Philos Stud (2014) 168:119–150 DOI 10.1007/s11098-013-0264-x

The use-conditional indexical conception of proper names

Dolf Rami

Published online: 20 December 2013 Ó Springer Science+Business Media Dordrecht 2013

Abstract In this essay I will defend a novel version of the indexical view on proper names. According to this version, proper names have a relatively sparse truth-conditional meaning that is represented by their rigid content and indexical character, but a relatively rich use-conditional meaning, which I call the (contex- tual) constraint of a proper name. Firstly, I will provide a brief outline of my favoured indexical view on names in contrast to other indexical views proposed in the relevant literature. Secondly, two general motivations for an indexical view on names will be introduced and defended. Thirdly, I will criticize the two most popular versions of the indexical view on names: formal variable accounts and salience-based formal constant accounts. In the fourth and final section, I will develop my own use-conditional indexical view on names in three different steps by confronting an initial version of this view with three different challenges.

Keywords Proper names Indexicals Determination of reference Reference to past bearers of a name Multiple bearers of a name Empty names

1 Setting the stage: Indexical views on proper names

Indexicals are linguistic expressions whose semantic reference depends in a certain way on specific parameters of the context of use. Therefore, an indexical expression can have different semantic referents relative to different contexts of use. Prototypical and uncontroversial examples of indexicals are expressions like ‘I’, ‘here’, ‘now’ and ‘this’. The view that proper names are indexical expressions is

D. Rami (&) Department of Philosophy, King’s College London, Strand, London WC2R 2LS, UK e-mail: [email protected]; [email protected]

D. Rami Department of Philosophy, University of Go¨ttingen, Humboldtallee 19, 37073 Go¨ttingen, Germany 123 120 D. Rami controversial. Several important and influential contemporary philosophers of language1 explicitly reject this view or at least hold a view that is incompatible with it; and, I think, it is safe to say that the majority of philosophers holds the view that proper names are non-indexical expressions. There is only a comparatively small group of people that have explicitly argued for an indexical view on proper names.2 We can distinguish between two different general varieties of an indexical view on proper names that can be found in the relevant literature: informal and formal variants. A defender of the first kind of variant identifies certain natural language expressions that are unquestionably indexical and claims that they are semantically equivalent to proper names. Typically, such an equivalence claim is put forward without providing a formal semantic representation of the proposed equivalent indexical expression.3 The following claim is an example of such a view: ‘Alfred’ is semantically equivalent to ‘the present bearer of ‘Alfred’’. A formal variant of the indexical view on proper names makes use of a formal semantic framework that is suitable for the representation of indexical expressions and conceives of proper names as indexical expressions according to this framework. The distinction between formal and informal variants of an indexical view on names is obviously not exclusive. There could also be mixed variants. A more interesting and exclusive distinction is the distinction between distinctive formal and informal or mixed variants. A distinctive formal variant holds that proper names are a distinctive kind of indexical expression in natural language such that there aren’t any other indexical expressions in natural language that are not proper names (like complex demonstratives4 or indexical definite descriptions5), but are seman- tically (or truth-conditionally)6 equivalent to names. I will provide some reasons in favor of such a distinctive view in the second, third and fifth section of this essay. A formal constant indexical account on names represents names formally as simple7 or complex8 non-logical constants with an indexical semantics. The best- known formal semantic framework amenable to such an approach is Kaplan’s logic of demonstratives—in the following abbreviated as KLD.9 I will defend a novel

1 Some examples: Bach, Fine, Donnellan, Kaplan, Kripke, Perry, Sainsbury, Salmon and Soames. 2 C.f.: Burks (1951), Burge (1973), Cohen (1980), Sommers (1980), Zimmermann and Lerner (1991), Recanati (1993), Pelczar and Rainsbury (1998), Dever (1998), Pelczar (2001), Elbourne (2005), Cumming (2008), Matushansky (2008), Sawyer (2010) and Tiedke (2011). 3 C.f.: Burks (1951), Cohen (1980), Sommers (1980) and Sawyer (2010). 4 E.g.: ‘That Alfred‘or ‘That bearer of ‘Alfred’’. 5 E.g.: ‘The present bearer of ‘Alfred’’. 6 Typically, two expressions are considered as semantically equivalent if they make the same contribution to truth-conditional content. But if we accept additional non-truth-conditional aspects of meaning, and I will argue for such an additional layer of meaning in this paper, truth-conditional equivalence is only a necessary condition for semantic equivalence. 7 Pure indexicals like ‘I’ are conceived of as simple individual constants in Kaplan (1978, 1989). 8 True demonstratives like ‘that’ are represented as complex expressions of the form ‘dthat[the /]’ in Kaplan (1978, 1989). 9 C.f.: Kaplan (1978, pp. 86–97; 1989, pp. 541–553). 123 Indexical view on proper names 121 version of a formal constant approach of the indexical view on names that is based on KLD and holds that names are a distinctive kind of indexical expression. It is reasonable to distinguish a truth-conditional from a use-conditional layer of meaning. The truth-conditional meaning of an expression concerns its contribution to the truth-conditions of a sentence that contains such an expression. The use- conditional meaning, on the other hand, concerns its contribution to the semantically correct use of a sentence relative to a context of use. According to KLD, every expression has two different aspects of truth-conditional meaning: content and character. Contents are a certain kind of generalization of what is classically called an intension. An intension is a (total or partial) function from possible worlds into extensions. Contents are functions from circumstances of evaluation into extensions. In the minimal sense of circumstances of evaluation, they are just possible worlds. But a circumstance of evaluation can contain additional parameters; for example, also a time parameter. In the case of a singular term, the extension is identical to the referent of this term. A content of a singular term is rigid if the output of this function is the same relative to every possible world. Character is a second layer of truth-conditional meaning additional to contents. It is a total function from contexts of use into contents. Kaplan conceives contexts of use as ordered quadruples consisting of (a) the agent, (b) the place, (c) the time and (d) the possible world of the context of use of a specific expression. It might be necessary to extend these parameters to capture the character of specific indexical expressions in an adequate way.10 A character is non-indexical if it has the same output relative to every context of use; it is indexical if not. In addition to these two truth-conditional layers of meaning, I will add a third use-conditional layer. I will call this additional layer (contextual) constraint.11 A constraint is a (proper) subset of the set of all contexts of use. That is, the contextual constraint of an expression e restricts the set of all possible contexts of use of e to the set of all contexts of use of e relative to which e is used in an acceptable or felicitous way. According to the indexical view on names that I will defend in this paper, proper names are non-logical constant expressions, with a rigid content, a specific indexical character, and a specific contextual constraint. It will be the main concern of this paper to characterize the specific indexical character and contextual constraint of proper names. My defence and development of the proposed use-conditional version of the indexical view on names will proceed in the following way: Firstly, I will provide two different general motivations for an indexical view on proper names. The first motivation will concern the so-called problem of shared names. The second motivation concerns the mechanisms that determine the referent of a proper name. Secondly, I will criticize the two most popular versions of the indexical view on

10 C.f.: Predelli (2005, p. 20; 2012, p. 550). 11 This view is inspired by Predelli (2012, p. 555). 123 122 D. Rami names: formal variable accounts12 and salience-based formal constant accounts.13 Apparent bound anaphoric uses of proper names are the main motivation for the former accounts. I will try to undermine this motivation and point out some problems of two different paradigm variable accounts defended in Elbourne (2005) and Cumming (2008). After that I will be concerned with salience-based accounts based on KLD as they are defended most prominently in Zimmermann and Lerner (1991), Recanati (1993), Pelczar and Rainsbury (1998) and Pelczar (2001). These views have different individual problems, but I will also try to show why salience- based variants are in general inadequate. Thirdly and finally, I will develop my own version of a formal constant indexical view on names that is based on pluralism about the determination of reference of a proper name, a view that I will introduce and defend in the second section of this paper, and a specific development of a suggestive analogy between names, complex demonstratives and third person personal pronouns. I will confront a preliminary version of this view with three problems and develop on this basis the proposed use- conditional variant of my formal constant indexical view on names.

2 Two general motivations for an indexical view on proper names

The main opponent of an indexical view on names is the so-called ambiguity view on names. These two views primarily disagree over two important issues: (a) the individuation of proper names and (b) the determination of the semantic referents of names. The ambiguity view holds concerning the individuation issue that there are no two different things that literally share a name, while the indexical view assumes that it is possible that one and the same name can have different bearers and different referents relative to different contexts of use. The ambiguity view holds that the semantic referent of a specific name is determined by a single and non- relativized relation that obtains between a name and its referent.14 The indexical view is more flexible concerning the issue of reference determination. There are indexicals, where one kind of relation that can be relativized to different contexts of use determines reference or there might also be completely different relations relative to different contexts of use at work that do this job. I will now try to show why there are reasons that concern these two issues that favor the indexical view over the ambiguity view on names.

12 A simple formal variable account represents names at the level of logical form by specific individual variables. Like the account defended in Cumming (2008). A complex formal variable account represents names at the level of logical form by means of specific complex singular terms that may contain an unbound individual variable. Burge and Elbourne, for example, have claimed that the name ‘Alfred’ is semantically equivalent to ‘The x:[Alfred (x) ^ x = y]’. C.f.: Burge (1973) and Elbourne (2005). 13 It is quite popular to use the notion of salience in connection with an indexical account based on KLD to solve the problem of shared names. One example of such a view claims that the name ‘Alfred’ is semantically equivalent to dthat [the most salient bearer of the name ‘Alfred’]’. 14 C.f.: Perry (2001, pp. 102–105). 123 Indexical view on proper names 123

2.1 The problem of shared names

The ambiguity view on names is mainly based on an analogy with homophonic ambiguous predicates. The English predicate ‘is a bank’ is in a non-systematic way ambiguous: it has different extensions relative to its different meanings. Not only is the representation of the predicate ‘is a bank’ in a formal language provided by two separate expressions, there also seem to be two different words in English that are spelled ‘bank’ and which have two completely different origins and histories. The ambiguity view interprets names in a similar way: The extension of a proper name is its semantic referent. Different referents correspond with different meanings of a name. Every act that assigns a new object to a certain name is an originating event of a new name. Names are individuated by their histories and histories of names by originating events. Therefore, there are no two different objects that bear the very same name. This consequence seems intuitively problematic. It concerns one important difference between names and non-systematically ambiguous predicates: Names can intuitively be shared by different objects. We accept sentences like the following as true: (1) There are many different bearers of the name ‘Alfred’. (2) In honor of David Hume, his parents gave David Kaplan the same first name as the former. The defender of the ambiguity view has to react to this data in some way. A radical way would be to deny that sentences like (1) and (2) have a true reading at all. According to this view, ordinary users of names and certain linguists and philosophers of language are massively ignorant about the true nature of names. Such a view strikes me as implausible. It might be defended in the case of natural kinds. But names are social artifacts and their nature essentially depends on how we think about them and how we use them. It is implausible to assume that we are massively ignorant about the individuation conditions of names. I think that is the main reason why most defenders of the ambiguity view aim to capture our intuitions about shared names and the sameness of names in some way. The most popular move to account for our data is based on an ontological distinction between generic and specific (or common currency15) names. Kaplan famously draws this distinction as follows: There is the generic name ‘David’, and then there is my [common currency] name ‘David’, there is David Lewis’ [common currency] name ‘David’, and so on. These are all three distinct words. The latter two 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).16

