic mode of presentation of the reference does not always uniquely determine the reference, even with respect to a context of utterance. In other words, the linguistic mode of presentation of the reference is not always a complete ‘character’ in Kaplan’s sense—a function from context to content[. . . ]. When the linguistic mode of presentation is ‘incomplete’ in this sense, a powerful pragmatic mechanism involving something like Grice’s maxims of conversation is required to determine the reference on the basis of the indication provided by the expression mechanism (see, in particular, Sperber and Wilson (1986)). (Recanati 1993, n. 6, p. 153) Recanati thus admits that further pragmatic factors, in addition to the notion of bearer-in-a-context, are required to determine the reference of a name in certain cases—and, notably, factors that other theorists regard as significant not for determining reference itself, but only for aiding hearers in understanding what is being referred to (e.g. Neale (2004, p. 85–87); Sainsbury (2015, p. 198)). However, it is clear that, even in cases in which Recanati wishes to maintain that only one name-convention is operative in the context, something must determine that that convention is operative. What determines this will—presumably—be a combination of factors such as circumstance of utterance, topic of conversation, what mutual acquaintances interlocutors have, perceptual or spatial proximity, salience, and the interests and aims of speakers. Again, Recanati’s indexical account looks suddenly very under-specified. Indeed, unless he actually intends the notion of being a unique bearer-in-acontext to be primitive—and his footnote regarding shared names, and comments later in the book, discussed below, suggest he does not—it is simply not a truth-conditional account of proper name reference since it does not provide determinate truth-conditions. Whether Recanati intended to offer something approaching a complete truth-conditional account, or whether he believes that there is, in principle, a way of codifying the pragmatic mechanisms involved in determining reference into determinate truth-conditions is unclear. Either way, it appears that, with his indexical mode of presentation for names, he has given us a constraint upon the reference of names: that the referent of a use of a name must be a bearer of that name. However, Recanati tells us little about what it is to be the bearer of a particular name. In §4.3, I will call 105 4.1. Proponents of Indexicalism into question both the idea that the bearer relation is a suitable candidate for playing the rôle of parameter in an indexical view of names, and the idea that it is a constraint or necessary condition on the reference of names. A Problematic Account of Bearers In the chapter that follows the introduction of Recanati’s (1993) indexical account of names, he offers an elaboration of the notion of bearing a name in a context that he initially left under-specified. The elaboration arises in an attempt to defend the meta-linguistic element of his account—that a name refers to the bearer of that name—from criticisms from Kripke (1981) of similar views. Specifically, Kripke’s criticism (p. 69) that a form of descriptivism that maintains that a proper name such as ‘Socrates’ should be analyzed as ‘the individual called “Socrates”’ is circular, since, superficially at least, it appears to say that what a speaker calls ‘Socrates’ is whatever is called Socrates.5 Recanati’s response is extremely complicated. He observes that his indexical view relies on, but does not explain, the notion of being a bearer of a name or being called by a name (Recanati 1993, pp. 159–160). If that notion were explained in terms of what token utterances of a name refer to, then clearly it would be circular to explain the reference of token utterances of names in terms of the notion. Recanati instead introduces name types, a pseudohomophonist idea that he explicitly links to the specific names of Kaplan (1990) (Recanati 1993, p. 166, n. 5). Rather than existing as individual lexical items in a whole language, name types refer to particular objects just amongst local communities of speakers. So, name-using practices centre around the reference of a name type within a community, and an object bears a name in a context if it is the referent of a name type that is appropriately related to the name (instantiates it, perhaps) in the community selected by that context. And so, an utterance of a name refers to an object if that object is the referent of the name type of which the utterance was a token, as determined by the context of utterance (ibid., p. 166, n. 5). This leaves us wondering how a name type gets a reference in a particular community. According to Recanati, who draws from Evans (1982) and Loar (1980), there are Putnam-style producer experts in each community who determine and sustain the reference of a name type with the reference of their token uses of the name. Recanati’s story about how objects bear names has severe problems, largely owing to its complicated nature, that I will not discuss in detail here. One problem is that the localized homophonism he introduces opens him up to 5 Kripke attributes this view to William Kneale, though it should be noted that it is very similar to the view of Bach (1987, 2002) and Katz (2001), and foreshadows recent predicativist views such as those of Fara (2015); Gray (forthcoming); Matushansky (2008). 106 4.1. Proponents of Indexicalism various problems for causal-homophonism. Another problem is with his reliance on producers to fix the referents of the name-types for the consumers in their communities, and the sharp distinction between them that this requires. It appears that producers do not themselves use name-types to refer, for this would risk circularity. They must use the names themselves, such that they could refer to anything with them. But this seems to require not only that they cannot be the producer for any more than one homophonic name-type, but that they cannot even be the consumer of a homophonic name-type. It is their uses of the name that create the name-type they produce, so nothing could plausibly distinguish their productive uses of the name from their consumptive uses of a name-type that is itself a specific form of the name. 4.1.2 Pelczar and Rainsbury The indexical account of names offered by Pelczar and Rainsbury (1998) is similar in many respects to that of Recanati (1993), though they explicitly state their view in terms of a Kaplanian-style logic of indexicals.6 The stated aim of the paper is to develop an account of proper names that has the benefits of existing direct reference account of names (by which Pelczar & Rainsbury mean causal-homophonism)—most obviously rigidity—but that is also able to deal with the various puzzles of propositional attitudes7 with the ease of descriptivist theories. The account produced treats names as rigid indexicals in the vein of ‘I’ and ‘you’. In the Kaplanian terms assumed by Pelczar & Rainsbury, this means that names have a constant content—the content returns the same referent in every world of evaluation (rigidity)—and a nonconstant character—the character can return different contents in different contexts of utterance (indexicality) (Pelczar and Rainsbury 1998, pp. 293–4). Dubbings-in-Force If names are indexicals, and indexicality is assumed to be explained using a Kaplanian framework, then there must be a feature (or features) of the context of utterance that governs what content is returned. Pelczar & Rainsbury propose adding the feature dubbings-in-force to the tuple forming the context: 6 Pelczar & Rainsbury cite Recanati in a footnote towards the end of their paper, but only as one example of a mode of presentation-type view of names that is, they claim, unable to accomplish all that their indexical view can. It appears that they overlook the similarity of the accounts, and how easily Recanati’s mode of presentation could be adapted in a Kaplanian-style character. 7 Problems such as Frege’s puzzle: If Hesperus is Phosphorus, and proper names just stand for their referents, why is that ‘Phosphorus’ cannot be substituted salva veritate for ‘Hesperus’ in an utterance of ‘The ancients believed that Hesperus is Hesperus’ (Frege 1892)? 107 4.1. Proponents of Indexicalism A dubbing is a speech-act whereby a name acquires a referent, and a dubbing is in force in a given context if in that context the item that was dubbed in that dubbing bears the name it received in that dubbing. (Pelczar and Rainsbury 1998, p. 294) Like the homophonists, then, Pelczar & Rainsbury are assigning semantic significance to naming events, and assuming that such events are what create and maintain name-bearer relations. And, like Recanati, they are appealing to the idea of an object bearing a name in a context such that it might not bear it in a different context. Indeed, the only significant differences between Pelczar & Rainsbury’s account of the meaning of names, thus stated, and Recanati’s account, seems to be that Pelczar & Rainsbury explicitly make use of a Kaplanian-style logic of indexicals, and that they provide a slightly more detailed view of what constitutes bearing a name. Problems for Pelczar & Rainsbury Since their accounts are so similar, Pelczar & Rainsbury run into essentially the same issues that Recanati does, and address them in the same way. They are thus subject to the same criticisms. They provide no indication of what might determine when a dubbing is (or isn’t) in force, although, unlike Recanati, they recognize this fact, saying: The dynamics of dubbings-in-force can be complex, and we shall not attempt to provide a systematic way to decide which dubbings are in force in a given context. (ibid., p. 295) It is not clear from this whether or not it is believed that, in principle, a systematic way to decide which dubbings are in force could be provided. Given the tenor of the sentence quoted, it seems that Pelczar & Rainsbury might believe that it could be, suggesting that they do indeed intend their account to be, in principle, fully truth-conditional. However, since they neither provide any indication of how this might be done, nor suggest that they take dubbings-in-force to be primitive, we are left, just as by Recanati, with an under-specified and indeterminate account of proper name reference that, at best, only provides constraints on reference, and lacks explanatory power. A further problem that Pelczar & Rainsbury’s indexical account has in common with Recanati’s is that there are clearly likely to be contexts in which dubbings are in force such that more than one object bears the same name in that context. Pelczar & Rainsbury use the example of using the name ‘George Bush’ in the context of talking about US presidents to illustrate the issue. Although they are significantly more forthcoming about it than Recanati, their response is extremely similar, claiming that, in such a situation, 108 4.1. Proponents of Indexicalism one of the competing dubbings must be brought to prominence in order to determine a unique referent for the name (in that use). This might be achieved by a variety of mechanisms. One important factor in this raising to prominence might be relevant features of the conversation (if any) of the context of utterance of the name. (Pelczar and Rainsbury 1998, p. 295) Just like Recanati, Pelczar & Rainsbury mention Gricean conversational maxims as potentially being particularly relevant in bringing a dubbing to prominence. But, again, it is unclear whether the ‘relevant features of the conversation of the context’ are supposed to be systematizable such that they could be added to the formal context to provide a fully-explanatory determinate truth-conditional semantics. If not, and this time it is perhaps less likely that this is the authors’ intention, then the aim of providing a semantics for proper names that determines their reference—as is suggested by the use of a Kaplanian-style framework in which context is a formal notion—is clearly undermined. The need to raise a particular dubbing-in-force to prominence in order to arrive at a determinate reference also raises the question of why prominence was not simply built into the framework to begin with: Given that some mechanisms, presumably pragmatic mechanisms, are required to determine which dubbings are in force for a particular name in a context, and then, in many cases, pragmatic mechanisms are also required to raise one such dubbing to prominence, it would seem simpler just to posit most-prominent-dubbingin-force as a feature of the context, determined by pragmatic mechanisms. That Pelczar & Rainsbury do not do this, preferring a two-stage approach, might suggest that they do indeed believe that the first stage, but not the second stage, could be systematized.8 It might alternatively be indicative that Pelczar & Rainsbury conceive of the fixing of reference in a very hearer-centric manner, in which it is clear to a hearer, apparently without having to make any conscious inferences, that only certain objects are possible referents of a 8 Pelczar (2001), which considers some general criticisms of indexicalism, acknowledges that his indexical formulation of a semantics for names is a ‘veneer of simplicity’ that ‘covers an immense complexity that is simply gestured at by an appeal to “dubbings in force”’ (p. 143). He goes on: Any effort to state precisely the conditions under which a given dubbing of a thing with a particular name is in force (in any context) quickly leads to the consideration of multifarious factors, semantic, pragmatic, and even extra-linguistic, that, to say the least, resist tidy encapsulation. (p. 143) However, despite appreciating the complexity of the context-sensitivity of name reference, Pelczar does not see this as reason to abandon the idea of there being systematic truth-conditions. Instead, he notes that many indexicals have equally untidy relationships with context. This is, of course, true, but it should not be assumed the contextual determination of those indexicals is systematizable. Later in the paper, Pelczar seems to acknowledge even that there may be under-determination in the rules determining reference for some names and some indexicals. He suggests that, in those cases, the under-determination is resolved by ‘speaker discretion’ (p. 149–150), by which, I presume, he means speakers’ intentions. 109 4.1. Proponents of Indexicalism speakers use of a name—these are the dubbings in force. Often, so it might seem, there will only be one such object, and the work is done. But at other times, ambiguity might remain, and in these cases it seems reasonable to invoke such contextual conversational mechanisms as Gricean maxims. Whilst this might be an accurate picture of how hearers respond to the use of a name on different occasions, if we are interested in describing the rôle of context (in the non-formal sense) on reference, it is unnecessary to divide things up in such a manner, especially if there is no pretence of offering a genuinely truthconditional semantic account. It would be far simpler to observe that various different contextual mechanisms play a rôle in determining which bearer a name refers to on different occasions. However, I will argue, in §4.3, that even the mention of bearers or dubbings in the character of a name is misguided. 4.1.3 Rami The indexical account of proper names developed by Rami (2014) is, in my view, by some distance, the most advanced indexical-type view in the current literature. This is primarily because Rami engages deeply with the prior literature, offering criticisms similar to those I have laid out above. He then specifically attempts to develop a view that avoids the problems of those prior accounts, though, as a result, it has somewhat more limited aims. He also develops his view slowly and painstakingly, addressing at every step the kind of subtle contextual issues that affect the use of proper names, and which I have tried to highlight as the undoing of semantic views of name reference. However, whilst my response to the powerful effects of pragmatics on the use of names is to abandon the idea that their (semantic) meaning has anything to do with the determination of their reference, Rami, influenced by Predelli (2012, 2013), introduces to the semantics of names a use-conditional level, in addition to the truth-conditional levels of character and content. Using this apparatus, Rami develops ‘a novel version of a formal constant approach of the indexical view on names that is based on [Kaplan’s logic of demonstratives] and holds that names are a distinctive kind of indexical expression’ (Rami 2014, pp. 120–1). Although we share somewhat similar aims and motivations, there is a certain sense in which Rami’s approach to the problems of reference determination is the opposite to that which I will take in chapter 6 of this thesis. I follow Travis in pushing reference, and other phenomena that are relevant to the truth- or correctness-evaluability of utterances (those things that go towards determining what thought an utterance expresses) into the pragmatic level, leaving in the semantics just those things that are context- 110 4.1. Proponents of Indexicalism invariant, to play some rôle in, perhaps, compositionality and/or logical form. Rami keeps whatever is truth-conditional within the semantic level, but largely rejects the strong distinction made by most other reference theorists between semantic and speaker reference, preferring to maintain a single phenomenon of reference (for the most part) that is truth-conditional. In order to distinguish between those referential uses of names that seem to be correct and acceptable—what would normally be considered their semantic reference—and those that are in some way incorrect, although they achieve reference—usually conceived of as mere speaker reference—Rami posits a use-conditional layer within the semantics of names, which need not affect the reference or truth-conditions of a use of a name, but distinguishes semantically correct uses from semantically incorrect uses by means of a contextual constraint. Learning from the Mistakes of Others The first few sections of Rami (2014) consist of discussion of existing views about the semantics of proper names, including criticism of both causalhomophonism, and various forms of indexicalism.9 In taking into account the various problems of those accounts, Rami develops a pluralist account of reference determination, similar in motivation to the pluralism I discussed in §3.4.10 In fact, Rami’s pluralism might well be described as a disjunctivism—a fact he gladly acknowledges in conversation—since, rather than allowing an indeterminate and indefinitely large class of mechanisms of determination, as I am inclined to, he admits just three mechanisms: demonstrative identification, descriptive identification, and parasitic identification (ibid., pp. 127–30). The first form of identification includes cases in which a speaker refers to an object using a name in virtue of some form of demonstration; the second form includes cases in which a description is used to fix the reference of a use of a name, either explicitly or implicitly, such as when a speaker says ‘Barry, the one who really likes wine’; and the third form includes mouthpiece-type cases in which a speaker fixes the referent of their use of a name simply in virtue of using the same that another speaker has used. Another point Rami takes from discussion of other positions, and the indexical positions of Recanati (1993) and Pelczar and Rainsbury (1998) in particular, is the non-viability of the notion of bearing a name in a context. 9 Rami defines indexicalism rather more broadly than I have, including various accounts that I have categorized as either predicativist or variablist. I have not discussed these views on the grounds that they do not commit themselves in principle to a distinctive form of reference determination for names. For example, they subsume the question into that of the reference determination of definite article or complex demonstrative expressions (Burge 1973; Elbourne 2005; Geurts 1997; Sawyer 2010), or simply by taking a stand only with regard to the logical form of names, and not about referential mechanisms (Cumming 2008; Dever 1998). 10 Which cited Rami (2016) extensively. 111 4.1. Proponents of Indexicalism Rami’s criticism of this notion is somewhat more specific than mine. Whilst I criticized other indexicalists, in §§4.1.1 & 4.1.2, for failing to say anything about how it is determined what bears a name in a context, Rami takes it that both Recanati and Pelczar & Rainsbury implicitly rely on salience to do this job (he takes Pelczar & Rainsbury’s prominence to be salience by another name). Rami observes that, if the other indexicalists are invoking salience, they are doing it in a rather odd way: instead of claiming that a particular bearer of a name is salient in a context, they claim—according to Rami—that a name-bearing convention is salient (Recanati), or that a dubbing-in-force is salient. Social conventions and acts of naming, unlike objects that bear names, are not the kinds of things that, in the ordinary way of things, language users are aware of being salient or not (Rami 2014, pp. 137–8). However, Rami goes on, even if it were objects themselves that were required to be salient in order to refer to them, it is clear that, in fact, an object need not be salient, in an ordinary sense of ‘salient’, in order to be the referent of a name. Nor is being a salient bearer sufficient to be the referent in many cases (ibid., p. 138). Problems with Truth-Conditions With these considerations in mind, Rami begins to develop his own account, making use, like Pelczar and Rainsbury (1998), of a Kaplanian-style logical framework. Rami’s last attempt, before introducing the notion of a useconditional level of semantics is as follows: (IXC.1) [[ Alfredn ]]c,hw′ ,t′ i = the object that is identified demonstratively, descriptively or parasitically in cw in respect to the occurrence n of ‘Alfred’ by c a and that is a bearer of ‘Alfred’ at ct . (Rami 2014, p. 140)11 Here, c is the context, hw′ , t′ i is the content, a world-time pair, cw is the world of the context, c a is the agent of the context, and ct is the time of the context. n is an integer that indexes occurrences of a name in a sentence, such that, at the level of logical form, names are represented syntactically as ordered pairs of a proper name and an integer (or numeral). So, for example, if a sentence contains two occurrences of the name ‘Paris’, whether they refer to the same object or to two different objects, the representations of those occurrences in the logical form of the sentence will be h‘Paris’, 1i 11 I have made some very small changes to Rami’s notation. 112 4.1. Proponents of Indexicalism and h‘Paris’, 2i respectively.12 Rami introduces this indexing of occurrences in order to be able to allow that multiple uses of the same name within a single context might have different bearers. If the interpretation of a name were simply the object identified in the context that bore the name, this would not be possible. Given this motivation, merely indexing occurrences of names within a sentence would be insufficient, they would need to be indexed throughout any discourse taking place within a single context. This would be an easy enough change on Rami’s part, however. By introducing the idea that the referent of a use of a name must have been identified in some manner by the speaker, Rami is able to avoid introducing anything like the bearer-in-the-context or the most prominent dubbing-inforce, and so proposes a character for names quite unlike the indexicalists already discussed. The only constraint on the object of the speaker’s identification is that they bear the name used at the time of the context. The introduction of identifications means that Rami appears not to rely on non-specific contextual mechanisms, such as Gricean conversational maxims, that are usually taken to be cues for a hearer’s comprehension, and that Rami criticizes Recanati and Pelczar & Rainsbury for using in their accounts of reference determination. However, whilst Rami might not rely on an unspecified notion of context to fill in gaps in the same way that Recanati and Pelczar & Rainsbury do, it is unclear how his identifications would work without either invoking a great deal of context, or relying on unstated metaphysical assumptions about the nature of demonstrations, descriptions and the repetition of names. Rami’s (IXC.1) provides an account of names that operates solely at the semantic level of character and content. He goes on to discuss how the account deals with semantically correct, and incorrect, uses of names. Given the account, just those uses of names, the reference of which accords with (IXC.1), are semantically correct uses. Rami identifies three forms of infelicitous, or otherwise imperfect, uses of names (Rami 2014, p. 141): incomplete uses, which are not accompanied by an act of identification; empty uses, which are accompanied by an act of identification that fails to identify a single object; and improper uses, which are accompanied by an identification of an object that does not bear the name used. Of these, Rami claims that improper and incomplete uses of names are semantically incorrect, but that empty uses can be perfectly acceptable, as is widely discussed in the literature. Rami suggests that the simplest way of accommodating semantically correct, empty uses of names is to adapt the Kaplanian-style framework he uses into a negative 12 The use of sequential positive integers is, presumably, arbitrary. Rami simply needs a means to differentiate each occurrence from any other. 113 4.1. Proponents of Indexicalism free logic in the manner of Sainsbury (2005). This would have the effect of rendering all positive utterances containing empty name false.13 However, Rami observes, that in terms of truth-conditions, there is no difference between incomplete and empty uses of names: neither have referents, and so utterances containing either will come out false. (IXC.1), as a wholly truth-conditional account of meaning, thus does not correctly distinguish between semantically correct and incorrect uses of names. A further problem for (IXC.1) in this respect is that it rules out two types of name use that Rami deems to be semantically correct, on the grounds that they appear to refer to objects that are not, in some sense, bearers of the name used at the time of use (Rami 2014, p. 142). The first of these uses are to refer to potential future bearers, such as when one tries out a new name for an object, with the intention of establishing a new name for it. Rami uses the example, borrowed from Ziff (1977), of an utterance of ‘Whiskers is in the kitchen’ in response to the question ‘Where’s the cat’, where the cat in question does not yet bear a name (as far as the interlocutors know). Rami claims that such a use of a name is perfectly correct, but need not be itself an act of naming, since a practice of calling the cat ‘Whiskers’ may not arise. The use is thus semantically correct, although it refers to an object that does not yet bear the name used. Rami’s second type of case involves uses of names that were formerly borne by their referents. Such cases are discussed in detail by Saul (1997) and Zimmermann (2005), and will feature largely in §4.3. Rami observes that sentences such as ‘Leningrad has more than one hundred thousand inhabitants’ seem unacceptable (though not, apparently, false), but both ‘In 1944, Leningrad had more than one hundred thousand inhabitants’ and, ‘In 1944, St Petersburg had more than one hundred thousand inhabitants’ seem fine.14 The problem seems to be that former names generally cannot be (or should not be, or are not) used to refer to their former bearers at reference times15 subsequent to the time they ceased to bear the name, whereas names that are currently borne can be used to refer at any reference time. This may, of course, be too much of a generalization, but it is clear that at least in some cases such restrictions on felicity exist. Rami observes that the issue does not appear to be truth-conditional, since it does not seem to be false that Leningrad has more than one hundred thousand inhabitants, but an 13 I will not discuss here the value of such a move. that the city currently known, officially, as ‘St Petersburg’ was called ‘Leningrad’ in 14 Note 1944. 15 The notion of reference time, as distinct from speech time and event time, introduced by Reichenbach (1947), indicates the time at which an object is being talked about, indicated by tense and aspect, and/or temporal adverbial phrases. So ‘In 1944’ indicates that the reference time of the subsequent occurrence of ‘Leningrad’ is 1944. 114 4.1. Proponents of Indexicalism utterance to the effect would still be incorrect or impermissible in some other way. Presumably, we might also want to allow certain identities to hold between Leningrad and St Petersburg, since the city has, in some ways at least, remained the same city throughout its many twentieth century name changes, and this would lead to problems if assertions that were true of St Petersburg were false of Leningrad, especially since many of those true of Leningrad also seem to be true of St Petersburg. Again, then, Rami claims that this question of semantic correctness and incorrectness must be dealt with at a use-conditional level. A Use-Conditional Semantics for Names For these two reasons, together with considerations about what the logical (truth-conditional) entailments of uses of a name should or shouldn’t be, Rami develops an account of the semantics of proper names that contains both a truth-conditional and a use-conditional level. He states: What we need [. . . ] is a specific minimal formal characterization of the rigid content and indexical character of names that allows us to transfer all components of the meaning of a proper name that concern the determination of semantically correct referents of proper names to the level of use-conditional meaning. (Rami 2014, p. 147) To this end, Rami proposes a partial assignment function, relativized to contexts and occurrences, to assign referents to occurrences of names at the truth conditonal-level (or not, in the case of empty names). This function, N, takes ordered triples of the world of context, cw , the name itself, and an integer marking the occurrence of the name, n, as arguments, and returns an object in the domain of discourse.16 This gives us: (IXC.2) [[ Alfredn ]]c,hw′ ,t′ i = N (hcw , ‘Alfred’, ni) (ibid., p. 147) The use-conditional level is introduced as a contextual constraint, (CC), on possible contexts of use, defining just the semantically correct ones for the use of a particular name: 16 Rami does not mention what the domain of discourse is for any particular use of a name, but since he is clear that non-salient objects can potentially be referred to using names, we might assume that every object is in the domain of every discourse. 115 4.1. Proponents of Indexicalism (CC) c is a semantically correct context of use of the name ‘Alfred’ in respect to its occurrence n if an act of demonstrative, descriptive or parasitic identification is performed by c a in cw in respect to n whose unique target is, if any, identical to N (hcw , ‘Alfred’, ni) and this target, if it exists, satisfies at least one of the following additional conditions: (a) it is a bearer of the name ‘Alfred’ relative to ct ; (b) it is not a bearer of the name ‘Alfred’ relative to ct , but it is used in cw by c a with the intention to introduce a new bearer of ‘Alfred’; (c) it is not anymore a bearer of ‘Alfred’ relative to ct , the loss of the target as bearer of ‘Alfred’ was not due to an (implicit) shift in bearer-hood from one object to another, and the occurrence n of ‘Alfred’ in cw is in a sentence whose truth-value evaluation relative to ct depends on a time s such that this target was a bearer of ‘Alfred’ at s in cw . (Rami 2014, p. 148–9) Problems for Rami Rami claims, correctly, I think, that his account avoids the problems that he identifies for the various other accounts of names that he considers, but in doing so it is unclear what we are left with. Semantic reference, as conceived of by homophonists (and probably other indexicalists) has effectively been done away with, since there are no constraints on reference at the truth-conditional level. All we have there is a stipulation that a particular use of a name returns a particular referent in a context. Anything resembling the speaker reference/ semantic reference distinction comes at the use-conditional level, and what we have there is effectively a list of conditions specifying which ways of using a name are acceptable in some way. This some way is semantic according to Rami, but presumably only in the sense that it is encoded in the contextinsensitive meaning of the name. The lack of a speaker/semantic distinction at the truth-conditional level is, in my view, a positive feature of the account. However, in separating any conditions on reference from truth-conditions, and making them constraints rather than criteria, Rami has effectively—and, presumably, intentionally—depleted his account of any story about how the reference of names is determined. Recanati and Pelczar & Rainsbury state that the semantic reference of a use of a name is the bearer-in-the-context or the object of the most prominent dubbing-in-force, respectively, even though they don’t specify what it is to be one of those things, but in Rami’s final 116 4.1. Proponents of Indexicalism formulation of the semantics of a name, we don’t get anything like that. He tells us when a use of a name is semantically correct, and gives us a function, which is presumably only a representation or model of reference, but he does not state why or when the function maps to what it does. My criticism, if it is a criticism, of Rami may seem odd here, since my aim in this thesis is to develop an account of proper name reference that does not impose upon it necessary and sufficient conditions (or strong constraints). Moreover, my primary criticisms of semantic views of reference have been that the conditions they provide impose on the reference of names have excluded good cases. Rami appears to impose no conditions on reference, only conditions on semantically correct reference. The problem is, however, that he puts reference in the semantic meaning of names at all. Unlike Recanati’s REF, which just states that there is a referent—that the truth-conditions are singular (Recanati 1993, p. 17)—Rami’s (ICX.2) is a function that provides a referent (except in the case of empty names) at the semantic level, but without any guide to how this happens. On the view I will develop in chapter 6, the semantics of a name provides no guide to the reference of a use of it, but reference is conceived of as a phenomenon at the pragmatic level: names have no truth-conditional content at the semantic level. It seems very likely that Rami intends that speakers’ own determinations of reference,17 mentioned during his discussion of pluralism, are what are in fact doing the work here— we know that Rami thinks that there are at least three semantically correct ways that a speaker can determine a name’s reference—and these might plausibly be considered either pragmatic mechanisms, or a feature of presemantics. The function (ICX.2), then, might simply be a means to represent that at the semantic level: to introduce the reference to the semantics such that it can contribute to truth-conditions, and can be interpreted at the useconditional level. Rami has informed me in conversation that he requires the reference to be introduced to the level of truth-conditional semantics in order that various inferences be possible. So, perhaps Rami does indeed intend such acts to occur purely at the pragmatic level—all to the good—but given that, within his framework, truth-conditions are semantic, and the semantics of names does not incorporate a means of involving such pragmatic mechanisms, specified or not, it is currently rather unclear how reference fixed in such a way could affect truth-conditions. Rami might respond to this kind of criticism by adding to his function N an argument for something very under-specified such as ‘pragmatic mechanisms that determine reference’. This might allow him the freedom to be very liberal about how reference contributes to the truth-conditional level, whilst 17 Something he has confirmed in conversation. 117 4.1. Proponents of Indexicalism providing a mechanism for getting it there. His account would then be in the state for which I criticized Recanati and Pelczar & Rainsbury in §§4.1.1 and 4.1.2, such that it gives the appearance of attempting to provide truthconditions for names, but fails to do so because the contextual input is underspecified. Of course, in Rami’s case this would be intentional. Moreover, for Rami, the truth-conditional semantics would be wholly under-specified, since all it would state is that there is a reference specified pragmatically by extra-contextual (in the formal sense of ‘context’, c) factors. This would be something similar to Recanati’s 1*, and, at this point, the claim is essentially that proper name reference is pragmatic, but there are certain use-conditional constraints upon what is semantically correct. What, then, of Rami’s (CC)? The first point to note is that (CC) is doubly disjunctive: there is a tripartite disjunction of permissible means of reference determination, and a tripartite disjunction of conditions on bearer relations. This leaves one with the worry that the nine total constraints Rami codifies might just be those that he happens to have thought of. The bearer conditions in particular seem to be the result of an admirable, but limited consideration of the vagaries of the use of past and future names. To this end, Rami acknowledges that his account may appear to be ad hoc, and to lack theoretical unity (Rami 2014, n. 93, p. 149). He responds that both disjunctive lists may be incomplete and require additions, but that the apparent lack of unity that this engenders is not a flaw in the account, but a symptom of the multi-faceted nature of language use. Indeed, Rami invokes Wittgenstein: It is one of the Wittgensteinian lessons about natural language that we should not in general expect some kind of unification or deep explanation if we are interested in how natural languages work and how their expressive tools function. (ibid.) With this point, I agree wholeheartedly. However, I take it to be motivation not to attempt to include the kind of considerations found in (CC) in the semantics of names at all. Whilst Rami is content for his lists to continue to grow, I find it implausible that anything approaching a determinate list could exist. Moreover, if the semantics of a word is, as I take it to be, the contextinvariant content that speakers learn when they learn that word, it seems unlikely to contain the highly contingent social factors to which Rami appeals. Consider, Rami’s Leningrad example. If the historical considerations that lead some sentences containing ‘Leningrad’ to be felicitous, and others not, are contained in the semantics of the word, then speakers who, for whatever reason, continued to use the name after it was officially decided to change the city’s name back to ‘St Petersburg’, were making a mistake with regard to the 118 4.2. Problems with Parameters semantics of the word. Old Party diehards who continued to use ‘Leningrad’, knowing the name had been changed, were not simply making a social or political statement, but making a semantic mistake. This seems implausible if one considers such speakers using the name between themselves: the name functions in exactly the way it did in the days of the Soviet Union18 and it would seem hopelessly naïve to inform a diehard using ‘Leningrad’ that the city’s name had actually been changed, so she should alter her practice. We might claim that such speakers still regard the city as bearing the name ‘Leningrad’, at least in a certain sense, and so regard their utterances as correct, but it is still not clear why this is an issue for semantics, rather than being merely a pragmatic social consideration. It is also unclear whether the temporal issues involving the temporal aspects of name bearing generalize to other languages, or even other dialects of English.19 Facing such an objection, Rami could claim just to be providing a semantics of proper names in English, or a dialect of English, but given the apparent universal occurrence of proper names playing very similar rôles, this would seem to significantly weaken his thesis. If we think of naming-conventions as social practices, or, at least practices that arise socially (or even legally (Rami 2016)), then it makes sense to think of the norms of these practices—which are presumably what (CC)’s (a)–(c) amount to—as norms of social practices, not of semantic or linguistic practices. Calling St Petersburg ‘Leningrad’ might be taboo in certain social interactions, or reveal certain political dispositions of the speaker, but this is quite different to claiming that a semantic mistake has been made. Refusing to use the name a person has chosen for themself, and instead using one they have asked you not to use, may well be a moral, personal, or social affront, but not obviously a semantic mistake.20 4.2 Problems with Parameters As mentioned in the introduction to this chapter, I maintain that there is a general and insurmountable problem with the programme of attempting to account for the reference of proper names by providing a character-like meaning that specifies particular contextual factors that will, on an occasion of use, determine reference. Indeed, I maintain that it is unlikely even that contextual constraints on the reference of names could be specified such that they held for every use of a name. I claim that, for any proposed set of contextual parameters alleged to determine the reference of names, they will 18 Zimmermann (2005) uses a similar example with Karl Marx Stadt and Chemnitz. is, of course, an empirical question. I have no evidence either way. 20 I will discuss this position further in §4.3 and chapter 6. 