Abstract
There are expressions called singular terms. As Salmon (1991, p. 835) puts it, “A singular term is any expression whose function, when used in a particular context, is to refer to (denote, designate), Reference Reference Denotation Designation ,i.e., to stand for, a single individual”. I would say that the typical function of singular terms is to allow us to single out one specific object in order to verbally attribute properties to it, i.e. to “talk about” it (in the terminology of Reimer 2003). There are three main kinds of them, definite descriptions such as “the director of La vita è bella”
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Notes
- 1.
More precisely, we should call “the director of La vita è bella” a singular definite description to differentiate it from plural definite descriptions such “the boys” or “the winged horses”. For brevity’s sake however we can call the latter plural descriptions. Following Russell , definite descriptions are usually distinguished from indefinite descriptions such as “a man” or “a brown table”. I shall often use “description” as short for “singular definite description”.
- 2.
To encompass also those cases in which this reference is not accompanied by speech we may more generally say thinker reference .
- 3.
As noted by Kaplan 1989, p. 492, there is a subtle difference between the notion of a rigid designator and that of a directly referential term . But we can ignore this for the time being. For discussions that help distinguish how “directly referential term” is to be understood here from other possible interpretations, see Marti 1995 and Voltolini 2004.
- 4.
Other expressions that are used as by and large equivalent to “descriptive content”, as this term is employed here, include the following ones of Fregean origin: “mode of presentation, mode of presentation ”, “sense of singular term ”, “individual sense”.
- 5.
See for example Carnap 1928, § 16 and 1934, p. 12 (cf. Farrell Smith 1989, pp. 123 and 134), Quine 1953, Chapter 1 and 1960, Katz 1972.
- 6.
At p. 49, n. 16 of his 1980, Kripke notes that he considers indexicals (“demonstratives ” in his terminology) as capable of being used as rigid designators just like proper names, but at p. 12, n. 10 he recognizes that “[t]he rigidity of demonstratives has been stressed by Kaplan ” (as noted, saying that a term is rigid is fairly close to affirming that it is directly referential, although there is a subtle difference made clear by Kaplan). As is well known, the referentialist account of singular terms typically goes hand in hand with a parallel referentialist treatment of natural kind terms fueled by the arguments in its favour put forward in Putnam 1975 and Kripke 1971, 1980 (as Kripke notes at p. 122 of his 1980, although his and Putnam’s views on natural kind terms have much in common, they were developed independently; in 1975, p. 232, Putnam however acknowledges that Kripke was the first to recognize the implications of such views for the theory of necessary truth). Since I concentrate on singular reference , natural kind terms fall outside the scope of this book.
- 7.
For example, it is well known that Kripke ’s modal argument has been challenged by Dummett and others (see § 8.4) and that “causal descriptivism ” has been proposed in various guises as an alternative to Donnellan ’s and Kripke’s causal account of proper names by Lewis (1984) and others (cf. § 5.9). But there is more and I shall mention some notable samples in the rest of this note. Although Gareth Evans is not in the end by my standards a descriptivist (see note 21 below), a form of descriptivism about proper names may perhaps be attributed to him in his 1973, where criticisms of the causal-referentialist standpoint can be found (see § 5.9). Castañeda 1974 takes all singular terms as directly referring to guises, understood as concrete particulars that make up ordinary objects by means of a relation of consubstantiation in a bundle-theoretic ontological picture (guise theory ). But from the point of view of this picture, Castañeda’s guises play the role of Fregean senses of singular terms (descriptive contents) and thus his position turns out to be a form of descriptivism. Not surprisingly, then, Castañeda 1977 contains reactions to Donnellan’s and Kripke’s anti-descriptivist arguments and Castañeda 1989a opposes the referentialist elements in Kaplan ’s theory of demonstratives . A distinctive feature of Castañeda’s position is that proper names are restricted variables that in context take different guises (or descriptive contents, if one wishes) as values (see Castañeda 1990 for a fully articulated defence of this view of proper names). Loar 1976 presents a rather thoroughly descriptivist view that in some measure assigns descriptive contents to Donnellan’s so-called referential descriptions and takes proper names to express metalinguistic descriptive contents (as I have done in Orilia 2000 and as I shall do in a different way here, by taking advantage, as we shall see, of token-reflexivity à la Reichenbach ). As regards indexicals, Loar takes self-ascriptive belief as primitive (pretty much as will later be done in Lewis 1979 and Chisholm 1981) and builds up a descriptivist account of indexicals on that (see Castañeda 1987 and Austin 1990 for some problems that a view of this kind must confront). Schiffer in his 1978 distinguishes between “Russell ’s description theory of singular terms” and “Russell’s description theory