15 That is Kaplan’s name for specific names. I follow Sainsbury (2012) and use the other label in the following. 16 Kaplan (1990, p. 111). 123 124 D. Rami

Generic names are the genera, or species, of our individual common currency names.17 The properties that names intuitively exemplify are on this basis divided between generic and specific names. Specific names capture the semantic and communicative role of names, while generic names are introduced to capture certain intuitions about the sameness and shared-ness of names. Such a strategy seems to be problematic in several ways. There are two methodological problems: Firstly, this distinction leads to an ambiguity of the ordinary notion of a name. Ordinary speakers are not aware of this ambiguity that concerns sentences like (1) and (2). The view seems to be in danger of multiplying meanings beyond necessity.18 Secondly, it seems to be also methodo- logically questionable to draw such an ontological distinction just to fill an obvious explanatory weakness of the proposed historic individuation of names. The view seems to be in danger of multiplying entities beyond necessity. It is not only necessary to postulate a distinction between generic and specific names to solve the outlined aspect of the problem of shared names, but also to postulate that generic names can also have name-bearers in a certain sense and that they can have more than one bearer in this sense. Only on this basis we can interpret sentences like (1) or (2) as true. The best way to fulfill this requirement is to postulate that the notion of a bearer of a name is systematically ambiguous19: A generic name x has a bearer (in a derived sense) iff at least one of all specific names that are instances 20 of x has a bearer.21,22 But such a view seems to face a dilemma. If we want to account for the intuitively true readings of (1) and (2), we have to regard generic names as names and words. The ambiguity view identifies the bearers of a name with the referents and extensions of names. If generic names have bearers in a derived sense that is systematically related to the bearers of specific names, then generic names also have in a derived sense semantic referents and extensions. But then they cannot be conceived in a meaningful way as names, because they have multiple referents and extensions and therefore would require disambiguation. But this would undermine the whole motivation for the distinction between generic and specific names. It is also unclear for

17 Kaplan (1990, p. 108). 18 That is, if we translate sentences like (1) and (2) into a formal language, we have to represent their different meanings by means of different formal representations with different truth-conditions. 19 Alternatively, one might claim that a quotation name like ‘‘’Peter’’’ can only be used to refer to generic names. But such a restriction seems to be implausible and ad hoc. The quoted passage from Kaplan at least suggests that Kaplan himself does not explicitly hold such a radical view, because he uses quotation names there to refer to both kinds of names. 20 I use the relation expressed by ‘x is an instance of y’ to capture the essential ontological connection between specific and generic names. 21 This view is similar to a view that holds that propositions are primary bearers of truth-values and that sentences that express propositions are only in a derived sense true or false. 22 We do not multiply the readings of (1) and (2) to four readings (three false and one true reading), if we regard the expressions ‘name’ and ‘bearer’ as systematically ambiguous. Two logically possible readings are excluded as semantically not well-formed. Therefore, a sentence like (1) now has a true and a false reading. 123 Indexical view on proper names 125 which semantic and communicative purpose we could use generic names qua words that have multiple referents and extensions. So maybe it is better to regard them not as words, as Kaplan also considers in our quote, but as some abstract unnatural kinds that function as templates for the creation of new specific names. But if generic names are conceived as unnatural kinds that function as templates concerning the linguistic form of expressions, it does not make sense to treat them as representational devices that have name-bearers or extensions. Therefore, on this basis we again cannot account for the desired readings of (1) and (2). Hence, the sketched view seems to face a dilemma and so it is questionable whether we can assign to generic names a coherent and meaningful explanatory status. Another problem concerns the nature of the relation between generic and specific names. It is completely unclear how generic and specific names are exactly related. What is clear is that generic names are individuated in a more coarse-grained way than specific names. Some have suggested that generic and specific names are related in the same way as species and specimen. Another suggestion might be that they are related like types and tokens.23 But this latter suggestion does not work if we individuate specific names by their origin, and therefore resemblance in form is neither a necessary nor a sufficient condition for their identity. Normally, different specimen of one species share a certain essential and common feature that is independent form their origin, but it is not clear what this feature should be in case of specific names if different specific names that are instances of the same generic name may completely differ in their form, might have different origins and might also have different bearers or no bearer at all. These observations show, I think, that the distinction between specific and generic names provides no plausible solution to the mentioned problem. Prima facie a historic individuation as it is proposed by the ambiguity view also has some desirable features. It can account for an important similarity between predicates like ‘bank’ and names. There are same-spelled names with completely different origins and they should therefore also count as different words. Typically, indexical views on names commit themselves to the popular individuation of words and names by their spelling or pronunciation. But there are good reasons against such a view. The very same name can have different spellings. There is, for example, a Russian tennis player whose name is sometimes spelled ‘Juschni’ and in other contexts ‘Youzhny’. The very same name can have different pronunciations. Different people from different regions with different accents or mother tongues can pronounce the very same name in different ways. Against this background, the individuation of names by their origin seems to be much more plausible and also the idea that names are in this respect similar to other nouns. But there are quite different ways to defend a historical approach concerning the individuation of names. Such an approach is not necessarily committed to the view that each act of naming or establishment of new bearers of a name leads to the creation of a new name.24 It seems to be possible and plausible to identify the originating events of names in a different way. The name that is nowadays spelled ‘David’ in English or German has a long tradition, whose origin is somewhere

23 C.f. Kaplan (1990, p. 111). 24 This was pointed out to me by Peter Sutton. 123 126 D. Rami in the Middle East in biblical times. There are different objects that bear this name with its long history. Therefore, the plausible idea that names are individuated by their origin can be incorporated into an indexical view on names and without conceiving of names in analogy to ambiguous predicates. There are also additional significant differences between names and ambiguous predicates that speak against the adequateness of the proposed analogy. Firstly, ambiguous expressions like the predicate ‘is a bank’ only have a relatively small number of meanings. Secondly, most ordinary speakers are perfectly aware of these different meanings and the knowledge of the meanings of these expressions is a requirement for linguistic competence concerning the use of these expressions. Thirdly, it is also not clear in the case of names whether a difference in reference/ extension is really a sufficient condition for a difference in meaning. One aspect of the problem of shared names is to account for intuitive truths like (1) and (2), another aspect is to explain how the truth of such claims and therefore the shared name thesis that names can have more than one bearer is compatible with an additional, intuitively plausible assumption, which we might call the singular reference thesis: It is the (semantic) function of a proper name to refer to one and only one thing. The indexical view, if it is combined with the mentioned alternative historic individuation of proper names, can provide a straight-forward solution to the first aspect of this problem that is in perfect agreement with common sense. But the view also has the resources to deal with the second aspect in a quite moderate way. The indexical view on names relativizes the reference relation in case of proper names to contexts of use and can in this sense account for the intuition that names have at most one semantic referent relative to a context of use. Additionally, it rejects the assumption of the ambiguity view that the name-bearer relation is necessarily equivalent to the relation of semantic reference in the case of names. Hence, such a view can account for the truth and compatibility of the shared name thesis and the singular reference thesis in a quite moderate way. We do not have to sacrifice any important assumption concerning the nature or semantics of names for this purpose. This shows, I think, that the indexical view has to offer the far best solution to the problem of shared names and it fairs on this issue much better than its main opponent, the ambiguity view. The two mentioned aspects of the problem of shared names and their solution therefore provide one good motivation for adopting an indexical view on names.

2.2 Pluralism about the determination of reference

Beside the analysis of the meaning and semantic status of proper names in natural languages, the analysis of adequate mechanisms for the determination of the reference of a name (or use of a name) relative to a context of use is a central theoretical concern. There are two different views about the determination of the referent of a proper name that have dominated the discussion in the last decades: the descriptivist and the communication chain conception of reference determination. Both views share the general assumption that there is a single general mechanism that accounts for the determination of the reference of a name. But they disagree about the nature of this general mechanism. According to descriptivism, there is an object-identifying descriptive meaning or content that is associated with a name (or 123 Indexical view on proper names 127 use of a name) and the referent of this name (or use of this name) is determined by means of the satisfaction of this descriptive content. According to the communi- cation chain picture, there is an initial act of naming that conventionally links a name with its bearer and after that initiating act a name is typically passed on from user to user. The referent of such a name (or use of this name) is determined by an act of participation in such a communication network that links the name (or use of this name) to the referent of the initial act of introduction. But both views seem to overgeneralize. In the case of descriptivism that is very well-known. It is possible to use a name and refer to an object even if one is not in the position to associate a specific definite description with a bearer of this name (ignorance case); or if the description that is associated with a specific name is not satisfied by a bearer of this name (error case).25 But the communication chain view seems to overgeneralize as well: It is also not true that a proper name with an established use can only be used in a parasitic way and therefore only with the intention to use this name like some person or group of people that have uses this name previously (in a certain situation) in a certain way. In fact, there seem to be quite different ways to determine the reference of a proper name in an adequate way. Therefore, I propose to reject the general assumption made by both mentioned views and hold instead a pluralist (or disjunctivist) view about reference determination. We can make use of the jargon of reference fixing referential intentions to draw distinctions that are required to formulate the version of pluralism that I aim to propose. According to this view, the referent of a name can be determined: (a) by means of object-related referential intentions, (b) by means of descriptive referential intentions, (c) by means of parasitic referential intentions, or (d) by means of mixed referential intentions that are combinations of at least two of the three other kinds of intentions. If there is an object x in the focus of attention of a speaker S and S intends to use the name ‘N’ to refer to x, then the referent of ‘N’ is determined by means of an object-related referential intention.26 If a speaker S intends to use the name ‘N’ to refer to the F, where ‘the F’ is a placeholder for complete attributively used definite descriptions, then the referent of ‘N’ is determined by means of a descriptive referential intention. If a speaker S intends to use the name ‘N’ in the same way as T, where ‘T’ stands for a certain person or a certain group of people, then the referent of ‘N’ is determined by means of a parasitic referential intention. A proper name can be used accompanied by an object-related referential intention if a bearer of this name (or at least a representation of this bearer) is present in the context of use of this name. Such intentions can be made manifest by a pointing gesture or something similar.27