19 This 119 4.2. Problems with Parameters be either too general, such that they cannot predict what the referent of a name will be on an occasion, and will not decide between potential referents in all cases, or they will be wrong, such that they predict the wrong referent on some occasions. Although it is, of course, nigh on impossible to provide a general argument to this end, it seems very likely that, for any proposed character of a name, cases can be provided in which the reference of the name shifts without the proposed parameters changing. In §4.3, I will consider a parameter very commonly proposed as either a determinant of name reference, or a constraint on it: bearerhood. In this section, however, I will provide arguments against some other plausible parameters. One such parameter, which Rami accuses Recanati and Pelczar & Rainsbury of appealing to, although it is only implicit in their accounts, is salience. Salience is often appealed to in contextualist accounts of meaning, in literature on indexicals, demonstratives, and other types of expression. For the time being, let us put aside worries about bearerhood, and imagine a proposed character for names such that the referent of a use of a name is the most salient bearer of that name, or simply the most salient object that is plausibly the referent of the name. I aim to show that such a character will fail because salience is neither a necessary nor sufficient condition on the reference of names. 4.2.1 Salience Salience is a particularly vague notion. With regard to contextualism, there seem to be at least two ways of interpreting salience. The first is as just one kind of contextual factor, perhaps among many, that might be relevant to the determination of the content of context-sensitive expressions. The second is as some kind of broad category for various more specific contextual mechanisms— a kind of placeholder for finer-grained factors. On the first interpretation, it is possible to show that salience, although presumably sometimes relevant to the determination of the reference of names, is neither necessary nor sufficient. And on the second interpretation, appealing to salience simply provides no kind of theory of name reference, amounting to little more than saying ‘context does the job’, but with the disadvantage that not all contextual factors can plausibly be covered by the term. It is relatively easy to come up with cases to show that salience, on the first, more specific, understanding, is not a necessary or sufficient condition on name reference. On this understanding, salience is something related to apparentness and pertinence. An object will be salient in a context if, for example, it is present, or conspicuous, or is obviously relevant to the topic of 120 4.2. Problems with Parameters conversation. But clearly, one feature of names is that, in certain circumstances, they can be used to refer to objects that are not present or already relevant. Consider a case in which Ben is talking to Tom about Stormont. They are alone in a room. Having tired of Stormont, Ben says, out of the blue, ‘I saw Lucy Campbell the other day’, referring to a mutual friend, who was in no way salient in the context or the discourse. This seems like a perfectly good case of reference to an object that was not salient prior to being mentioned (though, of course, she would be afterwards). The fan of salience may respond that, it is not salience alone that informs name reference, but rather a scalar parameter like ‘the most salient bearer of the name’, in the manner of Pelczar & Rainsbury’s ‘most prominent dubbing-in-force’. The Lucy Campbell referred to is thus the referent because she is the most salient of all Lucy Campbells in virtue of being well-known to both Ben and Tom, and them both knowing this. But we can construct a case in which the apparent most salient Lucy Campbell fails to be the referent: Suppose that instead of saying ‘I saw Lucy Campbell the other day’, Ben says ‘Have you read anything by Lucy Campbell? She’s the new columnist for The Point.’. Tom and Ben’s mutual friend must still be the most salient Lucy Campbell at the time the name is used, even if the salience shifts during the second sentence, so should be the referent according to the proponent of the salience parameter. But it is clear from the second sentence that Ben is meaning to talk about a different Lucy Campbell, and it seems fair to assume that he did refer to that different Lucy, even though it may not become clear to Tom until after Ben has uttered the name. This example shows that salience is not a necessary condition upon name reference, at least not in any simple or obvious way, since the object referred to was not salient, and also that it is not a sufficient condition, since there was a more salient Lucy Campbell who was not referred to. A view by which salience determines reference on the second interpretation of salience, whereby it is treated as something like a conflation of a variety of more specific contextual factors, is harder to address by cases, simply because it does not make any specific claims about how reference is determined. As such, however, it is clear that it is also too general to make any kind of predictive claims about reference, and so does not contribute to anything resembling a genuinely truth-conditional semantics for names. Claiming that such a broad notion of salience is involved in every case of reference determination, and that it is the kind of thing that could play the rôle of a semantic parameter in an indexical account of names, must either be some kind of promissory claim—that the rôle of context could be unpacked and codified—or it is to reject the idea that such a semantics determines reference at all, at which point the exercise becomes redundant. An indexical account 121 4.2. Problems with Parameters of names based on this notion of salience would have nothing to say about the second Lucy Campbell case, for example, unless it was simply assumed that the notion gave the right result in every case. Mount (2008) develops a salience-based account of demonstrative reference that seems to skirt somewhere between the two interpretations of salience. According to her account of the reference of demonstratives, the referent of a demonstrative upon an occasion of use is the object that is mutually recognized as maximally salient by conversational participants (p. 154). Mount does not apply this thought to proper names, and indeed appears to think that names are not context-sensitive (p. 156), but it will be instructive to consider whether her account of maximal salience amongst interlocutors could plausibly be applied to names.21 Mount claims that ‘an object is mutually recognized as maximally salient by conversational participants when all interlocutors have focused their attention on it’ (ibid., p. 154). This seems like a plausible account of what maximal salience might be, and Mount presents a variety of cases in which demonstratives are used whilst speakers and audiences are in various states of awareness or ignorance about various objects in the vicinity, and about which her salience-based view makes convincing predictions. However, her notion of salience is not plausible as a determinant of the reference of names because it gets the order of explanation wrong: proper name reference is, at least in some circumstances, a means of focusing the attention of interlocutors on an object, or otherwise bringing it to mind in some way. To claim that an object must be maximally salient in order to be able to refer to it with a name is either to get things the wrong way around, or it is redundant, since, in many cases, it is the referring to an object that focusses one’s audience attention on it. Names seem to be unlike demonstratives in this respect:22 An effective use of a demonstrative often requires that the intended referent be made salient by the speaker at around the same time, i.e. with a demonstration, or that the object already be salient in some way. But, whilst names can be used with a demonstration to indicate their reference—and I do think that such a demonstration can determine name reference in some circumstances—and can be used such that a salient object is their referent simply in virtue of being contextually compelling, they can also be used to refer to objects that were not, prior to their use, salient to the interlocutors. Clearly, given this difference between demonstratives and names, claiming that the referent of a name is the maximally salient object at the time of utterance cannot work. Instead, one of the options I suggested above, that the referent of a use of a name is the most salient bearer of that name, or the most 21 I will not consider whether it provides a good account of the reference of demonstratives. at least, names are unlike the way Mount claims demonstratives to be. 22 Or, 122 4.2. Problems with Parameters salient object that is plausibly the referent of the name might be considered. But it is unclear how these could be implemented. We might simply modify Mount’s account and say that the referent of a name is the object that is made the focus of the interlocutors’ attention when the name is used. But this tells us nothing about reference. There is no more reason to think that this is the order of explanation than the converse: that it is referring to an object that focuses attention upon it.23 Indeed, if certain pragmatic factors—which may vary from occasion to occasion—determine what interlocutors’ attention is focused on, then those same factors might also be what determines reference, but it adds nothing to say either that reference determines or is determined by attention focus. Furthermore, this account of name reference is unlikely to give any kind of clean results since, given any particular utterance of a name within a longer utterance, any number of objects might fleetingly be the focus of interlocutors’ attention, especially if more than one name is uttered, and it seems unrealistic to suppose that appreciating what a name has referred to requires focusing one’s attention on that object, however briefly. Alternatively to this modification of Mount’s view of demonstrative reference, we might turn to a different idea of salience: one that does not explicitly appeal to the focus of interlocutors’ attention following the use of a name. Such an account might entail something along the lines of ‘the most salient bearer of the name is the bearer of the name that, given the topic of conversation, location, interests and purposes of the interlocutors, proximal objects, and other contextual factors, is most likely to be appreciated by interlocutors as the referent of the name’. However, whilst this might be considered a perfectly adequate account of the most salient bearer of a name in a context, it clearly won’t do if it is also supposed to do duty as an account of proper name reference, since it appeals to likely name reference in the definition. Moreover, it is unclear to me how the salience of a bearer of a name could be accounted for without either requiring that a bearer already be present or apparent to interlocutors, which will clearly rule out good cases of name use, or mentioning what a use of the name is likely to achieve in the minds of the interlocutors. 4.2.2 Other Parameters What other contextual parameters might an indexicalist consider appealing to in the semantic character they provide for names? Recanati and Pelczar & Rainsbury both mention Gricean conversational maxims playing a rôle in differentiating between rival referents, though it does not appear that they 23 Thanks to Alex Clark for highlighting this point. 123 4.2. Problems with Parameters intend these to be codified into the semantics. Other options might include conversational topic, proximity, the interests and purposes of interlocutors, linguistic context, previous reference, general relevance, and knowledge or beliefs of interlocutors about various aspects of each other’s knowledge or beliefs. It is clear, I think, that none of the items in this list would, by themselves, be necessary or sufficient for name reference. It is easy either to think of potential counter-examples, or to see that a factor is not sufficiently determinate to play the rôle. Rather than stating that any particular contextual factors are necessary or sufficient for determining reference, the indexicalist might instead propose a disjunctive character, such that, although no one factor is necessary or sufficient, the whole disjunction is. This is a similar idea to Rami’s conditions upon semantic correctness for a use of a name. My main objection to this strategy is that such a list of possible contextual factors could, in principle, never be complete, unless the items on the list were extremely general, at which point, clearly the strategy in uninformative. Again, it impossible to demonstrate that no finite, determinate list of factors could be complete, but the strategy for showing that any particular list is incomplete is, as ever, to provide counter-examples. There are further problems with the idea of a disjunctive character, however. Firstly, there must, presumably, be some principle or principles contained in the character that adjudicate in cases in which particular factors are in conflict with each other with regard to determining a referent. Such a principle or principles may have to be quite complex. It is unlikely that something such as a simple majority system, or a weighting of the factors would be adequate, since in different circumstances, different pragmatic factors will be relevant in different ways, or not relevant. Indeed, once this is acknowledged, it is clear that there is a second-order problem of context-sensitivity: the particular way that context affects the reference of a name varies from context to context. A second problem with the disjunctive character approach is that the semantics of a word, according to the notion of semantics I am employing, is the kind of thing that must be encoded in some manner in the minds of language-users: it must be learnt by speakers when they learn the word. But an indefinitely long disjunction of pragmatic mechanisms, together with rules for their implementation, does not seem like a plausible candidate for what is learnt when a word is learnt. The ability to see how a name or other word interacts with the contextual circumstances in which it is uttered seems much more like something that comes with domain-general knowledge of the world and how things are with it on particular occasions than knowledge of the specific meanings of names or words. 124 4.3. Problems with Bearers 4.3 Problems with Bearers Most indexicalist views of proper names, and many other kinds of view besides, consider a minimal constraint on the semantic reference of a name to be that the referent bears that name, or has been so dubbed, or something very similar (e.g. Matushansky 2008; Pelczar and Rainsbury 1998; Recanati 1993; Sawyer 2010). This presents a problem if the views are supposed to offer a determinate and specific account of the determination of name reference, or at least to provide a constraint on it. The relation of bearing a name is highly occasion-sensitive: there are many ways to bear a name, and there are many ways to have been dubbed. Indeed, different ways of acquiring a name might well give rise to different ways of bearing it. This state of affairs might be ok if it appeared to be the case that name reference tracked the occasionsensitivity of bearerhood, such that a name only ever referred to objects that counted as bearers of that name on an occasion of use. However, I will show that this is not the case. Moreover, it is also not the case that names can only refer to objects that on other occasions count as bearers of them, or will, at some point, so count. 4.3.1 Many Ways to Bear, and Not Bear, Names One can bear a name legally, because it appears on one’s birth certificate; one can bear a different name because one’s parents use it to refer to one; or because one’s school friends or maths teacher used it.24 Similarly, one can fail to bear a name in many ways. And whilst on some occasions one might be the bearer of a particular name, on others one might not be. One can bear a name because one chooses to, and, conversely, not bear a name because one has chosen not to. Such decisions can lead to moral and social limitations on the use of certain names, but it is not clear whether they can create semantic or linguistic limitations except after the fact, in virtue of altering general usage. In this section, I am interested in how names can be used to refer. I will try to show that this is a different issue to how they should be used with regard to considerations that go beyond simply referring. We have already seen, in §4.1.3, examples in which it appears that a name can refer to object that does not count as bearing it. These are the cases discussed by Rami regarding past and future bearers: 24 Dolf Rami has objected to me that there are actually only two ways to bear a name presented here: bearing a name officially in some way, and bearing a name because there is a practice of referring to one with that name. This is, of course, one way to divide up ways. But, as Rami also pointed out to me, we can simply recast ‘ways of bearing’ in terms of what we would plausibly say that an object is correctly called in particular kinds of circumstance or situation. This type of locution captures what I will say below just as well. 125 4.3. Problems with Bearers # Leningrad is on the Baltic. # Byzantium is a major tourist destination. # Cassius Clay suffers from Parkinson’s syndrome.25 X Karl Marx Stadt is now called ‘Chemnitz’. X Whiskers, as I think I might call her, is in the kitchen. As Rami recognized, the first three of these cases are problematic in some manner, apparently because the objects referred to no longer bear those names. But, on pain of contradiction, this cannot be because the names used fail refer to anything. The problem is not that nothing is referred to in these cases so that the inscriptions come out false, or neither true nor false. Whatever is at fault with these cases, and I acknowledge that something is, the names used refer, or have the potential to refer, to objects that used to bear those names. In the last two cases, there seems to be nothing at all wrong with the inscriptions. One might imagine that, in the case of ‘Karl Marx Stadt is now called “Chemnitz”’, a prior, perhaps infelicitous, use of ‘Karl Marx Stadt’ might have occurred, to which this inscription is a correction, but this only seems to highlight that the name can be used referentially, although the city no longer bears it. One way to respond to this kind of data would be to claim that it is the existence of a name/bearer relation at reference time rather than speech time that is significant. Thus ‘Leningrad was the second largest city in the Soviet Union’ is felicitous because the city referred to was called ‘Leningrad’ for most of the time that the Soviet Union existed. However, this does not explain why current names can be used felicitously to refer to their bearers at reference times prior to their coining: ‘St Petersburg was besieged during the Second World War’. It might be that it is fine to use current names at any reference time, but past names can only be used at reference times at which they were current. However, not only is such an account rather convoluted, it still cannot account for entirely felicitous utterances such as ‘Karl Marx Stadt is now called “Chemnitz”’, in which a past name’s reference time coincides with its speech time. Moreover, it is hard to see how the proponent of such an account could explain how reference seems to work fine whatever the name or reference time, but felicity is significantly affected. Another way to respond might be to claim that names do not easily lose bearers, and that in all of these cases of apparent former name use, the names 25 This section was written prior to Muhammad Ali’s death. I have not adjusted tenses where it would affect the point of my examples. Though, in fact, this particular example might now be even more infelicitous: # Cassius Clay suffered from Parkinson’s syndrome. 126 4.3. Problems with Bearers are still in fact borne by their referents, at least in the contexts in which they are being used26 —as evidenced by the fact that their use is still possible in these contexts. However, this response seems unsatisfactory. It appears that the notion of name-bearing being appealed to by indexicalists is the same as, or very close to, the ordinary language kind of conception of bearing a name, or having a name, or being called by a name (as in ‘The city is called “St Petersburg”’). According to Recanati, the name-bearing conventions he appeals to are social conventions, and presumably are supposed to be the same social conventions that speakers are aware of as they track namings and the use of names. But, as far as I am aware, according to this ordinary conception of naming conventions, it is quite possible for the names of objects to change, such that one name stops being the name of an object, and another takes its place as the object’s name.27 Indeed, this seems like the natural explanation for what happened in the case of St Petersburg—a city whose name changed three times during the Twentieth Century. If this is not the kind of convention that Recanati or other indexicalists have in mind, then they must provide some account of the nature of the social conventions they are appealing to, such that objects can gain, but rarely lose, names, and which are largely insensitive to the authority both of recognized namers, and of general use. The bearer-reliant indexicalist might claim that objects can lose names, but only when their use has completely ceased. Or, at least, that they only lose names in all contexts when their use has completely ceased. We might then reasonably ask in what kinds of contexts objects still bear names that are widely considered to be merely former names. The answer clearly cannot be: those in which the name still refers to the object, on pain of begging the question. The idea that a name/bearer relationship is simply dead once its use has ceased is also strange. Presumably, there must be a long period in which the name is not used to refer to the object in question before it can be considered forgotten as a name for that object. Consider an ancient settlement, the remnants of which are buried deep underground, and which has been forgotten for so long that no one has referred to it by its name for thousands of years. It appears that the settlement can still bear the name: if an archeologist were to discover the ruins of the settlement, and find written evidence of its name, she could speak that name and refer to the settlement, and it would be natural to say that she had discovered its name. It might no longer be regarded as bearing the name, however: if the settlement had been discovered well 26 Recanati 27 Though, has indicated, at least in conversation, that this would be his preferred route. of course, objects can have multiple names simultaneously. For example, Derry/ Londonderry. 127 4.3. Problems with Bearers before the written evidence was discovered, it might have already been given a name by the archeologists, and if, when the original name was discovered, it was particularly hard to pronounce in modern English, the new name might stick and we would say ‘the settlement was called “such-and-such” by its inhabitants’, but not necessarily ‘the real name of the settlement has been discovered’. Of course, this may be very sensitive to context, and there would be perfectly good ways of talking by which the settlement did bear the name. It seems clear, however, that simply going out of use is not sufficient for a name to stop having a particular bearer. So this is not a route that the indexicalist can take to explain when a name loses a bearer, or vice versa. A further problem that the indexicalist must face if they wish to claim that the referent of a name must bear it at the time of use, is that they must come up with an explanation for why utterances such as ‘Leningrad is on the Baltic’ so often sound infelicitous, without appealing to names changing. They could perhaps appeal to different ways of bearing names and state that although St Petersburg changed its official name, the City officials (or the Kremlin) had no power to effect its status with regard to the social name-using convention associating it with ‘Leningrad’. It must then be claimed that the existence of any variety of name-using practice is sufficient for allowing reference, but, for some reason, current official naming practices are relevant to whether or not a reference to a city sounds admissible or strange. However, if this line is adopted, the indexicalist needs to say something about what it is about official names that is significant to the felicity—but not the reference—of utterances in some contexts, but not others. Note that there will be contexts, such as those in which the interlocutors are old Party diehards, in which there is no infelicity in using ‘Leningrad’ in present tense utterances. There also appears to be nothing special about names being ‘official’ if this means something recognized by law. Consider Muhammad Ali. He stopped bearing the name ‘Cassius Clay’ in popular discourse at some point after he adopted ‘Muhammad Ali’. Ali explicitly renounced his former name, and it appears that it is normal to talk about him changing his name, and—as just demonstrated—referring to ‘Cassius Clay’ as a former name. Certainly, utterances such as those of the form ‘Cassius Clay suffers from Parkinson’s syndrome’ will sound strange in many contexts, whilst those of the form ‘Muhammad Ali was born in Louisville’ won’t. If a racist sports reporter had continued referring to him as ‘Cassius Clay’ after Ali changed his name, and particularly after his new named had been widely adopted, the reporter would have been doing something incorrect: presumably committing a social and moral transgression. However, Ali never legally changed his name. So, in legal contexts, Ali never stopped being the bearer of ‘Cassius Clay’. If he had 128 4.3. Problems with Bearers given his name to the US Court of Appeals as ‘Muhammad Ali’ when he was appealing his indictment for draft dodging, he would have said something incorrect in the eyes of the law. It is clear that in both these Ali cases—the sports reporter and the courtroom—there is no issue with reference: ‘Cassius Clay’ can be used by the reporter to refer to Ali, even though this is no longer regarded by the public as his name, and people in the courtroom would have been able to refer to Ali using ‘Muhammad Ali’, even though that was not his name in the eyes of the court. The indexicalist who would maintain that the infelicity of uses of particular names in certain contexts is not due to referents not bearing those names in those contexts, must provide an explanation of these cases that does not simply appeal to official names being felicitous. Even if ‘official name’ is taken to include something like ‘the most widely used name’, assuming that legal names of people are also official, a further distinction must be made. Claiming something along the lines of ‘in non-specialist contexts, the most widely used name is the most felicitous’ is also problematic. Firstly, it is unclear how non-specialist contexts could be identified adequately. Secondly, it seems implausible that such a claim is true: infelicity is, presumably at least in part, subjective, and hearing it does not generally require awareness of how most other people use a name, though it does seem to require awareness of whether or not a name has been changed in one of various different ways. 4.3.2 Reference Without Bearers The problems for appealing to name-bearing as a parameter that contributes to the determination of reference of a name go further than the fact that objects can bear names in different ways and count as bearers in some contexts and not others in ways that do not affect reference. It also appears that there are perfectly good cases of reference that do not merely amount to a speaker willing that whatever name they are using refers to what they want it to. There may be many cases in which a speaker has simply mislearned a person’s name, but nevertheless, they may succeed in referring in many—though perhaps not all—situations. If the mistake carries on for long enough, and it becomes recognized by a variety of hearers that the speaker has an idiosyncratic way of referring to the person, we might wish to say that the mislearned name has become a name for the person, but this is by no means required for reference. Consider a case in which Lucy mistakenly believes that Harris is called ‘Farris’, simply in virtue of having mis-remembered his name, not because of any mistake about identity involving someone actually named ‘Farris’. Lucy attempts to introduce John to Harris by saying ‘John! Have you 129 4.3. Problems with Bearers met Farris here?’ and indicating Harris. It is clear that Lucy can have referred to Harris with her use of ‘Farris’, even though this is not a name that Harris bears. Lucy may have committed an embarrassing social faux pas, and she has failed in at least one thing she set out to do with her words—informing John of Harris’s name—but it is not clear that she has made any semantic mistake, or a mistake with regard to reference. For another example, consider a group of speakers who give a friend a form of nickname by which they use a 6-digit numeral sequence to address and refer to her, perhaps as a cruel joke about her prison spell. This nickname can have exactly the same syntactic use as an English proper name, and be understood amongst a reasonably large group as a name for the friend. But now suppose that instead of using a stable numeral sequence, the users of the nickname generate a sequence at random each time they use it: ‘763549’, ‘477863’, and ‘968732’, for example. This practice can go on for a long time with very little repetition, as it is relatively easy to spontaneously generate such apparently random sequences, and there are one million possible combinations (though some, such as ‘000000’, might never occur). It is easy to imagine that this naming practice could be effective: amongst those speakers who are aware of it, it will generally be obvious, at least from linguistic context, that the nickname is being used, rather than some unrelated 6-digit sequence being spoken. The nickname(s) clearly refer to the object of the practice since they are used very effectively to communicate and talk about the object within a community. However, although we can talk about ‘the object of the practice’, none of the individual sequences is borne by the friend, or is her name, and we could even imagine that no sequence is ever used more than once. Here again, then, we have a case in which uses of terms that strongly resemble names clearly refer to an object which is not a bearer of those terms, and do so in a way that clearly does not merely depend upon the intentions of the particular speaker. It also seems plausible that, although there is a convention or practice associated with the referential uses of these terms, it is a social phenomenon, not a semantic one, and one that does not seem to form any essential part of the semantics of the terms. The kinds of example I have considered in this section are supposed to lend credence to the idea that notions such as name/bearer relations, naming conventions, and dubbings-in-force, are not the kinds of thing that can play the rôle of determinate contextual parameters that will, on any occasion determine, or contribute to the determination of, the reference of a name. At least, not in the way that seems to be envisaged by most indexicalists. It would be possible to add in piecemeal additions or exclusions to the semantic character of names in order to account for the various kinds of 130 4.4. Indexicalism: Not all Bad consideration I have offered, and to continue to do so as more are thought of or arise, much as Rami seems to propose for his conditions on semantic correctness. However, even more so than for Rami’s conditions, this would seem to undermine something in the nature of a semantic character that determines reference. It would be to attempt to incorporate contingent and highly contextual considerations about how names can be made to work on particular occasions into the general, context-insensitive, meaning of names. These complex social considerations about reference that would have to be codified and systematized—something that is anathema to such phenomena—in order for them to be contained in semantics. Far simpler, then, to postulate that the meanings of names do not essentially involve them, reference occurring in a pragmatically determined way on occasions of use, and it may or may not be affected by social conventions pertaining to the use of names. If certain other pragmatic factors outweigh the usual social pressures, and, for example, a name is used to refer to something that does not bear that name, though the error may be significant, it is a social, or even a moral error, not a semantic one. 4.4 Indexicalism: Not all Bad So far, my criticisms of indexicalist accounts of proper names have focussed on a particular understanding of indexicalism: that characterized by the claims (i), (ii) and (iii) or (iii*), as introduced at the beginning of this chapter. This does seem to be the view of most indexicalists, and it is inspired by a particular conception of indexicals: one in which the semantic meaning of the indexical specifies how its reference will be determined, relative to a context of use. However, if one were to conceive of indexicals differently, it might be that the comparison between indexicals and proper names would be more compelling. If one didn’t imagine that the semantics of an indexical was, in principle, supposed to provide truth-conditions for it, then many of the problems I have raised for the indexicalists would be less pressing.28 And, indeed, if one were to generalize the position I espouse with regard to proper name reference to the reference of other terms, one would be inclined to abandon that view of the semantics of indexicals. My primary objection to indexicalist accounts of proper names has been that there could, in principle, be no set of determinate contextual parameters such that they could, together, determine the reference of a name on every 28 As I have stated, Rami, apparently unlike Recanati and Pelczar & Rainsbury, doesn’t do this. However, his placing of semantic constraints on the use of names presents different, though related, problems. 131 4.4. Indexicalism: Not all Bad occasion of use. Clearly, however, if indexicalists were to claim that the reference of indexicals is not determined by their semantics in the manner just alluded to, but that names are like indexicals in some other manner, then my arguments would have no teeth, and my main grievance with the indexicalists would evaporate. It may well be that many indexicalists would cease to have a view on how proper name reference is determined, and would instead, like many predicativists and variablists, simply be interested in the semantic form of names. Under such circumstances, I might be inclined to wonder how like the meanings of indexicals the meaning of names really is, since a number of indexicals appear to have more detailed meanings than names, but this worry would depend heavily on just what was being said about the semantics of indexicals, and indeed semantics in general. ‘ 132 Chapter 5 Reference as Speaker Reference: Intention-Based Accounts Inten[t]ions are creatures of darkness, and I shall rejoice with the reader when they are exorcised W.V.O. Quine, ‘Quantifiers and Propositional Attitudes’ (modified) T here is a view of proper name reference available in logical space that is not, strictly speaking, an example of a semantic view, but also does not correspond to the pragmatic view I will espouse in part III. Though it might be considered pragmatic upon some interpretations of the term. This is the view that proper name reference of the sort generally considered to fall into the category of semantic reference—reference that contributes to the truthconditions of an utterance—is essentially determined by speakers’ intentions, rather than by factors external to speakers. At its very simplest, such a view would be use-relative, like indexicalism, and my pragmatic view, and would simply state that what is referred to by, or with, a name on an occasion of use is whatever the speaker intended to refer to with their use of the name (if anything). Whether such a simply-stated position could actually result in a viable account is unclear since it says nothing so far about what constitutes an intention to refer. However, even if it did produce an account of reference that were able to give an indication of what uses of names referred to, it would be inadequate in accounting for intuitions about the reference of certain uses of 133 names. This is because it is essentially just a version of the radical internalist approach to meaning known as Humpty-Dumptyism. Widely regarded as wholly untenable, this is the position whereby it is possible for a speaker to express a thought or proposition using any words they wish—or, in our case, refer to whatever they like using whatever name they like—regardless of how those words are commonly used, simply in virtue of their intention to do so. Such intentions are unconstrained by (beliefs about) how effective the words will be at communicating the thought, proposition, or reference. Travis (1980) addresses just this kind of view—that intentions determine reference—and observes its obvious flaws whilst making an important point about the external nature of reference: [D]o a referrer’s intentions always determine what was referred to? Apparently not. Suppose, e.g., those intentions are bizarre. [Footnote suppressed] Sam walks up to a stranger (Max) on the streets of Azuza, points at Max’s Afghan and says, ‘The dog is hot.’ with the intention of commenting on a certain Rottweiler in Arnhem. Or, Esmeralda finally takes her whippet to the vet’s, after which Herbert, who has been treated to daily reports on the dog’s health, calls and asks, ‘How’s the dog?’. To this, Esmeralda replies ‘The dog is fine.’, with the intention of reporting on her cousin Harry’s spaniel (of which Herbert has never heard). Two things seem clear: first, Sam didn’t refer to the Rottweiler, nor Esmeralda to the spaniel (though in one sense each may have been referring to that intended dog1 ), and second, Sam did tell Max that his Afghan was hot, hence referred to the Afghan, and similarly Esmeralda told Herbert (perhaps without meaning to) that her whippet was fine. With making references, as with anything else people do, good intentions are not enough. (ibid., p. 148–149) To my knowledge, this Humpty-Dumpty view of name reference has never been seriously defended in print, though Predelli (1998, 2002) comes close to proposing a Humpty-Dumpty view of the reference of indexicals. Accounts of names that invoke speaker intentions as referential mechanisms for truthconditionally relevant reference (i.e. not ‘mere’ speaker reference) generally place constraints on the way that intentions can determine reference. This is done either by specifying that in addition to being the object of a speaker’s referential intention, a referent must also fulfil certain criteria, or by placing constraints directly upon the kinds of intention that are relevant to reference. This latter option is likely to be adopted by advocates of Gricean or neoGricean approaches to meaning, whereby the intentions that are relevant to 1 This is the sense in which a speaker can report their intentions or supposed intentions making use of the imperfect aspect of a verb, mentioned in n. 14, p. 83. 134 5.1. Constraint-based Intentionalism meaning and communication take a particular reflexive form: the speaker must intend to express something by their utterance, and intend that their audience comes to appreciate that that is their intention by means of their utterance. However, the view that Gricean-type communicative intentions are the sole determiner of name reference would require a particularly strong kind of Griceanism. In this chapter I will briefly discuss the idea that proper name reference is determined by a speakers referential intentions together with certain constraints, by addressing the view of names proposed by Michaelson (2013). I will then, in a digression from the rest of the thesis, discuss the strong kind of Griceanism not in the context of names, but much more generally. In a largely free-standing section, I argue against a form of neo-Griceanism found in certain works by Stephen Neale, amongst others, on the basis of certain data from experimental psychology. Whilst it is not clear that Neale actually holds that reference is solely determined by speakers’ communicative intentions in these works, it is certainly true that a neo-Gricean holding such a view of reference would have very similar commitments to Neale, and so face the same problems. 