of de re thoughts” and explicitly defends the latter rather than the former. Yet, he seems to be at least inclined to a form of descriptivism that relies on ideas from Castañeda 1977 (Schiffer 1978, p. 194) and perhaps from Loar 1976 (Schiffer 1978, n. 27). Certainly, he is committed to theses that a descriptivist about singular terms might like, such as the idea that beliefs are not propositions that can have ordinary objects as constituents (a view to which, as we shall see, a referentialist might be driven). Roughly speaking, according to Schiffer, when we might think an ordinary object is a constituent of a belief, the constituent is really a corresponding descriptive content, apart from the following exceptions: beliefs about oneself and the present moment need not be mediated by descriptive contents. (Schiffer has also criticized referentialist positions in later writings, to be considered in § 8.13, below.) Plantinga 1978 criticizes referentialism about proper names and by building up on his essentialist standpoint in ontology takes them to express individual essences. Searle 1983 explicitly defends descriptivist and internalist positions about proper names and indexicals against referentialist attacks (roughly, internalism is the view that takes propositional attitudes to be narrow; cf. § 1.8 below). According to Searle, the meaning of a proper name is provided by something like a definition and its reference-fixing content (for a given speaker) is given by the totality of the “intentional content ” that a speaker associates with the name (this may make it impossible always to equate the meaning of the statement uttered in earnest by a standard speaker with a belief of the speaker in question, as in the kind of descriptivism that I would like to defend). Moreover, Searle recognizes that indexicals are essential in the sense proposed by Perry (see § 4.2) and thus takes sentences with indexicals to express indexical propositions that somehow show a self-referentiality. As I understand Searle, this amounts to the idea that, e.g., the proposition expressed by a sentence like “that man wears a red hat” is a proposition perspicuously representable along the lines of “there is exacly one x that is a man causing g and x wears a red hat”, where g is a visual presentation in the mind of the speaker (Searle 1983, p. 212; Russell held a similar view, as discussed in § 3.8 below). There are elements of descriptivism in Burge ’s position, as he rejects the view that proper names and indexicals “simply import a referent into a proposition” and admits that “all reference is perspectival, mediated by cognitive factors” (1983, p. 89). Burge is however committed to externalism about propositional attitudes (1977, 1979), in a way that makes his position akin to referentialism (cf. § 1.8 below). In fact it seems to me that he allows for ordinary objects as constituents of propositions expressed by means of demonstratives (and thus possibly of proper names, since Burge views them as involving a hidden demonstrative (1973)). For he views demonstratives as free variables that in a context are assigned a referent, which may well be an ordinary object. Jubien 1993 takes proper names to be descriptions rigidified by means of an implicit “actually”. For example, “Hesperus” must be understood as celestial body that actually occupies position p at time t (p. 500). Geurts 1997 presents a form of descriptivism about proper names and sees the need for descriptivism for indexicals as well. Brinck 1997 defends descriptivism for the first-person pronoun, by taking any token of it to express at the same time a stable self-concept and a de re sense that shifts with context (p. 121) and consists in information gained from the latter (p. 123). Although Jackson inclines towards two-dimensionalism (see note 10 below), the position defended in his 1998a can be counted as a form of descriptivism about proper names (as well as general kind terms, which are not our concern here). Jackson admits there that speakers and hearers may well associate different descriptions to the same proper name, but argues that this does not hinder communication as long as the descriptions are co-extensive. King 2001 defends the view that complex demonstratives should not be treated as directly referential, although it seems to me that all the data he considers have to do with cases in which a term of the form “that F” can be understood as the description “the F” (e.g., “that hominid who discovered how to start fire was a genius”) or with anaphoric uses (as we shall see, I myself treat anaphoric complex demonstratives as descriptions). Finally, some challenges for referentialism based on data regarding plural indexicals can be found in Palma 2004.
- 8.
Interestingly, there may be cultural differences regarding how some of the referentialist and descriptivist data are appreciated (Machery et al. 2004). Unfortunately, it falls outside the scope of this book to discuss the philosophical issues that this raises.
- 9.
Attempts to deal with the problems of co-reference and no-reference of course pre-date Schiffer ’s concern. For example, Kaplan 1989 (circulating at least since 1977) tries in some degree to address the former by appealing to characters and can perhaps be said to address the latter by admitting merely possible objects. See also the distinction between thoughts and senses in Perry 1977. Moreover, Kripke 1973 in a sense addresses the no-reference problem.