25 C.f.: Kripke (1980, pp. 82–92). 26 I have slightly modified Sainsbury’s general distinction between descriptive and object-related intentions for my purposes. Sainsbury uses this distinction to clarify a distinction made by Kripke between two ways of introducing a proper name into use. See: Kripke (1980, p. 96) and Sainsbury (2005, pp. 106–107). 27 A person may enter a stage, a spot-light may shine on him and he gets announced by someone who utters ‘We are very proud to present you Peter James’. In this case it is very plausible to assume that the 123 128 D. Rami

Often proper names are used accompanied by a descriptive referential intention if their user is in the position to identify a specific bearer of this name (or at least thinks he is able to) and this bearer is not (visually) present in the context of use of this name. A descriptive referential intention can be made manifest by descriptive appositions.28 Furthermore, it is very likely that a proper name will be used accompanied by an parasitic referential intention if someone aims to participate in a conversation, where people talk about a person or thing he does not know, and he aims to gain more knowledge about this person or thing and therefore asks certain questions about it.29 There might also be cases where a person uses a name accompanied by different kinds of intentions and in certain case they may also be incompatible such that we have to decide which intention trumps over the others.30 If one dislikes the jargon of referential intentions or thinks that it commits one to a too narrow and implausible internalist framework of reference determination, it is also possible to make our distinction between different mechanisms of reference determination on the basis of a more neutral terminology.31 The proposed version of pluralism about the determination of reference of a proper name can also be expressed as follows: The referent of a proper name can be determined either by a demonstrative, a descriptive, or a parasitic act of identification of a certain object or a mixed identification that is based on these three kinds of acts of identification. And we can leave it against this background, for example, in the demonstrative case open whether a certain pointing gesture is sufficient for a demonstrative identification of an object or whether it must be accompanied by a specific object-related referential intention or some other sort of intention.32,33

Footnote 27 continued name is used accompanied by an object-related referential intention. Furthermore, in a meeting someone may point to a specific person and say ‘Peter James is our representative in London’. Again, it seems to be plausible to assume that ‘Peter James’ is used with the mentioned kind of intention. 28 Like in the following example: Lance Armstrong, the guy who won the Tour the France seven times,is a bold liar. 29 There are people that are talking about the famous philosopher John Perry. Someone overhears the conversation and says: Who is John Perry? 30 Cases of this kind that concern the introduction of a name are presented and discussed in: Sainsbury (2005, pp. 107–122). 31 Thanks to my colleagues at King’s College London, Peter Ridley, Peter Sutton and Mark Textor, for pressing me on this point. 32 I am also sympathetic to the view that descriptive acts of identification can have different forms. The use of a descriptive apposition might be sufficient in a certain contextual setting for this purpose, but a speaker might also determine the reference of a name in a descriptive way by means of a so-called individual concept. It is plausible to assume that we store information about different individuals with whom we interact in a direct or indirect way. We might call these stores of information individual concepts. There are now different ways to determine the referent of a name by means of an individual concept: The original causal source, the dominant causal source or the object which satisfies most of the stored information can be determined as the referent of a name by means of a specific descriptive act of identification. 33 Parasitic acts of identification are acts where a speaker identifies the referent of his use of a name by means of exploiting some preceding act of reference by himself or other speakers. In general, these are acts of identification where the reference is ‘‘borrowed’’ from some preexisting use of a name. 123 Indexical view on proper names 129

The central question now is whether our pluralism can be combined with an ambiguity view and an indexical view or whether there are any reasons why this doctrine fits better with one of these different views. Or more specifically: what is the exact role of the mentioned three different mechanisms? Is it really necessary to conceive of them as reference fixing mechanisms or could we also conceive them as disambiguating mechanisms? Acts of disambiguation are acts that make a specific meaning of an object salient relative to a certain context of use. Our distinguished acts of identification are acts that identify individual objects. So these can only be regarded as acts of disambiguation if the meaning of a proper name is identical to its referent. But such an identification is not plausible. Firstly, it is not necessary to know the referent of a specific name to be a competent user of a name. Secondly, there seem to be meaningful names without a referent. The best examples are names that were introduced by means of a descriptive act of naming and used under the false assumption that this act of naming was successful. Such names can nevertheless be used in meaningful sentences that might be true or false.34 Against this background, it is more plausible to conceive of our mechanisms as reference fixing mechanisms and exploit an analogy between proper names and demonstratives. It is common to distinguish automatic from non-automatic indexical expressions. A paradigm example of an automatic indexical is the expression ‘I’.35 This expression has an automatic character; the determination of the referent of this expression relative to a context of use does not require that the user of it has specific intentions or performs specific additional acts. If ‘I’ is used relative to a certain context of use, it refers automatically to the user of ‘I’. In the case of non-automatic indexicals, on the other hand, the determination of the referent relative to a context of use requires that the user of such an expression has specific intentions or performs certain additional acts. A paradigm example for a non-automatic indexical are bare-boned demonstratives like ‘this’ and ‘that’. The utteranceofsuchanexpressionaspartof a complete speech act is not sufficient to determine its referent, a specific additional action or intention is required for this purpose. In the case of bare-boned demonstratives a demonstrative act of identification of an object is required to determine the referent of such an expression. Against this background, the following prima facie specification of the non-automatic indexical character of ‘that’ seems plausible: The referent of the bare-boned demonstrative ‘that’ relative to a context of use c is the object that is identified demonstratively in c by the utterer of c. If we apply this distinction in the light of pluralism to proper names, it seems to be suggestive to conceive of proper names in analogy to demonstratives as non- automatic indexical expressions. Prima facie there seem to be two main differences between bare-boned demonstratives and proper names. Firstly, proper names seem to have a more tolerant non-automatic character than bare-boned demonstratives, because there are different legitimate mechanisms to determine the referent of a name relative to a context of use. Secondly, there seems to be an additional constraining factor in play in connection with the determination of reference. In this

34 The name ‘Vulcan’ is a standard example of this sort. C.f.: Sainsbury (2005, pp. 87–90). 35 Some complications are provided for such a conception by certain recorded or written utterances that contain the expression ‘I’. C.f.: Predelli (2005, p. 48). 123 130 D. Rami respect names seem to resemble complex demonstratives like ‘that Alfred’ and third person personal pronouns like ‘he’ more than bare-boned demonstratives. Prima facie it seems to be plausible to conceive the character of complex demonstrative of the form ‘that F’ in the following way: The referent of a complex demonstrative ‘that F’ relative to a context of use c is the object that is identified demonstratively in c by the utterer of c and that is F in c. The additional constraining factor is expressed by the additional descriptive information provided by means of the expression ‘F’. In the case of proper names, the additional constraining factor seems to be the name-bearer relation. If we now exploit the proposed analogy, we can formulate the character of a proper name as first approximation in the following way: The referent of a proper name ‘N’ relative to a context of use c is the object that is identified demonstratively, descriptively or parasitically in c by the utterer of c and that is a bearer of ‘N’ in c. Therefore, I think that pluralism about reference determination in combination with the observation that it seems to be possible to determine the referent of a proper name in a demonstrative way, provides suggestive reasons to conceive of proper names as a distinctive kind of non-automatic indexical expressions that resemble complex demonstratives and third person personal pronouns at least in certain respects.

3 Against the two most popular formal versions of the indexical view on names

Before we explore the proposed possibility of developing an indexical view on names in more detail, let me demonstrate why I think that the two most popular versions of formal indexical views on names that can be found in the relevant literature are not viable alternatives to the proposed approach. In this context, I will also show why the analogy between names and complex demonstratives or names and third person pronouns should be taken too seriously. Furthermore, I will argue against the view that the notion of salience or prominence can be used to capture different adequate reference determining mechanisms in the case of names in a bundled and uniform way.

3.1 Against formal variable indexical views on names

The main motivation for a formal variable indexical view on names is provided by a certain examples of apparent anaphoric uses of proper names. (I use the expression ‘anaphoric use’ in a very broad sense such that it covers proforms of laziness, prototypical bound anaphora, discursive anaphora, and donkey anaphora.36)Itisa well-known datum that third person personal pronouns and complex demonstratives can be used in a referential and an anaphoric way. There also seem to be anaphoric uses of proper names in addition to the more familiar referential uses. Prima facie the following two discourse fragments seem to have readings according to which the

36 C.f.: King (2012, pp. 367–369). 123 Indexical view on proper names 131 name ‘Mary’ is used in an anaphoric way; namely in (3) as a discursive anaphora and in (4) as a donkey anaphora: (3) A Mary and a Paul joined the Diogenes Club yesterday. Mary is a very nice person.37 (4) If a boy loves a girl with the name ‘Mary’, Mary ends up being admired by him.38 There are people who have used examples like (3) to argue for a view that is called variablism and that holds that names can be represented by individual variables in a formal semantic framework. According to this view, the use of ‘Mary’ in (3) is an example of a bound use of such a name-variable.39 Other people have used examples like (4) to argue for the view that a proper name like ‘Mary’ can be represented by a complex descriptive phrase like ‘(ix)(Mary(x) ^ x = y)’ that contains the predicate ‘Mary’, the free variable ‘y’ and such a phrase is interpreted on the basis of a Fregean40 theory of the description- operator ‘(ix)’.41 According to this view, the use of ‘Mary’ in (2) can be interpreted as a use of ‘(ix) (Mary(x) ^ x = y)’, where the free variable ‘y’ is bound by a universal quantifier.42 The first view is based on an analogy between proper names and third person pronouns.43 The second view is inspired by an analogy between proper names and complex demonstratives.44 Do examples like (3) and (4) provide good motivations for the proposed variable formal indexical views on names? I don’t think so. Let me argue for this claim in detail. Firstly, there is a significant difference between third person pronouns, on the one hand, and complex demonstratives and proper names, on the other hand. Third person pronouns have prototypical bound anaphoric uses. These are uses where the pronoun is in the scope of a quantifier expression and is bound by this quantifier.45 Third person pronouns also have deviant bound anaphoric uses, where the pronoun