5.1 5.1.1 Constraint-based Intentionalism Michaelson Michaelson (ibid.) proposes an account of proper name reference whereby the reference of a name on an occasion of use is just whatever object the speaker intends to refer to with that use, so long as that object is a bearer of that name in the context, and so long as only one bearer is so intended. Nothing will be referred to if there is no object that the speaker intends to refer to that is a bearer of the name, nor if there is more than one object that the speaker intends to refer to that bears the name. Specifically, Michaelson says: The use of a name N refers to o1 when (i) the speaker intends to refer to o1 , (ii) o1 bears the name N in context, and (iii) there is no object o2 such o1 6= o2 and both (i) and (ii) also hold for o2 . Otherwise, it fails to refer. (ibid., p. 68) Michaelson’s account is thus somewhat similar to the account of Recanati (1993), but with an additional—and foundational—intentionalist element, and with the difference that, apart from the notion of being a bearer in context, which Michaelson does not elaborate on, context is not left the task of selecting between rival bearers. 135 5.1. Constraint-based Intentionalism Michaelson’s account of name reference appeals to intentions because, like Griceans, he believes that language can be conceived of ‘as being in part a system for signalling our intentions for other agents to those agents’ (Michaelson 2013, p. 64). To transmit a thought, then, involves transmitting information about a speaker’s intentions. And to communicate about an object, it is necessary for a speaker to get an audience to recognize that they intend them to entertain a thought about the object (p. 65). Michaelson maintains that, in using a name, a speaker means to indicate that they have some object in mind and that it is a bearer of the name. Moreover: ‘by signalling to the listener that her intended referent bears the name N, the speaker thereby commits herself to her intended referent bearing that name’ (p. 66). So, Michaelson’s view is that, in using a name, a speaker expressly commits themself, on pain of misusing the name and potentially failing to refer to anything, to there being some object that they intend to refer to, and—as a logical consequence of them succeeding in referring—to that object bearing the name. 5.1.2 Problems with Michaelson Michaelson acknowledges that, on his view of reference, there is a gap between what thought an utterance can be used to communicate in a context, and its truth-conditional content. This is partly because, according to him, and unlike on the Gricean picture, a speaker can refer to an object in such a way that they know that their audience is very unlikely to appreciate what is being referred to. But the problem will also clearly arise in cases in which, intentionally or not, a speaker uses a name to communicate something about an object that does not bear that name. Michaelson mentions a case (the leafraking case from Kripke (1977)) in which interlocutors mistake Smith, who is raking leaves, for Jones, and one says to the other, ‘Jones is raking leaves’. According to Michaelson’s account of names, the speaker has semantically referred only to Jones, and asserted something false of him (assuming he is not also, quite by chance, raking leaves). This is in spite of the fact that it might appear that the speaker has succeeded in communicating something true about Smith to their hearer. Michaelson claims that the speaker’s confused thoughts may have been communicated, but this is irrelevant to the truthconditional content of their utterance. Presumably, in a simpler case in which a speaker has simply got their intended referent’s name wrong, rather than mistaking them for someone else, Michaelson would also be inclined to say that whatever is communicated, reference has failed. Michaelson claims to leave room for some account by which a speaker’s intention to refer has a 136 5.1. Constraint-based Intentionalism bearing on communication, even if the intended referent is not a bearer in the context. However, he does not say what that bearing is, nor does he provide any clue about how he would account for unintended communication about non-bearers, or apparently true utterances about non-bearers. I discussed bearer constraints at length in §4.3, and indicated why limiting the reference of names just to bearers in context eliminates perfectly good cases of reference. All such arguments apply to Michaelson as to indexicalists, if not more so, since his constraint is so strong. Further problems arise with regard to his claims that speakers’ intentions are inherently relevant to reference. Michaelson is well aware that his account produces results that some theorists—myself included—find counter-intuitive, particularly with regard to cases in which a speaker’s intentions trump the communicative potential of their use of a name. I will not concern myself here with tubthumping about the intuitions surrounding such cases. However, I believe that there will be cases both in which speakers refer to objects that they do not intend to, and in which reference fails because a speaker uses a name in such a way that their intended reference is opaque to their audience. Michaelson may simply reject that there are such cases, maintaining again that the communicative potential of an utterance is a wholly separate issue to the truth-conditional content that that utterance expresses. Michaelson suggests that it acceptable for communication to come apart from truth-conditions, given a particular way of understanding the rôle of truth-conditional meaning, whereby the thoughts that might be transmitted by means of an utterance need not be directly or literally expressed by that utterance (Michaelson 2013, p. 74). This leads him to suggest that truthconditions enable us to hold one another accountable for the reliableness or unreliableness of the things we say ‘independent of any particular communicative aims we might have in context’. This seems to get things the wrong way round. What is said with an utterance, i.e., what its truth-conditional (or truth-evaluable) content is, as I conceive of it, is constrained by what it would be reasonable to expect, or what it would be reasonable to take to have been said, given the particular circumstances of utterance. A speaker may fail to say or communicate what they intend to with an utterance because they are not reasonable. Similarly, an audience may take a speaker to have said something they did not intend, or that they did not in fact say, because either speaker or audience is not reasonable. But it does not seem plausible that a speaker can successfully communicate a thought or thought-like content to their audience with an utterance except in virtue of having expressed that content with the utterance. Therefore, it is unclear how we could hold each other reliable or unreliable independently of the thoughts we communicate, 137 5.2. Against a Strong Neo-Griceanism or have reasonably tried to communicate. 5.2 Against a Strong Neo-Griceanism In the last half-century, Paul Grice’s speculative work on the nature of meaning has been developed, in a variety of ways, into a rich, and apparently productive, research programme. At the heart of the Gricean and neo-Gricean programme stands an analysis of meaning that connects our everyday notion of ‘meaning’ to a particular sort of intentional structure, aimed at conveying certain information to one’s listener. This analysis appears to provide a satisfying picture of the connection between meaning and communication, and a way of distinguishing semantic from non-semantic intuitions about language-use. In this section, I argue that the particular intentional structures posited by Griceans stand in conflict with findings in empirical psychology. Communication, as conceived of by Grice and his followers, is characterized by various (more or less complicated) descriptions of an audience’s recognition of a speaker’s intentions (their speaker meaning2 ) both to transmit information and for the audience to recognize that intention to transmit the information (e.g. Grice 1969; Sperber and Wilson 1986). If that model of communication were correct, it would entail that, when speakers intend to communicate something, they must, as a matter of course, intend to make those communicative intentions transparent to their audience. It would also entail that for there to be any question of communication, audiences must have the capacity to identify the intentions of a speaker. This latter consequence, in particular, has been challenged by linguists, psychologists and philosophers, on the basis that not all language-users are capable of the high-level intention-recognition required by Griceans, yet appear to be capable of communicating (Breheny 2006; Glüer and Pagin 2003; Gregoromichelaki et al. 2011; Keysar 2007; Laurence 1998; Millikan 2005). I will appeal to related data in order to attack the former consequence, that speakers, as a matter of course, form utterances on the basis of intending that their audience recognize their meaning. To this end, I initially take Neale (2004), and (ms.), as representative of the neo-Gricean programme, since they expound one of the best developed and philosophically sophisticated entries in that programme. These papers present an interesting and original contemporary approach to Grice, which the author takes to be faithful to Grice’s own project (Neale ms., pp. 2425). They argue for a particular interpretation of how speaker meaning is 2 Referred to by Grice himself as ‘utterer’s meaning’. 138 5.2. Against a Strong Neo-Griceanism determined, and how this relates to other notions that arise in languageoriented disciplines. A significant tenet of Neale’s project is that speaker intentions are the sole determiner of what is said with an utterance—its truthevaluable, or propositional content. Speaker intentions are thus claimed to metaphysically or constitutively determine an utterance’s content, in contrast to mechanisms which merely enable epistemic determination by the audience of what a speaker said (working out what was said), but play no rôle in fixing it (Neale 2004, pp. 76, 78, 88; ms., p. 48). Variations on the this move—placing the burden of providing content on to a speaker’s intentions—have been adopted by other theorists of language to similar ends, and so the problem I present for Neale will generalize to their positions as well. This move presents a problem because it requires that certain constraints be placed upon a speaker’s intentions such that they do not fall into Humpty-Dumptyism and have speakers able to express anything they please with any utterance. These constraints often amount to stating that a speaker cannot intend to express with an utterance what they believe could not be communicated with that utterance. For neo-Griceans, this means that speakers must, as a matter of course, take into account (their beliefs about) their specific audience’s perspective and capacity to understand what they say. However, this is precisely what is brought into question by the psychological data to which I will appeal: Horton and Keysar (1996) and Keysar (2007) claim that when forming utterances, speakers do not, as a matter of course, take the perspective of their audience into account. Neale and other neoGriceans seem to be trapped, then, between allowing that speakers cannot say whatever they like with any utterance, and accounting for the way in which speakers actually appear to form utterances. Having introduced this problem, I consider various ways in which Neale might avoid it and conclude that none of them will work. I then discuss how other philosophers in the neo-Gricean tradition will be affected by the problem in a manner similar to Neale. Neale (ms.), which contains the most explicit statement of the strong neo-Gricean view that propositional content is determined solely by communicative intentions, does not specifically mention reference. However, since reference appears to have a bearing on the propositional content of utterances containing names and other referring terms, we might suppose that their contribution to this content is also determined solely by speakers’ communicative intentions. Neale (2004) explicitly states: Who or what A is referring to by uttering some expression X is determined by A’s referential intentions in uttering X. (ibid., p. 80) However, he goes on: 139 5.2. Against a Strong Neo-Griceanism Nonetheless, a distinction between what A referred to and what A intended to refer to is not one obviously lacking a point. So, in the first instance we should separate (i) who or what A intended to refer to by an expression X on a given occasion, and (ii) who or what a rational, reasonably well-informed interpreter in B’s shoes thinks A intended to refer to by X on that occasion. In cases where (i)=(ii), we can talk freely about what the speaker referred to. (In cases where (i)6=(ii), we could argue about which of (i) or (ii) or some third thing has the ‘right’ to be called the person or thing referred to, but what would be the point? First, what third thing distinct from (i) and (ii) could be of any significance to a theory of interpretation? There is simply no role for a transcendent notion of what was referred to upon which (i) and (ii) converge when all goes well. Second, why is a choice even needed in cases where (i)6=(ii)? Conceptually they are distinct, and they are both needed in a theory of interpretation. When all goes well, they coincide, and it’s just too bad they don’t always do so. Surely there is no philosophical payoff in bestowing the honorific ‘what was referred to’ on one rather than the other when they diverge.) (Neale 2004, p. 80) These quotations appear somewhat in tension with each other: if what a speaker is referring to with an expression is determined by their referential intentions, why is it relevant to what they referred to what a rational, reasonably well-informed interpreter in the shoes of the audience takes them to have referred to? In what way is the latter needed for a theory of interpretation? Neale clarifies by stating that referential intentions are constrained, just as general communicative intentions are, by what the speaker believes they can communicate using the expression in question. Indeed, Neale uses almost identical language to describe the nature of communicative and referential intentions. In the case of referential intentions, he says: A cannot (intend to) refer to some particular person α by uttering some expression X on a given occasion if [s]he believes it is impossible for h[er] audience B (or at least any rational, reasonably well-informed interpreter in B’s shoes) to construe him as referring to α. (ibid., p. 80) It appears, then, that (ii), from the quotation above, is involved in the interpretation of referring expression in virtue of this constraint upon referential intentions. However, despite his talk of rational, reasonably well-informed interpreters, this comes into the account only parenthetically, in an act of apparent equivocation. Exactly what Neale intends here is discussed for the general case of communicative intentions below. As I have made clear, however, I do not intend to suggest that Neale is strongly committed to the view that proper name reference is determined solely by speakers’ Gricean 140 5.2. Against a Strong Neo-Griceanism intentions. Rather I claim that anyone making such a claim would likely be committed to a version of neo-Griceanism at least as strong as Neale’s, and thus subject to the objections I make to his account of speakers’ general communicative intentions. Neale on speaker meaning 5.2.1 For Grice, the speaker meaning3 of an utterance is, at its most basic, what the speaker intends to communicate with the utterance. It is what follows expressions of the type ‘Speaker S meant by uttering X that. . . ’ (Grice 1969, p. 91). The notion is spelt out in various, increasingly complex ways by Grice, and those who took up his project, but the following vague generalization incorporates many of them: “A meant something by x” is (roughly) equivalent to “A intended the utterance of x to produce some effect in an audience by means of the recognition of this intention”. (Grice 1957, p. 220) Neale (2004) makes the following statement regarding the connection between speaker meaning, and intentions, and what is said: What [S] meant by uttering X can be factored into what [S] said (or asked) by uttering X and what [S] only implied. Thus, again following Grice, what [S] said and what [S] implied are determined by, and only by, certain very specific interpreter-directed intentions [S] had in uttering X. (ibid., p. 78) He goes on to say, with regard to truth, that What [S] says and implies are the sorts of things that are true or false. (ibid., p. 79) And, with regard to semantic content, If talk of the ‘semantic content’ of a sentence X relative to a context C is just a snazzy way of talking about what the speaker said by uttering X on a particular occasion—the occasion that C is being used to partially model—then of course we can accept its empirical significance.(ibid., p. 79) From these statements, it seems very plausible to take Neale to hold that the truth-evaluable, or propositional, content of an utterance is solely determined by the audience-directed intentions of the speaker. Neale (ms.) makes this position even more explicit, providing the following principles: 3 Or the utterer’s (occasion-)meaning. 141 5.2. Against a Strong Neo-Griceanism T1 Audience-directed intentions [speaker] S had in uttering a sentence X (on a given occasion) are the sole “suppliers” of any propositional content there is to what S says and what S implicates by uttering X (on that occasion). T2 Thinking that the linguistic meaning of X, the context in which X is uttered, or indeed, anything other than audience-directed intentions S had in uttering X (on a given occasion) [are] “suppliers” of any content there is to what S says by uttering X (on that occasion) is a mistake typically engendered by (a) conflating constitutive and epistemic notions of the “determination” of the content of what S says, and (b) conflating facts that constitutively determine the content of a communicative intention with facts that formatically determine it, by which I mean facts that causally contain the formation of a communicative intention with that content, (c) misunderstanding how sentence meaning, speaker, saying, implicating, and various other notions fit together in a nexus of theoretical concepts, or (d) a misunderstanding of what semantic and pragmatic theories can plausibly hope to explain. (Neale ms., p. 28) The key speaker intentions in Neale’s account of meaning are meaning or communicative intentions. He characterizes their formation thusly: S thinks that by producing an utterance of X [she] will likely get [her] audience, A, to see that S intends A to think that p at least partly on the basis of recognizing that S uttered X intending A to think that p. (ibid., p. 46) But Neale warns that the formation of such intentions is moderated by the speaker’s beliefs: [A]ssuming [s]he is being co-operative S cannot mean that p by uttering some sentence X if [s]he believes it is impossible for [her] audience A (or at least any rational, reasonably well-informed interpreter in A’s shoes) to construe [her] as meaning that p. (Neale 2004, p. 77). So, communicative intentions are moderated by the speaker’s beliefs about what it is possible to communicate with their words, which are themselves moderated (or constituted) by their beliefs about how the hearer(s) will interpret the speakers intentions, given the conventional meanings of words, the context of utterance, etc. (ibid., pp. 77, 78; ms., p. 47). It is this idea, that a speaker’s beliefs moderate their communicative intentions, that is supposed to prevent Neale’s view from collapsing into Humpty-Dumptyism, and is thus key to making the account plausible at all. 142 5.2. Against a Strong Neo-Griceanism By appealing to experimental psychological data regarding the formation of utterances, I will undermine Neale’s idea that a speaker’s beliefs about their audience’s capacity to interpret an utterance moderate their meaning intentions. He is committed to this idea because, according to his account of speaker meaning, speakers must intend to make their communicative intentions transparent to their audience. Thus, I will argue that, unless Neale’s account is significantly adapted, in a manner he is likely to find unacceptable, it is either incompatible with empirical data, or collapses into Humpty-Dumptyism. This result will, however, not only apply to Neale. The idea that Humpty-Dumptyism can be escaped by appeals to speakers’ intentions being moderated by their beliefs about what can be communicated is also put forward by Donnellan (1968) and Bach (2012), and is appealed to by Buchanan (2013) and Harris (2014), amongst others. Their accounts will therefore also be affected by the psychological implausibility of that idea. 5.2.2 Data on utterance formation Various experimental results (see, e.g., Mitchell, Robinson, and Thompson 1999; Wellman, Cross, and Watson 2001; Wimmer and Perner 1983) have been taken to show that children do not consider that the mental states of others might differ from their own (they do not mind-read) until around four years. One paradigmatic experiment is explicated in Breheny (2006), which surveys such studies, thus: 1. Puppets Sally, Anne and subject play with sweetie. Sally places sweetie in container A. 2. Sally departs. Anne moves sweetie from container A to container B. 3. Sally returns. Subject is asked, ‘Where does Sally think her sweetie is?’ [or ‘Where will Sally look for her sweetie?’]. (ibid., p. 76) Children under the age of four systematically give the ‘wrong’ answer to the question in 3., responding that Sally thinks sweetie is in (or will look in) container B, but after that age, they begin to get the ‘right’ answer systematically. The researchers carrying out and interpreting these and similar experiments generally take them to show that the children ‘failing’ the test do so because they are not yet capable of reasoning about the mental states of others, as distinct from their own, or do not consider that others may have different beliefs and mental states, or at least different evidence, from themselves (cf. Lewis, Hacquard, and Lidz 2012). As is the case for most (neo-)Griceans (cf. Gregoromichelaki et al. 2011), Neale’s claim that speakers 143 5.2. Against a Strong Neo-Griceanism intend, when they have communicative intentions, to elicit inferences in their audience about their intentions already looks psychologically implausible in light of the data mentioned. Young children, and other speakers who have a limited theory of mind, seem unlikely to be able to form such complex intentions about the mental states of others, yet are apparently well able to communicate, albeit with certain limitations. However, various studies cited by Keysar (2007) suggest that even adult speakers do not, as a matter of course, take their audience’s perspective or ‘shared context’ into account when forming utterances, but rather, form utterances on an ego-centric basis. Horton and Keysar (1996) present experimental data supporting the idea that some utterances are formed on an ego-centric basis. Indeed, Horton and Keysar claim that their data suggest that speakers plan all their utterances ego-centrically (that is, only taking their own perspective into account), and then monitor them and update them if they do not fit the audience’s perspective. However, this update occurs only if there are sufficient time and memory resources.4 The experiment is designed to test two contrary models of utterance formation. On the Initial Design model, which is comparable to the Gricean view, when forming utterances, speakers take their audience’s perspective into account from the outset. On the Monitoring and Adjustment model, utterances are initially planned ego-centrically, and then monitored for consistency with the audience’s perspective.5 The experiment and its results, are described in Keysar, Barr, and Horton (1998), thus: [W]e asked participants to describe simple figures to addressees. The figures were presented in the context of other figures. For example, the participants described a circle in the context of a larger circle. We informed some participants that the addressees shared those context figures, whereas other participants were told that these figures could not be seen by the addressees (the figures were privileged to the speakers). The crucial measure was the extent to which speakers’ descriptions relied on context, which was indicated by their use of adjectives. For example, if they described the circle as a “small” circle, it suggested that they relied on the larger context figure. 4 Whilst the details of this two-stage account are challenged by some other psychologists and psycholinguists, notably Susan Brennan and colleagues (e.g. Brennan, Galati, and Kuhlen 2010; Brennan and Hanna 2009), the claim that speakers do not invariably or as a matter of course take their audience’s perspective into account—as is shown by Horton and Keysar’s data—is not disputed. I return to this issue below. 5 Horton and Keysar (1996) do not go into detail about what mechanisms are involved in monitoring utterances for compatibility with an audience’s perspective. They indicate that it involves checking utterances against the common ground, but do not specify, for example, whether this might involve Gricean-style mind-reading. The precise details will not effect my argument. 144 5.2. Against a Strong Neo-Griceanism (a) Time 0 LISTENER SPEAKER (b) Time 1 LISTENER SPEAKER Fig. 5.1: An example of a trial in the “privileged context” condition. The lower object cannot be seen by the listener and constitutes the speaker’s privileged contrast set. [In Time 0 the speaker described the upper object, which moved behind the visual barrier into the listener’s part of the screen in Time 1.] (Horton and Keysar 1996, p.99) There were 24 subjects, who were asked to describe figures (pieces of clipart) to a listener who, although believed to be another subject by the speaker, was always a confederate of the experimenter. There were 16 experimental item sets, which appeared on a computer monitor divided by a close-fitting perpendicular screen that separated the speaker and listener. The item sets consisted of a rectangular box, divided vertically into two equal parts by the screen, the right part visible to the speaker, and the left to the listener. On the speaker’s side appeared two figures: a moving figure, to be described, which moved from the speaker’s side, across the screen to the listener’s side. Below the moving figure was a context figure which differed in either shape, shade or size from the moving figure. Half the subjects were told that their listener would never see the context object—privileged context cases, represented in simplified form by Fig. 5.1—and half were told that the listener 145 5.2. Against a Strong Neo-Griceanism (a) Time 0 LISTENER SPEAKER (b) Time 1 LISTENER SPEAKER Fig. 5.2: An example of a trial in the “shared context” condition. In Time 0 the speaker described the upper object, which moved behind the visual barrier into the listener’s part of the screen in Time 1. (Horton and Keysar 1996, p.97) could see the context object—shared context cases, represented by Fig. 5.2.6 Additionally, half of each of those groups were told to take as much time as they liked in describing the figure, whilst the other half were told to begin describing within 1.5 seconds of the figure’s appearance. In addition to the 16 experimental sets, there were 24 filler sets, identical in set-up except that when the moving figure crossed the divide, it changed shape (represented by Fig. 5.3). The listener’s task was to say whether or not the moving figure appearing on their side was the same figure described by the speaker. The experiment was divided into two presentations separated by a five-minute break. Each moving object was presented once in each presentation, though with a different context object on each occasion (Horton and Keysar 1996, p.101–104). The Gricean-type Initial Design model of utterance formation predicts that when the subject believes their audience shares their visual context, they 6 All representations of the item sets are reprinted from Horton and Keysar (1996), with permission from Elsevier and the authors. 146 5.2. Against a Strong Neo-Griceanism (a) Time 0 LISTENER SPEAKER (b) Time 1 LISTENER SPEAKER Fig. 5.3: An example of a negative-trial filler. The object that moved into the listener’s screen is different from the one that moved out of the speaker’s screen. (Horton and Keysar 1996, p.102) will describe the moving figure in terms of the context figure, for example, by saying ‘the smaller circle’. However, when the subject believes that the audience does not share their visual context, the model predicts that the subject will not make reference to the other figures. The Monitoring and Adjustment model makes the same predictions under normal circumstances, but predicts that if the subject is not given the opportunity to adjust their initial plan before speaking, if they are under time pressure for example, then they will describe the figure in terms of their own perspective, demonstrated by their making reference to other figures, whether or not they have been told that the audience shares their visual context. This—the prediction of the second model—is just what the experimenters found: The results were straightforward. First, speakers relied on context more often when it was shared than when it was privileged to them. This result is predicted by both models, and suggests that the final descriptions were sensitive to the perspective of the addressees. The critical test came when we asked the speakers to perform under time constraints (i.e., to 147 5.2. Against a Strong Neo-Griceanism start talking 1.5s after they saw each figure). Under time pressure, their descriptions were just as likely to rely on a privileged as on shared context. This is precisely what the monitoring-and-adjustment model predicts: Under pressure, speakers do not have sufficient time and resources to monitor and correct their utterances, and consequently they fall back on their initial plans. These plans are egocentric in the sense that they are not sensitive to the common ground with addressees: The speakers rely on their own context regardless of whether it is part of common ground. (Keysar, Barr, and Horton 1998, p. 47) The experiment had two context conditions and two initiation speed conditions. The results were coded according to whether or not the speaker appealed to context in their description by, for example, using a relative adjective. Coders were aware of what the context object was for each description, but unaware of either of the experimental conditions (Horton and Keysar 1996, p. 106). The results are represented in Table 5.1 and Fig. 5.4, reproduced from Horton and Keysar (ibid., pp. 107–8). Initiation speed Unspeeded Speeded Context-related Shared context Privileged context 0.74 0.24 0.48 0.43 Context-unrelated Shared context Privileged context 0.76 1.35 0.82 0.68 Table 5.1: Mean number of context-related and context un-related adjectives per description as a function of context condition and initiation speed (during second presentation). Unspeeded Shared Privileged Speeded Shared Privileged 0 0.05 0.1 0.15 0.2 0.25 0.3 Fig. 5.4: Mean ratio of context-related adjectives to the total number of adjectives plus nouns per description as a function of context information and initiation speed, for the second presentation. 148 5.2. Against a Strong Neo-Griceanism Horton and Keysar’s results convincingly suggest that, between the two models proposed, the Monitoring and Adjustment model is to be preferred over the Initial Design model, assuming that they are reliable and generalizable. Proponents of the Initial Design model, which aligns with Gricean notions of utterance formation, are unable to explain why time pressure makes any difference to how likely speakers are to take their audience’s perspective into account, since they claim that speakers take audience perspective into account as a matter of course when initially planning utterances. There is no scope for the position that speakers, by default or as a rule, take their audience into account when initially planning utterances but, in performance, sometimes fail to do so for various reasons. Assuming that Horton and Keysar’s results are not anomalous,7 they appear precisely to show, by the contrast between the time-constrained and the unconstrained results, that allocentric planning—planning involving the perspective of others—is not the default when forming utterances. Brennan and Hanna (2009) and Brennan, Galati, and Kuhlen (2010) challenge the two-stage model of utterance formation proposed in Horton and Keysar (1996), Keysar, Barr, and Horton (1998), and Keysar (2007), among others, on the grounds of criticism of the methodology of certain apparently supportive experiments, and of evidence found in other studies suggestive of spontaneous interaction between interlocutors (Brennan, Galati, and Kuhlen 2010, p. 325). Brennan and her colleagues accuse theorists who deny that speakers mentalize about their audiences of being unable to account for ubiquitous incrementality, co-ordination, and apparent joint action in dialogue. However, Gregoromichelaki et al. (2011) present a model of dialogue which is able to account such phenomena—indeed, is entirely driven by them— whilst being directly informed by the research criticized by Brennan, and propounded directly in opposition to (neo-)Gricean accounts of communication. Gregoromichelaki et al. claim that interlocutors develop individual representations of a dialogue in tandem, without the need for representing each other’s mental states or creating joint structures (ibid., pp. 225–6). However, even if we were to lend weight to the worries of Brennan and her colleagues, their specific criticisms of the two-stage model, and the details of Brennan’s positive views, will not affect the thrust of my attack on the (neo-)Gricean framework. The two-stage model involves the strong view that all initial utterance formation is wholly ego-centric, modular, and encapsulated against consideration of information about audience perspective, and it is this strong claim that Brennan and her colleagues attack. However, they do not 7 Keysar (2007) cites similar findings in other studies. 149 5.2. Against a Strong Neo-Griceanism claim that findings such as that of Horton and Keysar (1996) are not evidence of ego-centric utterance formation, they merely claim: [P]rocessing need not be encapsulated from relevant partner-specific information that is straightforward and known in advance. Under some circumstances, speakers can adjust immediately to their addressees’ needs or perspectives, even when different from their own. (Brennan, Galati, and Kuhlen 2010, p. 335, my emphasis) This limited claim, that speaker’s awareness of their audience’s perspective can play a rôle in utterance formation, is significantly weaker than that being made by Neale and other neo-Griceans. Neale requires that a speaker’s beliefs about their audience’s perspective moderate their communicative intentions as a matter of course. The data presented by Keysar (2007) suggests that speakers do not take the perspective of their audience into account as a matter of course when forming utterances. Therefore, their beliefs about what their audience will infer seem irrelevant. This is a significant problem for Neale. In §5.2.3, I consider ways in which Neale might try to avoid this problem, and in §5.2.4 I expand the problem to other neo-Griceans. 