- 10.
In addition to referentialist and descriptivist theories, the current literature also acknowledges approaches to reference that try to combine referentialist and descriptivist aspects in order to have the best of both worlds. Depending on their distinguishing features and one’s classificatory purposes, they may be considered as somehow both descriptivist and referentialist or as neither. I have in mind here (i) the so-called “two factor” or “dual aspect” semantic theories developed in the late 1970 s and early 1980 s primarily in order to deal with propositional attitudes in the light of Putnam ’s distinction (cf. § 1.8 below) between narrow and wide content (Field 1977, Loar 1981, McGinn 1982, Block 1986) and (ii) the “two-dimensional” approaches mainly inspired by the desire to make clearer sense of Kripke ’s claims (cf. § 4.6 below) that there are necessary a priori and contingent a posteriori truths (Stalnaker 1978, 2004, Davies and Humberstone 1981, Jackson 1998, Garcia-Carpintero 2006, Garcia-Carpintero and Macià 2006, Chalmers 1996, 2002, 2004; see Davies and Stoljar 2004 and Chalmers 2006 for surveys and Chalmers 2007 for an attempt to deal most explicitly with propositional attitudes from this perspective). Roughly, in these theories one factor or dimension of meaning (the primary intension, in Chalmer’s terminology ) takes care of Fregean intuitions concerning cognitive significance , while the other (the secondary intension, in Chalmers’s terminologyIntension ) deals with Putnamian or Kripkean intuitions regarding wide content and metaphysical necessity . These two levels must be appropriately independent, i.e., not systematically related in the way sense and reference are in the Fregean framework (where sense determines reference; see § 2.9 below), for otherwise these dual theories, as we may call them, would hardly be distinguishable from the latter (Marconi 2005). Since I try to counter the referentialist trend by proposing a purely descriptivist alternative, an examination of these hybrid proposals falls outside the scope of this book. I would like to claim, however, without any pretense of doing full justice to them, that the descriptivist approach that I defend here can deal, as we shall see, with both the descriptivist and the referentialist data , without the quite formidable technical and theoretical complications that these approaches appear to require when fully developed, as they appeal to such things as possible worlds, centred worlds, different kinds of intensions, diagonal propositions and the like in order to bring cognitive significance to the fore (as is well exemplified by Chalmers’ works). In requiring such complications to enjoy the benefits of cognitive significance, the dual theories are quite similar to the explicitly referentialist attempts to deal with the descriptivist data , which I criticize in § 8.13. Needless to say, the dual theories encouter resistance in referentialist quarters as well, because they incorporate elements of descriptivism (see Soames 2005a and, for a reply, Chalmers 2006a). For a criticism of dual theories that presupposes neither a descriptivist nor a referentialist perspective, and that I am inclined to endorse in its essentials, see Marconi 2005, based on the idea that the dual theories have trouble in accounting for the “articulation problem”, namely the problem of explaining how the two aspects of meaning precisely relate to each other (Marconi focuses on two-dimensionalism and Chalmers in particular and relies on Fodor 1987 to criticize the two-factor theories). It is worth noting that Kaplan ’s semantics for indexicals, because of its distinction of character and content, is often seen by two-dimensionalist s as a forerunner of their approach (Davies and Stoljajr 2004, Chalmers 2007). By my lights, however, Kaplan is a clear-cut case of referentialist and I shall treat him as such in line with what is typically assumed in current literature.
- 11.
The way I use the expression “contextual parameter ” here should not be confused with how I use it in § 2.10 in order to refer, roughly speaking, to the speaker, receiver, time or place of an utterance.
- 12.
Other authors, such as Manuel Garcia-Carpintero and John Perry have similarly made use in recent years of Reichenbach ’s token-reflexivity in order to account for singular reference . However, they do not do this in an attempt to defend descriptivism as I try to do here. Garcia-Carpintero 2000 does not take the token-reflexive items as pragmatic meanings in the way I do from my descriptivist viewpoint, but as reference-fixing devices involved in propositions presupposed by the propositions expressed by sentences with indexicals or proper names. Such expressed propositions may well contain ordinary objects referred to by the indexicals or proper names, pretty much in line with referentialism. The token-reflexive items are however used by Garcia-Carpintero in order to account for the co-reference and no-reference problems, in something of the spirit of a two-dimensional semantics . As we shall see in § 8.13, Perry 2001 takes advantage of token-reflexive “reflexive contents ” to deal with the co-reference and no-reference problems from a clearly referentialist point of view and thus without taking them as “official contents” (which is what I do from my descriptivist perspective, when I take them to be pragmatic meanings). Of course, my token-reflexive descriptive contents differ from Garcia-Carpintero’s and Perry’s token-reflexive items in that mine involve the relation @, to be be discussed at length as we proceed.