37 A similar example is provided in: Burge (1973, pp. 435–436), Geurts (1997, p. 321) and Cumming (2008, p. 526, 535). 38 Similar examples are provided in: Geurts (1997, p. 321). 39 According to this view, (3) has the logical form: AxAy(Mary (x) ^ Paul (y) ^ Joined the Diogenes Club yesterday (x) ^ Joined the Diogenes Club yesterday (y) ^ Nice person (x)). C.f.: Cumming (2008, pp. 535–536). 40 A Fregean theory of definite descriptions holds that definite descriptions are complex singular terms that can either refer to a single object or to no object at all. An atomic sentences that contains a definite description that does not refer to a (single) object is according to this view neither true nor false. C.f.: Hawthorne and Manley (2012, pp. 181–184). 41 C.f.: Elbourne (2005, pp. 97, 169–173, 180–181). 42 According to this view, (4) has the logical form: VxVy(((Boy (x) ^ (Girl (y) ^ Mary (y))) ^ Love (x,y)) ? Admire (x, (iz)(Mary(z) ^ z = y))). C.f.: Burge (1973, p. 433), Elbourne (2005, pp. 97, 169–173, 180–181). 43 C.f.: Cumming (2008, pp. 525–526). 44 C.f.: Burge (1973, p. 432, 435–436). 45 An example of this sort is: At least one man thinks that he has lost someone. 123 132 D. Rami seems to be outside of the scope of certain quantifier and nevertheless seems to be bound by this quantifier.46 Discursive or donkey anaphora are examples of this sort.47 If names have anaphoric uses at all, they share with complex demonstratives the feature that they only have deviant bound anaphoric uses. This shows that variablism is in danger of overgeneralizing the possibility of anaphoric uses of names. It needs to provide a rationale why certain bound uses of name-variables are appropriate and others are not. It seems to be difficult to provide such a rationale on the basis of the proposed formal representations of proper names. Secondly, the view is also not restrictive enough concerning so-called unbound uses of proper names. In a formal theory, we can assign any arbitrary object via an assignment function to an unbound individual variable. Intuitively, we cannot use any name in a correct and felicitous way to refer to any object, but prima facie only to the bearers of a name. Therefore, an additional restriction of the assignment function is needed. But such an adaption would lead to an ambiguity, because the semantics of names relative to their unbound uses would differ from the semantics relative to their bound uses. In the case of unbound uses the values of a name- variable would then always be restricted to the bearers of a certain name relative to the domain of discourse of the actual world. In the case of bound uses the values of a name-variable would then be restricted to the bearers of a certain name relative to the domain of discourse of the quantifier that binds such a variable. And these domains can differ between a quantifier that is modally unembedded and a quantifier that is modally embedded.48 Thirdly, there seems to be a significant difference between examples like (3) and examples like (4). In the case of examples like (3) it can reasonably be doubted that the expression ‘Mary’, whether treated as a name or not, has a bound anaphoric use at all. There seems to be a straight forward pragmatic explanation of why it may appear that ‘Mary’ is used in an anaphoric way. The expression ‘A Mary’ is used in (3) as a specific indefinite. It is plausible to assume that indefinites can be used to refer to single objects according to their specific use.49 We can use ‘A Mary’ according to its specific use and the name ‘Mary’ in the very same context to refer to the very same object. Such an interpretation gives the most obvious interpretation of (3). Furthermore, it can be doubted that a correct semantic analysis of specific indefinites is provided by an unrestricted existential quantifier.50 If we substitute ‘A Mary’ in (3) for an expression that has an existential reading only like ‘At least one Mary’, the use of the name ‘Mary’ in (3) cannot be accommodated in a plausible way and its use appears to be odd. The truth of the two sentences that (3) contains depends on the fact that a specific individual named ‘Mary’ has the ascribed properties, but according to the logical form that Cumming assigns to (3), the truth of this form does not depend on any specific or single individual. It also has to be

46 An example of this sort is: Every farmer who owns a donkey, beats it. 47 C.f.: King (2012, pp. 368–369). 48 The view on names proposed by Elbourne has to face an analogous problem that is discussed below. 49 C.f. Hawthorne and Manley (2012, pp. 93–94, 99–100). 50 C.f. Hawthorne and Manley (2012, pp. 122–136). 123 Indexical view on proper names 133 noticed that the two sentences that (3) contains have intuitively independent truth- values. It is intuitively possible that the first sentence is true while the second is false. But we cannot account for this data on the basis of the proposed analysis. Therefore, examples like (3) do not seem to provide good evidence for variablism. Fourthly, the really challenging apparent anaphoric uses of proper names that are provided by examples like (4) are relatively rare and restricted to very specific sentential contexts.51 But it seems to be methodologically questionable, to make use of such a rare and special use of an expression—as Elbourne does—to motivate a general semantic analysis of proper names. Fifthly, the semantic analysis proposed by Elbourne is problematic if we apply it to non-anaphoric, referential uses of a proper name. A phrase like ‘(ix)(Mary(x) ^ x = y)’ is only a weakly rigid designator.52 It only refers to the very same object in respect to those possible worlds relative to which this object is a bearer of the name ‘Mary’.53 But intuitively ‘Mary’ refers to the very same object in respect to every possible world relative to which this objects exists. Hence, we cannot account for the true reading of a sentence like ‘It is possible that Mary is not identical to any bearer of ‘Mary’’ according to the proposed analysis. Therefore, we need to rigidify the expression ‘(ix)(Mary(x) ^ x = y)’ for this purpose in some way. Either by using an actuality- or a dthat-operator. But this kind of rigidity is not welcome, if we use an expression like ‘(ix)(Mary(x) ^ x = y)’ in a bound way and in a context like (4). In such cases, a non-rigid interpretation of ‘Mary(x)’ is desired. And it is even required if we use (4) within the scope of a modal operator like ‘It is possible that’ and want to account for the correct truth-conditions of such a sentence. Therefore, Elbourne’s analysis cannot provide a uniform analysis of so-called anaphoric and referential uses of a proper name that he aims to provide. If he wants to makes use of the given analysis of (4), he must accept that names have different semantic interpretations relative to so-called anaphoric and referential uses. Sixthly, as we have already noticed, names seem to have more in common with complex demonstratives than with personal pronouns. In the light of this similarity, some people have claimed that a proper name like ‘Alfred’ should be treated as semantically equivalent to a complex demonstrative like ‘This Alfred’ or ‘That Alfred’.54 There is one observation that seems to provide some evidence in favor of this view. We can substitute a proper name ‘N’ relative to every apparent deviant anaphoric use with the corresponding complex demonstrative ‘this N’ or ‘that N’ salva veritate, and without changing the content of such a claim. But things are different, if we focus on non-anaphoric, referential uses of names. In these cases, it is not in general possible to substitute a name salva veritate for a complex

51 There are also sentences that have a similar structure to (4), but intuitively do not have the proposed anaphoric reading: Every man, who loves a girl named ‘Mary’, will invite Mary for dinner. For most people this sentence only has a reading relative to which ‘Mary’ refers to a single person. C.f.: Elbourne (2005, p. 180). 52 C.f.: Hawthorne and Manley (2012, p. 11). 53 In the context of our examples the expression ‘is a Mary ‘and ‘is a bearer of the name ‘Mary’’ can be conceived as semantically equivalent. 54 C.f.: Burge (1973) and Sawyer (2010). 123 134 D. Rami demonstrative ‘this N’ or ‘that N’. There are significant differences between these two kinds of expressions: The determination of the reference of a complex demonstrative of the form ‘this F’ or ‘that F’ requires an act of demonstrative identification. Such a requirement does not exist in the case of names.55 Furthermore, a sentence like ‘That Alfred is a nice guy, but the other one isn’t’ is perfectly meaningful, but if we substitute ‘That Alfred’ with the name ‘Alfred’ the result is an odd sentence.56 These differences also seem to confirm the thesis that if names are indexicals, they are indexicals of a distinctive kind. Our six observations suggest being skeptical whether examples like (3) or (4) provide a good motivation for a formal variable indexical view on names. Only examples like (4) seem to provide at all a challenge for a formal constant indexical view. Some of our observations suggest how this challenge can be met without a commitment to a formal variable indexical view on names. We could claim that the proposed anaphoric use of the name ‘Mary’ in (4) is not a genuine use of a proper name. (4) has a reading relative to which ‘Mary’ is used as a name, but in this case the expression is used in a referential and non-anaphoric way. Apparent anaphoric uses of a name like ‘Mary’ are in fact elliptical and anaphoric uses of the corresponding complex demonstrative ‘this Mary’ or ‘that Mary’. This suggestion can be vindicated by an analogy with certain other apparent uses of proper names. In a sentence like ‘Hilary can be a man or woman’ the expression ‘Hilary’ can have two different uses: According to one use, it is used as name that refers to a specific individual. According to a second use, it is used as an elliptical version of the unspecific indefinite ‘A Hilary’.57 A similar explanation can be given in case of our rare and apparent anaphoric uses of proper names in example sentences like (4). On this basis, we can capture the data provided by examples like (4), but we thereby avoid the outlined pitfalls of developing an adequate version of a formal variable indexical view.

3.2 Against salience-based formal constant indexical views on names

There are at least three different formal constant indexical views based on KLD that have been proposed in the relevant literature. According to KLD, every expression has an extension that is relativized to two different kinds of parameters: contexts of use and circumstances of evaluation. Following Kaplan, we consider circumstances of evaluation as ordered pairs consisting of a possible world and a time parameter. I use the abbreviation ‘[e] c, \w, t[’ for ‘the extension of ‘e’ relative to the context of use c and the circumstance of evaluation consisting of the possible world w and the time t’. We can specify the conditions u that determine the extensions of

55 C.f.: Cohen (1980, pp. 149–150) and Bach (1993, p. 140). 56 C.f.: King (2006, pp. 148–149). 57 The following well-known example given in Bach (1993, pp. 146–147) has two similar readings: If presidents were elected by alphabetical order, Aaron Aardvark might have been president. Firstly, there is a reading of this sentence relative to which ‘Aaron Aardvark’ is used to refer to a specific person. Secondly, there is a reading of this sentence relative to which ‘Aaron Aardvark’ is used as an elliptic version of the indefinite expression ‘an Aaron Aardvark’. 123 Indexical view on proper names 135

58 expressions in KLD by certain equations of the general form ‘[e] c, \w, t[ = u’. From such statements one can also read off the character and content of an expression in KLD, because the following additional equation holds: [e] c, \w, 59 t[ = char(e)(c)(w,t). Against this background, we can now capture the indexical semantics of a name like ‘Alfred’ according to the three mentioned indexical views in the following way:

(IC1) [Alfred]c,\w, t[ the object of the originating event of the most salient name- using practice that concerns the name ‘Alfred’ relative to 60 cw. (IC2) [Alfred]c,\w, t[ the object of the most salient naming-convention relative to 61 cw that concerns the name ‘Alfred’. (IC3) [Alfred]c,\w, t[ the object of the most salient dubbing in force relative to cw that concerns the name ‘Alfred’.62 There are obvious similarities and differences between these three views. Firstly, these are three at least superficially different variations of a specific kind of formal constant indexical view. All three theories can be generated from the general schema ‘The object of the most salient F relative to cw that concerns the name ‘N’’, where ‘F’ is placeholder for an expression that denotes some common factor about the use of a name ‘N’ that relates ‘N’ to its bearers. The three accounts use at least superficially different expressions to classify these common relating factors. Secondly, all three views make use of the notion of salience (or prominence)to provide a functional correspondence between a name and a bearer of this name relative to a certain context of use. Before we can evaluate these three views, we have to clarify the contents of the three listed reference fixing conditions. A name-using practice consists of individual uses of a specific proper name that are in a specific way connected with each other. According to the orthodox conception of name-using practices or so-called chains of communication, introduced by Kripke, each name-using practice has an originating event that assigns a specific name to a specific object.63 Prototypically, these acts are acts of naming. The first explicit use of such a name is parasitic on the originating act: It is used with the intention to refer to the named object. Subsequent uses of a name are parasitic on their preceding uses. According to this view, the referent of an individual use of a name is the object of the originating act to which this use is connected via a chain of uses of this name that constitute a certain name- using practice. Prototypically, this kind of view about reference determination is

58 C.f.: Predelli (2012, p. 548; 2013, pp. 6–7) and Zimmermann (2012, p. 2361). 59 char(e) = the character of the expression ‘e’, char(e)(c) = the content of ‘e’ with respect to c. C.f.: Predelli (2013, p. 13). 60 C.f.: Zimmermann and Lerner (1991, pp. 353–355). 61 C.f.: Recanati (1993, pp. 138–143). 62 C.f.: Pelczar and Rainsbury (1998, pp. 294–298), Pelczar (2001, p. 138) and Tiedke (2011, pp. 715–718). See also: Matushansky (2008, pp. 591–595) for a cross-over of the views given by Recanati (1993) and Pelczar and Rainsbury (1998). 63 C.f.: Kripke (1980, pp. 90–97), Zimmermann and Lerner (1991) and Sainsbury (2005, pp. 106–124). 123 136 D. Rami combined with an ambiguity view on names and therefore a fine-grained individuation of names that holds that there are literally no two different objects that share the same name. As we have seen, one motivation for an indexical view is the rejection of such a fine-grained individuation of names. The three listed standard versions of a formal constant indexical view respect this motivation and are committed to a more coarse-grained individuation of names that allows that a single name can have more than one bearer. But this combination of a coarse-grained individuation of names and a Kripkean determination of reference leads to a specific problem for the view (IC1). Such a view has the consequence that there can be different name-using practices that involve the very same name. Therefore, fission- and fusion-cases concerning name-using practices are possible and these lead to trouble. There are fission-cases where one name belongs to a name-using practice that has a single origin relative to which a name is assigned to more than one object. Single acts of naming that assign one name to different objects are possible.64 But that means that there can be originating events of name-using practices with more than one object or target. Furthermore, it is also possible that a single name-using practice has more than one origin, because such a practice might be the product of a fusion of two name-using practices with two different origins with different or identical objects.65 This shows that in certain cases the salience of a specific name- using practice might not provide a sufficient condition for determination of the reference of a proper name relative to a specific context of use. The view (IC3) shares this problem with (IC1), but this view also has an additional problem if dubbings in force are conceived of in a certain way. The notion of a dubbing in force can be understood in a narrow and in a wide sense. In the narrow sense, a dubbing is identical to a single speech act of naming. In the wide sense, the notion of dubbing describes an act or series of acts that has the same consequence as a single act of naming, namely that a certain object becomes the bearer of a certain name. A dubbing D concerning a name ‘N’ is in force relative to a time t if ‘N’ still has the bearer or the bearers at t that were assigned to ‘N’ by D and therefore no other act was performed between the time of the performance of D and t that was able to change the naming status of ‘N’. Pelzcar and Rainsbury proposed to understand the notion of dubbing in the narrow sense.66 But that is

64 C.f.: Textor (2010, p. 112). A variation of an example suggested by Hans Kamp is the following: The tyrant utters: I hereby name every male baby that was born in my empire yesterday Vladimir. 65 An example of a name with two different origins that link this name to one and the same object is the following: By coincidence two different tribes, which have no contact with each other, name the very same mountain ‘Ateb’. The name is used by both tribes independently for a certain while, but then it happens that the two tribes get in contact and a war breaks out. One of the tribe wins and the remaining people merge to one tribe. Against this background, the two existing name-using practices concerning the name ‘Ateb’ merge to one practice. An example of a name with two different origins with two different objects is the following: There are twins that live in two different villages. The inhabitants of both villages are not aware of the existence of the other twin. One village introduces and establishes the use of the name ‘Goldilocks’ for one twin. The other village uses the same name for the other twin. At a certain time one of the villages is destroyed completely in a war and one of the twins dies unnoticed during this war. The remaining inhabitants of the destroyed village become inhabitants of the other village and the two name-using practices concerning the name ‘Goldilocks’ merge to one practice with one referent. 66 C.f.: Pelczar and Rainsbury (1998, p. 254). 123 Indexical view on proper names 137 problematic, because not every name is introduced by means of an act of naming. Probably, most names are introduced in a more implicit way. But that is only an additional problem. Whether we understand dubbings in a wide or narrow sense, such a view faces a similar problem as (IC1). A single act of naming can make it the case that a certain name receives more than one bearer. Therefore, the salience of a specific dubbing in force might not be a sufficient condition for the determination of the referent of a proper name relative to a context of use. A naming-convention is a convention that links names with their bearers. In agreement with Perry, Recanati holds the view that there might be different individual naming-conventions that link the very same name to different bearers of this name. On this basis, the indexical view based on (IC2) does not face the mentioned problem of (IC1) and (IC3) concerning the insufficiency of reference determination. There is a dispute between Perry and Recanati whether these conventions are semantic or non-semantic conventions.67 An indexical view like (IC2) is committed to the view that these conventions are non-semantic, because otherwise it would collapse into a specific variant of the ambiguity view. These conventions can be established by acts of naming or acts with the same institutional consequences as acts of naming. Against this background, it makes sense to call such conventions non-semantic. But on the other hand, these non-semantic conventions could be constitutive for semantic naming-conventions. Therefore, a defender of the view (IC2) must not only reject the thesis that those naming- conventions that are constituted by certain institutional acts are semantic, he must also hold that these conventions do not constitute specific semantic naming- conventions that might be identified with the meanings of a proper name. And it is not clear whether there are any good reasons to resist such a kind of view if these conventions really constitute the name-bearer relation between names and objects. Therefore, the indexical view based on (IC2) seems to be in danger to collapse into a certain version of the ambiguity view.68 Apart from these specific problems of the three mentioned views, there are also certain general problems that our three salience-based views share. Firstly, all these accounts use the notion of salience in a rather peculiar way. They do not apply the notion of salience to objects that are bearers of a certain name and therefore potential referents, they apply it to certain instances of relations that relate names and their bearers. It might be doubted whether on this basis our ordinary acts of the determination of a referent of name relative to a context of use can be captured in an adequate way. A specific F can only become the most salient F relative to a certain

67 C.f.: Recanati (1993, pp. 138–140) and Perry (2001, pp. 102–113). 68 I think, Perry is right if he conceives naming-conventions as permissive conventions. C.f.: Perry (2001, pp. 103–105, 109) and Korta und Perry (2011, p. 75). If a certain object x is the bearer of a name ‘N’, then it is possible to use the name ‘N’ to refer to the object x in felicitous way. Naming-conventions are on this basis certain norms that regulate the felicitous use of specific names. But it might be doubted whether instances of such a convention or norm can be exploited either on the basis of the ambiguity view or the indexical view to determine the referent of a single proper name. This explanatory use of these conventions does not make sense to me, because I do not see that there is a reference determining link that can be provided by such a convention. I think, they can at most be used to constrain adequate referents of a proper name. 123 138 D. Rami context of use if there is a certain mechanism that raises the salience of F to the maximum. Most ordinary competent users of a proper name aren’t aware of the existence of name-using practices, naming-conventions and dubbings in general. It is unclear which kind of mechanisms ordinary speakers may use to raise the salience of abstract entities like name-using practices or naming-conventions. It is even more mysterious how concrete events in the past like dubbings can be raised to salience. It seems to be more plausible to hold that the only kind of salience that matters in the case of the determination of the referent of a name is the salience of a certain bearer of a name. Therefore, a more plausible variant of the salience-based approach would be a variant that identifies the referent of a proper name relative to a specific context of use with the most salient bearer of this name relative to this context of use. Such a variant would also avoid the mentioned individual problems of the three mentioned views. Secondly, it seems to be necessary to restrict the mechanisms that determine the salience of a bearer of a name or instances of a certain relation that connects names and their bearers in a certain way. Intuitively, the user of a proper name has the control to which object he refers by means of his use of a name. Even in the case of a parasitic use of a name, such a use is controlled by the user of this name. He can decide to use the name in a parasitic way and he can determine which pre-existing use of a name he aims to exploit. This important controlling aspect is missing in the case of our distinguished variants of the indexical view on names. According to them, the most salient bearer of a name relative to a context of a name is conceived of as the referent of this name, whether the salience of this object depends on actions of the user of this name or not. That is implausible. Thirdly, and most damagingly for salience-based approaches concerning the determination of reference of name: The salience of a certain bearer of a name, or the salience of an instance of a certain relation that connects names and their bearers, is neither necessary nor sufficient for the determination of reference of a name. It does not seem to be the case that the referent of a name relative to a context of use needs in any plausible sense to be salient. I can use a name in a context, where no bearer of this name is salient at all, and refer to a specific object in a felicitous way, although my audience might not immediately be able to figure out to which object I am referring, because no object is raised to the maximum of salience relative to such a context of use by me or anyone else.69 Furthermore, there might be a certain bearer of a name that is made most salient by a user of this name or someone else in a context of use, but it is nevertheless possible that the user of this name refers to a completely different object in this context. One may think of a situation where a user of a name takes his audience on a ride concerning the referent of this name, before he reveals the real referent. Or a user might, for example, refer to different objects by different uses of the same name in the very same context and only one of these objects might be the object whose salience is raised to the relative maximum by the user of this name. Salience-based indexical accounts seem to confuse facts about the determination of the reference of a name with evidence for such facts that is available to an interpreter.

69 This objection was pointed out to me by an anonymous reviewer. 123 Indexical view on proper names 139

These observations show, in my opinion, that salience-based indexical views on names provide a fundamentally wrong view of the determination of the reference of names relative to a specific context of use. Let us investigate in the last section of this paper whether an approach based on the mentioned variant of pluralism about the determination of reference fairs better.