5.2.3 Ways Out for Neale Consider a speaker, Elaine, who is taking part in an experiment like that of Horton and Keysar (1996). On her monitor is displayed a grey boot (the moving object), with a paler grey boot below it (the context object). Elaine is told that her listener, George, does not have the context object on his screen. Elaine is asked to start describing the moving object within 1.5 seconds of it beginning to move. When it does begin to move, Elaine describes the object as ‘the darker boot’, even though she knows George does not have a lighter boot to compare it to. She is sincerely attempting to enable George to correctly identify whether or not the moving object on his monitor is the same as that on her own. We know from Horton and Keysar’s results that such a situation can occur. Call this scenario the ‘ego-centric case’. Neale and other neo-Griceans are unable to account for such a case because it appears that Elaine’s beliefs about her audience’s perspective have not affected the formation of her utterance. There are at least three things Neale could say with regard to his intention-moderating beliefs in response to the the ego-centric case. I will consider each in turn. Co-operative and Ego-centric Utterances One approach would be to simply observe that on many occasions (when there is sufficient time and memory) speakers do take their audience into 150 5.2. Against a Strong Neo-Griceanism account—Neale mentions in the quotation from (2004, p. 77), given above, that he is assuming S is being co-operative. On those occasions, speakers’ beliefs about what they can get the audience to construe them as meaning are precisely relevant. However, Elaine is not being unco-operative in the ego-centric case, and speakers are generally not being unco-operative in real life cases, though this may not entail that they are being co-operative (cf. Gregoromichelaki et al. 2011). It is unclear what Neale intends when he says “assuming [S] is being co-operative” (Neale 2004, p. 77). He might mean to be stipulating a definition of what it is to be a co-operative speaker, so specifically ruling out the ego-centric case. So it seems ‘ego-centric speakers’ fail to have any communicative intentions (and thus can’t express propositional content). Then the account is lacking since these ‘unco-operative’ cases appear to be, if not the standard, as Keysar and his colleagues claim, then at least not exceptional. Alternatively, he might be offering a characterization just of co-operative speech, in which case his results are extremely limited with regard to communication as a whole, since, again, ego-centric communicative behaviour appears not to be exceptional. Neale appears to be allowing that his account might not incorporate all communicative intentions when he says The question of what constitutively determines what S means on a given occasion and the question of what is involved in epistemically determining (i.e. ascertaining or identifying, or at least forming a hypothesis about) what S means on that occasion are conceptually distinct, even though the formation of S’s meaning intentions is typically formatically determined [their formation is causally constrained], in part, by S’s conceptions of the sorts of things S may reasonably presume to be potentially involved in the process of A’s forming a hypothesis about what S means. (Neale ms., p. 48) His use of the word ‘typically’ suggests that sometimes a speaker’s beliefs about how their audience is likely to interpret them do not determine their communicative intention. But again, if Neale is intending to give an account only of communication in which speakers take their audience’s perspective into account, then he is simply choosing to ignore a significant facet of human communication. If Horton and Keysar’s two-stage interpretation of their data is taken seriously, he is ignoring the default means by which speaker’s form utterances, as well as, arguably, all the communication of small children and autistic people (Baron-Cohen, Leslie, and Frith 1985; cf. Yergeau 2013). Neale is, of course, free to account only for a specific sub-class of communication, but it certainly appears that he is trying to provide a general account, and a general account of communication that fails to account for a significant portion of examples is not a general account of communication. Moreover, if 151 5.2. Against a Strong Neo-Griceanism there were a plausible account of communication and utterance content that could account for all occurrences, that would be preferred to a partial account. Intentions towards a general audience A second approach would be maintaining that a speaker’s intention-moderating beliefs are not (or need not be) aimed at a specific audience, but rather are about how one might in general best communicate. This would be a significant departure from the traditional (interpretation of the) Gricean picture, although one might think it is hinted at in a parenthetical remark of Neale’s: ‘or at least any rational, reasonably well-informed interpreter in A’s shoes’ (Neale ms., p. 46). Whether or not this really is a hint at the approach under consideration would depend on how generally we took the idea of being in A’s shoes. Intention-moderating beliefs are specifically supposed to be about context, topic and linguistic conventions (presumably word meanings), etc. According to certain alternatives to the Gricean picture of language (e.g. Gauker 2008; Travis 1997), these are the kinds of thing that directly affect what is said, because they are what effect in general how utterances are interpreted. Even if a speaker is ego-centric in their formation of an utterance in the sense that they do not take their specific audience’s perspective into account, their utterance presumably will be sensitive to linguistic convention, and various non-audience-specific contextual mechanisms, such as topic, their own purposes, and environment. In the ego-centric case, it is specifically reliance on contextual comparison that indicates a lack of allo-centrism, as the particular visual cue Elaine appeals to is unavailable to George. To make this (hypothetical) interpretation of Neale explicit we might have: S thinks that by producing an utterance of X she will likely get competent users of the language to whom the same contextual information is available to see that S intends them to think that p, at least partly on the basis of recognizing that S uttered X intending them to think that p. and Assuming she is being co-operative, S cannot mean that p by uttering some sentence X if she believes it is impossible for any rational, reasonably well-informed interpreter to whom the same contextual information is available to construe her as meaning that p. These adapted characterizations of a speaker’s communicative intentions seem to allow for the ego-centric case, at least given certain assumptions about it. Since it is stipulated that the general intended audience has access to the same contextual information as the speaker, the fact that a speaker fails to attend to 152 5.2. Against a Strong Neo-Griceanism their actual audience’s perspective, and bases their utterance formation only on their own, does not rule such a speaker out of the account. However, in allowing for the ego-centric case, the general audience approach threatens to exclude some allo-centric cases. The position, as stated, will have trouble dealing with cases in which an audience does not share certain elements of the context with the speaker, and the speaker takes account of this and moderates their utterances appropriately, but, in the process, renders it inaccessible to a general audience who shared more of the speaker’s context. Moreover, even if the approach under discussion were to overcome this dilemma, the rôle still played by intentions about others’ intentions may make the account implausible for pre-theory of mind children, though Neale may be inclined to accept this result, ruling them a special case. There is good reason to think that the general audience approach is not the path that Neale or other neo-Griceans take, or would take. The position is at odds with Griceanism, and in tension with his descriptions of audiencedirected intentions. Neale very much has in mind beliefs about a specific audience’s ability to construe the speaker’s meaning. Dispositions A third interpretation of Neale, to account for the ego-centric case, would be to appeal to a competence/performance distinction, and claim that speakers are always disposed to take audience’s perspective into account, but in some circumstances, such as when under time pressure, they fail to actually do so. Thus, when speakers act competently, they act just as Neale suggests. In the ego-centric case, in which Elaine fails to act as Neale suggests, she can succeed in expressing propositional content in virtue of her disposition to act in the correct way: she would have taken George’s perspective into account had she had sufficient time to adjust her utterance. In order to assess this position, we need to clarify how intentions fit into it, since these are essential to Neale’s framework. It appears that there are two ways of clarifying the position: either (i) the speaker has the communicative intention just in virtue of having the disposition, or, less plausibly, (ii) they have the communicative intention only in allo-centric cases, but, nonetheless, can express content in the ego-centric case in virtue of their disposition to be allo-centric. Both of these options represent a significant departure from the position stated in Neale (2004, ms.). Clarification (i) would involve adapting Neale’s characterization of communicative intentions so that S’s intention to communicate p to A with an utterance X does not require that S believes it is possible to do so, but merely 153 5.2. Against a Strong Neo-Griceanism that S be disposed to use an utterance with which it would be possible. To see why, consider the ego-centric case: Elaine is presumed to believe that George does not share her visual context, on the basis of having been so told. Yet that belief does not affect whether or not she appeals to the visual context in describing the boot figure. Thus, given clarification (i), Elaine had a communicative intention to describe the figure to George by appealing to another figure, even though she believed that George couldn’t see that other figure. This puts a great deal of pressure on a central premise of Neale’s paper: that one cannot intend to do what one believes to be impossible. Of course, we do not have a direct contradiction of this premise; it may be that Elaine believing that George cannot see the other figure does not entail that she believes that George cannot understand a description that appeals to the other figure. However, the spirit of the premise is certainly undermined. The premise that seems to replace it, given clarification (i)—that one cannot intend what one believes would be impossible if one acted as one is disposed to act—is much less convincing. As a constraint on communicative intentions it allows too much. It entails, when slotted into Neale’s broader account, that a speaker can express any propositional content using any words or sounds, as long as they are disposed to use words which they believe can express that propositional content. So if, in a stupor, a speaker named ‘Sheila’ mutters some incomprehensible sounds—or even says ‘Marjorie’—when asked what her name is, assuming that she is disposed to say ‘Sheila’, then her sounds have in fact expressed that her name is Sheila. I find this result highly implausible; we have essentially arrived once more at the Humpty-Dumpty position whereby a speaker says (expresses) whatever they intend to, regardless of the words they actually use, so long as they have certain dispositions to use appropriate words. It may be possible to constrain the kinds of dispositions that allow intentions to come apart from beliefs about the possibility of their fulfilment, but I cannot see how this would be achieved, particularly in a non-ad hoc manner. Clarification (ii), that speakers have the communicative intentions only in allo-centric cases, would involve Neale giving up another central premise in order to accommodate the ego-centric cases, namely T1 of Neale (ms.) (stated on p. 142), that a speaker’s communicative intentions are the sole suppliers of propositional content. If this premise remained, utterances formed without taking the audience’s perspective into account could not have any propositional content since, ex hypothesi, the speaker has no communicative intention in these cases. It is not entirely clear what T1 could be replaced with. Perhaps something like “A speaker’s communicative dispositions are the sole 154 5.2. Against a Strong Neo-Griceanism suppliers of the propositional content of an utterance” or “An utterance’s propositional content is solely supplied by the communicative intentions the speaker would have had, had they spoken in ideal circumstances”. The former looks unpromising because it says very little—presumably a speaker has many diverse communicative dispositions at any time—and the latter looks unpromising because it threatens to radically shift the propositional content of all utterances, unless a precise and workable spelling-out of ‘ideal’ is struck upon. Moreover, even though embracing clarification (ii) and jettisoning T1 would allow Neale to retain the intention/belief premise lost with clarification (i), T1 is a major thesis of Neale (ms.), and is at least strongly intimated in Neale (2004), and it seems highly unlikely that he would be willing abandon it. The dispositional interpretation of Neale, then, faces two related problems. Firstly, there appear to be severe difficulties in working out a plausible account of it whilst attempting to stay within the spirit of Neale (ms.). Secondly, even whilst attempting to stay within the spirit, major elements of Neale’s project have to be dropped or radically altered in a manner which would presumably be unacceptable to him. Conclusion 5.2.4 I have considered three routes Neale might attempt to take in order to make his account of the rôle that speakers’ beliefs play in moderating communicative intentions consistent with psychological evidence about how speakers actually form utterances. All have either been found wanting, or to be undesirable for Neale. 1. The idea that the account is only dealing with co-operative cases of speech results in a dilemma: either it is willfully ignoring the nonexceptional, ego-centric cases of utterance and communication, or it is failing to account for these cases, though it means to. 2. The idea that a speaker’s communicative intentions are moderated by beliefs about a general audience, rather than their actual, specific audience produces a problem for dealing with utterances which specifically do take a particular audience into account. This route also appears to be in tension with Neale’s Gricean commitments. 3. The idea that speakers are disposed to act in the way Neale specifies, and the data regarding ego-centric utterance formation can be explained away as an issue of performance rather than competence, faces severe difficulties. It is unclear how the idea can be fully spelled out without 155 5.2. Against a Strong Neo-Griceanism falling into Humpty-Dumptyism. It also appears that, even if it were workable, it would involve Neale giving up major premises of his view. As an innovative and influential member of the neo-Gricean programme, Neale is well-placed to represent and defend its core analysis of meaning and communication as involving recognition of speakers’ intentions. I have endeavoured to show that Neale’s interpretation of this core analysis will not stand up to empirical data regarding utterance formation without abandoning either plausibility or significant Gricean assumptions. Given that Neale is incapable of accounting for Horton and Keysar’s findings, I suggest that they provide a very serious challenge to the Gricean programme in general. Daniel Harris commits himself to a Neale-style position about the nature of communicative intentions, so his work will be affected by the problem I have put forward for Neale, as will be the work of Ray Buchanan and Kent Bach, other prominent neo-Griceans. In showing that a variety of strong neo-Gricean accounts of meaning are unable to account for psychological data about the formation of utterances, I hope to have shown that any Gricean account of proper name reference that attempts to claim that the reference of names is determined solely by speakers’ referential intentions is very likely to be undermined in the same way. In order to avoid falling into the Humpty-Dumpty position with regard to name reference, someone putting forward such a Gricean account of names would have to place constraints on what referential intentions a speaker could have. Doing so in the manner that Neale does for communicative intentions generally—and appears to for referential intentions in Neale (2004)—seems the most likely option, especially when it is observed that other Griceans attempt similar strategies. 156 Part III A Pragmatic View of Reference 157 Chapter 6 The Pragmatic View Stated And this can be expressed as follows: I use the name “N” without a fixed meaning. (But that impairs its use as little as the use of a table is impaired by the fact that it stands on four legs instead of three, and so sometimes wobbles.) L. Wittgenstein, Philosophical Investigations P receding chapters have attempted to show why certain existing views of proper name reference are deficient in various ways. Through these attempts, the general form of my positive view will have become clear. In this chapter I will go into more detail about the pragmatic view of proper name reference that I propose is superior to those views I have already discussed. In order to reinforce the impression of the view provided in the previous chapters, I begin by providing some general principles of the view before going on to discuss it in more detail. In my discussion, I cover each of the principles, offering either reason, motivation, or argument for it. I begin by discussing reference and its relation to names in §6.2. I reiterate and elaborate on my thoughts on the connection between reference and truth from §2.1, and briefly cover empty reference and reference over time, which present potential complications to an account of reference. In §6.3, I move on to the context-sensitivity of names and their reference, and the relation between semantics and reference. I make clear, in §6.4, why I do not think that semantics determines reference, and why it is that name reference cannot have any necessary and sufficient conditions. The lengthiest section of the 158 chapter, and that containing perhaps the most significant work, §6.5, deals with names and bearers, and continues the discussion from §4.3. I cover why it is that I claim that bearer relations are not required for name reference, and argue that my conception of the rôle of such relations can better account for our use of and intuitions about them. I also observe that there is a rôle within my view for some kind of defeasible, probabilistic constraint that also helps to account for the appeal of naming-using practices. §6.6 moves away from the principles of §6.1, and briefly covers what can be said about the individuation of names on my pragmatic view of their reference. In §6.7, I return to Travis, and spend some time comparing my view of reference to his more general account of occasion-sensitivity. I deal with the apparent similarities and differences, covering the contribution of names to thought, and the possibility of Travis cases for names. Finally, in §6.8, I consider two likely objections to my view: that it fails to make predictions about name reference, and that it is implausible and unworkable for there to be no systematic criteria for reference. I show that neither objection is especially troublesome. The view I present cannot be said to be wholly novel, though I believe no one has ever argued for it at such length, or with engagement with the contemporary literature before. In the early Eighties, Travis wrote at least two papers (Travis 1980, 1981a) in which he engaged specifically with reference, largely in response, so it seems, to the influence Kripke was having at that time (Kripke 1977, 1981). In these papers, Travis attacks what I have described as semantic views of reference—those that propose that there are stateable conditions that determine the reference of a referring term—and proposes the idea that such systematic conditions are not necessary either for terms to be used to refer, nor for language-users to recognize what they have been used to refer to. However, Travis provides little detail about this proposal, and says very little specifically about proper name reference. My view then, although influenced by Travis’s arguments in those papers, is developed largely independently, and in much greater detail.1 Although his work has not influenced me directly, in the way that Travis’s has, Wettstein (1984) addresses a very similar question to that addressed by this thesis, though applied to the reference of indexicals and demonstratives, rather than proper names, and interpreted slightly differently: What bridges the gap between the lexical meanings of referring terms and their reference? Like me, Wettstein’s answer is that features of the context of utterance bridge the gap, thus determining reference. Specifically, his answer is that the 1 Travis has briefly addressed the reference of proper names in particular in a much more recent paper (Travis ms.). In it, he appears to endorse a view similar to mine, but, again, he provides little detail. 159 6.1. Principles of the Pragmatic View same features that make the reference of a term apparent to an audience determine its reference (Wettstein 1984, p. 64). It is clear that Wettstein intends that a variety of extra-linguistic, contextual factors—or what he calls ‘extra-contextual cues’—can come into play to contribute to the reference of demonstratives and indexicals, though it is less clear to what extent this is supposed to be represented in the semantics. It appears that Wettstein is, unlike me, conceiving of reference as a semantic phenomenon, rather than considering that his claims about what determines reference push it into the realm of the pragmatic. Another difference between Wettstein’s claims about indexicals, and mine about proper names, is that Wettstein says that the contextual cues that determine reference are those that convey a speaker’s intended reference, and which the speaker is responsible for and appears to be exploiting. On my view, the contextual factors that audiences rely on to appreciate what has been referred to will very often be those that determine reference, and for good reason—this is clearly an effective way to work out what is being referred to—but I do not specifically claim that those factors determine reference because the audience uses them to recognize the reference. Furthermore, unlike Wettstein, I do not claim that the referent of a use of a name is, or must be, the object that the speaker intends to refer to with it. 6.1 Principles of the Pragmatic View These principles are intended to offer a broad introduction to my pragmatic view. They should not be viewed as axioms or seen to represent a manifesto. Rather, they are something akin to a precis of the most significant aspects of the view, and in particular those points at which it differs from other views. 1. Names, as used in a certain syntactic manner—in English, in subject or object place, without an article—carry an expectation of reference. a) This expectation of reference does not require that objects of reference exist. 2. If an utterance contains a referential use of a name, then the truth of that utterance will turn, in some manner, on the way that the object referred to by the name is. This is part of the nature of reference. 3. Names are context-sensitive expressions. Their reference depends upon the context in which they are used. a) Thus, names do not refer tout court. Only uses of a name, or a name as uttered on an occasion, or speakers using names, can refer. 160 6.1. Principles of the Pragmatic View b) The context-insensitive semantic value of a name is not constituted by its reference. 4. The semantic rules governing the language are silent with regard to the reference of names. a) There may be certain probabilistic constraints on the use of names, which could plausibly hold across occasions of use. But these are essentially different to context-insensitive, determinate meanings, and are, by their nature, defeasible. 5. There are no necessary and sufficient conditions for the reference of proper names. a) The reference of a name is determined on an occasion of use by pragmatic factors of the context. b) There is no systematic way to specify in advance what specific factors will effect the reference of a use of a name. They may well vary significantly from occasion to occasion. Though this is not to say that certain factors will not be relevant significantly more frequently than others. c) There is no determinable collection of all determinate pragmatic factors that can effect the reference of a name. d) Neither speaker nor hearer have any special authority over what a name refers to on an occasion of use: speakers’ intentions do not constrain the reference of names they use, so they can refer by mistake if they fail to appreciate the vagaries of context; hearers can fail in the same way, and so fail to understand what has been referred to with a name. 6. Whilst it might be true that names are used to name things, and that this is a discriminating characteristic of names amongst referring expressions, it is not a feature that governs their referential use. a) Name bearing is an extra-linguistic, social phenomenon that can bear on contexts of utterance. This is not to say that such social conventions cannot arise as a result of a name being used to refer to an object. b) Names can have huge social significance. There might be significant social or moral consequences for transgressing the social norms of name use, but this is a social or moral issue, not a semantic issue. 161 6.2. Names and Reference 7. There are no analytic truths of the form ppNNq refers to NNq, where pNNq is a name mentioned. a) A true utterance of the form pNN is Fq does not entail pNN is a bearer of pNNqq. The reference of an utterance of a name is determined by pragmatic factors of the context. These pragmatic factors are not specified in advance by the semantics of the term. There are no necessary and sufficient criteria for reference independent of an occasion of use, and on an occasion of use there is just what is referred to. If singular reference occurs, it is because the weight of pragmatic factors determine that there is a way to best understand the utterance such that the speaker, or the name they used, counted as having referred to a particular object.2 There is no way to stipulate a set of all the factors that might, on any occasion, affect reference, nor how they might affect it, though we can point to some factors that are likely to play a rôle in a range of circumstances. For any particular utterance of a name, we may be able to point to some or all of the factors that contributed to the determination of the reference of the name, but this will only be relevant to that particular utterance. It need not be the case that any of those factors would contribute to an utterance of that name having the same reference under different circumstances on a different occasion. Nor that further factors might not have changed or undermined the effect of the original factors, had things been different. The reference of a proper name cannot be made sense of independently of the broader utterance containing it. This is because the appropriate way to understand an utterance, which includes interpreting any referential terms, is dependent upon the complete context of utterance, which will include the linguistic context. 6.2 Names and Reference Principle 1. says that when singular proper names occur in utterances referentially, in a grammatically referential position, as in sentences such as Lucy did indeed go to Texas; Tom drove around Swindon for three hours; The story touched Luke and Kate in new ways; 2 See Travis (2008, p. 9) for the way in which these considerations bear on truth rather than reference. 162 6.2. Names and Reference Have you seen Jupiter recently?; Bruntsfield is part of Edinburgh; they carry a certain expectation that they refer in some manner. This has certain consequences for the truth of the utterances containing them, as made clear in principle 2. Principle 1. is perhaps the least controversial of all, and would be accepted by most of the theorists discussed in this thesis. Exceptions might be Russell (1905), who thought that, as disguised Russellian definite descriptions, proper names found in ordinary language do not, strictly speaking, refer. Others who are influenced by Russell in various ways may also deny that, in a certain sense, names refer. Bach (1987, 2002) and Katz (2001) hold that names are equivalent to Russellian descriptions of the form pthe bearer of NNq, but that in the case of any name that is shared, the uniqueness condition of the Russellian description fails and so the name fails to literally (semantically) refer. Barwise and Cooper (1981) also treat names (and all nouns) as quantifier expressions, and so not referential in a certain sense, but their interest is primarily model theoretic, not in analyzing how words of English actually function. 6.2.1 Reference and Truth As I discussed at the beginning of §2.1, I take proper name reference to be the phenomenon that leads utterances containing uses of proper names to be about a particular object (or objects, where an utterance contains multiple uses of names). So, the names ‘Lucy’ and ‘Texas’ in an utterance of ‘Lucy did indeed go to Texas’ refer to the things that the sentence says something of, namely that the former went to the latter. Something is the referent of ‘Lucy’ in that sentence if and only if the truth of the utterance turns on whether that thing in particular did indeed go (or counted as going) to Texas, and something is the referent of ‘Texas’ if and only if the truth of the utterance also turns on whether the referent of ‘Lucy’ did indeed go (count as going) to that thing in particular. This idea that, if an utterance contains a name or other referring term (and is not reporting upon someone’s belief or attitude), then the truth of the utterance turns on how the particular thing referred to is, gives us what is sometimes known as direct reference: The way in which the reference is determined does not appear to play a rôle in how the name contributes to the truth of the utterance. The fact that we can state, in more or less vague terms (what is it for truth to turn on the way an object is?), what the relation between reference and truth is for any given utterance containing a proper name is not an indication that 163 6.2. Names and Reference we have something approaching a definition of reference. Reference enables us to say things about particular objects, but it is not something separate from saying something about a particular object. We can say things about particular objects, and to do so is, generally, to refer to them. If I say something about a particular object, then whether or not I have spoken the truth will depend upon whether things are the way I said they were with regard to the object. None of this is very deep, but then, neither is the connection between reference and truth—which is not to say that truth is not essentially connected to reference. Utterances containing more than one name, such as the one mentioned above about Lucy and Texas, provide one reason that we cannot create a general rule of the form pa name NN refers to an object x on an occasion of utterance if and only if the truth of the utterance containing NN turns on how x isq3 since the truth of sentences containing multiple names is likely to rely on how more than one object is, and we do not want ‘Texas’ to refer to Lucy just because the truth of an utterance containing ‘Texas’ turned on how Lucy is. Something like pa name NN refers to an object x on an occasion of utterance if and only if the truth of the utterance containing NN, p. . . NN . . .q, where the content of p. . . _ . . .q is φ(_), turns on whether φ( x )q might work as a general principle or heuristic for how to create descriptions of the relation between reference and truth of the kind given above, but the inadequacy of the formalization of disquotation, and of the content of an utterance, mean that this could not provide any thing like a definition of reference in terms of the truth of utterances, nor should we expect it to. The relationship between reference and truth, although significant, and useful for elucidating reference and its importance in using language to express thoughts, does not explain everything that there is to explain about proper name reference. Proper names can, of course, be used in a whole range of utterances for which truth is not a relevant condition of evaluation because they do not aim at expressing truth evaluable thoughts. Names can be used to refer in questions, jokes, imperatives, wishes, and a variety of other nondeclarative utterances. What all these uses have in common, however, is that 3 Read as a plain conditional from left to right, however, we do get a true schema: pa name NN refers to an object x on an occasion of utterance only if the truth of the utterance containing NN turns on how x isq. 164 6.2. Names and Reference the name will make the utterance about its referent in some way. If there is some condition of correctness or evaluation, that will turn on the referent. There are also declarative (and other) utterances that contain referential uses of names in such a way that it is unclear that the truth of the utterance turns on the referent, even if we allow that the notion of truth turning on something applies to the objects of multi-place predicates as well as their subjects. For example, complex referring terms that contain names may present a quandary: does the truth of ‘Greenberg’s girlfriend went to Edinburgh’ turn on how Greenberg is? In a sense, no, it turns on how his girlfriend—and Edinburgh—are. Such an utterance is not obviously about Greenberg. But it still seems that the occurrence of ‘Greenberg’ refers to Greenberg (whomever he is), indeed, it must do so in order for the description containing it to refer to his girlfriend. Although such a sentence is not equivalent to something like ‘Greenberg is such that (the person described as) his girlfriend went to Edinburgh’, something like this does seem to be entailed by the initial utterance.4 The truth of this utterance certainly does turn on how the referent of ‘Greenberg’ is. 6.2.2 Reference and Empty Names The somewhat more controversial part of principle 1. is part a), which says that the referential expectation of names does not require that their referents exist. I include this in order to allow for the use of names to talk about nonexistent things, including fictional characters, within a unified account of names. As Sainsbury has widely observed (2005, 2009, 2015), names that have an accepted empty use within a community, such as fictional names, or names such as ‘Vulcan’, can be used and understood in very similar ways to the names of existent objects. My interpretation of this feature of names will be that names can be used to talk about things that do not exist5 in just the way that they can be used to talk about existent objects. Exactly what to say about such cases is unclear to me, however. These look like case of reference, but we might not want to say that the uses of names in question refer. Certainly, my conception of reference is not one of metaphysical connection between word and object, as it has been for some reference theorists, and this might 4 The nature of reference using descriptions means that the referent of ‘Greenberg’s girlfriend’ may not be the girlfriend of someone named ‘Greenberg’, and indeed need not be anyone’s girlfriend. We must therefore imagine that the context of utterance can be maintained such that whomever was the referent of ‘Greenberg’ in ‘Greenberg’s girlfriend went to Edinburgh’ is also its referent in ‘Greenberg is such that (the person described as) his girlfriend went to Edinburgh’. 5 My use of the word ‘thing’ should not be taken to have any kind of existential import. By using the expression ‘things that do not exist’, I am not meaning to commit myself to a Meinongian realm of the non-existent. 165 6.2. Names and Reference make things easier for me. I will touch on uses of empty and fictional names with regard to my pragmatic view of name reference again in §8.3.3. 6.2.3 Reference Over Time Another controversial claim about reference that bears consideration is that it need not be thought of as one-time, momentary relational event. Referential terms such as names may be what brings reference into a sentence or utterance—what makes them apt for being singular referential sentences or utterances—but this does not entail that the reference comes exactly at the same time as the occurrence of the referential term and is thereby fixed indefinitely. Gregoromichelaki et al. (2011) argue that there is linguistic and psychological evidence that, in dialogue, anything that resembles (what philosophers think of as) propositional content tends to emerge incrementally over the course of the dialogue, rather than being expressed by each single utterance. They appeal, in particular, to split utterances (one utterance being split between two or more interlocutors) and clarification requests (one interlocutor asking another to clarify a previous utterance). If the propositional content of an utterance can emerge over the course of a dialogue, in virtue of the interaction between speakers (who, for Gregoromichelaki et al. are each constructing their own representation of the content), rather than all at once with each speaker’s utterance, then perhaps the referential content can also emerge incrementally. Consider the following plausible dialogues: A: I saw Lucy today. B: Which Lucy? A: Lucy Horowitz. A: What do you think of Lucy. . . B: Brown? I don’t like her. She’s so mean. A: No, no. Lucy Horowitz. In both these cases, whilst one could maintain that the first speaker succeeded in referring, given the rôle of their interlocutor in bringing out who is being talked about, I think it makes as much sense to consider the reference as having emerged during the dialogue, rather than at any particular time. If this idea is taken seriously, then the initial utterances of speaker A should not be considered as individual utterances for the purposes of interpretation, or rather, interpretation is incremental over the course of the discourse or split utterance. So, just as we might not think that anything truth-evaluable has been expressed until someway into a discourse, we might think that what the 166 6.3. Context-Sensitivity utterance is about, and thus what has been referred to by the occurrences of names, does not emerge until after the speaker B has spoken. This position is clearly at odds with those views of names that maintain that they refer tout court, or entail that the reference of a name essentially occurs just at the time it is uttered. 6.3 Context-Sensitivity I mentioned briefly in chapter 3 that there was no obvious reason for assuming that names refer tout court. This arises partly from the idea, influenced by Wittgenstein, Austin, and Strawson, amongst others, that reference is something that is done with words, not something that words somehow carry with them. I have talked throughout the thesis of names, or uses of names, or utterances of names, referring. When I am talking about my own view, or about what actually happens with name reference, rather than outlining a rival view, these locutions should all be taken to be reducible to something that a speaker is doing with a name, just as a bucket carries water only to the extent that an agent uses it to carry water. Certainly, and very obviously, names cannot be used to refer independently of a particular utterance or inscription (or, just perhaps, something like a mental tokening), and so it is unclear what would be lost by denying that names refer independently of either a particular referential use, or when playing a part in a larger referential expression that requires them to pick out an object. The idea that names refer tout court appears to be simply an artifact of a certain assumption about what their semantics must be like. That assumption, based in truth-conditional semantics, is that words have meanings that are the kinds of things that can be composed into a truth-condition for a sentence. If one assumes that names are not context-sensitive, and that all they contribute to the truth-condition for a sentence is their reference, then it makes sense to think of a name’s reference as its meaning. As was made clear in §2.3, I do not think that the meanings of words compose into truth-conditions for sentences. Thus, even if I do think that a name contributes just its reference to the truth or correctness of an utterance in which it is used, there is no need for me to maintain that its semantics specifies what its reference should be. I have not argued for the position that word meanings are not truthconditional, nor will I, though I discuss it further in §6.7. I have argued and indicated, however, that names are context sensitive. The most obvious argument for this is that names are shared, as I argued in §3.1 (at least, in virtue of arguing against its converse). My arguments for that position were not definitive, since they essentially amount to arguing for a particular 167 6.4. Semantics and Reference conception of names on the basis of intuitions. A stronger argument for the context-sensitivity of names is found, I think, in the generally problematic nature of both of the most plausible and prominent types of accounts of proper names that maintain that they are not context-sensitive: causal-homophonism and intention-based accounts. Whilst I do not claim, in chapter 3, that I have refuted causal-homophonism, I do show that it is a substantially more complicated and less attractive account than its supporters claim. If we reject causal-homophonism on this basis, as I believe we should, being, as it is, both counter-intuitive and overly complicated, we are left with little option but either to accept an intention-based account of proper name reference, or to accept that names are context-sensitive with regard to their reference. In chapter 5, I discussed intention-based accounts and found the two most prominent versions of such an account—constraint-based intentionalism and neo-Griceanism—to be implausible, either as accounts of proper name reference, or in general. Again, my arguments here were not definitive, but raised significant enough concerns about plausibility that another account of name reference is to be preferred. This leaves us with just one option: that proper names are context-sensitive: principle 3. The question we must then ask is: in what manner they are context-sensitive? In chapter 4 I considered indexicalism, the most prominent family of views of proper name reference according to which it is determined in a contextsensitive manner. I argued that, whilst indexicalists are quite correct that it is only upon particular occasions of utterance that names refer, and that the reference of an utterance is determined by features of the context of the utterance, they are wrong to suppose that the way in which this happens is specified by the semantics of the name. In §6.4, I will discuss the relation between semantics and reference further, and reiterate my arguments that there is nothing in the semantics of a name that determines the reference of its uses. 6.4 Semantics and Reference My arguments against indexicalism in chapter 4 took the general form of arguing that there is no way of constructing a semantics for a proper name such that it will be able to determine the reference of the name on every occasion of use. This provides principle 4. Whatever features of context are appealed to to determine reference, they will be either too general, such that they cannot predict what the referent of a name will be on an occasion, and will not decide between potential referents in all cases, or they will be wrong, such that they predict the wrong referent on some occasions. As I made clear, 168 6.4. Semantics and Reference it is very hard, if not impossible, to make a general argument to this effect. The best that one can hope for is to point out the ways in which any proposed account that claims to provide a semantics for names that determines their reference falls foul of the dilemma just posited. It is worth noting that my position is not essentially a sceptical or pessimistic one. My proposal that the reference of proper names is determined on each occasion of utterance by pragmatic mechanisms in a way that is not specified by their semantics is a positive thesis. As I observed in §2.3, I have positive motivation for my view: providing an account of proper name reference that is consistent with a radical contextualism such as that proposed by Travis. It so happens that the correctness of this kind of radical contextualism, about both language and thought, is most clearly seen against the backdrop of a mainstream of truth-conditional semantics and propositional conceptions of thought.6 But this does not mean that rejecting the notion that there are necessary and sufficient conditions for the correctness of words and thoughts is motivated primarily by a pessimistic meta-induction on failed attempts to provide them. Of course, it is by observing the way that language is used, and how it relates to the world, that one reaches conclusions in the Wittgensteinian, Austinian, or Travisian vein, but it would be quite wrong to think that, in doing so, one must have postulated and rejected categorial rules, as if that were the only way to do philosophy of language. My claim that the semantics of names does not determine their reference stems from the idea, stated at the beginning of §2.3 that semantics does not aim at providing the kind of content that is subject to correctness conditions. So, whatever the semantics of the words of an utterance provides, it is not anything truth-evaluable, or fit for expressing thoughts about the world. Since the reference of a name is something that contributes directly to the truth- or correctness-evaluability of an utterance containing that name—or a thought expressed by such an utterance—it seems reasonable to think that the reference of proper names is not determined at the semantic level. Moreover, semantics also does not appear to be in the business of specifying the way in which words are to be understood, given particular contexts, and so it does not specify how the reference of proper names is determined. As I made clear in §2.3, I have not argued that these premises are correct, or that my claims follow from them in the ways I assume. Rather, I have argued for my claims by arguing against all the ways that it has been proposed that semantics determines reference, or specifies how reference is determined. I discussed, in §6.3, how I have argued for the claim that proper names are context sensitive in respect of their reference by rejecting causal-homophonism 6 The kind of views Travis (2006b) calls psychologistic. 