- 13.
To make for a lighter reading, I often take advantage of the ambiguity of “sentence” and other metalinguistic terms such as “name” or “singular term” and use them to refer either to a type or to a token, as the case may be. In connection with the type/token ambiguity, there is an ambiguity in reporting speech or in syntactic-grammatical classifications, that can be illustrated as follows. When one says for example that Tom has uttered “snow is white”, this may mean either, so to speak, that a concrete uttering relation links Tom to a token of the type “snow is white” or that a somewhat more abstract uttering relation links Tom to the type itself in virtue of his more concrete link to the token. Similarly, that an expression E is, say, a proper name may mean either that (i) the token E is classifiable as proper name in that it is a token of a type of the kind proper name or (ii) E is a type of the kind proper name. We may leave these ambiguities unresolved unless something crucial hinges on them. Similarly, I may sometimes use a numeral in parentheses used as label for a sentence type, e.g., “(1)” above, to also refer to a (hypothetical) token of the sentence type rather than to the sentence type itself. In these cases the context is meant to make it clear which token has been referred to.
- 14.
The example “winged horse”, together with countless others, such as “gentle young man”, may suggest that when adjectives and then a noun are put in sequential fashion, a conjunctive property is expressed. But, as the classic counterexample, “fake diamond”, testifies, this is not always the case. It would be improper of course to represent the meaning of this predicate as [fake & diamond]. At any rate, for simplicity’s sake, I shall avoid being fussy about these details unless the context suggests otherwise.
- 15.
From an even more formal point of view, it should be noted that complex properties such as [winged & (horse ∨ donkey)] can be appropriately represented by recourse to the lambda operator familiar from second-order logic and often used in natural language semantics in the Montague tradition. More on this in note 25 below.
- 16.
I neglect as far as possible the issue of what precisely in a proposition corresponds to tense. More on this later.
- 17.
This and many other examples in the following were formulated when George W. Bush was the president of the USA.
- 18.
I shall neglect this option, although many philosophers and in particular many referenialists seem to prefer it. At any rate, nothing crucial in my attempt to assess the dispute between descriptivists and referentialists will depend on this point.
- 19.
As we shall see, there are cases (having to do with anaphora) in which the property component of a descriptive content expressed by a definite description is “open” in the sense that, as we may put it, it is expressible by recourse to a variable. In these cases, because of the variable, the descriptive content cannot determine any object in particular. More on this below.
- 20.
Although this picture may be faithful to what Frege would have said, it may not be faithful to how Russell would put the matter, for according to his doctrine of definite descriptions as “incomplete symbols”, there is no single constituent, |the F|, contributed by “the F” to the proposition expressed by “the F is G”. However, for present purposes, we may ignore divergences among descriptivists on such fine-grained details.
- 21.
In his 1981 exegesis of Frege ’s account of indexicals, Gareth Evans seems to admit senses of indexicals that somehow involve as constituents the referents of the indexicals themselves, referents which in typical cases are of course ordinary objects (see also Peacocke 1983). Such an approach is thus by my standards not a form of descriptivism (which is in line with taking Evans as wanting to render Frege somehow compatible with referentialism rather than as wanting to defend a Fregean descriptivism). In fact, it is an approach that runs into problems when an indexical happens to have no referent, contrary to descriptivism as I understand it (cf. § 3.5). For in such a case it is unclear how the indexical can have a sense, given that the sense must involve the referent and there is no referent. This problem remains even if the sense of the indexical is not taken to have the referent as constituent, but is still seen as dependent on the referent for its existence (Evans 1982, McDowell 1984). The kind of descriptivism I want to defend here does not face this problem and it falls outside the scope of this work to address the issue of how the problem can be dealt with in a framework à la Evans (see on this McDowell 1990, Peacocke 1991, Dummett 1993, Chapter 7). For more details on Evans’ position, cf. Coliva and Sacchi 2001.
- 22.