4 The use-conditional version of the formal constant indexical view on names

At the end of the second section of this paper we briefly introduced an indexical view on names that was based on pluralism about the determination of reference of a name and a specific development of the suggestive analogy between proper names and complex demonstratives. Based on KLD we can formulate a first and provisional version of this approach in the following way:

(ICX) [Alfred]c,\w, t[ the object that is identified demonstratively, descriptively or parasitically in cw by ca and that is a bearer of ‘Alfred’ at 70 ct. In this section, I will focus on three different kinds of problems that concern each of the mentioned variants of the formal constant indexical view on names. These are: (i) the problem of multiple occurrences of the same proper name with different referents, (ii) the problem of an adequate distinction between felicitous und infelicitous uses of proper names, (iii) the problem of unwelcome logical consequences of sentences that contain proper names. In the following, I will only discuss these problems in application to (ICX), because my main aim in this section is to develop this view against the background of these problems into a use- conditional version of a formal constant indexical view on names. 4.1 The problem of multiple occurrences of the same proper name with different referents

Intuitively, different occurrences of a single name can be used in a single sentence or utterance to refer to different things relative to a single context of use. A nice example of this kind is provided by the following sentence71: (5) Alfred could not hear what Alfred was saying. This sentence can be used relative to a single context of use in such a way that its truth depends on the reference of the two occurrences of ‘Alfred’ in (5) to two different people. We cannot account for this kind of data on the basis of the current formulation of (ICX). If we interpret (5) relative to a single context of use, we have to assign on the basis of (ICX) to both occurrences of ‘Alfred’ in (5) the very same extension. Furthermore, the current formulation of (ICX) is problematic, because it

70 ‘cw’ refers to the possible world of the context c; ‘ca’ refers to the agent of the context c; ‘ct’ refers to the time of the context c. 71 A similar example is used in Korta and Perry (2011, p. 75). 123 140 D. Rami is not compatible with the possibility that two different objects are determined relative to a single context of use as referents of two different names by two different acts of identification of the same kind. Against this background, it is, for example, not possible to account for the correct truth-conditions of an utterance like ‘Alfred is not identical to Peter’ if the referent of both names is identified by the same kind of act of identification. There would be a relatively easy fix for this second kind of problem if we substitute the reference fixing-conditions in (ICX) by the following: the bearer of ‘N’ that is identified demonstratively, descriptively or parasitically in cw by ca. But even against this background, the problem could be restated for two different kinds of singular terms, whose referents can be identified by the same kind of act of identification. Therefore, a uniform solution would be welcome that could solve both kinds of problems in all their variations. The best way to solve these two related problems in a uniform way seems to be provided by (a) a numeral indexing of different occurrences of a proper name in a sentence at the level of logical form in combination with (b) a relativisation of those kinds of acts of identification that are mentioned in (ICX) to occurrences of a proper name in a sentence.72 Therefore, we can reformulate (ICX) in the following way:

(ICX.1) [Alfredx]c,\w, t[ the object that is identified demonstratively, descriptively or parasitically in cw in respect to the occurrence x of ‘Alfred’ by ca and that is a bearer of ‘Alfred’ at ct. According to this proposal, names are syntactically represented at the level of logical formas ordered pairs consistingoftheordinary proper name ‘N’ anda numeral that marks the terminal nodes of a syntactic tree that represents the structure of a certain sentence. The idea of relativizing the acts of identification that are required for the determination of reference of certain referring expressions to the occurrences of these expressions in sentences also seems to be empirically adequate. Ordinary speakers are implicitly aware of the fact that the use of different occurrences of such expressions requires certain actions, intentions or decisions on the side of the user of such an expression. Nevertheless, there are different default expectations concern- ing multiple uses of different kinds of expressions that depend on the different, specific nature of their characters. In the case of demonstratives there is the default expectation that different occurrences of the same demonstrative in different subsequent sentences correspond with different demonstrative acts of identification that identify different objects.73 In the case of names there is the default expectation that if the first occurrence of a name in a discourse situation is identified in a certain way, subsequent uses of the same name in different subsequent sentences are identified in a parasitic way that relies on the identification of the first occurrence.74

72 I am following Predelli (2012, pp. 557–560; 2013, pp. 192–196), who uses a very similar strategy to solve the corresponding problem in the case of bare-boned demonstratives. 73 C.f.: Taylor (2003, pp. 6–8). This default expectation concerns a specific range of relative closeness: But the use of multiple occurrences of a single demonstrative expression as arguments of the identity predicate are exceptions. (Thanks to Christian Beyer for reminding me of this point). 74 C.f.: Dever (1998, §2.3.2.4.2). This default expectation depends on a specific closeness constraint: It concerns occurrences that are relatively close to each other, but not too close to each other. If, for 123 Indexical view on proper names 141

These default expectations can be undermined by contextual counterevidence. A use of (5) relative to a contextual setting where it is obvious that a speakers aims to refer to two different bearers of the name ‘Alfred’ is an example of this kind. The proposed solution of the problem of multiple occurrences of a name is compatible with the possibility that one and the same act of identification is performed in respect to different occurrences of a name and the possibility that different acts of identification in respect to different occurrences of a name might identify the very same object as referent.75,76

4.2 The problem of an adequate distinction between felicitous und infelicitous uses of proper names

There are different sorts of felicitous und infelicitous uses of a proper name. According to the constraints set by (ICX.1), we can distinguish three different cases of infelicitous uses of proper names. Firstly, a use of a name might be called incomplete if a name is used without any act of identification. Secondly, a use of a name might be called empty, if a name is used accompanied by an act of identification that fails to identify a single object. Thirdly, a use of a name might be called improper, if a name is used accompanied by an act of identification that identifies a single object that is not a bearer of this name relative to time of the use of this name. According to this proposal, every improper use is a semantically incorrect use of a name. A use of a name is immaculate according to (ICX.1), if a name is used accompanied by an act of identification that identifies a single object that is a bearer of this name relative to time of the use of this name. Every immaculate use of a name is a semantically correct use, but not every semantically correct use is an immaculate use. There are two notable sub-classes of empty uses of names. (i) There are accidentally empty uses of a name: these are uses where a speaker aims to refer with a name to a bearer of this name, but fails to identify any object. (ii) There are systematically empty uses of a name: these are uses of a name that can be traced back to an unsuccessful act of naming, where someone aims to name an object and failed without immediately recognizing his failure; or these uses can be fictional uses of a name, relative to which someone only pretends that a name refers to something. Therefore, in these systematically empty cases there are no objects that could be identified in a correct way.

Footnote 74 continued example, two occurrences of a single name are arguments of the very same predicate like in ‘Robert hates Robert’ or (5), then the default expectation is different. (Thanks to Peter Ridley for reminding me of this data). 75 It seems to be more difficult to apply such an approach to salience-based indexical views. Because it might be questioned that it is meaningful to relativize the salience of a certain object to an occurrence of a proper name in a sentence. 76 This conception is also compatible with so-called mid-sentence shifts of the context of use, but it is not committed to such shifts to account for the mentioned data. 123 142 D. Rami

Only incomplete uses are semantically incompetent uses. Certainly systemati- cally empty uses of names, but maybe also accidentally empty uses of names, should be conceived of as semantically correct uses.77 From a truth-conditional point of view there is no difference between incomplete and empty uses,78 but there clearly is a difference from a use-conditional point of view. A use of a name performed without an act of identification is an incorrect and incompetent use, but a use of a name performed with an act of identification that fails to identify a single object is semantically correct, but in a certain sense imperfect. Therefore, the first problem concerning the adequate distinction of felicitous und infelicitous uses of proper names consists in drawing a difference between semantically correct and semantically incorrect uses of a name without a referent. A solution to this problem seems to require the introduction of an additional, use-conditional layer of meaning that concerns the use of a proper name in argument position.79 Furthermore, there seem to be semantically correct uses of a name that are not captured, but excluded on the background of (ICX.1). In this sense (ICX.1) seems to be too restrictive and the class of improper uses according to (ICX.1) is too large. There are at least two kinds of semantically correct uses of proper names that are excluded by (ICX.1): Firstly, there is a specific sub-class of uses of proper names that refer to potential future bearers of these names. Secondly, there is a specific sub-class of uses of proper names that refer to past bearers of these names. These excluded semantically correct uses of a name lead to a second problem concerning the adequate distinction of felicitous und infelicitous uses of proper names. Before we will propose a solution to these two kinds of problems, let me outline the second problem in more detail by giving some examples of excluded semantically correct uses of a name. Let us at first introduce an example of a semantically correct referential use of a name that refers to potential future bearer of this name. There are different ways to make it the case that a certain object becomes the bearer of a certain name. The most prominent way to achieve this goal is provided by explicit acts of naming. But there are also more implicit ways to reach the same goal. One can start to use a name with the intention to establish a new name for a certain object and if one is lucky enough, a specific series of uses of a certain name may lead to the introduction of a new name for a certain object. Imagine a case where someone finds a cat and decides after a while to keep the cat. The new owner of this cat does not have a name for the cat. He is undecided which name he should give the cat. Someone asks the new owner ‘Where is the new cat?’. And he responds by uttering the following sentence80:

77 In this essay, I will leave the question open whether there are any good reasons to restrict the class of semantically correct empty uses of names to systematically empty uses. 78 In the case of atomic sentences both uses produce either false or neither true nor false sentences. 79 C.f.: Predelli (2012, p. 557). 80 This is an example borrowed from Ziff (1977, p. 321). It was pointed out to me by Mark Textor. 123 Indexical view on proper names 143