169 6.4. Semantics and Reference and intention-based accounts of names. Indeed, in chapter 3, I argued that detailed consideration of causal-homophonism should lead one to think that there is no one particular way in which the reference of proper names is determined: pluralism about proper name reference. Discussion of a whole range of accounts of how the reference of names is determined, and the observation that each account does indeed work in a range of cases— frequently different kinds of case—might well lead one further towards pluralism about the determination of proper name reference. This leaves us with the position that the reference of proper names is context sensitive, and that it can be determined in a variety of different ways. In chapter 4, I considered indexicalist accounts of proper names, which I took primarily to be accounts that maintained that the reference of names is context sensitive, but it is still determined in a way specified, or at least constrained, by their semantics.7 As I mentioned at the beginning of this section, my approach to these views was to work through the literature and show that none of it is successful in showing that the semantics of names determines their reference, and then work through the most relevant and plausible contextual factors that might affect reference, and show that none of them could either determine or constrain reference on all occasions. Given my view of semantics, as the context-insensitive contribution that words make to each utterance of them, if semantics were to determine or constrain the reference of names, and names were occasion sensitive, then some particular contextual factors would have to constrain or determine the reference of every utterance of a name. Thus, having shown that none of the contextual factors that seem relevant to proper name reference do constrain or determine the reference of every utterance of a name, I have shown that it is implausible that the semantics of names do determine or constrain the reference of names. But, since I also claim to have shown that it is highly plausible that name reference is context-sensitive, then it must be determined on each occasion by pragmatic mechanisms, independently of semantics. 6.4.1 Necessary and Sufficient Conditions The positions regarding proper name reference that I am opposed to are essentially those that maintain that there are necessary and sufficient conditions for it. I regard one of the primary problems of philosophy—that is, a stumbling block for the discipline, rather than a major object of its investigation—to be the assumption that there are necessary and sufficient 7 As discussed in that chapter, Rami (2014) seems to be an exception to this characterization of indexicalism. 170 6.5. Reference and Bearers conditions for membership of the extension or satisfaction of terms, concepts, properties, etc. If we want to capture the way in which speakers actually use proper names to refer to things, then we must attempt to observe just that, and claim just what the evidence allows. We should not develop a theory (in the sense of providing conditions) about what names refer to, based on a few limited cases, or a conception of what reference should be like, given certain preconceptions, and then try to make all cases fit the theory, dismissing those that do not as deviant or defective. There may be problems with the evidence available, as noted in §2.2, but this does not justify picking and choosing within it. As made clear by principle 5., It is my strong positive thesis that there are no necessary and sufficient conditions for proper name reference, and I believe the evidence bears this out to the extent to which it is able. As I have made clear elsewhere, in claiming that there are no necessary and sufficient conditions upon the determination of proper name reference, I am not claiming that there is nothing to say about what determines reference in particular cases. Nor am I claiming that all we can say about cases is that context or pragmatics does the job in some wholly opaque manner. Indeed, my claim is that both speakers and hearers are sensitive, albeit not necessarily consciously, to the factors that determine the reference of an utterance of a name on an occasion. Travis (1980) expresses this position with regard to reference determination well, having the following to say about the idea that there are statable conditions that determine reference on every occasion of utterance: [It] may well be that for any particular case of reference making, there is some identifiable (and finite) set of facts about it such that it can be said correctly that, in virtue of those facts, such-and-such a reference was made, and such-and-such referred to. It may further be that when we have recognized those facts, or factors, we can also be in a position to be certain that, in that case, there aren’t any more relevant facts to be recognized—none, at least, that might alter the results we could otherwise reach. Nevertheless, there may be novel cases in which all those facts hold, but where we can, e.g., learn that there are other things one ought to attend to in figuring out what reference was made. In other words, though we may be finite organisms with creative capacities, perhaps the finite devices we actually rely on simply don’t determine, by themselves at least, what we will (or ought to) do in every case. (ibid., pp. 154–155) 6.5 Reference and Bearers I discussed bearers at length in §4.3, and offered there reasons why namebearing relations should not be considered essential to proper name reference, 171 6.5. Reference and Bearers and cannot play the rôle of contextual parameters that contribute to the determination of name reference. Indeed, since names can be used to refer to things that do not bear them, either at all, or in the particular context of utterance—as was made clear by various examples, such as those of ‘Leningrad’, and Lucy’s introduction of Harris to John—it cannot be necessary that the referent of a name bears that name. However, clearly, in many cases of proper name use, the referent will be a bearer of the name, in some manner. We do use names to name things, and complex social structures have grown up around this, such as having a stock of names which are used and re-used to name things. Frequently, different names are used to name male- and femalegendered people, settlements, other geographical features, planets, buildings, etc., and such practices are highly culture- and community-dependent. None of these things are encoded in the semantics of names. I am in agreement with Recanati (1993) that naming practices are social, non-linguistic conventions. However, contrary to Recanati, I maintain that such practices do not provide non-defeasible limits on reference, and are not specifically appealed to by the semantics of names. 6.5.1 Social Practices Learning I find convincing Recanati’s arguments that naming and name-using practices are social, rather than linguistic conventions, mentioned in §4.1.1. He observes that, in learning to be a competent speaker of a language, one need not learn the particular bearers of any particular names, but just how, in general, to utilize names to refer. This is in contrast to other kinds of word: in learning how to use the word ‘shoes’, one must learn that it names shoes, but in learning how to use the name ‘Harry’, one need not learn that it names any one in particular. A parallel point can be made by observing that dictionaries do not contain proper names with lists of their bearers, nor would it seem appropriate for them to, although they contain attempts at characterizing the ways that most other types of word are used. It is worth noting that it seems very likely that children in fact do learn how to use names by exposure to particular objects that bear them. However, by the time they are mature speakers, they will (in general) be able to recognize when words are proper names, and how to use any proper name, even if it is novel to them—either a novel name, or a novel referent or bearer of the name. As I mentioned in §4.1.1, it is conceivable—though very unlikely—that a generally competent speaker might never have been exposed to any naming practices at all, and thus not use names to refer to any bearers of them at all. 172 6.5. Reference and Bearers Complexity A further argument that name bearing and naming practices are social phenomena that are not encoded in semantics might be found in their complexity. As indicated above, the ways that names are used to name things—to create continuing practices of use—involve complex and contingent social factors. If naming practices were linguistic conventions, then these complexities would have to be contained within the language. The fact that the boundaries of London substantially changed in 1923 would have to be reflected by a change in the English language, and whatever governmental body that made the decision to revise the boundary was unwittingly also ruling on the language. Indeed, it was not merely English that was ruled upon, but every language in the world, since they all have the capacity to refer to London, and so must have a linguistic convention associating a word with the place. Apparently, in Denmark, and, until recently, in Germany, it was a legal requirement that a person’s assigned gender be ascertainable from their given name. According to the view under consideration, these would be laws governing the German language, rather than laws governing social practices. Of course, there being such laws would not mean that, in Danish, it is impossible to create a linguistic convention of using a traditionally male name to name a girl, merely that such a move would be illegal. It might mean, however, that, as a matter of the way the language is used, the language itself does entail that certain name-bearers are male and certain name-bearers are female. In the same way, a propounder of the linguistic convention position would maintain that the language of the Wagera Indians, discussed by Evans (1973), and mentioned in §3.3,8 allows one to infer a speaker’s name by knowing certain facts about their family. It seems clear that such inferences are not about linguistic conventions. They are not of the same kind as inferences that if somebody says that something is dry, then, according to them, it is not wet, which would seem to be an inference about linguistic conventions. Rather, they appear to be inferences involving the social practices of a community. Names that have previously always been given to boys can be used to name girls without changing the function or meaning of the names. Furthermore, names that are usually assigned only to places, or have, as a matter of fact, only been used to name male dogs, can be used to refer to people, given the right circumstances. This brings me on to further arguments to the point that not only are name/bearer relations and naming practices 8 Note that the internet seems to contain no mention of Wagera Indians, except in relation to Evans’s discussion of them. 173 6.5. Reference and Bearers social, rather than linguistic phenomena, they are not encoded in any way into the semantic meaning of names. And here I depart abruptly from Recanati. 6.5.2 Nicknames and One-Off Names Though, like me, Recanati (1993) claims that practices that associate particular proper names with particular objects are social conventions, he also claims that there is a general linguistic convention such that, for any use of a name, the referent is, in virtue of social convention, a bearer of that name. It is this latter linguistic convention, contained in the meaning of each name, that I deny exists. In addition to the arguments I provided in §4.3 for the position that bearer relations cannot play a rôle in the semantics of a name such that they contribute to or constrain the determination of the reference of the name, I now consider how the use of nicknames might also lead one towards such a position. How Nicknames are Introduced Many nicknames appear to function in the same way as more official forms of proper name. Though a nickname could have a different grammatical form to most proper names in English, for example, by having the form of a definite description, many are bare nominals like paradigmatic proper names (of people), and can be used referentially, vocatively, and probably even predicatively, just like other proper names. Like other proper names too, nicknames can be introduced in a range of different ways: by conscious and intentional naming, by simply starting to use the name referentially, and by various forms of accident. However, bearing a nickname does seem to involve an established practice of the name being used to refer to an object. This makes them somewhat different to, say legally bestowed names, because, arguably, simply having been assigned a legal name is enough for it be an object’s name, whether or not it is used to refer to it with any frequency. Indeed, it appears to be part of what makes a nickname a nickname that it is unofficial in some sense, and receives its status as a name for something from the fact that a practice of using it has arisen. This does not mean that a nickname must have arisen without an explicit dubbing—it is perfectly conceivable that someone might say to someone or something else ‘I’m going to call you “Squirt”’—but, unlike a parent filling in a birth certificate, or a civil servant filling out a town charter, a practice of using the nickname must emerge in order for the dubbing to have made a bearer of the object. Or so it strikes me.9 9 See Ziff (1977, p. 320) for a related discussion of nicknames. 174 6.5. Reference and Bearers How Can a Nickname be Borne? This is what might be involved in an object having a nickname: there being an established practice of using the name to refer to the object. This, in itself, presents a problem for accounts of proper name reference that maintain that the referent of a name must bear that name. If an object must bear a nickname in order for it to genuinely be used to refer to it, then how is a practice of using the name to refer to the object ever supposed to emerge, such that the bearing of the name is grounded. This kind of problem has been addressed at various points previously, most obviously in §§3.2.3 & 4.1.1. The answer is generally that speaker reference takes care of these cases by providing a means for speakers to refer to things with names—in some sense of ‘refer’—even if an account of the meaning of the name doesn’t allow it. Indeed, nicknames, thought of as I have suggested, at first appear to suit the producer/consumer dichotomy put forward in Recanati (1993) rather well: those users who initiate the use of a nickname to refer to a particular object are speaker referring to it, and when their practice becomes ingrained enough, they become the producers of a practice within a community, and the object comes to be a bearer of the name. Anyone who subsequently comes to use the name because it is a name for the object is a consumer, whose use gets its reference in a context because of the practice. However, it is very unclear that this picture really gets right what actually happens when a nickname is initiated and then crystallizes into a naming practice such that the referent of the nickname can be said to bear it. Suppose, one evening, one friend calls another ‘Julio’, no one ever having done so before. Suppose this use is vocative, rather than referential in the sense I have been using ‘referential’ up to now. The addressee of a vocative use of a term is often easy to ascertain, since they must generally be pragmatically perspicuous. Now imagine that another friend, who is present that evening, and enjoys this use of ‘Julio’, also uses it vocatively to address the same friend. A couple more friends, having observed these vocative uses, join in, and ‘Julio’ is used both vocatively and referentially for the rest of the night. The use of ‘Julio’ might stop there. If it did, we would probably not say that ‘Julio’ was now a name for its referent, or, perhaps we might say that they bore it briefly, just for that evening. But certainly the name was being used to refer to a particular person on that night, and it was being used in that way because other speakers had already used it in that way. We might say that everyone on that night was merely speaker referring, but if they were, their speaker reference looks pretty similar to the reference of people using ordinary names to refer to established bearers. 175 6.5. Reference and Bearers Now suppose that the ‘Julio’-use does not stop there, but carries on. Friends who were not present for the first uses observe the use of the name and begin using it to refer. At some point, though very obviously not at any definite point, we can say that a name-using practice is established, and ‘Julio’ is now a nickname for the object of the practice. They now bear that nickname. Are we to suppose that, at that point, everybody’s use now changes from speaker reference to semantic reference? Or perhaps just those initiated to the practice after it is established refer semantically, and those who were there at the beginning continue to speaker refer. Even before that can be said to have happened, however—when the practice is still nascent—someone who does not know the referent can hear it being used when the referent is not present, and, under the right circumstances, can use ‘Julio’ parasitically to refer in the way that the existing users do. An established bearer is not required for such reference. One-Off Names The uses of nicknames, or nickname-like terms present further problems for the idea that proper names require bearers and established practices, however. In a way similar to that in which ‘Julio’ was used on the very first occasion, words that strongly resemble proper names in their grammatical form and use can be used referentially without there being any pre-established practice. Indeed, names that do already have bearers can be used to refer to other things without simultaneously ‘semantically’ referring to their bearer. Consider someone wearing a t-shirt that says ‘BROOKLYN’ on it. The inscription refers to the place, Brooklyn, a borough of New York City. Suppose someone else, seeing the t-shirt wearer across the room, says to a third person, who can also see the t-shirt wearer, ‘Brooklyn’s got a silly haircut’. They can refer to the t-shirt wearer using the name ‘Brooklyn’, making use of the perceptual circumstances they share with their interlocutor. Despite the fact that they are using a name that is borne by a borough of NYC, they do not also refer, in some different way, to that borough. This is also not a case of deferred reference, referring to the person in the t-shirt via the borough, in the way that it is postulated that one can refer to the US Federal Government using ‘Washington’ via the city in which the centre of that institution is based. Of course, it seems that it is largely a coincidence that it was an existing proper name used to refer in the scenario. Any term could have been used. For example, the speaker could have said ‘Look at Mullet’, making use of another mutually-perceptible attribute of the object they are trying to bring to the attention of their interlocutor. But, whether or not a word that already exists 176 6.5. Reference and Bearers as a proper name is used, it appears that the reference occurs—or does not—in just the same way. Devotees of semantic views of proper name reference will claim that these kinds of case are simply of a different kind to the use of proper names to refer to bearers of those names in line with established practices. In the cases I have provided, they will say, what I am calling reference is simply the communication of the speaker’s intention to refer. Uses of established names refer not merely because of a speaker’s intention, but in virtue of a conventional meaning of the name, whether that meaning is the reference itself, or it is a character that achieves reference by selecting a social practice with a reference. I do not deny that there are social conventions and practices of using certain names to refer to certain objects. To do so would be absurd. People and pets and pubs and places and a variety of other kinds of things that are significant and worthy of repeated identification, are given names, or achieve names, in some way. And those things having those names enables speakers to refer to those objects in certain circumstances more easily than they otherwise would have. However, this does not mean that there is a special kind of reference consisting of using names in accordance with these practices. Recanati argues that name-using practices are features of contexts of utterance. I agree. But, unlike Recanati, I do not think that the meanings of names need to appeal explicitly to these particular contextual features. Rather, I claim that names are referential terms whose reference is determined by the context of utterance in a way not pre-specified by their meaning. Various different pragmatic mechanisms and features of the context can contribute to the determination of the reference of a use of a name. In the ‘Brooklyn’ case, the speaker was able to use a visual cue available both to them and their interlocutor in order to refer to a particular person, together, perhaps, with their direction of gaze, or other demonstrative cues, and the fact that the purpose of their utterance was clearly not to refer to any bearer of the name ‘Brooklyn’. In the initial ‘Julio’ case, the fact that all interlocutors had observed the name being used vocatively to obviously address a particular object enabled those interlocutors to use the name to refer to the same object. In cases in which a name is used to refer to an established bearer, a range of pragmatic mechanisms may still play a rôle in determining reference, but existing practices of using that particular name to refer to particular objects will, very probably, be prominent amongst them. As I have stressed throughout this thesis, naming practices cannot be the only pragmatic mechanism at work, because there are shared names. But, the fact that names are so often used in line with established practices, means that naming practices are very likely to 177 6.5. Reference and Bearers be a prominent factor in the determination of the reference of many referential uses of names. 6.5.3 A Defeasible Constraint? One thing that my discussion of the relation of bearers to the use of names raises is that names lend themselves rather easily to reuse. In the example of ‘Julio’, once the name had been used with a particular reference, it was natural for it to be used again to refer to the same object. In the ‘Brooklyn’ case, it’s easy to see how a similar, albeit short-lived, practice could have arisen. This seems to be the intuition behind the causal-picture of reference: once a name has been used to refer to something once, or a few times, the very fact that it has been used before in that way seems to make it easier to use again in the same way. Unlike for the causal-theorist, however, my claim is that this fact is not in itself either sufficient or necessary to determine reference, nor does the fact have an effect on the kind of reference that speakers or their words can exhibit. As I have tried to show, there is no requirement that the referent of a name is a bearer of it, or even that it has been used to refer to it before. However, it is also clear that established social practices of name use frequently do play a rôle in determining the reference of a use of a name, together with other contextual factors. As just indicated, ‘Julio’-type cases also indicate that it is not just established name-using practices that can affect the reference of names: short-term practices can arise in which a name gets used to refer to an object for a short time, in virtue of having been introduced by one introductory use. These cases are rather reminiscent of the introduction of a discourse referent in Discourse Representation Theory (Kamp and Reyle 1993), or even the use of anaphoric pronouns. It should be noted, however, that the phenomenon is not limited to proper names, as is indicated by the reference to discourse referents and anaphora: once a particular definite description, for example, has been used to refer to a particular object, even if—perhaps especially if—it does not provide a particularly accurate description of its referent, then that definite description can be reused to refer to the same object in part in virtue of the initial use of the description. If this feature of proper names—that one use with a particular reference sets things up for further uses with the same reference—cannot be accurately captured either with appeals to bearers and social practices in the semantics, or by simply putting referents in the semantics, then we might look for another way to characterize it. So far, my line has been that we should not try to characterize such things in any kind of systematic way. It is a fact about the 178 6.5. Reference and Bearers way we organize our societies, not our language, that we give some objects proper names and then make use of those practices in referring to those objects. It is a fact, I claim, about language that proper names can be used to refer to things, and what they refer to is determined by the context of utterance in a way not specified by their semantics. If we wanted to capture the feature of names I have been discussing, we might want to add to this fact about language in the following way: Proper names can be used to refer to things. What a name refers to is determined by the context of utterance in a way not specified by their semantics, but is likely to be something that it has been used to refer to previously, as repetition increases correlations between names and particular referents and particular situations.10 This is the idea indicated in principle 4. a). This likelihood is not limited simply to what has been referred to with the particular name in the immediate linguistic or temporal context, as is the case for anaphoric pronouns, but can accommodate previous uses within the linguistic community or simply any prior use, given the right context. This additional caveat, about what is likely, captures something of the idea that uses of names are apt to be repeated, but does not impose anything resembling necessary or sufficient conditions or constraints upon their reference. Such a caveat will help explain the expectation of the repeated use of names with the same reference in Julio-type cases, but also the existence of name-using practices, which not only can arise through repeated use of a particular name with a particular reference, but enshrine in social conduct that a particular name has been used with a particular reference before, and so is apt for repetition. However, such a caveat about what is likely cannot be captured by traditional truth-conditional semantics. There is no way to effectively express that something is likely or unlikely in classical model theory or the (categorial) semantic frameworks that have emerged from it and around it. There are only indefeasible rules, and meaning postulates to constrain the admissible models. Although I am not presenting my views within any particular semantic framework or formalism, my assumptions about what a semantics can look like have so far assumed that it will make binary distinctions, as it appears that my opponents have. If those assumptions are maintained, then I claim that the semantics of a name goes no way towards determining its reference. If those assumptions are relaxed, and I can move towards probabilistic semantics, then I believe that it is plausible that the semantics of a name make it more likely that a use of it refers to something it has been used to refer to previously. Though discussion of it is beyond the scope of this thesis, work is being done in developing semantic frameworks that would allow for the meanings of 10 Thanks to Peter Sutton for highlighting this feature of names to me. 179 6.5. Reference and Bearers names to incorporate probabilistic ‘constraints’ on their reference (e.g. Cooper et al. 2014, 2015; Sutton 2013, 2015). Sutton (2013), in particular, discusses at length how his particular Bayesian approach to semantics and pragmatics can provide a non-truth-conditional picture of language in which a Travisian approach to language and meaning can be accommodated, and is even to be expected (pp. 211–212). Sutton observes, with regard to the effect of repetition on the likelihood of particular reference determination: [R]epeated uses [of a name (or other word)] entrench correlations, and correlations between uses of words and described situations give rise to streams of information that can be used by language users to perform certain communicative actions. (Via correspondence) 6.5.4 Analytic Truths There is an apparently widely (though, of course, not universally) accepted adage or axiom that, for any proper name pNNq,11 it is true, or analytically true that pNNq refers to NN (see, e.g. Sainsbury 2005, p. 73). There may be good reason for thinking that utterances with this kind of form are (or can be) frequently true. For example, one can easily imagine an utterance of ‘When I said “Alex”, I was referring to Alex’ being true. As I have discussed widely, a referential use of a name usually licenses further uses with the same reference (given the right circumstances). So, if someone is using the name ‘Katharine’ to refer to someone, other speakers can use ‘Katharine’ to refer in the same way, and then, from their perspective, ‘Katharine’ refers to Katharine. According to causal-homophonism, which is largely inspired by these parasitic uses, as long as it is a specific name with a referent (i.e. not empty) occurring inside the quotation marks, then such sentences will always be true, because each name refers to a particular object tout court. However, whilst these kinds of utterances can be be true, and perhaps often are, according to the pragmatic view of proper name reference, they have nothing of the analyticity that some theorists have ascribed to them. Indeed, if they are true, it is entirely contingent upon circumstance. This is principle 7. On any view that maintains that names only refer upon occasions of use, rather than tout court, if ppNNq refers to NNq is to be true, it must be interpreted as meaning something like pWhen pNNq refers, it refers to NNq, or pUses of pNNq refer to NNq. However, if the second occurrence of the name, outside the 11 Note that this term is inside quotation marks, such that, whilst any name may be substituted for the term, the double ‘N’ variable itself ranges over referents of names. Often, the double ‘N’ notation is used without quotation marks as a variable ranging over names themselves, rather than their referents, however, that use of the notation provides no way of moving from mention to use of a name. 180 6.5. Reference and Bearers quotation marks is supposed to be referential—as it appears to be—then such sentences seem at odds with the idea that names are context-sensitive, a view held at least by indexicalists and myself, and presumably some constraintbased intentionalists. If the name occurring within the quotation marks is capable of referring to multiple objects, then it cannot always be true that it refers to one particular object. There are at least two ways that proponents of context-sensitivity about names might get around this, though I think neither of them work. One way might be to claim that upon any particular utterance of the sentence ‘“Katharine” refers to Katharine’, the context will fix a referent of the use of ‘Katharine’, and so it will be true that, in that case, the name ‘Katharine’ referred to the object that was the referent of the use of ‘Katharine’, who is, trivially, in that situation, Katharine. But this reading does not make the sentence or the utterance an analytic truth, at least, not in the right way. Indeed, if it renders it true at all, it is probably because it simply expressed the idea that the use of the name referred to what it referred to. The simple present tense of ‘refers’ suggests a habitual relation that is absent in the example. A more convincing case of this type, that does not suggest analyticity, might occur when discussing an inscription of a name, the reference of which, might be taken to be habitual: ‘“Mike”, here, refers to Mike. You know: Mike!’. A more likely route for a proponent of context-sensitivity who wanted to maintain the analytic, or axiomatic, or logically true nature of the sentence would be to claim that, in its timeless, sentential form, the occurrence of the name outside the quotation marks is not properly referential. Rather, the whole sentence presents some kind of schema that reflects the fact, as claimed by Recanati (1993), Pelczar and Rainsbury (1998), and Michaelson (2013), that names refer to bearers of those names. Thus, it will always be true that a (non-empty) referential use of ‘Mike’, will, for some Mike, refer to Mike. This principle is indeed enshrined as analytic or axiomatic by some theorists. It is clear from Recanati’s account of the meaning of names that it is entailed by a use of the name ‘Clare’ that the referent of that use is a bearer of that name. Katz (2001) also maintains that there is an analytic entailment from ‘John Stuart Mill wrote A System of Logic’ to ‘a bearer of “John Stuart Mill” wrote A System of Logic’ (p. 139). However, as I have argued at length, it is possible to use a name to refer to an object that is not a bearer of that name, and so there cannot be an analytic entailment of the sort just mentioned from a use of a name. Rami (2014) also argues against such entailments on the grounds that they are unwarranted. 181 6.6. How to Individuate Names 6.6 How to Individuate Names In chapter 3, I considered a way in which not to individuate names: by causal connection to the baptism of a single object. There must be a way of individuating names, otherwise there would be no non-trivial notion of having the same name. However, so far I have simply assumed that intuitions about names can provide all that I need. In this section, I will briefly consider what there is to say about the individuation of names. As for the individuation of anything, that of names is not a cut-and-dried matter. As I have argued, there are reasonably clear cut cases of names being the same, and there are clear cut cases of names being distinct (for example, ‘Patrick’ and ‘Donnatella’). But there are also grey areas, in which we sometimes might say names are the same, and sometimes might not. For instance, do Peter, Pieter and Pierre share a name? In some circumstances the answer might be ‘yes’, and in some ‘no’. In others it might be ‘who cares?’. Kaplan (1990) provides some good reasons for not individuating words by phonological or orthographical similarity, which might be a natural starting point. We might usually want the phonologically identical ‘Jeffrey’ and ‘Geoffrey’ to be the same name with distinct spellings, but perhaps the phonetically similar surname ‘Joffrey’ to be a distinct name. We usually also do not want differences in pronunciation of orthographically identical names to entail differences in identity. For example, I have had some difficulty in making my name, which I pronounce /"pi:t@/ (‘peetuh’), understood in parts of the US, where it is pronounced /"pi:t @~/ (‘peedurrr’). But this difference in ˇ pronunciation does not make for two distinct names. Etymology seems to provide a more workable guide to the identity of names.12 Whilst speakers are often not aware of the detailed etymological history of the names they use, close etymological ties frequently line up with closeness in phonology and orthography. These are things considered by ordinary speakers, and etymology will shut out the misleading cases. Appeals to etymology can therefore allow that, in some sense, ‘Peter’, ‘Pieter’, and ‘Pierre’ are the same name, whilst there is a sense in which two names that just happen to sound very similar, but are from entirely unconnected languages, are not. And, since etymological closeness is a matter of degree, it can account for the variable standards employed in individuating names—on some occasions names may be required to have a high degree of etymological relatedness to count as being the same, and on others, a lower degree. Employing etymology in such judgements also ensures that individuation 12 Thanks to Peter Sutton and Dolf Rami for early discussion of this idea. I believe it was Peter’s suggestion. 182 6.7. Names and Occasion-Sensitivity is about the words employed, and not about the reference of any particular utterance of a name. Of course, it may emerge that etymology cannot, in fact, account for the ways in which people generally individuate words. But once the notion is separated from the determination of reference, it seems likely to be of little philosophical interest. 6.7 Names and Occasion-Sensitivity The picture of proper names and their reference that I have now painted is heavily influenced by Travis, as I have previously indicated it would be. The way that I am claiming name reference is determined—in virtue of pragmatic mechanisms that cannot be systematically codified—is essentially the same way that Travis claims throughout his work on philosophy of language that the truth of utterances, and, most famously, the extension of predicates on an occasion, is determined. As I discussed in §2.3, the determination, in this way, of truth-values, predicate extensions, and other correctness conditions, on an occasion, is what Travis terms ‘occasion-sensitivity’. My position is effectively, then, that the reference of proper names is occasion-sensitive. Occasion-sensitivity is a phenomenon that affects the truth-relevant assessment of linguistic items: Sentences are occasion-sensitive because the truthvalues (or other correctness-values) of utterances of them are determined by occasional pragmatic factors, and predicates are occasion sensitive because whether or not their application is correct—whether or not something falls within their extension—is determined by pragmatic factors on an utterance-byutterance basis. Being able to tell whether or not a predicate is being correctly or appropriately applied on an occasion is something that requires domaingeneral knowledge about the ways the world is on the occasion, not merely knowledge of the semantics of the predicate. If proper names are occasion sensitive, then their reference should be determined by pragmatic factors on an occasion of utterance, as reference is a truth-relevant phenomenon associated with names. In this section I will consider further what it is for a name to be occasion-sensitive, and the way in which they are, as compared to other things. 6.7.1 Occasion-Sensitivity in Thought So far, both here, and, largely, in §2.3, I have described occasion-sensitivity as if it were a phenomenon that only applied at the level of language: to linguistic expressions such as predicates, sentences, and names. Such a position about language would be consistent—at least up to a point—with 183 6.7. Names and Occasion-Sensitivity a form of contextualism that maintained that word meanings do not specify what the world must be like in order for uses of those words to correctly describe it or apply to it, but that, on an occasion of utterance, words can still express that the world is a particular, fully determinate way. That is, that there are determinate, proposition-like thoughts that fully describe the world as being a certain way, and that, on an occasion, an utterance can express such a thought. But this would mean that there are fully determinate, discrete ways that one could say that the world is, and that the world would be, if one were correct in so saying: Consider an assertion that some particular leaves are green. Even if we were to claim that exactly what that assertion asserts or expresses depends upon the context of utterance in a way not specified by the semantics of its words, if we were to claim in addition that, in the context, the utterance asserted that the world (or a part of it) is a certain determinate way—expressed a determinate and discrete thought—then there would have to be a particular, discrete way for something to be green that it was asserted that the leaves were. But if there were such discrete ways to be green, we could presumably enumerate them, and observe that on this particular occasion, the word ‘green’ meant ❣r❡❡♥276 , and it now appears that the contextualism that was being proposed can be redescribed as a mass-ambiguity type view like that I compared to causal-homophonism on p. 59. Travis explicitly rejects the view that there are determinate, discrete thoughts expressed by utterances, and that there are particular determinate ways that things must be for an assertoric utterance to have correctly described them (e.g. Travis 1990). He says: When Pia said ‘Sid grunts’, she spoke on a certain understanding of being a grunter. When Sid said ‘The room is dark’, he spoke on a certain understanding of a room being dark. ‘What understanding?’, one might be inclined to ask. Which may seem to call for an answer of the form, ‘The understanding such that. . . ’, where what filled the blank would uniquely identify some one understanding there is to have of being a grunter, or of a room being dark. ‘It is that understanding on which someone would count as a grunter just in case (he were) such-and-such.’