Given this understanding of definite descriptions, a sentence of the form “the F is G”, such as (2), turns out to be false if no object is F or more than one object is F. As is well known, Frege 1892 has a different view, according to which in such cases the sentence lacks a truth value, and Strawson 1950 argued against Russell by defending a position similar in this respect to Frege’s. Whereas in Russell’s view the sentence entails that there is exactly one F and thus it is false, when this is not the case, in the Frege-Strawson view, the sentence presupposes that there is exactly one F and thus cannot reach the level of having a truth-value when this is not the case. The discussion on which standpoint is the right one still continues (Ludlow 2009), but it seems to me that Neale 1990 offers a very good defence of the Russellian line, which I shall take for granted here.
- 23.
The issue arises of how we should treat the “a” that follows the copula in sentences such as “John is a man” or “the morning star is a planet”. Following Montague 1974 (Chapter 8), and taking the “is” to express identity rather than simply a predicational link, we may take the “a” as expressing the very same determiner, |some|, that it expresses in a sentence such as (9), understood as having truth condition (9b). It seems to me that this is the right path to follow, since it minimizes assumptions of ambiguity. However, since nothing crucial depends on this for our main purposes, I shall follow the common practice of viewing expressions such as “is a man” or “is a planet” as ways of attributing properties without involving identity or the determiner |a|.
- 24.
In presenting (10a) and (11a), I neglect to take into account, at least from the perspective on anaphora that I intend to defend in the following, the linguistic meaning of the pronouns “it” and “he”. This is not important at this juncture, but will be taken care of below.
- 25.
In other words, the locution “such that” is used in (10a), (11a) et similia pretty much as the lambda operator functions in formalistic approaches to natural language semantics (see, e.g., Partee et al. 1990, Chapter 13 We can add, for the more formalistic-minded philosopher, some detail on how we could represent more precisely propositions involving denoting concepts. For example, the proposition |{every man} is mortal|, in which the denoting concept |every man| occurs as active, can be represented, by assuming a second-order logical language with the lambda operator, as [λf ∀x(man(x) ⊃ f (x))](mortal). From the point of view of this representation, a denoting concept such as |every man| is seen as a complex property , [λf ∀x(f(x) ⊃ mortal(x))] (expressed by a so-called lambda abstract ), which can be predicated of another property, being mortal in our example (and could thus be called a property of properties ) (cf. Montague 1974, chs. 6-8). The formal principle of lambda conversion grants that the proposition [λf ∀x(man (x) ⊃ f(x))](mortal) is logically equivalent to ∀x(man(x) ⊃ mortal(x)). The principle of lambda conversion can be stated as follows (where A(a 1/b 1 … a n /b n ) is the wff that results from the wff A after simultaneously replacing each occurrence of a i with b i , for 1 ≤ i ≤ n, where b i is free for a i in A):
$$[\lambda a_1 \ldots a_n A](b_1 ,{\rm{ }} \ldots ,{\rm{ }}\ b_n ) \equiv A(a_1 /b_1 \ldots {\rm{ }}a_n /b_n ).$$The other propositions involving active denoting concept s can be understood in the same fashion. For example, |the president-of-the-USA is American| should be seen as [λf ∃1 x(president-of-the-USA(x) & f(x))](American), which is equivalent, by lambda conversion , to ∃1 x(president-of-the-USA(x) & American(x)). As these examples suggest, the determiner meaning components of denoting concepts are relational properties such as [λgf ∃1 x(g(x) & f(x))] and [λgf ∀x(g(x) ⊃ f(x))]. They are relational in that they have, so to speak, two “holes” to be filled, corresponding to the two variables bound by the lambda operator in their formal representations. By filling the first hole of a denoting concept with a monadic property we create by composition a denoting concept just as (roughly speaking) we create a noun phrase by combining a determiner and a predicate. For instance, by combining “the” and “table” we get “the table” and similarly by filling in the first hole of [λgf ∃1 x(g(x) & f(x))] with |table| we get [λf ∃1 x(table(x) & f(x))]. The difference between active and non-active occurrences of denoting concepts can be captured by taking non-active denoting concepts as occurring qua logical subjects rather than qua properties of properties. Thus, for example, the proposition |%every man% is a concept| in which the denoting concept |every man| occurs non-actively can be represented as concept([λf ∀x(man(x) ⊃ f(x))]), which makes it clear that the proposition in question should be understood as the attribution of the property of being a concept to the denoting concept [λf ∀x(f(x) ⊃ mortal(x))]. It should be noted that the principle of lambda conversion is not applicable in this case, because “[λf ∀x(man(x) ⊃ f(x))]” does not occur in predicate position, but rather in subject or argument position. As is well known, once we allow for lambda conversion and for predicate terms such as lambda abstract s to occur in both predicate and subject positions, we have to face Russell ’s paradox and similar conundrums (the problem does not arise for Montague , since he assumes type theory ). For two different ways of dealing with these issues and further formal details on representing propositions with denoting concepts in the way sketched in this footnote see Orilia 1999, 2000a, 2006. A similar approach to denoting concepts, from which I have drawn inspiration, can be found in the work of Cocchiarella (1989, 2008 and others). Cocchiarella however deals with Russell’s paradox and the like in a way that I find problematic (Orilia 1996; see Landini 2009 and Cocchiarella 2009 for a discussion of this). Before bringing to a close this rather technical note, it is worth pointing out that lambda abstracts offer us a way to represent not only denoting concepts but all complex properties in general. Thus, for example, the property [winged & (horse ∨ donkey)] used above as an example can be represented by the lambda abstract “[λx (winged(x) & (horse(x) ∨ donkey(x))]”. In the light of lambda conversion, to predicate this property of an object, a, so as focus on the proposition [λx (winged(x) & (horse(x) ∨ donkey(x))](a), is equivalent to asserting the conjunctive proposition (winged(a) & (horse(a) ∨ donkey(a)).