(6) Whiskers is in the kitchen. Such an utterance is perfectly acceptable if it is accompanied by the intention to establish a new name for the new cat and it is also true if the new cat is in fact in the kitchen. But this utterance need not to be conceived of as an implicit act of naming. There might be a dispute about the name of the new cat after the utterance of (6) and it may take a while until there is some agreement about the new name of the cat. And even if it turns out that the cat in the end receives another name than ‘Whiskers’, the initial utterance of (6) remains flawless and the use of the name ‘Whiskers’ remains a proper use of the name although the cat to which the utterer of (6) referred was not a bearer of this name relative to the time of the utterance. Therefore, there seem to be acceptable uses of a name that refer to potential future bearers of a name. On the other hand, a use of ‘Whiskers’ in a sentence like (6) would be improper if it is not accompanied by the mentioned intention or if the user of (6) either erroneously thinks that the cat is already a bearer of the name or if the user of (6) confuses the cat with a cat that in fact bears the name ‘Whiskers’. These would be examples of improper uses of a name based on a misconception or a misidentification.81 Let us now focus on semantically correct uses of names that refer to past bearers of these names. ‘Leningrad’ was once the name of St. Petersburg, but it is not the official name of this city anymore. A proper use of ‘Leningrad’ in a present tense sentence is not possible. An utterance of a sentence like the following seems to be unacceptable or odd: (7) ? Leningrad has more than hundred thousand inhabitants. But interestingly there are also semantically correct uses of the name ‘Leningrad’. It can be used in past tense sentences that refer to times relative to which ‘Leningrad’ was a name of the second largest city of Russia. The following sentence is an example of an acceptable and true sentence of this kind: (8) In 1944, Leningrad had more than hundred thousand inhabitants. Furthermore, there is a significant difference between the present name of the second largest city of Russia ‘St. Petersburg’ and its former name ‘Leningrad’. We can substitute the name ‘St. Petersburg’ for ‘Leningrad’ in (7) and (8) and we get two true and acceptable sentences (7*) and (8*) as result.82 Another interesting aspect of the phenomenon of reference to past bearers of a certain name concerns the use of the name ‘Madagascar’ that presently has a completely different bearer than it had a long time ago. Let us focus on the following example sentence that contains this name. (9) In the fourteenth century, Madagascar was part of the African mainland. This sentence has a default reading relative to which it is false. It cannot be used in a semantically correct way to refer to a certain part of the African mainland, although

81 See Kripke (1977, p. 263; 1980, p. 25, Fn.3) for a famous example of an improper use of a name based on misidentification. 82 C.f.: Saul (1997, p. 103) and Zimmermann (2005, p. 55–62). 123 144 D. Rami it once was a name for a certain part of the African mainland. Therefore, there seems to be some significant difference between the name-bearer change that concerns the name ‘Madagascar’ and the name-change that concerns the second- largest city of Russia. These different aspects of the phenomenon of reference to past bearers of a name need some explanation. Furthermore, the sentence (8) provides a challenging example for our proposed semantics of names based on (ICX.1), because we cannot account on this basis for the intuitively true reading of (8). A semantic solution to the problem of reference to past bearers has been proposed that postulates a specific ambiguity concerning proper names and distinguishes restricted from unrestricted names.83 An unrestricted name does not interact with any sort of binding operator and it refers to the very same object relative to every embedded or non-embedded use. A restricted name interacts with time-operators and if it is in the scope of a time operator it refers to a bearer of a name relative to the time that is specified by this operator. If we apply this solution to our account (ICX.1), then (ICX.1) can be used to specify the semantics for unrestricted names, and a slight modification of (ICX.1) that contains an altered name-bearer-condition that is not relativized to the world of the context of use,but to the time parameter of the circumstances of evaluation can be used to capture the semantics of restricted names. This solution treats ‘Leningrad’ as a restricted name and ‘St. Petersburg’ as an unrestricted name. Against this background, we can explain why the names ‘Leningrad’ and ‘St. Petersburg’ cannot be substituted salva veritate. But the solution has also some unwelcome consequences. Firstly, it solves a problem by the postulation of an ambiguity that seems to be ad hoc and without an independent motivation. Secondly, in application to the name ‘Madagascar’ it leads to the implausible result that this name is ambiguous and (9) has two different readings. Intuitively, there is no restricted name ‘Madagascar’ that is still in use. Therefore, we additionally need to explain, which changes that concern instances of the name-bearer relation lead to the creation of a restricted name and which do not. Thirdly, according to this solution the odd sentence (7) comes out as a perfectly acceptable sentence that is either false or neither true nor false, depending on which logical framework for names without a referent one uses. Fourthly, it would be desirable to have a solution strategy that we can also plausibly apply to the phenomenon of reference to potential future bearers of a name. An application of the same strategy to examples like (6) would lead to the postulation of another class of names—they might be called intermediate names—and another kind of ambiguity of names. But such a strategy seems to be even more implausible in application to these other cases. Is there a more moderate, plausible and uniform alternative solution? In my opinion, the oddness of (7) points into the direction that we should not locate the difference between ‘Leningrad’ and ‘St. Petersburg’ at the truth-conditional level of meaning, but on a use-conditional level. A comparison of (7) with (10) might help to clarify this point:

83 C.f.: Zimmermann (2005, pp. 56–59; 61). 123 Indexical view on proper names 145

(10) Alfred, you are my favourite roommate. Relative to its vocative use, a name like ‘Alfred’ has a use-conditional meaning, but no truth-conditional meaning. This use-conditional meaning can be specified in the following way: A vocative use of ‘Alfred’ relative to a context of use c is only then a semantically correct use if the user of ‘Alfred’ in c addresses a bearer of this name by means of his use of ‘Alfred’ in c.84 Therefore, if I use a sentence like (10) and address someone by the name ‘Alfred’ that is not a bearer of the name ‘Alfred’, I have used the name ‘Alfred’ in an incorrect way. Nevertheless, (10) can come out as true relative to such an improper use. There is no truth-conditional difference between (10) and ‘You are my favourite roommate’. The situation seems to be similar in the case of (7). A use of the name ‘Leningrad’ in a present tense sentence like (7) is a semantically incorrect use, nevertheless it also seems to be possible in this situation to make a true statement with the utterance of (7) from a purely truth- conditional point of view. Against this background, there is a certain use-conditional difference between ‘Leningrad’ and ‘St. Petersburg’, but no truth-conditional difference. A similar strategy can be used to distinguish the mentioned semantically correct uses of names for potential future bearers from the improper uses of this kind. If the name ‘Whiskers’ is used in a sentence like (6) with the intention to introduce a new name for a certain object, then such a use of the name ‘Whiskers’ is semantically correct, in so far as it satisfies certain requirements that are specified by the use- conditional meaning of the name ‘Whiskers’ relative to the use of a name in argument position. But if, on the other hand, the name ‘Whiskers’ is used in a sentence like (6) with the intention to refer to an object of which the user of this name erroneously thinks that it already bears this name although it does not, then such a use is improper. Nevertheless, there does not seem to be a significant difference between these two kinds of cases from a purely truth-conditional point of view. We can make true or false statements relative to both situations with an utterance of the sentence (6).85 Such a use-conditional solution of both aspects of our second problem can nicely be combined with a use-conditional solution to our first mentioned problem. The introduction of an additional use-conditional layer of meaning seemed to be the only possible way to solve the latter problem, because from a truth-conditional point of view it was impossible to distinguish incomplete from empty uses of a proper name. Therefore, we receive the desired uniform and moderate solution by means of the postulation of an additional use-conditional meaning that specifies certain constraints for semantically correct uses of proper names.

84 C.f.: Predelli (2013, pp. 133–137). 85 It is important to notice that the situation is significantly different in the case of misidentification examples pointed out in Kripke (1977, p. 263; 1980, p. 25, Fn.3). In such cases there are two different relevant objects in play that might be conceived as referents of a proper name and only one of them can plausibly be conceived of as the semantic referent of a name. 123 146 D. Rami

Let me now focus on another problem that provides additional reasons for the introduction of such a use-conditional layer of meaning and the exportation of the name-bearer-condition that (ICX.1) contains to this use-conditional layer.

4.3 The problem of unwelcome logical consequences of sentences that contain proper names

Intuitively, neither the sentence (11) logically implies the sentence (12) nor the propositional content of (11) implies the propositional content of (12): (11) Alfred is a nice guy. (12) Alfred is a bearer of ‘Alfred’. Such inferences seem to be incorrect, because there are possible worlds in respect to which (11) or the propositional content of (11) is true, but (12) or the propositional content of (12) is false. Against the background of KLD, we can define different notions of entailment. Firstly, there is a notion of entailment that resembles our intuitive notion of sentential entailment: D entailss a iff for every circumstance of evaluation s and every context of use c: If D is true relative to s and c, then a is true relative to s and c. Secondly, there is a character-guaranteed notion of entailment that resembles our intuitive notion of content entailment: D entailsc a iff for every context of use c: If D is true relative to c,86 then a is true relative to c.87 If we apply the first of these two notions on the basis of (ICX.1) to (11) and (12), our intuitive verdicts are confirmed. But the situation is different if we apply the second character-guaranteed notion of entailment on the basis of (ICX.1) to our example sentences and aim to thereby capture our pretheoretical notion of propositional entailment. Against this background, (11) logically implies (12). Hence, (12) is an unwelcome logical consequence of (11). This is especially problematic, if we take into account that the sentences (6) and (8) according to the discussed problematic readings have similar logical consequences as (11)—if we assume (ICX.1). Furthermore, sentences like the following are contingent analytic truths according to KLD and (ICX.1): (13) If Alfred exists, then Alfred is a bearer of ‘Alfred’. This again seems to be unacceptable in the light of the mentioned semantically acceptable uses of (6) and (8). Therefore, there are good independent reasons to assume that the name-bearer condition that (ICX.1) contains should not be part of the truth-conditional meaning of a name. We can easily block these unwelcome logical consequences if the name-bearer condition that (ICX.1) contains is transferred to a use-conditional layer of meaning of a name.88

86 S is true relative to the context of use c iff [s]c, \ cw,ct [ = T. C.f.: Predelli (2013, pp. 7–8). 87 C.f.: Predelli (2013, pp. 13–14, 187). 88 C.f.: The same strategy is proposed for similar reasons concerning the analysis of bare-boned demonstratives in Predelli (2012, pp. 551–556; 2013, pp. 189–192). 123 Indexical view on proper names 147

The given reasons do not seem to be the only reasons that speak against the correctness of (ICX.1) from the perspective of logical entailment. Against the background of (ICX.1) a sentence like (11) also has the following character- guaranteed logical consequences89: (13) There is either some demonstrative, descriptive or parasitic act of identification. (14) Somebody is the utterer of ‘Alfred’. But intuitively the content of sentence like (11) does not have these logical consequences. This shows that an adequate solution to the problem of unwelcome logical entailments cannot just be solved by an exportation and modification of the constraining name-bearer condition, it also seems to be necessary to formulate the truth-conditional contribution of a proper name in a completely different way to block all unwelcome logical consequences. The central question against this background is now the following: can we find a plausible reformulation of the truth-conditional meaning of a proper name that allows us to transfer the reference fixing conditions that (ICX.1) contains to the use-conditional layer of meaning of a proper name?