[. . . ] There is another way of understanding ‘a certain way [to understand an utterance]’. On it, understanding being a grunter in a certain way would just be understanding it as it was to be understood on Pia’s speaking of it—understanding it so as to be that of which she spoke. When would one be doing that? Just when one was understanding it as it ought to have been understood in the circumstances of her speaking of it. There is determinacy in what one would have to do to be doing that. In the circumstances, some things would be misunderstandings, others would not. It would be a misunderstanding, say, to think that anyone would 184 6.7. Names and Occasion-Sensitivity be a grunter on the understanding on which Pia spoke of this if, hit hard enough in the solar plexus, he would grunt. One understands Pia’s words in the right way—as they ought to be understood—if one does not lapse into misunderstanding. What one would have to do to do that, on the present story of a truth-bearer, is fixed by nothing less than the significance of the circumstances in which she spoke: that is, by their having the significance they did. Fine. But which understanding is that certain one on which Pia spoke of being a grunter? On the view I recommend, this last question is ill-formed. It has no answer. One cannot count understandings like that. (e.g. Travis 2008, pp. 7–8) Occasion sensitivity is, then, essentially, a phenomenon that arises not just at the level of language, but at the level of thought. 6.7.2 Names’ Contribution to Thought However, with respect to occasion sensitivity at the level of what (utterances of) words express or how they say things are, it might appear that proper names, as I have characterized them, are different from expressions such as sentences and predicate terms. Throughout this thesis, I have focussed on proper name reference in the sense that names refer to particular objects (of various kinds), and I have argued that each proper name is capable of referring to multiple objects, and that what object a name refers to on an occasion of utterance is determined by pragmatic mechanisms. If we consider the way in which ‘green’ is occasion sensitivity, we can say that whether or not something counts as being green is fixed only on an occasion of utterance by the context, but there are not discrete ways of counting as being green such that we could systematically say ‘it’s green in these ways, but not these’. The case of names is different, at least superficially. We can also say that what ‘Lucy’ refers to is fixed only on an occasion of utterance by the context, but in this case there are discrete things that could be referred to as ‘Lucy’, and thereby be Lucy, at least in the sense that we would usually have an idea of how to count the kinds of objects to which we give names. There is a certain sense in which it must be settled what object I am referring to with a name before it can be assessed whether what I am saying about that object is a correct thing to say. And, moreover, that once the reference is settled, which, I claim, requires a certain understanding of an utterance, the contribution of the name does not obviously depend on further nuances of understanding.13 But this does not quite seem to be the case for predicates: it is not necessary to fix upon some particular understanding of ‘green’ before assessing whether something is 13 Travis (1995) makes a similar point (p. 253), but goes on to argue that it is wrong in a certain way. I will consider ways in which things are more complex in subsequent paragraphs. 185 6.7. Names and Occasion-Sensitivity green. So, perhaps proper names are occasion sensitive with regard to their reference at the level of language, but not of thought, because once we know what an utterance of a name refers to, there is a particular object that is related in a predictable way to the truth or correctness of the utterance or the thought it expresses. Is the way that the reference of proper names is fixed therefore significantly different to the way in which the correct understanding of the use of other terms is fixed? It does appear to be different in a certain respect, but I do not think that this difference is so significant that we should think of it as a wholly different phenomenon. It is not surprising that reference and predication should interact differently with the thoughts to which they contribute the capacity for correctness, since they play different rôles.14 Similarly, one would expect utterances of sentences to behave differently to utterances of subsentential parts of speech. But, in fact, there are more ways that names might be considered to be occasion-sensitivity than the way that I have so far considered. Individual objects of the sort that we give proper names, as they are differentiated from one another, are relatively determinate, but the ways that they can be are less determinate. If the occasion-sensitivity of ‘green’ is characterizable as there being ‘many ways to be green’, then one may attempt to simply apply this characterization to names, and maintain that there are ‘many ways to be London’. However, such a use of the name ‘London’ appears to be referential, and so we would presumably be saying that there are different ways to be the particular geographical or political entity referred to (cf. Lewis 1993). Intra- and Inter-Object Occasion-Sensitivity This is a different kind of variability from the variation of discrete referents of a name. This is the idea that a name, even as it refers to one object, or, at least, closely related objects, can be used to refer to slightly different things, or different portions or parts of the same thing, or one thing under different understandings of it, and that this variation is determined pragmatically. As Travis (2006b) maintains, ‘every interpretation admits of interpretations’ (p. 38). For example, London is a city, but its name can refer, correctly and truthfully, to the geographical area the city occupies, or its population, or both, or its government, or its infrastructure, or just Inner London, or Inner London plus a bit. It does not appear that all this is encoded in the semantics of the 14 Travis (2006b), following Wittgenstein (1953), is clear that predicates should not be thought of as behaving like proper names, where names are thought of as straight-forward labels of objects, but this in itself need not rule out that proper names in fact behave more like Travis’s conception of predicates. 186 6.7. Names and Occasion-Sensitivity name, even on homophonist accounts. We might call this phenomenon ‘local’ or ‘intra-object’ occasion-sensitivity. This variability in the reference of names, unlike the inter-object determination of reference that I have so far focussed on, need not admit of determinate, discrete fixings at the truth-evaluable level. Whether these two levels of occasion-sensitivity for names are viewed two distinct phenomena, or a degrees of the same phenomenon, and, indeed, whether one accepts that either or both exist, may depend on how one wishes to properly individuate names. Homophonists, who individuate names in such a way that no name has more than one bearer, may accept local occasionsensitivity, and this would presumably be consistent with that individuation, although it may put pressure on it at edges. Indeed, homophonists presumably must accept that, at least in some sense, the reference of some names varies in the manner indicated for London in the preceding paragraph. Although some uses of place names to refer to their administration or population, for example, may be put down to metonymy, which is often considered a systematic case of deferred reference (cf. Nunberg 1993), not all variation in their use can be so described, and metonymy, as it is standardly understood, requires a core object of reference. Whether ‘London’ refers to the whole area inside the administrative boundary of the London Authority, or the built-up area inside it, or that area separated from the distant centres of population that have a lasting identity independent of London, such as Croydon and Dagenham, or just the inner boroughs, cannot be addressed by metonymy. Since, according to homophonists, the semantics of a name consists just of its reference, any variation in that reference may be accounted for either by positing more names, or by pragmatic mechanisms, though these pragmatic mechanisms need not be of the un-systematizable variety described by proponents of occasion-sensitivity. The variable extent of London, for example, which might be used to provide Travis-style cases, appears to be similar to widely accepted cases of the vagueness of proper names. Vagueness is not deemed to be a threat to the truth-conditional semantics of names, because there’s a sense in which it does not affect what a name refers to (Schiffer 2010). However, there is also clearly a sense in which it does—there are several possible answers to the question ‘what is the population of Los Angeles?’, for example, because the question could be understood as being about the City of Los Angeles, Los Angeles County, or some portion of the wider metropolitan area (cf. Lewis 1988).15 15 Chomsky (2000) makes a somewhat similar point about London, though to rather different ends. 187 6.7. Names and Occasion-Sensitivity Identity and Occasion-Sensitivity Travis himself provides an example of what I have called intra-object occasionsensitivity in his (1995), though for referential descriptions rather than names. He uses the example to show that, counter to Frege’s view of the effects of different senses on co-referring terms in non-opaque contexts, referring terms, with particular referents, do admit of different understandings that can impact upon the truth-value of utterances containing them, and in particular, upon identity statements. In claiming that referring terms admit of different understandings, even after they have had their reference fixed upon a particular object, Travis is claiming precisely that referring terms are occasion sensitive in a way comparable to predicates and sentences, or, put another way that amounts to the same thing, but clarifies where this particular occasion-sensitivity lies, that the identity of objects, or the concepts of objects, is or are occasion-sensitive. Travis’s example is as follows: In the seventeenth century, Vermeer painted a picture, ‘Het Straatje’, whose central subject is a house. Call it Vermeer’s house. Around 1980, rubble was discovered in the cellar of a house. Call this Kok’s house. Kok’s house looked nothing like what Vermeer painted. The rubble, though, was shown to be the old façade of Vermeer’s house. Kok’s house, it seems, though different in nearly every respect, was the result of a series of renovations of (at the start at least) Vermeer’s house. The intuitive view, I think, is that there are two views one could coherently take (though not both at once) as to whether Kok’s house was Vermeer’s house. One could view it as Vermeer’s house, and would sometimes speak truth in saying so (as long as one’s words were rightly understood as describing how things are on that view of the matter). Equally, one could view it as not Vermeer’s house, and similarly sometimes speak truth in saying that (again, in words bearing a suitable understanding of what they thus say). (ibid., p. 256) Travis’s proposal is that on some occasions, given certain understandings of the words, the identity statement ‘Kok’s house is Vermeer’s house’ can be true, and on others, given certain other understandings, it can be false. In both kinds of case, the reference of each respective referring expression is the same, that is, ‘Kok’s house’ refers to one particular thing, and ‘Vermeer’s house’ refers to one particular thing, and in one case that is the same particular thing, and in the other it is not. It might well be that exactly what constitutes each house, materially or otherwise, differs on the different occasions—presumably it must—but that does not appear to mean that there are different houses. Travis is very clear, however, that this apparent occasion-sensitivity of identity does not pose a threat to Leibniz’s law, or other standard principles of the 188 6.7. Names and Occasion-Sensitivity logic of identity. Rather, they simply require clarification that they can only be applied within particular occasions, understandings or perspectives (Travis 1995, p. 270): It will never hold on any one occasion of utterance, in any one context, according to any single perspective, both that Kok’s house is and isn’t Vermeer’s house. Thus it cannot be inferred that Kok’s house isn’t Kok’s house, or Vermeer’s house isn’t Vermeer’s house, as it could be if one was held to be both the other and not the other simultaneously. As Travis’s example employs descriptions, one might reasonable wonder (prior to having read the whole paper) whether it turned on an ambiguity between referential and non-referential uses of descriptions, or on non-rigid reference. However, an analogous example can be created using proper names, as will be required anyway for my purposes, simply by replacing the descriptions with names: Consider an apparently lost ancient city called Sreemrev, known from ancient texts to have existed, but whose precise location has long been forgotten. Suppose it is discovered, through archeological excavation in the modern city of Skok, that the remains of ancient Sreemrev lie under Skok, and that, in fact, contrary to received historical wisdom, Sreemrev was never abandoned for any significant period, but rather, was simply developed and rebuilt over the millennia in the manner of any continuously populated city. It seems reasonable to apply exactly the same reasoning that Travis applied to his Amsterdam houses to Sreemrev and Skok. Under certain circumstances it will be absolutely correct to say that Sreemrev is Skok: It was discovered that this was the case. There is clear evidence of near constant occupation, and of the continuous rebuilding and development typical of a single city. But under other circumstances, and for different purposes, it would be correct to say that the ancient city and the modern city are distinct: they are populated by different and entirely unrelated civilizations (say), one adopting the site and its structures a few years after the other abandoned it, and the two have distinct and disjoint administrative histories; one lies under the other, which is significantly larger; and the location of one was forgotten for a thousand years, whilst the other is a well known holiday location. Reference and Identity I will not spend time rigorously defending the position that identity is occasion-sensitive, as it is not essential for the wider purpose of this thesis. Clearly, I agree with Travis that there is intra-object occasion-sensitivity, and that the pragmatic mechanisms that effect the understanding of a particular use of ‘Vermeer’s House’ or ‘Sreemrev’ may well be the same kinds of mechanism that effect what metonymic or vague understanding a use of 189 6.7. Names and Occasion-Sensitivity ‘London’ or ‘LA’ or ‘Princeton’ receives. But one need not accept this kind of position in order to accept that pragmatic mechanisms determine the reference of a name between wholly distinct objects in the way that I have described. Assuming we do accept the position, however, it might be reasonable to ask whether or not inter-object and intra-object occasion-sensitivity are one and the same phenomenon, or distinct phenomena. The answer to this question, if it is even well-formed, is that it is unclear, but it probably doesn’t matter very much. Certainly, the two kinds of thing can be described differently: one as the determination of reference on an occasion, and one as the determination of identity. But determination of identity presumably involves or is involved in the determination of reference—consideration of the city-type cases certainly suggests so—simply at a more fine-grained level. It may be that, on an occasion, different mechanisms are involved in fixing upon a referent amongst objects and fixing upon a particular understanding of that object, but given pluralism about reference determination, this does not entail that these are different kinds of process or two completely separate phenomena. What the idea of intra-object occasion-sensitivity, or occasionsensitivity of the concepts of objects, shows is that, if one does buy into the idea of occasion-sensitivity being ubiquitous to both language and thought, the kind of occasion-sensitivity I have described for reference is not distinguished from that of predicates and sentences (for example) by being limited to language. Names provide a referent for a thought, but that referent, at the level of language, is not thereby presented with a fixed understanding. 6.7.3 Reference and Travis-Cases If proper names are occasion-sensitive with regard to their reference, then we might expect there to be Travis cases illustrating this, just as there are for predicates, and for the intra-object variability of proper names. What would such cases be like? As I discussed in §2.3.1, Travis cases designed to demonstrate the occasion-sensitivity of predicates hold the meaning of an expression constant demonstrating that the truth-value of utterances containing the expression can be varied by shifting an element of context that is not connected to the meaning. However, there is little or no intuitive grasp on what the meaning of a proper name is, such that it could be held constant whilst the reference is varied. Moreover, the idea that (intuitively) single names can refer to multiple objects is extremely familiar, so generating interesting cases that mirror those for predicates is difficult. Given a slightly different way of understanding the familiar Travis cases for predicates, however, things seem slightly easier. The meaning is held constant 190 6.7. Names and Occasion-Sensitivity while the truth-value varies, which is supposed to show that meaning doesn’t determine truth-values or -conditions. But, exactly what holding the meaning constant means depends on what kind of view is being attacked. If it is the view that meanings are contents (see p. 48) that is at issue, then the intuitive idea of the content of the expression in question is held constant. And if it is the view that meanings are characters (see p. 50), then whatever contextual parameters that a particular proposed character for an expression states contribute to its truth-conditions are held constant. The idea is to use cases to attack more or less particular ideas about what the truth-conditions for particular expressions are. In the case of names then, the task is not to simply hold their meaning constant whilst varying their reference, but to hold whatever a particular theory says determines their reference constant, and then still vary their reference. This is precisely Travis’s strategy in his papers that discuss reference (primarily Travis (1980) and (1981a)). This is a fairly familiar task and, although presenting explicit and fully-formed cases may not have been my primary mode of argument, examples and arguments following this pattern can be found throughout part II of this thesis. For good measure, here is another case that aims both at causal-homophonism and indexicalism: Consider a supervisor at a provincial sorting office who calls out ‘Do we have all the post for London?’. She refers to a very particular area with her use of ‘London’: the London postal area. This is because of her particular post-specific purposes. The boundaries of the London postal area are (mostly) within the administrative boundaries of London, but distinct from anything like Inner London, or London-minus-the-less-Londony-bits. Though, they are themselves quite specific: every address either has, or does not have, a London post code. Now imagine that the supervisor has, as a result of Royal Mail’s privatization, gone to work for Father Christmas in Lapland. She is now in charge of getting all the presents into the right sacks on the sleigh. She calls out ‘Do we have all the post for London?’ (not having got used to the idea that packages and parcels are anything but post). Since Father Christmas pays no heed to post codes or post towns, ‘London’ here has a different referent to the supervisor’s previous use. Perhaps, for Father Christmas’s purposes, London is everything inside the m25, or even all of southeast England. Whether or not one thinks that the fact that the name ‘London’ can refer to the London postal area as well as smaller and larger areas of the UK can be explained by vagueness or by metonymy, it is clear that it did so refer in this case because of the context in which it was uttered. Unless London the city and the London postal area are conceived of as completely distinct bearers of ‘London’, then, for a homophonist, any such context-sensitivity must be extra-semantic, and it seems likely that this would be the case too for an indexicalist. 191 6.8. Objections to the Pragmatic View 6.8 Objections to the Pragmatic View Having described the pragmatic view of proper name reference in some detail, I now turn to two potential objections that might be made to the account. Although the objections might appear significant initially, once they are carefully considered, they do not present a significant problem for the view. 6.8.1 Prediction One objection that seems likely to be leveled at my account of proper name reference is that it says almost nothing about proper name reference, at least in the sense of saying what actually goes on. Rather, it says that proper name reference is not determined in a variety ways that other people have suggested, and then essentially just says that it happens, and that’s all we can say about it. To some extent, this kind of objection is accurate, I simply deny that it presents a problem for me. My position is exactly that there can be no theory of name reference, where that means a list of determinate criteria—necessary and sufficient conditions—for when a name will refer to a particular object. The reference of an utterance of a name depends entirely upon the circumstances in which it is uttered in a way that cannot be specified in a systematic way in advance. But that is not to say that there is nothing to be said about it. For any particular use of a name, or imagined use, we can point to a range of factors and mechanisms within the circumstances that will contribute to the determination of reference. We can even observe that certain kinds of factors are very frequently relevant. Factors such as relevant social name-using practices, the aims, interests and purposes of the speaker and audience, the topic of conversation, the objects that interlocutors or other relevant agents are attending to, or have been attending to, the actions and demonstrations of interlocutors, etc. It is thus not true that my pragmatic account of name reference says nothing about name reference, it simply says something different to what other accounts aim to say. But possibly the most significant part of my account is that what those other accounts set out to do is unviable. A related objection, and possibly the root of the thought that I say nothing about reference, is that, by its nature, my account has no predictive power. Such an objection may assume that theories of proper name reference should be able to make predictions about reference, for, if they do not, they cannot be judged to be accurate or inaccurate. This objection strikes me as rather misplaced, or at least, to hold philosophy of language up to unrealistic standards that, in fact, my opponents also do not meet. It is true 192 6.8. Objections to the Pragmatic View that, by its nature, faced with a scenario in which a name is used under particular circumstances, my account does not provide a theory such that those circumstances can be fed in and a conclusion produced about what the name referred to, or would refer to. In this way, my view appears to be unlike causal-homophonism and indexicalism, because they have the resources to say things like ‘the referent of the use of the name is the referent of the name, which is the object of the baptism that the name is causally connected to, unless the referent of the name has shifted’, or ‘the referent of the use of the name is the most salient bearer of the name in the context’. However, these are clearly only de dicto descriptions that go very little way to actually identifying a referent, and provide little that could, in a genuine case, indicate whether a proposed referent actually fulfilled the criteria. My pragmatic view of name reference can, in fact, produce de dicto descriptions of the referent of name that are not significantly less informative than those just provided: ‘the referent of the use of the name is the object that, in the context of utterance, given the particular circumstances of the occasion, is best understood to be what the speaker referred to with that name’. Like the homophonist and indexicalist cases, this goes little way to identifying a particular referent, but unlike those, it seems like it might offer some idea of whether it was fulfilled by a particular candidate for referent in a real case, even if it cannot itself select candidates. There is another, closely related, way in which my pragmatic view of proper name reference also behaves differently to other accounts. And this could be viewed either as a flaw or a feature of the view. Whilst, as I have pointed out, other accounts have limited genuinely predictive power, especially in real cases, one can of course construct cases in which it is specified that a particular object fulfils their criteria, and one can then attempt to construct cases in which their criteria come apart from an intuitive notion of what is referred to in the case. As I argued in §2.2, ordinary speakers’ judgements about what is referred to in any particular case is the best data about reference that is generally available, even if it has its flaws. In such a case, the theorist proposing criteria for reference has two options. They can either accept that their theory has produced an incorrect prediction, and attempt to modify it, or they can stick by their theory and insist that their prediction is in fact correct, given the notion of reference they are describing—this is often the motivation, or a large part of the motivation, for the distinction between semantic reference and speaker reference—and that any intuitions that something else has been referred to are either mistaken, or they are tracking a different notion of reference. However, if we were to ask what my account of reference says about the very same cases, it would be difficult to know exactly what to say. 193 6.8. Objections to the Pragmatic View The most accurate answer would be that, if the cited intuitions about what was referred to tracked what rational language users who were actually involved in the case as interlocutors or observers would judge, then whatever those intuitions judged to be the referent is the referent. However, my pragmatic view provides no way to judge whether any particular ordinary speaker judgement about a hypothetical case does line up with what language-users actually in the case would judge. Indeed, by the very nature of these kinds of case, and as discussed in §2.2, there is no way to tell what language-users actually involved in the case would judge, because there are no such people. It looks, then, rather like whatever our best data says the reference of a name is, that’s what my account must be assumed to say it is. So it is hard to see how my account could be tested. There are two points to consider with regard to this result. The first is that this is not so unusual a position to find an account of proper name reference in. The second is that this is not actually a bad result for my account, since, although it may not make predictions about cases, it offers an explanation both for why reference is as it is, and why speakers judge things to be as they do, and this will explain why it gets all the cases right. The first point is related to my claim that other accounts of reference are also bad at making predictions. Indexical accounts, in particular, tend to offer a very minimal description of the character or mode of presentation that they ascribe to names, such that, whether or not it is imagined that it could in principle be fleshed out into a complete and determinate truth-condition, they in fact provide rather slim constraints on reference. For example, for Pelczar and Rainsbury (1998), the only constraint that could really be applied to a case is that the referent should have been dubbed using the name in question. The notion of a dubbing being in force (or being prominent) is a largely unexplained contextual mechanism that must be assumed to track language users’ judgements about which objects that are presented in a context are likely referents. Similarly for Recanati (1993): he provides the constraint that the referent of a name must bear that name, but the idea of being the bearer in a context and how to distinguish between multiple bearers in a context are left for context to explain. So, in terms of making predictions about cases, both these versions of indexicalism can distinguish between possible referents of a name on the basis of whether they have been dubbed with or bear the name—and, as I have made clear, this is hardly a categorical distinction—but beyond that, they must rely on an unspecified assessment of the context of the case to judge which of those objects that are dubbed with or bear the name are the referent. And presumably those judgements are going to be based on the same kinds of things that language-users are basing their intuitive judgements on. So, it 194 6.8. Objections to the Pragmatic View looks like these indexicalist accounts of names are only going to differ from the data in their predictions about cases if the data says that the referent is a non-bearer, or is not the object of a dubbing. Their ability to make predictions is thus only very slightly better than that of my account. The second point to be made with regard to the fact that my account appears to be untestable, is that it is unclear whether this really is a point against it. I have acknowledged that it does not have the capacity to make predictions, except by the same means that any data is produced, which leaves nothing to judge it against. However, even if the account doesn’t have predictive power, it does have explanatory power. It has an explanation for why what names refer to is likely to coincide with the judgements that language users make about what names refer to. What names refer to, and what we expect them to refer to, given the circumstances in which they are used, are inextricably linked. Speakers, generally, use the names that they do to refer to particular things in particular situations because those are appropriate names to use to refer to those things in the particular circumstances of those situations. Suppose a speaker uses a name, and that name refers to a particular object upon that use. That it refers to what it does is determined by pragmatic mechanisms of the context, such as the speaker’s purposes and interests, and those of their audience, and relevant name-using practices. These mechanisms are what determine the reference of the name because, as a matter of fact, they are what affect members of the linguistic community’s expectations of what the world is like given that certain words—in this case a name—have been used in the circumstances of those mechanisms. The speaker used the name knowing, in some sense, that particular factors would affect its reference, and how, and they used it for that reason—because it would refer to the object they wanted to refer to in the circumstances. The audience is also attuned to the factors that will determine the reference of the name, and are able to recognize the effect that they have, and thus they can appreciate the reference of the name as it was used by the speaker. Note that this claim is not that the reference of a name is determined by whatever factors allow a hearer to appreciate what was referred to, or not because those factors play that rôle, but rather, hearers are, in general, attuned to particular factors in particular contexts because those factors are likely to contribute to the determination of reference. Given all this, it is unsurprising that the intuitive judgements of language-users about cases of proper name reference are likely to track the same kinds of factors that determine reference, and so be a good guide to what has been referred to in those cases. The fact that my account of name reference cannot provide independent criteria that can be 195 6.8. Objections to the Pragmatic View tested against those intuitions in every case is not a fault with the account, given that the account maintains that there are no such criteria. 6.8.2 Do We Need Criteria? A criticism levelled frequently at the idea that there are no determinate criteria specifying what words contribute to the content of utterances of sentences containing them is that it somehow makes language excessively mysterious, or makes communication impossible. Travis (ms.) quotes Stanley (2007) in this regard: On the pessimistic view, there is stability to word meaning and the significance of the syntactic structure of sentences. But in general there is no systematic way of going from the meanings of the words in a sentence and its syntactic structure to the intuitive truth-conditions of its various utterances.[. . . ] If any version of the pessimistic view were correct, significant facts about linguistic communication would be inexplicable.[. . . ] The pessimistic view is difficult to accept because language users (even at very young ages) smoothly grasp information about the world from sentences they have never previously encountered. Furthermore, they do so given only knowledge of the meanings of a (relatively) small number of individual words and their modes of combination into sentences. If there is no systematic way of proceeding from knowledge of the extra-linguistic context and knowledge of the meaning of individual words and modes of combination into sentences to a grasp of the information about the world that is conveyed by an utterance of a sentence, it is mysterious how language users could so smoothly move from linguistic comprehension to action. In short, if the pessimistic view were correct, the connection between speech and action would be inexplicable. (Stanley 2007, pp. 10–11, quoted in Travis ms., pp. 28–29) This general criticism could be applied to my pragmatic account of names in the following way: if there were no determinate criteria for the determination of proper name reference such that there is a systematic connection between (a use of) a name and its referent, then it would be inexplicable that language users could recognize the reference of names. One initial observation about this objection to my view is that it is only available, on pain of hypocrisy, to those whose own accounts of proper name reference—if they have them—maintain that reference is determined in ways that are recognizable to language users. This rules out proponents of causal196 6.8. Objections to the Pragmatic View homophonism and intention-based views on the grounds that neither causal chains of name uses nor speakers’ intentions are inherently, or even generally perspicuous to audiences (or speakers, in the case of causal chains). Both these kinds of accounts maintain that it is factors other than those that actually determine reference that allow audiences to identify the reference of speakers’ uses of names. However, I have committed myself, in my discussions of communication, to the idea that whatever determines the reference of names should be perspicuous to language-users. So the objection at hand is one that, by my own lights, should be of concern to me. Assuming the objection is brought in good faith, then, there are two ways to respond to it. One is to offer, or indicate, an account of the ways in which word meaning and pragmatics works, upon which the objection is misplaced with regard to the kind of thing my account of name reference might be. The other is to offer an explanation of what can be expected of an account of meaning with regard to communication, appealing to Travis’s own response to the objection. I will briefly appeal to both options, which are, I think, perfectly compatible, and represent different kinds of presentation of different aspects of a correct view of language and communication. The first option is to observe that there are statistical, probabilistic accounts of semantics and pragmatics that are specifically intended to explain how communication can be possible given minimal word meanings that specify very little information about how they can contribute to the content of an utterance in a context. As mentioned in §6.5.3, Sutton (2013) explicitly explains how learning and communication are entirely consistent with occasionsensitivity, given the probabilistic account of semantics and pragmatics that he presents. Again, I will not go into the details of Sutton’s account here, but it will suffice to observe that the kind of objection Stanley presents only arises given a particular conception of communication that goes hand-in-hand with a categorial, truth-conditional semantics. Once that is abandoned in favour of, for example, the information-theoretic view of communication presented by Sutton, together with a probabilistic conception of how utterances should be interpreted that abandons anything resembling traditional propositions, the objection ceases to have teeth. The second option is related, in that it involves observing that, given a different view of the nature of how language is used to communicate thoughts, the objection no longer arises. Indeed the view of language that Travis provides to counter the objection is, I believe perfectly consistent with Sutton’s, though I will not attempt to show that here in any detail. One similarity between Sutton and Travis is the appeal to language users as rational agents for whom utterances can form evidence, not just for what a 197 6.8. Objections to the Pragmatic View speaker is trying to do with their words, but, through that, for how things are. One key point with regard to this, which I have mentioned before, is that interpreting what was said with some words involves not just knowledge of what those words mean, but also general knowledge about how the world is: the circumstances surrounding an utterance, but also the ways in which circumstances might affect how things might be described as being. For example, it might be relevant to understand that a botanist requires different things from a green leaf than a photographer does in some circumstances of recognizing that a leaf is green. Although, in other circumstances, it may not be. In at least three papers in which Travis discusses reference (1980, 1981a, ms.), he appeals to Descartes’s assertions that humans are rational. Travis states: [A] rational being has an unbounded capacity to recognize and evaluate novel ways of accomplishing goals, of solving problems. Suppose one stated a determinate, executable, method for discovering such ways; some set of principles fixing how solving some given domain of problems is to be done. Then a rational being would be able to consider, and evaluate rationally, alternative ways of doing that, and to recognize the need for such should need arise. [. . . ] [S]uch flexibility, or plasticity, in our thought about the environment in which we find ourselves—in our ways of making thought mesh with the things there are in it for one to think about—is central to an ability to have coherent thought about the world at all. (Travis ms., p.29). Travis maintains that it is this kind of capacity that allows language-users to recognize what a term has been used to refer to, or whether a particular object is correctly described in the way that it was, without there being any systematic set of determinate conditions that determine these things. Rather, as rational beings, we are able to employ domain-general knowledge, and general capacities, in order to solve specific problems of recognition. And, moreover, we are able to apply relevant knowledge and methods to specific cases, even though that specific knowledge and those specific methods may not be suitable for other, similar recognitional problems. Whether or not certain methods or areas of knowledge are relevant under certain circumstances is not something that is specified in advance by, for example, the semantics of words. In chapter 7, I provide an example of a class of referring terms, apparently very similar to proper names, for which it is hard to see, when their use is considered, how their semantics could provide conditions that determine their reference, so it must be fixed entirely pragmatically. Consideration of this 198 6.8. Objections to the Pragmatic View class of terms should therefore make it plausible that the reference of a range of terms can be fixed wholly pragmatically. This should allay worries that it is implausible that the reference of proper names could be determined in this way. 199 Chapter 7 Who’s Mum? Seal up your lips and give no words but mum. Henry VI, Part 2 I n this chapter I will attempt to show that the pragmatic view is coherent and plausible as an account of proper name reference by considering another class of referential terms, namely familial terms such as ‘Mum’. These appear to be examples of what Pelczar and Rainsbury (1998) call ‘quasi-names’ and possibly what Recanati (1993) calls ‘pseudo-names’. I will suggest that the most obvious account of the determination of their reference is something like the pragmatic view, whereby nothing in their semantics determines their reference, nor strongly constrains it. 7.1 ‘Mum’ Words such as ‘Mum’ and ‘Dad’ are, at least superficially, similar to proper names in that they can be used bare (i.e. without being preceded by a determiner or possessive pronoun) to refer to objects (I will exclude vocative uses from my discussion, because it appears to be relatively easy to address someone using almost any term). However, unlike names, their use is generally not connected to a dubbing-type event, but rather is connected to satisfying the common noun ‘mum’ or ‘dad’ just in virtue of being a parent (in some sense). In Southern British English there is a convention to use (or be able to 200 7.1. ‘Mum’ use) ‘Mum’ to refer to one’s mother, and ‘Dad’ to refer to one’s father.1 I propose that these referential uses can very plausibly, or even most plausibly, be accounted for in a manner in which their reference is determined by purely pragmatic means. ‘Mum’ can be used as a common noun that is fairly synonymous with ‘mother’, but a common noun is not the type of thing that can easily refer to a particular object, when used bare. ‘My mum’ is a very commonly used referential term to refer to a particular mother, but it cannot be that all uses of bare ‘Mum’ are elliptical for ‘my mum’, since, as I’ll argue, speakers can use it to refer to other people’s mothers. ‘The mum’ might be a more plausible candidate, but this simply pushes the problems into the semantics of so-called incomplete definite descriptions. A Kaplanian indexicaltype view is an alternative, but it is very hard to see what kinds of contextual parameters would do the work required, unless they are very indeterminate. Moreover, it is clear that there are not multiple, homophonous terms ‘Mum’, each of which refers to a different person, as homophonists maintain is the case for names. 7.1.1 Constraints There are some constraints on the reference of uses of ‘Mum’. Generally, it refers to a mum. And one wouldn’t expect it to be able refer to just any mother—the mother must, presumably, be related in someway to the dialogue or utterance, but just how it must be related seems to be very much a matter of context. It is also not a necessary condition upon reference that the referent be a mother. One can well imagine cases where a non-mother is referred to as ‘Mum’, perhaps for humourous purposes, though it seems likely that doing so will be highlighting some property or other the referent has that is, in some sense, motherly. There are other kinds of quasi-names, such as ‘thingumybob’ and ‘what’s-their-name’, mentioned by Recanati (1993), that appear to have almost no constraints on their reference. Some such terms may be more suggestive of (e.g.) an animate or inanimate referent—‘what’s-theirname’ versus ‘thingumybob’ and ‘whatsit’—but these appear to be largely defeasible. 1 There are parallel conventions to use related terms for the same purposes in other dialects of English. Many other languages have equivalent terms that can be used both bare, referentially, and as common nouns. 201 7.1. ‘Mum’ 7.1.2 My Mum or Yours? Anecdotal evidence suggests2 that bare, referential ‘Mum’ is used primarily to refer to the/a mother of a speaker when the speaker is addressing members of their close family, particularly siblings and (other) parents, but also amongst wider family, and people well-acquainted with the mother. When addressing others, it seems that ‘my mum’ would often be a more appropriate way of referring to one’s mother, although, in many contexts it is likely that bare ‘Mum’ could still succeed in referring to the speaker’s mother. All this is suggestive that ‘Mum’ refers to the speaker’s mother. But this is not always the case. It is entirely normal, at least in my dialect (or perhaps just my family), for parents to refer to mothers of their children as ‘Mum’ when addressing those children (particularly if the parent also has a familial relationship with that mother).3 It is also quite possible for non-parents to refer to the mother of their addressee using bare ‘Mum’, though in my experience this type of use is suggestive (though not perhaps requiring) of a high degree of familiarity with the referent, and is far more common if the speaker is of the same generation as, or previous to, the referent. For example, if I’m at my parents’ home and a friend of my mother calls, they might say ‘Is Mum in?’, though note this is somewhat infantilizing (to me). However, it’s unlikely that I would succeed in referring to anyone if I walked into my supervisor’s office and asked him ‘How’s Mum?’, since neither my supervisor nor I know each other’s mothers, and, in this context, ‘Mum’ would be a very unusual way for me to try to refer either to my mother or to his, suggesting undue familiarity with both him and her. It is implausible that a reasonable language-user would recover a referent from such a use. These phenomena appear to be very contingent social conventions, that may well differ in other communities, without the meaning of ‘Mum’ differing.4 Thus, I find it highly implausible that these kinds of social contextual factors play a rôle in determining the reference of a use of ‘Mum’ as part of its semantics. To demonstrate how much context sensitivity is at play in the referential use of ‘Mum’, consider the expression ‘How’s Mum?’, when spoken by a person addressing her sororal niece. If the utterance occurs in a context in which the niece is more likely than the aunt to have salient information regarding the wellbeing of the niece’s mother, it is likely that the 2 Unfortunately, I do not have the resources to carry out any rigorous data collection to support my assertions of anecdote, but I hope readers will not find my assertions unfounded. 3 Though generally such a parent would not, unlike their children, use ‘Mum’ vocatively. 4 Eliot Michaelson, who grew up in California, tells me that his parents would not refer to each other using bare ‘Mum’-type expressions, preferring ‘your mum’-type constructions, something that to my ear would be suggestive of a lack of familial-ness. 202 7.1. ‘Mum’ referent is the niece’s mother. However, in a context in which, for example, the aunt’s own mother is ill, and staying with the niece, then it is likely that the speaker’s mother is the referent. Of course, many other contextual factors are likely to affect the reference, such as what terms the aunt has previously used to refer to each mother in particular situations. It can also be the case that ‘Mum’ refers to someone who is neither the speaker nor the addressee’s mother. Consider the sentence ‘My boyfriend’s parents are coming to visit. Mum wants to go to the Museum, but Dad wants to visit the Gallery’. Whilst such a sentence might strike one as slightly odd, it is easy to see that, in certain contexts, it is the boyfriend’s parents that are being referred to by ‘Mum’ and ‘Dad’. Alternatively, consider a doctor who attends the hospital bed of a woman who has recently given birth. Suppose that the new mother is dozing, but her partner is sitting by her bedside. The doctor might say to her—the partner—‘How’s Mum doing?’, referring, very clearly, to the woman in the bed, who is neither the speaker nor the addressee’s mother.5 7.1.3 No Parameters In each case of ‘Mum’ usage, whilst what it is reasonable to infer about the speaker’s purposes seems to be playing a rôle in what she refers to, it does not seem to be that it is anything like ‘the most salient mother’ being referred to, as might be proposed by a theorist who takes salience to be very important in reference determination (e.g. Mount 2008). In a situation in which both interlocutors have mothers, and both know each other’s mother, no mother need be more contextually salient than the other in a way that is significant for reference. Consider a case in which someone is talking to their boyfriend about his (the boyfriend’s) mother and father coming to town. The speaker says ‘So, we’re having dinner with your mum and dad on Saturday. And Mum’s going to be around in the morning too.’ Although the referent of ‘Mum’ in the second sentence could be the boyfriend’s mother (as in the previous example), I think it is clear that it can also be the speaker’s mother, particularly if the speaker is in the habit of referring to his mother, but not his boyfriend’s mother using bare ‘Mum’. This is in spite of the fact that the boyfriend’s mother has been made salient in the first sentence. It may be that there are understandings of salience according to which there is no reference without salience, as discussed in §4.2.1, p. 122, but I think that these understandings are explanatorily insignificant with regard to reference, since they turn out to be either circular or redundant with regard 5I believe that this example was suggested by a colleague at the King’s Advanced Research Seminar. 203 7.1. ‘Mum’ to explaining it: A sure-fire way of making an object salient, in one sense, is to refer to it. Thus it may be true that the referent of ‘Mum’ is always the most salient mother at the time the word is uttered, but this salience clearly cannot be involved in determining the reference, on pain of circularity. In a similar vein, it might well be that the kinds of pragmatic factors that go towards determining the reference of an utterance of ‘Mum’ are also the kind of factors that are likely to make an object more salient than others. If that is the case, there will certainly be a strong correlation between reference and salience, but it does not entail that salience is itself involved in determining reference. Given all of this anecdotal data, it seems very reasonable to conclude that context must play a very great part in determining the reference of ‘Mum’. Plausibly, to the extent that it cannot be supposed that the kind of necessary or sufficient conditions are recoverable that could play the rôle of a determinate semantic character. But the problems for semantic determination are larger than this. The reference of a use of ‘Mum’ can vary not just between the unitary mothers of the interlocutors, but also between different people who might count as being the mothers of the interlocutors, or who are called ‘Mum’ by the interlocutors. For instance, someone might have two mothers in virtue of having one biological, and one step-mother, both of whom they sometimes refer to as ‘Mum’, or they might have three fathers, all of whom can plausibly be referred to as ‘Dad’ in certain contexts. This shows that even if it were settled which interlocutor’s (or other person’s) mother was at issue, something more may well be needed to settle upon who in fact is the referent. 7.1.4 Familial Practices It may well be that people tend to develop particular practices for using words like ‘Mum’, either individually or within a family unit. In a family unit with more than one father or mother it is easy to imagine two colloquial forms of ‘mother’ or ‘father’ being used, one for each parent, e.g. ‘Daddy’ and ‘Papa’, as reported in Guasp et al. (2010, p. 6). This is also seen with terms for grandparents: different grandparents may be assigned different colloquial forms of ‘grandmother’ or ‘grandfather’ in order to more easily unambiguously refer to them. These kinds of established practice make the terms seem, on the lips of practice-inducted family members, more like proper names. However, it is important to note that even when these practices exist, they are not insurmountable. Suppose someone has two mothers and calls one ‘Mum’ and the other ‘Ma’. Whilst it might be hard for her to refer to Ma as ‘Mum’ amongst others who are familiar with the practice, it would 204 7.2. From ‘Mom’ to ‘Tom’ clearly be possible for her to so refer amongst non-inductees, given the right context, just as it is still possible for her to refer to other people’s mothers with ‘Mum’ in the kinds of circumstances described above. Note also that it is very easy for people unfamiliar with such family-specific practices to refer using different terms. Consider an undertaker who consistently refers to a deceased matriarch as ‘Mum’ when addressing her children. Reference can succeed perfectly well in such a situation, even if none of the children have ever referred to their mother as ‘Mum’.6 7.2 From ‘Mom’ to ‘Tom’ I have now argued that there is good reason to take seriously the idea that familial terms such as ‘Mum’, when used bare, have their reference fixed wholly by pragmatic means. The primary aim of this chapter has not been to show that words like ‘mum’, when used referentially, are the same as names in every way—clearly they are not—but rather to show that it is not implausible that a class of referential terms can have their reference determined solely by pragmatic mechanisms that are not encoded in semantics. Though, of course, if it were true that names and terms like ‘mum’ have more in common than meets the eye, this can only be grist to my mill. One superficial difference between names and familial terms is that, whilst the most common way in English to refer using a name is to use it bare, it seems the most common way to refer using a familial term is to use a possessive pronoun, or the possessive form of a name, preceding it. Using a proper name preceded by a possessive is rare in English, but it does occur in Northern English dialects, where (e.g.) ‘our Rita’ or ‘my George’ are common ways of referring to family members (Wales 2006). The secondperson possessive (‘your Mavis’) also appears to occur, though perhaps less frequently, and the third-person might also be acceptable (‘Vera’s Ken’). Such constructions are easily understood by speakers of other dialects, and even seem natural and likely to occur, where both speaker and audience have a close friend or family member of the same name and ambiguity would otherwise ensue, just as one would use ‘my mum’ where ‘Mum’ would be inappropriate or unclear. For example, my girlfriend’s name is ‘Katharine’, and my sister’s name is ‘Kathryn’. This obviously has the potential to result in confusion when speaking to my parents, and so they frequently attempt to make use of constructions such as ‘your Katharine’ and ‘our Kathryn’. 6 Thanks to Robyn Carston for this example. 205 7.3. Conclusion 7.3 Conclusion I have argued that it is highly plausible that a certain class of referential terms have their reference determined wholly pragmatically. If it is the case that reference can be determined in this way for one class of terms, it is possible that something similar is true for other classes of referring terms—reference can be a pragmatic phenomenon. Whether or not they are precisely the same kind of term, it is clear that there are some significant similarities between proper names and familial terms, even if there are some differences in their use. It is therefore not implausible to think that the reference of both types of term might be determined in a similar fashion, and since, as I have argued, it is implausible that familial terms have their reference determined in the ways specified by prominent semantic accounts of name reference, it is at least plausible that proper names have their reference determined pragmatically. 206 Chapter 8 Conclusions and Considerations A good name is better than precious ointment Ecclesiastes, 7:1 I n this, the final chapter, I will offer a summary of what has gone before, highlighting those parts that I believe are most significant, or that contain the most original contributions to the debate. I will also discuss briefly some topics pertaining to proper name reference that have not arisen at any length previously, but that offer scope for further work in developing my pragmatic view. In several cases, I will suggest the options and directions for this further research that seem most promising. The first is what the semantics of proper names might be like, and what can be said about it, given my pragmatic view. The second topic I will discuss in this regard is how my pragmatic view can respond to various classical problems and issues relating to proper names. 8.1 Summary In what has preceded, I have made some preliminary remarks about the methodology and philosophy that are assumed by the thesis; I have offered extensive criticisms of various semantic views of proper name reference. I have introduced and discussed at length a pragmatic view of name reference that attempts to do justice to the insights garnered from both my background philosophical assumptions, and the problems of semantic views. Here, I recapitulate my findings. 207 8.1. Summary 8.1.1 Background Chapter 2 introduced the question of the determination of proper name reference that is addressed by the rest of the thesis. It then went on to discuss what could count as data by which to judge an account of proper name reference, and compare it to other accounts. I noted that judgements about what had been referred to in cases of name use, real or imagined, provide the most plausible, usable data on name reference. However, there was a significant debate over whether the judgements of so-called experts, i.e. philosophers of language and other semanticists, or the pre-theoretic judgements of ordinary language-users were to be preferred. I concluded that the judgements of ordinary language-users about cases of proper name use were significantly preferable to the judgements of experts, if experts’ judgements were affected by their theorizing. In order to be genuinely significant, such judgements would have to be collected carefully, making use of rigorous experimental methodology. However, unfortunately, such methodology is largely beyond the capacities of most philosophers, and is certainly unavailable to me. Throughout most of the thesis therefore, although I made extensive use of cases designed to draw the pre-theoretic judgements of my readers, the conclusions I drew from them were largely my own, without the support of experimental data, rigorous or otherwise, since it was simply unavailable to me. I believe that my judgements were not unduly influenced by my theoretic commitments, though, as for any work making use of such judgements within arguments, the success of those arguments will depend on the acquiescence of my readers. Following the discussion of data, in chapter 2, I introduced my thoughts on the semantics/pragmatics divide, which played a key rôle in my claim that proper name reference is a wholly pragmatic phenomenon. I suggested that the contribution that words make to every utterance in which they occur—their context-insensitive contribution—is their semantic content, and that anything that words contribute to an utterance that goes beyond this, anything that depends upon the context of their utterance, is pragmatic. Being heavily influenced by Travis, I take any truth-evaluable content to be pragmatic, on this understanding of pragmatics. I did not argue for this general claim, but I spent some time outlining and discussing Travis’s general account that makes this claim: that the semantics of words and sentences—their meanings— radically under-determine their content, or what is said or expressed with them. Following this exposition of Travis, I briefly discussed communication, and its rôle with regard to language and reference. 208 8.1. Summary 8.1.2 Semantic Views Part II introduced the semantic view of proper name reference: those accounts that maintain that the reference of names is in some way semantic, because it is determined by, or constitutes, their semantics. I considered two main groups of accounts within this view: causal-homophonism and indexicalism. I also discussed a third group of accounts, intention-based accounts, some, but not all of which might be considered to fall within the semantic view of name reference. I did not, generally, attempt to offer refutations of any of these types of account, as such, but rather, argued that each is subject to significant complications and limitations that damages its credibility, particularly when compared to an account that is better able to deal with the problems I present for those accounts, as I claim my pragmatic view is. I approached each of the three types of account differently. Chapter 3 provided four, largely distinct arguments against causal-homophonism, which I treated as a single, cohesive account of name reference. Against indexicalism, in chapter 4, I proposed a single, over-arching form of argument, but then discussed in detail the accounts of three different indexicalists, before applying my general argument form to some specific cases. I discussed two distinct forms of intentionbased accounts of name reference in chapter 5. With regard to the first, I simply demonstrated some significant problems. With regard to the second, I offered an extended argument to the effect that the kind of broad approach to language that must contain it, is essentially problematic. Causal-homophonism is the view that the reference of a proper name just constitutes its semantics, such that there is nothing more to its meaning than what it refers to. Just what it is that a name refers to is determined by a causal chain that connects each use of it to an initial baptism of the referent with the name. In order to square this view about meaning with the truth-conditional semantics espoused by its proponents, it is required that each name can only have one referent. Thus, it is claimed that any case of a single name having multiple referents, which appears to be a ubiquitous phenomenon, is in fact a case of homonymy or homophony: there are multiple names, each with a distinct referent, that simply sound the same and have the same orthographic form. I presented four main arguments against this view in chapter 3: the argument from shared names, the argument from communication, the argument from family names, and the argument from pluralism about reference determination. The argument from shared names was, perhaps, the least effective, just because it is unlikely to convince anyone already inclined toward accepting the homophonist individuation of names. I appealed to ordinary language claims that proper names are 209 8.1. Summary shared, in order to reject the contrary claims of the homophonist, and argued that any attempts they make to accommodate the ordinary language claims as being anything other than simply false will fail. I maintained that we should attempt to accept the ordinary language claims, and thus should be illdisposed towards the homophonist individuation of names. The argument from communication took aim primarily at the causal explanation of the determination of name reference. I claimed that it fails to properly account for the way that language-users use names, and in particular, how they use them to communicate. Perhaps the most original argument against causalhomophonism that I presented was that from family names, since very little has previously been said about what homophonists can or should say about family names. I observed that family names have at least two primary rôles, at least in English-speaking communities in the West. The first is to refer, in various ways, to families. The second is to refer to individuals whose full names incorporate the family name that names their family. Although there are various routes that the homophonist could take in order to accommodate these different rôles, none is simple, and each adds significant complications to the homophonist picture. My final argument, from pluralism, drew on the previous arguments, particularly that from communication, to observe that there appear to be a variety of ways that names are used, and their reference determined. Whilst the causal theory lends itself well to explaining one particular way of using names, it fails to do justice to others. Indexicalism is a group of views that suggest that proper name reference is in some way akin to the reference of indexicals, that is, it is contextsensitive, but in a way specified by the meanings of names. Thus, their semantics determines their reference because it provides a rule that specifies the reference, given certain contextual factors. My broad form of argument against indexical views about names in chapter 4 owed much to Travisianstyle arguments against similar views about the meanings of words generally. I maintained that, for any particular specification of how context affects the reference of a name, it will have one of three serious problems: it will be too general, such that it will fail to distinguish between rival possible referents; or it will specify the wrong contextual factors, such that it returns the wrong referents; or it will be incomplete, such that it fails to provide any specific account of reference determination. I discussed the indexicalist accounts of Recanati (1993), Pelczar and Rainsbury (1998) and Rami (2014) in depth, finding significant problems with each, primarily that they were, essentially, incomplete, and failed to provide a fully truth-conditional account of proper name reference, despite this being a clear intention. I then went on to discuss the general problem of trying to specify particular contextual factors that 210 8.1. Summary might effect the reference of a name upon every use, and attempted to indicate that several factors that seem plausible candidates will fail; in particular, I discussed salience. I then discussed in detail why appealing to name bearing, and name-using practices, as a necessary or sufficient condition for proper name reference is a doomed endeavour. The complications and nuances of what counts as bearing a name and being part of a name-using practice are brought out in this section. These discussions of the very possibility of providing criteria-based rules for the reference of names form a novel approach to the question of proper name reference. Chapter 5 discussed two different forms of intention-based account of proper name reference: constraint-based, and (neo-)Gricean. These have in common that they incorporate the intentions of the speaker into the determination of (semantic) reference in some manner. As I had made clear in chapter 3, intentions per se are not on their own suitable as determinants of reference, as it appears that speakers can refer to things that they do not intend to using names, and that simply intending to refer to something goes very little way towards actually referring to it. Constraint-based intentionalism denies that speakers can refer without intending to, and maintains that speaker intentions do determine name reference, but that what a particular name can refer to is constrained. I associated this view primarily with Michaelson (2013). I argued both that the determining rôle of speaker intentions, and the involvement of particular constraints, especially involving name-bearing relations, renders the account highly problematic. The kind of Gricean intentionalism about name reference that I envision places constraints not on what a name can be used to refer to, but on what referential intentions a speaker can have, in line with the kind of strong neo-Griceanism espoused by Neale (2004, ms.). I argued against this type of view by attempting to show that the general account of the constraints on communicative intentions required by Neale and other Griceans are implausible in light of data from experimental psychology. 8.1.3 A Pragmatic View Part III, and chapter 6, contained a detailed exposition of a pragmatic view of reference, the primary positive original contribution of this thesis, that had begun to take shape through the criticisms I offered of semantic views in part II. I began by mentioning how my view has been influenced by Travis’s work on reference. I then introduced seven broad principles that covered the major points of the view, and went on to discuss these principles in detail. I discussed my general conception of names and name reference, and how this 211 8.1. Summary relates to truth, touching on empty names and reference over time. I then went on to deal with what the mechanisms of name reference might be, and what they are not. I discussed the context-sensitivity of name reference, highlighting the reasons for holding reference to be context-sensitive that I have provided throughout the thesis, and the rôle of semantics in the determination of name reference. As I have made clear, the claim that the semantics of proper names do not play a significant rôle in the determination of their reference is the primary claim of this thesis. A significant part of this claim is that there are no necessary and sufficient conditions for proper name reference, such that a list of determinate criteria could ever be enumerated. This point also received elucidation. Perhaps the most controversial claim of my pragmatic view is that the referent of a proper name need not be a bearer of that name. This claim also received the most extensive discussion of the chapter, in which I detailed why I make it, and how the intuitions that generally lead theorists of name reference to make the converse claim can be better accommodated. One better way of accommodating such intuitions about bearers, as well as the apparent feature of names that they can be reused, repeated, and passed on, is to maintain that the referent of a name is merely likely to be something that has already been referred to with that name. Having provided discussion of the major features of the view, I then moved on to some clarifications. I mentioned how names might be individuated, given my conception of their reference, and I related that conception at length to Travis’s conception of occasion-sensitivity. I focussed, in particular, on whether the occasion-sensitivity that I appear to be proposing for name reference extends beyond the level of language, to thoughts, as Travis maintains it does with regard to other kinds of expression. I suggested that there was a good sense in which it did, although the comparison raised some interesting issues. Finally, in chapter 6, I considered some likely objections to my pragmatic view, concluding that they do not present significant problems for it. One of these objections, that it is implausible that there are no determinate criteria for proper name reference, is taken up again in chapter 7, where I considered referential uses of familial terms such ‘Mum’—a variety of term that, like family names, seems to have received very little attention in the philosophical or linguistic literature. I argued that observation of the ways that such terms can be used referentially makes it highly implausible that there are criteria contained in their semantics that determine their reference. This lends significant plausibility to my claim that proper name reference can also function without there being necessary and sufficient criteria encoded in their semantics. I have made a strong case for the claim that the reference of proper names 212 8.2. Semantics is a pragmatic, and not a semantic, phenomenon, and developed a detailed account of proper name reference based upon that idea. I have shown the weaknesses of other accounts of name reference, which maintain that reference is determined, as least in part, by the semantics of names, and contributes to the semantics of larger linguistic structures. I hope I have shown that my pragmatic view is a worthy competitor to these semantic accounts, and that, given their weaknesses, it is plausibly to be preferred to them. Having thus discussed what I have done over the course of the thesis, I now turn to what I have not yet done, and what will need doing in the future to shore up my view. 8.2 Semantics An issue that I have frequently gestured at and passed over in this thesis is what the semantics of proper names can be like, given my view of their reference. If there is to be some kind of compositional logical form at the level of semantics, what is it that names contribute? This question, rather than how name reference is determined, appears to have dominated discussion of names in the last few years, spurred on by renewed interest in predicativist, or neodescriptivist, accounts of names. I will, very briefly, say what the semantics of names cannot be, given my view of their reference, and then what they could be. As I discussed in the introduction to part II, accounts of proper names can be divided into accounts that take a stand on the semantics of names; those that take a stand on the determination of reference; and those that take a stand on both. So far, I have discussed only those views that take a stand on the determination of reference, whether or not they also take a stand on semantics. I rejected causal-homophonism, which states that the reference of each proper name is determined causally, and that their semantics is constituted just by their reference. In terms of semantics at the logical or morpho-syntactic level, causal-homophonism most naturally adopts the view that proper names behave like non-logical constant terms. I also rejected indexicalism, which maintains that names have a semantics like that of indexicals, which, in some way, specifies the way in which context is to contribute to the determination of reference. This could be done in a variety of ways, but most obviously would be modelled within a representation of logical or morpho-syntactic form as the value a function. It is clear from the nature of my view about proper name reference that I cannot accept a story about the semantics of names which involves either the determination of their reference, or an assumption that their reference is already determined at the semantic level. In common with many proponents 213 8.2. Semantics of semantic views, I maintain that a name contributes just its reference to the truth-evaluable content of whatever larger structure is the bearer of that content—it is directly referential. However, whilst for them, this is a contribution to truth-conditional semantic content, for me, all content of this kind, including reference, occurs at a pragmatic level, which presumably operates in some way on whatever comes from the semantics of uttered expressions. Thus, assuming that there still is a compositional semantic level, which is what I am calling ‘logical form’, whatever names contribute must just allow that, for example, it can be composed with predicates, and that reference can be attached to it at the pragmatic level. In the introduction to part II, I mentioned two views on the semantics of names that did not assume anything objectionable about reference determination: predicativism and variablism. I suggest that either of these accounts might provide what my pragmatic view of reference needs from semantics, though variablism might prove the most natural. However, working out in detail how either a predicativist or indexicalist semantics would work together with my pragmatic view of proper name reference, and whether one should be preferred to the other is clearly a task for further research. 8.2.1 Predicativism Predicativism is the view that proper names are predicates. Their semantic type is that of a common count noun, and this is observed most clearly in sentences such as ‘I know four Michaels’, and ‘I’ve never met a mean Giulia’, in which the occurrences of names are clearly predicative rather than referential. But predicativists maintain that even when occurring referentially, proper names are predicates. Most contemporary versions of the view maintain that, when names are used referentially, they occur—at the morpho-syntactic level— with a definite article preceding them, whether or not this is pronounced (Fara 2015; Gray 2014, forthcoming; Matushansky 2008). I will consider here just descriptivist versions of predicativism. In English, in many—though not all cases—names occur without any apparent determiner when used to refer singularly, though there are several natural languages in which a definite article is either permissible or obligatory with uses of proper names that would be bare in English. There has been much said on the topic of whether or not predicativism is plausible, and whether it can account for uses of names as well as accounts of names that treat them as essentially referring terms. I will not get into this debate in any detail here, but will just consider how well the predicativist view of name semantics would fit with my pragmatic view 214 8.2. Semantics of their reference.1 I claim that predicativism is not a view about proper name reference because it simply pushes the question of how reference is determined into how it is determined for definite article expressions. There is much debate over the nature of definite descriptions, whether they ever are referential, and how their reference is determined if so. I am of the opinion, as I have indicated previously, that descriptions can refer, and that this reference is the same kind of phenomenon as that of proper names, for example, it is not merely ‘speaker reference’, in contrast to the ‘semantic reference’ of names. Furthermore, as was also clear from my mention of Donnellan cases, I maintain that definite descriptions can refer, in the robust sense that names refer, to objects that do not satisfy their descriptive element. These claims are, of course, controversial, and I will not argue for them here. However, I suggest that they would be required of definite article expressions in order to square my pragmatic view of name reference with predicativism about name semantics. The standard view of the meaning of name predicates, in the sense of what is required to satisfy them, is that they are satisfied just by bearers of the name in question. So, the predicate ‘Lucy’ is applied correctly to objects that bear the name ‘Lucy’ (Fara 2015; Gray 2014). This seems to borne out by the way that names are used in predicate position, non-referentially: If I say ‘I know four Michaels’, this seems to mean something closely akin to ‘I know four people named “Michael”’. I have made a great deal of fuss about the fact that a name such as ‘Katharine’ can be used to refer to objects that do not bear that name. But this is consistent with referential uses of ‘Katharine’ having the morpho-syntactic form ‘thenull Katharine’, which is equivalent to ‘the bearer of “Katharine”’, if definite descriptions can be used to refer to things that do not satisfy them. Indeed, one might even think that this treatment provided a nice explanation of why ‘Katharine’ need not refer to a Katharine, and of the apparent symmetry between Donnellan-type cases and cases of name reference to non-bearers. I could claim that the descriptive portion of a definite description simply provides another pragmatic factor that can, but might not, affect the reference of the term in some manner. There would, of course, be some problems with adopting the predicativist view of semantics. Firstly, I am unhappy with the notion of there being unarticulated determiners and denuded definite descriptions. This seems to be an unduly theoretical posit that is difficult to make sense of: what are these null articles? In what manner do they exist? It may be that I can simply say that when used referentially, ‘Katharine’ behaves just like a definite 1 Thanks to Delia Graff Fara for highlighting to me that predicativism might be compatible with my view. 215 8.2. Semantics description of the form ‘the Katharine’, where ‘Katharine’ is a predicate of the kind described. However, this story might well run into complications. A further problem would be incorporating my suggestion that the semantics of a name should include that it is likely that a referent has been referred to with the name before. I could argue that name predicates are not satisfied just by bearers of the names, but by things that have been referred to with the name before. However, this does not seem to be justified by predicative uses of names: something does not become a Lucy just because it has been referred to with that name once, though it might be that it becomes more likely to be so described the more often it is so referred to. An alternative would be to claim that this is simply a feature of the reference of definite descriptions. This seems to have some plausibility. It is, at least, true that within relatively small communities, or contexts, or discourses, referential definite descriptions can set up practices of repeated use. For example, if a friend says to me one day, ‘I think the man in the hat over there is following me’, referring to a mutually apparent man in a hat, then the next day she can say ‘I saw the man in hat again today’, and succeed in referring in virtue of having used the description to refer the previous day. This phenomenon seems even more likely if a description has been used to refer to something that does satisfy it: the mere fact that it has been used to refer to that object makes it easier to refer to that object again with the same description. However, the phenomenon seems much more localized than that that applies to proper names, since my thought with regard to proper names—which may well require modification—is that a use of a name is likely to refer to something that it has been used to refer to before, at some point ever, with almost no constraint. 