4.4 Content, character and contextual constraint of proper names

What we need for this purpose 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 a proper name to the level of use-conditional meaning. The most natural formal solution to this problem makes use of a partial, context- and occurrence-relative assignment function in our meta-language. Such a function assigns to every occurrence of every proper name of our language relative to every context of use either a particular object of the domain of discourse or no object at all. There are no specific constraints concerning these assignments from a logical point of view. Therefore, it is possible to assign every object to every proper name relative to every context of use at the truth-conditional level. This assignment function maps ordered triples consisting of the possible world of the context of use, an integer that represents the occurrence of a name and the name itself to objects of the domain of discourse of KLD. I use the letter ‘N’ to represent this kind of function. We can now add this function to the standard parameters that formally represent a context of use to indicate thereby that the character of certain expressions is sensitive to this additional parameter. Against this background, we can now specify the character of the proper name ‘Alfred’ in the following way:

(ICX.2) [Alfredx]c,\w, t[ N(cw ,’Alfred’, x) This reformulation blocks the mentioned unwelcome logical consequences and nevertheless treats proper names as constant expressions with a constant content and

89 C.f.: Predelli (2013, p. 187) 123 148 D. Rami an indexical character. The fact that N is a partial function also allows to capture empty and incomplete uses of a name at the truth-conditional level. The easiest way to incorporate such names without an extension into the bivalent truth-conditional machinery of KLD is to treat atomic sentences that contain at least one name without an extension (following negative free logic90) as false. On the basis of this formal treatment of the truth-conditional meaning of names, we can now introduce the contextual constraint of a proper name as an additional use-conditional layer of meaning. The main function of this layer of meaning is to distinguish semantically correct from semantically incorrect uses of a proper name. One part of this explanatory role is to distinguish semantically correct from incorrect acts of determining the referent of a proper name. At the truth-conditional level it is possible that each proper name refers to each individual in our domain of discourse. In this sense, there is no difference in truth- conditional meaning between different proper names. The intuitive difference in meaning between different proper names is captured by the use-conditional meaning of a name. The contextual constraint of a proper name is the set of all contexts of use according to which this name is used in a proper or semantically correct way. This set of adequate contexts of use may differ from proper name to proper name. On the basis of the given outlined data about semantically correct (=empty or immaculate) and incorrect (=incomplete or improper) uses of proper names, we can now specify the contextual constraint91 for a name like ‘Alfred’ in the following way: (CC) c is a semantically correct context of use of the name ‘Alfred’ in respect to its occurrence x if an act of demonstrative, descriptive or parasitic identification is performed by ca in cw in respect to x whose unique target is, if any, identical to N(cw,’Alfred’, x) and this target, if it exists, satisfies at least one of the following additional conditions:

90 C.f.: Sainsbury (2005, pp. 64–75) and Rami (2013). 91 An anonymous reviewer raised the question how the layer of contextual constraint relates to pragmatic presuppositions. In Hawthorne and Manley (2012, pp. 230–231), for example, an analysis of names is very briefly sketched that holds that a sentence like ‘Alfred is a nice guy’ pragmatically presupposes contents that can explicitly be expressed by ‘Alfred is a bearer of ‘Alfred’’ and ‘Alfred exists’. They distinguish a gap-happy version of their approach, where a presupposition failure induces a truth-value gap on the truth-conditional level, from a gap-hostile version, where a presupposition failure has no influence on the truth-conditional layer of meaning, and a mixture of both approaches. If we aim to translate our use-conditional approach into the terminology of presupposition theory, the result would be a mixed approach; because presupposition failures in the case of those presupposed contents that are expressible by sentences of the form ‘N is a bearer of ‘N’’ do not affect the truth-value of a sentence that triggers this presupposition; while in the case of presupposed contents that are expressible by sentences of the form ‘N exists’ do affect the truth-value of a sentence that triggers this presupposition. Nevertheless, it is not clear whether our use-conditional account is only a presuppositional account in disguise. According to the use-conditional account, people might be completely ignorant about the contents that are triggered by names according to the presuppositional account. They might not possess the concept of a bearer of a name and might nevertheless be in the position to use a name in a meaningful way if their uses objectively satisfies the requirements set by the contextual constraint. The outlined account does not set any requirements for shared contents (=common ground) between speaker and hearer to explain meaningful, felicitous and truth-apt uses of proper names. 123 Indexical view on proper names 149

(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 ca 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 92 object to another, and the occurrence of x of ‘Alfred’ in cw is in a sentence whose truth-value evaluation relative to ct depends on a time s that is such that this target was a bearer of ‘Alfred’ at s in cw. The sub-conditions (CC)(a)–(c) are formulated in such a way that on their basis the three distinguished classes of semantically correct referential uses of a proper name relative to any context of use are captured: (a) all uses that refer to actual and presented bearers of a name, (b) specific uses that refer to potential future bearers and (c) specific uses that refer to past bearers of a name. This completes my outline of the proposed use-conditional indexical conception of proper names that characterizes proper names as constant expressions with a rigid content, a specific indexical character and a specific contextual constraint.93

Acknowledgments I would like to thank the following people for helpful discussions and comments: Stephen Barker, Christian Beyer, Louis DeRosset, Stacie Friend, Daniel Gutzmann, Hans Kamp, Wilfried Keller, Ruth Kempson, Wilfried Meyer-Viol, Peter Ridley, Wolfgang Spohn, Peter Sutton, Mark Textor and Ede Zimmermann. I also profited from helpful and interesting comments of two anonymous reviewers. Very special thanks to Mark Textor, Peter Ridley and Christian Beyer for their help and support.

References

Bach, K. (1993). Thought and reference (2nd ed.). Oxford: Clarendon Press. Burge, T. (1973). Reference and proper names. The Journal of Philosophy, 70, 425–439. Burks, A. W. (1951). A theory of proper names. Philosophical Studies, 2, 36–45.

92 This second condition of (CC)(c) seems to distinguish the Leningrad-case from the Madagascar-case. But probably there is alternative and more systematic way to distinguish these two kinds of cases. I leave this question for future research. 93 An anonymous reviewer objected to the proposed formulation of the contextual constraint of proper names that it seems extremely ad hoc and that it leaves one without much sense of theoretical unification and explanation. Granted, the disjunctive nature of (CC) is undeniable: Firstly, (CC) specifies three completely different mechanisms of reference determination. Secondly, it captures three different cases of proper referential uses of names by means of different sub-conditions. I think, I have argued for the corresponding distinctions in a convincing way. Maybe, both lists are still incomplete. Nevertheless, there doesn’t seem to be a systematic way to provide a unification. Proper names aren’t instances of a natural kind, they are linguistic artefacts that are used for specific communicative purposes. The disjunctive nature of the contextual constraint only reflects the various purposes that names fulfil in natural language. In the light of the institutional nature of natural language it should not be a surprise that unification is not always possible. 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. The only kind of unification in the case of names that seems to be possible is at the truth-conditional level. At this level names have a uniform behaviour and they resemble in this respect the truth-conditional behaviour of demonstratives and other indexicals that are rigid designators. 123 150 D. Rami

Cohen, J. L. (1980). The Individuation of proper names. In Z. Van Straaten (Ed.), Philosophical subjects (pp. 140–163). Oxford: Oxford University Press. Cumming, S. (2008). Variablism. Philosophical Review, 117, 525–554. Dever, J. (1998). Variables. PhD Dissertation, University of California at Berkeley. Elbourne, P. (2005). Situations and individuals. Cambridge, MA: MIT Press. Geurts, B. (1997). Good news about the description theory of names. Journal of Semantics, 14, 319–348. Hawthorne, J., & Manley, D. (2012). The reference book. Oxford: Clarendon Press. Kaplan, D. (1978). On the logic of demonstratives. Journal of Philosophical Logic, 8, 81–98. Kaplan, D. (1989 [1977]). Demonstratives, In J. Almog, J. Perry, & H. Wettstein (Eds.), Themes from Kaplan. Oxford: Oxford University Press. Kaplan, D. (1990). Words. Aristotelian Society Supplement, 64, 93–119. King, J. C. (2006). Singular terms, reference, and methodology in semantics. Philosophical Issues, 16, 141–161. King, J. (2012). Anaphora. In G. Russell & D. G. Fara (Eds.), The Routledge companion to the philosophy of language (pp. 367–379). London: Routledge. Korta, K., & Perry, J. (2011). Critical pragmatics. Cambridge: Cambridge University Press. Kripke, S. (1977). Speaker’s reference and semantic reference. Midwest Studies in Philosophy, 2, 255–276. Kripke, S. (1980). Naming and necessity. Cambridge, MA: Harvard University Press. Matushansky, O. (2008). On the linguistic complexity of proper names. Linguistics and Philosophy, 21, 573–627. Pelczar, M. (2001). Names as tokens and names as tools. Synthese, 128, 133–155. Pelczar, M., & Rainsbury, J. (1998). The indexical character of names. Synthese, 114, 293–317. Perry, J. (2001). Reference and Reflexivity. Stanford: CSLI Publications. Predelli, S. (2005). Contexts. Oxford: Clarendon Press. Predelli, S. (2012). Bare-boned demonstratives. Journal of Philosophical Logic, 41, 547–562. Predelli, S. (2013). Meaning without truth. Oxford: Oxford University Press. Rami, D. (2013). Existence as a property of individuals. Erkenntnis, published online first. doi:10.1007/ s10670-013-9505-9. Recanati, F. (1993). Direct reference. Oxford: Blackwell. Sainsbury, R. M. (2005). Reference without referents. Oxford: Oxford University Press. Sainsbury, R. M. (2012). The same name. In D. Rami & M. Textor (Eds.), A special issue of Erkenntnis on proper names. Retrieved from https://webspace.utexas.edu/rms9/www/Publications/RMSSame Name.pdf. Accessed 1 Dec 2013 Saul, J. (1997). Substitution and simple sentences. Analysis, 57, 102–108. Sawyer, S. (2010). The modified predicate theory of proper names. In S. Sawyer (Ed.), New waves in philosophy of language (pp. 206–225). Palgrave, NY: MacMillan Press. Sommers, F. (1980). The logic of natural language. Oxford: Oxford University Press. Taylor, K. A. (2003). ‘‘What’s in a name’’, in his: Reference and the rational mind (pp. 1–32). Stanford, CA: CSLI Publications. Textor, M. (2010). Proper names and practices: On reference without referents. Philosophy and Phenomenological Research, 81, 105–118. Tiedke, H. (2011). Proper names and their fictional uses. Australasian Journal of Philosophy, 89, 707–726. Ziff, P. (1977). About proper names. Mind, 86, 319–332. Zimmermann, T. E. (2005). What’s in two names? Journal of Semantics, 22, 53–96. Zimmermann, T. E. (2012). Context dependence. In C. Maienborn, K. von Heusinger, & P. Portner (Eds.), Handbook of semantics (Vol. 3, pp. 2360–2407). Berlin: de Gruyter. Zimmermann, T. E., & Lerner, J.-Y. (1991). Eigennamen. In A. V. Stechow & D. Wunderlich (Hrsg.), Semantik. Ein internationales Handbuch (pp. 349–370). Berlin: de Gruyter.

123