8.2.2 Variablism Variablism is the view that proper names have the semantic form of variables, or are like variables in some significant way, and when they are used referentially, they behave as free variables. It is propounded most prominently by Dever (1998) and Cumming (2008). Such a view can be contrasted, primarily, with the view that names are like non-logical constant terms. A major difference is that constants have their value assigned by an interpretation function that is part of a model, whereas variables have no specific denotation with regard to any particular model, but only upon a particular variable assignment, which is distinct from the model, and can be varied with regard to it. This makes indexicalism a natural semantic story for contextualist and pragmatic accounts of name reference, connecting names with certain (uses of) pronouns that are frequently treated as variables. Indeed, to the extent 216 8.2. Semantics that Dever and Cumming put forward views about how the reference of names is determined, it is largely pragmatic (in some sense) or contextual. Treating proper names as variables at the semantic level has, for me, the major advantage that there is no suggestion that a variable has a particular reference, without some assignment being applied, which can be at the pragmatic level, given a particular utterance. This will give us enough to say, as Dever and Cumming do, that (natural language) sentences containing (nonbound) proper names are effectively open sentences that require some kind of completion in order to express truth-evaluable content, just as expressions of first-order logic containing free variables have no interpretation on a model without being relativized to a variable assignment. I can then claim that, given an utterance of such a sentence, pragmatic factors provide a variable assignment (or a partial assignment) in the ways I described in chapter 6. The question now arises of what the variables are like. The simplest explanation would be just that each name is a variable, but this would create the problem that any two occurrences of a name in the same utterance, if they were deemed to fall within the same context, occasion, or circumstances, would be co-referential. But, as I have mentioned previously, it appears that if the same name is used multiple times within a short linguistic space, then, in many contexts, it will be likely that the uses are not co-referential. For example, if I say ‘I haven’t seen John for ages, but John said he’ll be at the party tomorrow’, rather than ‘I haven’t seen John for ages, but he said he’ll be at the party tomorrow’, then there is the impression that the second occurrence of ‘John’ refers to a different person to the first. I can adopt Rami’s (2014) solution to this problem, and add a sequential index to each occurrence of a name, which can be represented as an integer. Every occurrence of a name is now effectively a distinct variable. This means that we no longer have the notion of co-reference at the level of semantics (at least between occurrences of names), but this might be fine—even in cases in which the fact that a name has already been used in a dialogue contributes to the determination of reference, this is a factor effecting its reference at the pragmatic level. A consequence of this will be that certain inferences will not be possible at the logical or semantic level, for example, but this shouldn’t bother us. Given my Travisian claim that everything truth-evaluable occurs at the pragmatic level, this is where we should expect truth-functional inferences to be evaluable. A further question, and potential problem, for this variablist view arises when we consider binding. If referential proper names are free variables at the semantic level, then it might be reasonable to suppose that they can be bound by placing a quantifier at the beginning of a sentence containing a name. Dever (1998) does not consider binding in the case of names, but Cumming 217 8.3. Problems of Names (2008) views the possibility of binding as feature of the account, allowing explanation of various predicative uses of names, such as ‘I’ve never met a mean Giulia’ or ‘All the Johns were there’. However, Cumming considers more complex examples, such as ‘There is a gentleman in Hertfordshire by the name of “Ernest”. Ernest is engaged to two women’, which he regards as a case of a bound anaphoric use of a name (Cumming 2008). Elbourne (2005), a predicativist, also suggests that names can be bound in ways not generally appreciated; for example, uses of names that resemble uses of Dtype pronouns or definite descriptions: ‘Every woman who has a husband called John and a lover called Gerontius takes only Gerontius to the Rare Names Convention’ (ibid., p. 181). I have mixed feelings about such binding of names. The non-quoted use of ‘Ernest’ in Cumming’s example strikes me as a straightforwardly referential use, not a case of anaphora. If that’s so, then we may not want names to be bound unrestrainedly. However, ‘Gerontius’, in Elbourne’s example, clearly cannot be referential. It appears that it needs to be explained as a bound use. It seems plausible that, if proper names are treated as variables, there is no need to assume that they must be unrestrainedly bindable. They might be treated as terms that behave like free variables, but cannot be bound. Or they might be bindable only in particular ways, by particular quantifiers, assuming that there is a systematic explanation for the way in which binding is acceptable. 8.3 Problems of Names It would be fair to say that, with regard to problems for accounts of proper names, the major preoccupation of this thesis has been the problem of shared names, both in being posed to other positions, and in showing how my view can deal with it better. I have spent almost no time discussing the kinds of problems that have traditionally been posed for, or inspired, accounts of names. I will now very briefly address four of these issues, pointing out whether or not they present significant difficulty for my pragmatic view, and indicating how they might be approached in future research, if so. These problems are rigidity; name change and reference shift; puzzles about attitudes and beliefs; and fictional and empty names. The first two of these, I am able to deal with fairly easily, simply in virtue of the nature of my account. The latter two deserve more consideration, and I will provide some thoughts on the issues. However, they certainly present no more of a problem for my view than they do for a range of orthodox accounts, so even if it were to emerge that my view does not provide any simple solutions, this should not be seen as a particular disadvantage. 218 8.3. Problems of Names 8.3.1 Rigidity Rigid designation, as characterized by Kripke (1981), is reference to the same object in all possible worlds, i.e. regardless of the world of evaluation. In modal logical terms, this means that there is no scope interaction between rigid terms and modal operators, so all scope readings are equivalent. I certainly acknowledge that proper names, or their reference, exhibit the property that Kripke was alluding to with his characterization of rigidity, and which has been so-called ever since, and so my pragmatic view of name reference must account for it. However, I think that Kripke’s description of rigidity is somewhat unhelpful to understanding what the phenomenon really consists in. Indeed, it maybe unhelpful to think of it as a phenomenon at all. In particular, I think that Kripke’s emphasis on the modal contrast between proper names and definite descriptions is unhelpful, an opinion that he himself seems to hold when, in the preface to Naming and Necessity he observes that rigidity is not essentially a modal notion, and that proper names exhibit it even in non-modal settings (ibid., pp. 10–12). My view will be that rigidity is simply a feature of reference, and, as observed by Dever (1998), it essentially falls out of the kind of characterization of reference that I have provided. To refer to an object using a name is to pick out a particular object, and then presumably say something of that object. There is not really anything more to rigidity than is contained in that idea. On my pragmatic view of proper name reference, rigidity cannot be a property of names per se, but must be a feature of their reference (which only occurs on an occasion). As I discussed in §6.2, if a name refers to a particular object, then the truth of an utterance containing (a referential use of) that name turns on how that object is. This fact about reference just comes with rigidity: If we want to know the truth of an utterance of ‘Frances might have been a famous architect’, then, since ‘Frances’ is a referring term, we look at whether the referent might have been a famous architect, and this depends upon her modal properties, not on what the referent of the utterance might otherwise have been. Indeed, if things had been such that the reference of an utterance of the same name at the same time and place had been different, the occasion of utterance would have to have been different, and so the utterance itself would have been different. So, there is a trivial sense in which particular referential uses of a name have the reference they do essentially. In modal logical terms, the scenario I’ve just described is basically that of a name taking wide-scope over a modal operator. This seems very plausible where modal auxiliaries are concerned, such as in ‘Frances might have been a famous architect’, because the English syntax already makes it appear that 219 8.3. Problems of Names the name is out-scoping the modal. For name reference to be rigid in the received modal-logical sense, my account must also show that uses of names within the scope of ‘sentence’-modifying modal adverbs are rigid. Take a rather un-idiomatic example: ‘Necessarily, Frances is the height she is’. From a modal-logical point of view, this is true just in case the embedded sentence ‘Frances is the height she is’ is true in every possible world, and on the view that the semantic content of ‘Frances’ is just Frances herself, the name has the same reference in every world (if it is treated as a constant term with the same interpretation at every world, say). From my point of view, we can consider whether what was said with a particular utterance of ‘Frances is the height she is’ would have been true in counter-factual situations. This gives us the correct result. Something an utterance of ‘Frances is the height she is’ might plausibly say is that some particular person, Frances, is the height that she is. We can now ask whether that is necessarily true. But, of course, this is to apply modal reasoning to an object that has already been obtained, just as for the ‘wide-scope’ case, and the truth of the utterance turns on how that object is, not on any other merely-possible referents of the names. This answer was, of course, fairly obvious if we think about natural language rather than modal logic: ‘Necessarily, Frances is the height she is’ is intuitively equivalent to ‘It is necessary that Frances is the height she is’, and in this sentence it is clear that we are talking about Frances, the referent, and what she is like. This gives us rigidity in both an intuitive sense—the reference of a use of the name stays the same, whatever modal reasoning we apply—and in the modal logical sense—a wide-scope reading of a name over modal operators is equivalent to a narrow-scope reading. 8.3.2 Name Change and Reference Shift Cases in which the name of an object changes, or the reference of a name appears to shift have been significant to debates about proper name reference since Evans (1973) introduced the Madagascar case as an alleged counterexample to Kripke’s (1981) causal picture of reference. Sainsbury’s (2005) discussion of names is also riven with examples involving reference shift, and one of the primary motivations for the position he presents in his (2015) appears to be accounting better for such cases. Madagascar-type cases generally take the following form: There is an existing practice of using a particular name to refer to a particular object. Whilst appearing or intending to be following that practice, and using the name in accordance with it, speakers begin to use the name to refer to a different object, which eventually becomes the object of the practice, or develops into a new practice, depending 220 8.3. Problems of Names on how one thinks of such things. Such cases are well known to present problems for causal-homophonism and related views, for obvious reasons. But they also present potential problems for other accounts, such as some versions of indexicalism, that rely on bearer relations. This is because, at the point where the reference is shifting, the new object may not be a bearer of the name in question (though it may become one), and because the story one tells about how it is determined which bearer of a name a speaker is referring to might appeal to their intentions to be using one name-using practice or another. Such cases are significantly less problematic for my pragmatic view of proper name reference as it stands, however, which gives me a major advantage over other accounts. Because, on this view, a whole range of factors can contribute to the determination of reference; there is no requirement that there be a particular name-using practice or bearer relation that determines that a name refers to a particular object. Moreover, there is no requirement that a speaker refers to what they intend to with a use of a name. So, even if a speaker has certain intentions to refer to whatever other people are referring to with that name, they may refer to something else, or vice versa. It also does not matter if the speaker has muddled intentions about what they are referring to: the circumstances of utterance may make it clear that one object is the referent of the use of a name, even though it does not currently bear that name, and there are salient naming practices that pair that name with other objects. I have said relatively little in the rest of the thesis about what constitutes naming or name-using practices, except to indicate that they are social conventions of using particular names to refer to particular objects, the existence of which can be a factor in determining that utterances of that name refer to that object. I am content to leave further questions about the nature of these practices open. Describing exactly what is going on in Madagascar and other reference and use shift cases may still present some problems for my view, to the extent that describing what is actually occurring in each case may be complex, and it may be that a very thorough description of each case would be required in order to provide that detail. Nevertheless, my options for undertaking this task, given the flexibility of my pluralism about reference determination, are significantly better than those of many other accounts of names. 8.3.3 Attitudes and Beliefs, Fiction and Empty Reference At its core, my view of proper name reference is a direct reference view, like most, if not all, of the other rival positions I have discussed: it maintains that 221 8.3. Problems of Names reference allows talk about objects in the world by, in some manner, making objects themselves part of the interpretation of utterances containing names. As such, my view is subject to the problems that tend to affect such views. Two well known problems for such a view are propositional attitude reports and related puzzles about belief, and fictional names and uses of names with empty reference. Whereas my account, as I have stated it so far, has no particular obvious recourse with regard to these problems, it also does not do any worse than other direct reference accounts, and, in general, any broad approaches taken to these problems by the purveyors of other accounts will also be available to me. My view also has an advantage in being flexible and pluralistic, which makes it naturally ecumenical in the solutions it can adopt. I will briefly introduce each of the problems for the purposes of identifying further areas of research, but I will provide no more than a cursory suggestion of how my pragmatic view of reference might approach them. Puzzles about reporting beliefs and attitudes containing proper names, and other referring terms, have been central to the discussion of names and reference ever since Frege, arguably the originator of such discussion, introduced the puzzle about the morning star and the evening star (Frege 1892). These puzzles usually revolve around co-referring terms, and the fact that it appears that a rational person, who appears competent in the use of two co-referring terms may, nonetheless, fail to realize that they are co-referring. This presents a problem for direct accounts of reference, because they appear to entail that any co-referring names can be inter-substituted in an utterance salva veritate. After all, if the purpose of a name is just to make the truth of an utterance turn on how its referent is, then it shouldn’t matter what name one uses, as long as the right reference occurs. However, it is clear that an utterance of ‘The ancients believed that Hesperus appears in the evening sky’ can be true in the same context in which an utterance of ‘The ancients believed that Phosphorus appears in the evening sky’ is false. Thus, it is not true that all occurrences of co-referring names can be inter-substituted salva veritate. The kinds of utterance in which this kind of substitution fails—opaque contexts—appear to be almost entirely those that report something about the mental lives of thinkers. Moreover, the names must be being used as if the report is of something that the thinker would ascent to containing the name. If we rephrase the report in the last sentence as ‘The ancients believed of Hesperus that it appears in the evening sky’, then we can substitute the coreferring name and get a true utterance: ‘The ancients believed of Phosphorus that it appears in the evening sky’. This appears to suggest that in these opaque (linguistic) contexts, the names are not behaving in the same way that they usually do. Indeed, Frege’s solution was to claim that in such contexts, 222 8.3. Problems of Names names refer to their senses, rather than their usual referents. However, it is usually thought that, in order to have a sense in the appropriate manner, names must have a semantic meaning that goes beyond their reference, such as that proposed by indexicalists and denied by causal-homophonists. So far, I have suggested that the meanings of names are extremely sparse, so it is unclear whether this option is open to me, or what it would entail. However, since I claim that the reference of names is distinct from their meaning, I certainly have more avenues open to me than the homophonist. Exactly what I can and should say about the use of names in opaque reference contexts and attitude reports must be the purview of further work on my pragmatic view of name reference. The felicitous use of names to talk about things that, apparently, do not exist is also a potential problem for direct reference accounts of proper names, and a frequent concern of their authors. If reference essentially involves objects, what is going on in the case of utterances containing names that appear to behave like other names—they are meaningful—but have no referent? The most obvious uses of so-called empty names are to talk about fiction of various kinds, which is strewn with proper names that do not appear to refer to objects that exist in the manner of those that populate the real world, and to talk about things such as ‘Vulcan’, Le Verrier’s theoretical planet postulated to explain perturbations in the orbit of Mercury, which turned out not to exist. These kinds of uses of names present me with two desiderata for my pragmatic view. Firstly, that they be accounted for as felicitous and explicable uses of proper names of the same kind as those that refer—indeed, it must be possible that a single name can be used on different occasions to refer to real things and to talk about fictional things. Secondly, it should not involve positing additional ways of using proper names that are of an essentially different kind to ordinary uses, as this would fail to explain why such uses appear so similar to ordinary uses, and indeed, why speakers and hearers need not realize that they are talking about non-existent things. Given my views on the nature of language generally, I also think that it is desirable that an account of these kinds of uses of names should not entail that all such uses simply result in false utterances. Empty uses of names are made use of in a great variety of meaningful and productive discourse, and it is simply implausible that all such utterances are literally false, especially since they can express things that are clearly considered to be true. The resources of occasion-sensitivity appear to provide the scope for being able to say more nuanced things about the truth of utterances about fiction and other non-real things. For example, one might think that being a fictional kangaroo is one way to be a kangaroo so that, although Skippy isn’t a real 223 8.4. The Final Summary kangaroo, there is a sense in which she is a kangaroo, albeit a non-existent, fictional one. However, this still leaves the problem of what is happening with regard to reference: how can the truth of an utterance turn on the referent of a name if there is no such referent? One response would be suggest that there is a way to be a referent, such that truth can be turned on one, but that this doesn’t entail existence in the same way that real things exist. Whether this is a plausible route to take or not remains to be seen. I would not particularly want it to devolve into a realist account of fictional characters, since they rarely give the right kinds of results with regard to intuitively true or correct things that one can say about fiction.2 Another possible route to take would be to make use of a free logic within my semantics, which allows that there can be constant terms that have no interpretation, in the manner of Sainsbury (2005, 2009, 2015) and Rami (2014). However, in order to allow for true sentences involving empty uses of names, I would have to adopt a positive free logic, in contrast to the negative free logics preferred by Sainsbury and Rami, though this would be more complex. It is clear, however, that I have options with regard to the problem of empty names, as I do with the problems of attitude and belief reports. 8.4 The Final Summary Having gestured towards the ways in which my research might be extended and expanded, my thesis ends. In it, I have offered a novel story of proper name reference, upon which it is determined, not in virtue of the semantics of names, but in virtue of pragmatic features of context that are unspecified and unconstrained by semantics. 2 See Sainsbury (2009) for detailed arguments against such positions. 224 Bibliography Ahmed, Arif (2007). Saul Kripke. London: Continuum. APA (2013). Minorities in Philosophy. Tech. rep. American Philosophical Association. Bach, Kent (1987). Thought and Reference. Oxford: Clarendon Press. — (1992). “Intentions and Demonstrations”. Analysis 52 (2), 140–146. — (2002). “Giorgione was So-Called Because of His Name”. Philosophical Perspectives 16, 73–103. — (2012). “Meaning and Communication”. In Routledge Companion to the Philosophy of Language. G. Russell and D.G. Fara (eds.). London: Routledge, pp. 79–90. Bach, Kent and Robert M. Harnish (1979). Linguistic Communication and Speech Acts. Cambridge, Mass.: MIT Press. Baron-Cohen, Simon, Alan M. Leslie, and Uta Frith (1985). “Does the Autistic Child have a Theory of Mind?” Cognition 21, 37–46. Barwise, John and Robin Cooper (1981). “Generalized Quantifiers and Natural Language”. Linguistics and Philosophy 4, 159–219. Beebee, Helen (2013). “Women and Deviance in Philosophy”. In Women in Philosophy: What Needs to Change? K. Hutchison and F. Jenkins (eds.). Oxford: Oxford University Press, pp. 61–80. Beebee, Helen and Jennifer Saul (2011). Women in Philosophy in the UK. Tech. rep. The British Philosophical Association and The Society for Women in Philosophy UK. Borg, Emma (2004). Minimal Semantics. Oxford: Clarendon Press. Bosch, Peter (2007). “Productivity, Polysemy, and Predicate Identity”. In Proceedings of the Sixth International Tbilisi Symposium on Language, Logic and Computation. B. ten Cate and H. Zeevat (eds.). Heidelberg, Berlin: Springer, pp. 58–71. 225 Bibliography Breheny, Richard (2006). “Communication and Folk Psychology”. Mind and Language 21 (1), 74–107. Brennan, Susan E., Alexia Galati, and Anna K. Kuhlen (2010). “Two Minds, One Dialog: Coordinating Speaker and Understanding”. In The Psychology of Learning and Motivation. B.H. Ross (ed.). Vol. 53. Burlington: Academic Press, pp. 301–344. Brennan, Susan E. and Joy E. Hanna (2009). “Partner-Specific Adaption in Dialog”. Topics in Cognitive Science 1, 274–291. Buchanan, Ray (2013). “Conversational Implicature, Communicative Intentions, and Content”. Canadian Journal of Philosophy 43 (5–6), 720–740. Burge, Tyler (1973). “Reference and Proper Names”. The Journal of Philosophy 70 (14), 425–439. Burks, Arthur W. (1951). “A Theory of Proper Names”. Philosophical Studies 2 (3), 36–45. Cappelen, Herman (2013). Philosophy Without Intuitions. Oxford: Oxford University Press. Cappelen, Herman and Ernest Lepore (2005). Insensitive Semantics. Oxford: Blackwell. Carston, Robyn (2002). Thoughts and Utterances: The Pragmatics of Explicit Communication. Oxford: Blackwell. — (2013). “Word Meaning, What is Said, and Explicature”. In What is Said and What is Not. C. Penco and F. Domaneschi (eds.). Stanford: CSLI Publications. Chomsky, Noam (1979). Language and Responsibility. New York: Pantheon Books. — (2000). “Language as a Natural Object”. In New Horizons in the Study of Language and Mind. Cambridge: Cambridge University Press, pp. 106–133. Cohen, L. Jonathan (1980). “The Individuation of Proper Names”. In Philosophical Subjects: Essays Presented to P.F. Strawson. Z. Van Straaten (ed.). Oxford: Clarendon Press, pp. 140–163. Cooper, Robin, Simon Dobnik, Shalon Lappin, and Stefan Larsson (2014). “A Probabilistic Rich Type Theory for Semantic Interpretation”. In Proceedings of the EACL 2014 Workshop on Type Theory and Natural Language Semantics. R. Cooper, S. Dobnik, S. Lappin, and S. Larsson (eds.). Gothenburg, Sweden, pp. 72–79. Cooper, Robin, Simon Dobnik, Shalom Lappin, and Stefan Larsson (2015). “Probabilistic Type Theory and Natural Language Semantics”. Linguistic Issues in Language Technology 10 (1), 1–43. Corazza, Eros (2004). “Kinds of Context: A Wittgensteinian Approach to Proper Names and Indexicals”. Philosophical Investigations 27, 158–188. Cumming, Sam (2008). “Variabilism”. Philosophical Review 117, 525–554. 226 Bibliography Davies, Alex (2011). A Purpose for Context Sensitivity. PhD thesis. King’s College London. — (2014). “Off Target Responses to Occasion Sensitivity”. Dialectica 68 (4), 499–523. — (forthcoming). “Communicating By Doing Something Else”. In Charles Travis on Language, Thought, and Perception. J. Collins, A. Davies, and T. Dobler (eds.). Oxford: Oxford University Press. De Cruz, Helen (2015). “Where Philosophical Intuitions Come From”. Australasian Journal of Philosophy 93 (2), 233–249. Deutsch, Max (2009). “Experimental Philosophy and the Theory of Reference”. Mind and Language 24 (4), 445–466. Dever, Josh (1998). Variables. PhD thesis. UC Berkeley. Donnellan, Keith (1966). “Reference and Definite Descriptions”. Philosophical Review 75, 281–304. — (1968). “Putting Humpty Dumpty Together Again”. The Philosophical Review 77 (2), 203–215. Elbourne, Paul (2005). Situations and Individuals. Oxford: Oxford University Press. Evans, Gareth (1973). “The Causal Theory of Names”. Proceedings of the Aristotelian Society Supplementary Volume 47, 187–208. (Reprinted in G. Evans (1985), Collected Papers. Oxford: Clarendon Press, pp. 1–24. Page numbers refer to this volume.) — (1982). The Varieties of Reference. J. McDowell (ed.). Oxford: Clarendon Press. Fara, Delia Graff (2015). “Names are Descriptions”. Philosophical Review 124 (1), 59–117. Frege, Gottlob (1892). “Über Sinn und Bedeutung”. In The Frege Reader (1997). M. Beany (ed.). Oxford: Blackwell, pp. 151–171. (Orig. published in Zeitschrift für Philosophie und philosophische Kritik 100, 25–50.) Garcia Ramirez, Eduardo (2011). Proper Names, A Cognitive-Philosophical Study. PhD thesis. University of Michigan. Gauker, Christopher (2008). “Zero Tolerance for Pragmatics”. Synthese 165 (3), 359–371. Geurts, Bart (1997). “Good News About the Description Theory of Names”. Journal of Semantics 14 (4), 319–348. Gines, Kathryn T. (2011). “Being a Black Woman Philosopher: Reflections on Founding the Collegium of Black Women Philosophers”. Hypatia 26 (2). Glüer, Kathrin and Peter Pagin (2003). “Meaning Theory and Autistic Speakers”. Mind and Language 18 (1), 23–51. Goffman, Erving (1963). Behaviour in Public Places: Notes on the Social Organization of Gatherings. Toronto: Collier-Macmillan. 227 Bibliography Gray, Aidan (2014). “Name-Bearing, Reference, and Circularity”. Philosophical Studies 171, 207–231. — (forthcoming). “Minimal Descriptivism”. Review of Philosophy and Psychology. Gregoromichelaki, Eleni, Ruth Kempson, Matthew Purver, Gregory Mills, Ronnie Cann, Wilfried Meyer-Viol, and Patrick G.T. Healey (2011). “Incrementality and Intention-Recognition in Utterance Processing”. Dialogue and Discourse 2 (1), 199–232. Grice, H. Paul (1957). “Meaning”. The Philosophical Review 66 (3), 377–388. (Reprinted in Grice (1989), pp. 213–223. Page numbers refer to this volume.) — (1969). “Utterer’s Meaning and Intentions”. The Philosophical Review 78 (2), 147–177. (Reprinted in Grice (1989), pp. 86–116. Page numbers refer to this volume.) — (1989). Studies in the Way of Words. Cambridge, Mass.: Harvard University Press. Guasp, April, Helen Statham, Sarah Jennings, and Susan Golombok (2010). Different Families: The Experiences of Children with Lesbian and Gay Parents. London: Stonewall. Guin, Ursula le (1993). The Earthsea Quartet. London: Puffin. Hales, Stephen D. (2006). Relativism and the Foundations of Philosophy. Cambridge, Mass.: MIT Press. Hansen, Nat (2011). “Color Adjectives and Radical Contextualism”. Linguistics and Philosophy 34, 201–221. Harris, Daniel (2014). Speech Act Theoretic Semantic. PhD thesis. CUNY. Haslanger, Sally (2008). “Changing the Ideology and Culture of Philosophy: Not by Reason (Alone)”. Hypatia 23 (2), 210–223. Healey, Kieran (2011). Gender Divides in Philosophy and Other Disciplines. Blog post. Horton, William and Boaz Keysar (1996). “When do Speakers Take into Account Common Ground?” Cognition 59, 91–117. Jackson, Frank (1998). “Reference and Descriptions Revisited”. Philosophical Perspectives 12, 201–218. Jeshion, Robin (2015). “Referentialism and Predicativism about Proper Names”. Erkenntnis Special Edition Proper Names: Philosophical and Linguistic Perspectives 80 (2 Supplement). M. Textor, and D. Rami (eds.), 363–404. Justice, John (2001). “On Sense and Reflexivity”. The Journal of Philosophy 98 (7), 351–364. Kamp, Hans and Uwe Reyle (1993). From Discourse to Logic. Dordrecht: Kluwer. 228 Bibliography Kaplan, David (1989a). “Afterthoughts”. In Themes from Kaplan. J. Almog, J. Perry, and H. Wettstein (eds.). Oxford: Oxford University Press. — (1989b). “Demonstratives”. In Themes from Kaplan. J. Almog, J. Perry, and H. Wettstein (eds.). Oxford: Oxford University Press. — (1990). “Words”. Proceedings of the Aristotelian Society, Supplementary Volume 64, 93–119. Katz, Jerrold (2001). “The End of Millianism: Multiple Bearers, Improper Names, and Compositional Meaning”. The Journal of Philosophy 98 (3), 137–166. Keysar, Boaz (2007). “Communication and Miscommunication: The Role of Egocentric Processes”. Intercultural Pragmatics 4 (1), 71–84. Keysar, Boaz, Dale J. Barr, and William S. Horton (1998). “The Egocentric Basis of Language Use: Insights from a Processing Approach”. Current Directions in Psychological Sciences 7, 46–50. King, Jeff C. (2014). “Speaker Intentions in Context”. Noûs 48 (2), 219–237. Kripke, Saul (1977). “Speaker’s Reference and Semantic Reference”. Midwest Studies in Philosophy II. — (1981). Naming and Necessity. Oxford: Blackwell. Laurence, S. (1998). “Convention-Based Semantics and the Development of Language”. In Language and Thought. Interdisciplinary Themes. P. Carruthers and J. Boucher (eds.). Cambridge: Cambridge University Press, pp. 210– 218. Lewis, David (1988). “Vague Identity: Evans Misunderstood”. Analysis 48 (3), 128–130. — (1993). “Many, but Almost One”. In Ontology, Causality and Mind: Essays in Honour of D.M. Armstrong. J. Bacon, K. Campbell, and L. Reinhardt (eds.). Cambridge: Cambridge University Press. Lewis, Shevaun, Valentine Hacquard, and Jeffrey Lidz (2012). “The Semantics and Pragmatics of Belief Reports in Preschoolers”. Proceedings of SALT 22, 247–267. Loar, Brian (1980). “Names and Descriptions: A Reply to Michael Devitt”. Philosophical Studies 38 (1), 85–89. Ludwig, Kirk (2007). “The Epistemology of Thought Experiments: First Person Versus Third Person Approaches”. Midwest Studies in Philosophy 31, 128– 159. Machery, Edouard, Ron Mallon, Shaun Nichols, and Stephen Stich (2004). “Semantics, Cross-Cultural Style”. Cognition 92, B1–B12. Mallon, Ron, Edouard Machery, Shaun Nichols, and Stephen Stich (2009). “Against Arguments From Reference”. Philosophy and Phenomenological Research 79 (2), 332–356. 229 Bibliography Malmgren, Anna-Sara (2011). “Rationalism and the Content of Intuitive Judgements”. Mind 120 (478), 263–327. Martí, Genoveva (2009). “Against Semantic Multi-Culturalism”. Analysis 69 (1), 42–48. Matushansky, Ora (2008). “On the Linguistic Complexity of Proper Names”. Linguistics and Philosophy 21, 573–627. Mercier, Adele (1998). “On Communication-Based De Re Thought, Commitments De Dicto, and Word Individuation”. In Philosophy and Linguistics. R. Stainton and K. Murasugi (eds.). Boulder: Westview Press, pp. 85–111. Michaelson, Eliot (2013). This and That: A Theory of Reference for Names, Demonstratives, and Things in Between. PhD thesis. UCLA. Millikan, Ruth (2005). Language: a Biological Model. Oxford: Clarendon Press. Mitchell, Peter, Elizabeth J. Robinson, and Doreen E. Thompson (1999). “Children’s Understanding that Utterances Emanate from Minds: Using Speaker Belief to Aid Interpretation”. Cognition 72, 45–66. Mount, Allyson (2008). “Intentions, Gestures, and Salience in Ordinary and Deferred Demonstrative Reference”. Mind and Language 23 (2), 145–164. Nagel, Jennifer (2012). “Intuitions and Experiments: A Defense of the Case Method in Epistemology”. Philosophy and Phenomenological Research 85 (3), 495–527. Neale, Stephen (2004). “This, That and the Other”. In Descriptions and Beyond. M. Reimer and A. Bezuidenhout (eds.). Oxford: Oxford University Press, pp. 68–182. — (ms.). “Determinations of Meaning”. Unpublished manuscript, CUNY/NYU Mind & Language Seminar, February 2013. Available at ❤tt♣✿✴✴♥❡❛❧❡✳ ❝♦♠♠♦♥s✳❣❝✳❝✉♥②✳❡❞✉✴❢✐❧❡s✴✷✵✶✷✴✵✾✴◆❡❛❧❡❴❉❡t❡r♠✐♥❛✲t✐♦♥s❴❋✐♥❛❧✳ ♣❞❢. Accessed on 22/05/2014. Nunberg, Geoffrey (1993). “Indexicality and Deixis”. Linguistics and Philosophy 16 (1), 1–43. Papineau, David (1996). “Doubtful Intuitions”. Mind and Language 11 (1), 130– 132. Pelczar, Michael W. (2001). “Names as Tokens and Names as Tools”. Synthese 128, 133–155. Pelczar, Michael W. and Joseph M. Rainsbury (1998). “The Indexical Character of Names”. Synthese 114, 293–317. Perry, John (1998). “Indexicals, Contexts and Unarticulated Constituents”. In Proceedings of the 1995 CSLI-Amsterdam Logic, Language and Computation Conference. CSLI Publications. Plato (1997). “Cratylus”. In Plato Complete Works. J.M. Cooper (ed.). C.D.C. Reeve (transl.). Indianapolis: Hackett, pp. 101–156. 230 Bibliography Predelli, Stefano (1998). “Utterance, Interpretation and the Logic of Indexicals”. Mind and Language 13 (3), 400–414. — (2002). “Intentions, Indexicals and Communication”. Analysis 62 (4), 310– 304. — (2012). “Bare-Boned Demonstratives”. Journal of Philosophical Logic 41 (3), 547–562. — (2013). Meaning Without Truth. Oxford: Oxford University Press. Putnam, Hillary (1975). “The Meaning of ‘Meaning’”. Minnesota Studies in the Philosophy of Science 7, 131–193. Rami, Dolf (2014). “The Use-Conditional Indexical Conception of Proper Names”. Philosophical Studies 168 (1), 119–150. — (2015). “Multiple Uses of Proper Nouns”. Erkenntnis Special Edition Proper Names: Philosophical and Linguistic Perspectives 80 (2 Supplement). M. Textor, and D. Rami (eds.), 405–432. — (2016). “Names, Naming and Name-Using Practices”. In Philosophical Approaches to Proper Names. P. Stalmaszczyk and L. Fernández Moreno (eds.). Frankfurt-am-Main: Peter Lang, pp. 55–92. Recanati, Francois (1993). Direct Reference: From Language to Thought. Oxford: Blackwell. — (2004). Literal Meaning. Cambridge: Cambridge University Press. — (2010). Truth-Conditional Pragmatics. Oxford: Oxford University Press. Reichenbach, Hans (1947). Elements of Symbolic Logic. London: Macmillan. Rothschild, Daniel and Gabriel Segal (2009). “Indexical Predicates”. Mind and Language 24 (4), 467–493. Russell, Bertrand (1905). “On Denoting”. Mind 14, 479–493. Sainsbury, R. Mark (2005). Reference Without Referents. Oxford: Oxford University Press. — (2009). Fiction and Fictionalism. London: Routledge. — (2015). “The Same Name”. Erkenntnis Special Edition Proper Names: Philosophical and Linguistic Perspectives 80 (2 Supplement). M. Textor, and D. Rami (eds.), 195–214. Salmon, Nathan (2005). “On Designating”. Mind 114 (456), 1069–1133. Saul, Jennifer (1997). “Substitution and Simple Sentences”. Analysis 57 (2), 102–108. Sawyer, Sarah (2010). “The Modified Predicate Theory of Proper Names”. In New Waves in Philosophy of Language. S. Sawyer (ed.). London: Palgrave Macmillan, pp. 206–225. Schiffer, Stephen (2010). “Vague Properties”. In Cuts and Clouds: Vagueness, its Nature and its Logic. R. Dietz and S. Moruzzi (eds.). Oxford: Oxford University Press, pp. 109–130. 231 Bibliography Searle, John (1958). “Proper Names”. Mind 67 (266), 166–173. Soames, Scott (2002). Beyond Rigidity: The Unfinished Semantic Agenda of Naming and Necessity. Oxford: Oxford University Press. Sperber, Dan and Deirdre Wilson (1986). Relevance, Communication and Cognition. Oxford: Blackwell. Stanley, Jason (2007). Language in Context. Oxford: Oxford University Press. Sutton, Peter (2013). Vagueness, Communication, and Semantic Information. PhD thesis. King’s College London. — (2015). “Towards a Probabilistic Semantics for Vague Adjectives”. In Bayesian Natural Language Semantics and Pragmatics. H. Zeevat and H.-C. Schmitz (eds.). Language, Cognition, and Mind. Berlin: Springer, pp. 221–246. Szabo, Zoltan (2001). “Adjectives in Context”. In Perspectives on Semantics, Pragmatics, and Discourse. R.M. Harrish and I. Kenesei (eds.). Amsterdam: John Benjamins. Tiedke, Heidi (2011). “Proper Names and Their Fictional Uses”. Australasian Journal of Philosophy 89 (4), 707–726. Travis, Charles (1978). “Meaning versus Truth”. Dialogue 17 (3), 401–430. — (1980). “Classical Theories of Reference”. Canadian Journal of Philosophy, Supplementary Volume 6, 139–159. — (1981a). “Reference, Speakers and Semantics”. Language & Communication 1, 13–38. — (1981b). The True and The False: The Domain of the Pragmatic. Amsterdam: John Benjamins. — (1985). “On What is Strictly Speaking True”. The Canadian Journal of Philosophy 15 (2), 187–229. — (1989). The Uses of Sense: Wittgenstein’s Philosophy of Language. Oxford: Clarendon Press. — (1990). “Attitudes as States”. In Proceedings of the Fourteenth Wittgenstein Symposium: Centenary Celebration. Haller and Brandl (eds.). Vienna: HölderPichler-Tempsky, pp. 265–284. (Reprinted in Travis (2008), pp. 227–252. Page numbers refer to this volume.) — (1991). “Annals of Analysis”. Mind 100 (2), 237–264. (Reprinted in Travis (2008), pp. 65–93. Page numbers refer to this volume.) — (1995). “On Concepts of Objects”. In Karlovy Vary Studies in Reference and Meaning. Hill and Kot’atko (eds.). Prague: Filosophia, pp. 365–389. (Reprinted in Travis (2008), pp. 252–270. Page numbers refer to this volume.) — (1996). “Meaning’s Role in Truth”. Mind 105 (3), 451–466. (Reprinted in Travis (2008), pp. 94–108. Page numbers refer to this volume.) 232 Bibliography Travis, Charles (1997). “Pragmatics”. In A Companion to the Philosophy of Language. B. Hale and C. Wright (eds.). Oxford: Blackwell, pp. 237–264. (Reprinted in Travis (2008), pp. 109–129. Page numbers refer to this volume.) — (2000). Unshadowed Thought: Representation in Thought and Language. Cambridge, Mass.: Harvard University Press. — (2006a). “Insensitive Semantics”. Mind and Language 21 (1), 39–49. (Reprinted in Travis (2008), pp. 150–160. Page numbers refer to this volume.) — (2006b). Thought’s Footing: a Theme in Wittgenstein’s Philosophical Investigations. Oxford: Clarendon Press. — (2008). Occasion Sensitivity. Oxford: Oxford University Press. — (2011). “Truth and Merit”. In New Essays on the Philosophy of J.L. Austin. Gustafson and Sørli (eds.). Oxford: Oxford University Press. (Reprinted in C. Travis (2011). Objectivity and the Parochial. Oxford: Oxford University Press, pp. 229–261. Page numbers refer to this volume.) — (ms.). “Sense and Sensitivity”. Unpublished manuscript, available at ❤tt♣s✿ ✴✴✇✇✇✳❛❝❛❞❡♠✐❛✳❡❞✉✴✶✶✾✾✸✽✵✻✴❙❡♥s❡❴❛♥❞❴❙❡♥s✐t✐✈✐t②. Accessed on 1/5/2016. Wales, Katie (2006). Northern English: A Social and Cultural History. Cambridge: Cambridge University Press. Weinberg, Jonathan M., Chad Gonnerman, Cameron Buckner, and Joshua Alexander (2010). “Are Philosophers Expert Intuiters?” Philosophical Psychology 12 (10), 331–355. Wellman, Henry M., David Cross, and Julianne Watson (2001). “Meta-analysis of Theory-of-Mind Development: The Truth About False Belief”. Child Development 72, 655–684. Wettstein, Howard (1984). “How to Bridge the Gap between Meaning and Reference”. Synthese 58, 63–84. Williamson, Timothy (2007). The Philosophy of Philosophy. Oxford: Blackwell. Wimmer, Heinz and Josef Perner (1983). “Beliefs about Beliefs: Representation and Constraining Function of Wrong Beliefs in Young Children’s Understanding of Deception”. Cognition 13, 103–128. Wittgenstein, Ludwig (1953). Philosophical Investigations. P.M.S. Hacker and J. Schulte (eds.). Translated Anscombe, Hacker & Schulte. Revised 4th edition, 2009. Oxford: Wiley-Blackwell. Yergeau, Melanie (2013). “Clinically Significant Disturbance: On Theorists Who Theorize Theory of Mind”. Disability Studies Quarterly 33 (4). Ziff, Paul (1960). Semantic Analysis. Ithaca: Cornell University Press. — (1977). “About Proper Names”. Mind 86 (343), 319–332. 233 Bibliography Zimmermann, T. Ede (2005). “What’s in Two Names?” Journal of Semantics 22, 53–96. Zimmermann, T. Ede and Jean-Yves Lerner (1991). “Eigennamen”. In Semantik. Ein Internationales Handbuch. A.v. Stechow and D. Wunderlich (eds.). Berlin: de Gruyter, pp. 349–370. 234