- 26.
More formally, propositions involving open denoting concepts may be understood (following Cocchiarella 1989, § 7) in a way that can be illustrated by focusing, e.g., on (10a). The idea is to view a proposition such as (10a) along these lines: [λF ∀x(diamond(x) ⊃ f(x)]([λx [λg ∃1 y(person(y) & find(y, x) & g(y)]([λy belongs-to(y, x))]) (i.e., roughly, as the result of predicating |every diamond| of the property of being an x belonging to the person who finds x. Clearly, by lambda conversion , the latter is equivalent to (10b) and thus has the desired truth condition. The open denoting concept , given this formal representation, is [λg ∃1 y(person(y) & find(y, x) & g(y)]. I shall not tackle here the difficult issue of the consequences, for an ontological account of propositions, of admitting that some denoting concepts (and, more generally some concepts) seem to call for free variables in order to be adequately represented. In other words, this is the issue of the ontological status of open denoting concepts and their open property components. This issue may be neglected here, for after all it is a problem for descriptivists and referentialists alike. For example, the need for a semantic account of (10) and (11) may invite both descriptivists and referentialists to appeal to open denoting concepts , even though in accounting for other sentences (involving singular reference ) the descriptivist, but not the referentialist, will appeal to denoting concepts. For what it is worth, let me simply state that my tentative way to go would be by appealing to operators in the style of combinatorial logic to get rid of variable binding operators (see, e.g., Fitch 1952). Note in fact that a variable that appears as free, when one considers an open determiner phrase in isolation, can be seen to be, when that phrase occurs in the context of a sentence, a variable bound “from outside the determiner phrase” by means of a preceding variable binder.
- 27.
According to referentialism, M is not always an abstract entity, for it could be a particular directly referred to, e.g., by a proper name. Moreover, according to “weak” forms of descriptivism, as we may call them (see § 3.8, below), in some special case M could be a peculiar kind of particular, e.g. a Russellian sense datum .
- 28.
The dispute between descriptivists and referentialists can hardly be understood without the assumption, plausible in my view, that there are such things as contents of propositional attitudes . Some may want to refuse this assumption. Perry ’s “The Search for the Semantic Grail” (unpublished) usefully comments on those who like and those who dislike content and on the advantages the former have.
- 29.
The dispute between descriptivists and referentialists is best understood with the assumption of structured propositions in the background. There are however philosophers who deny it. As is well known, there is a well-developed tradition according to which propositions are sets of possible worlds or functions from possible worlds to truth values (Hintikka 1969, Montague 1974, Stalnaker 1984, Lewis 1986). Moreover, there are those who even deny that there are propositions understood as language-independent entities, for they think that linguistic (or language-related) items such as sentences (Carnap 1958, Quine 1956), sentence tokens (Davidson 1968) or interpreted logical forms (Harman 1972, Higginbotham 1991, Larson and Ludlow 1993) can play the roles that propositions are called for, in particular as accusatives of propositional attitudes . I think that these approaches are unsatisfactory (see McKay and Nelson 2005, § 7 for a survey and criticism of these views). But in any case talk in terms of them can presumably (at least for the most part) be translated into structured propositions talk. For example, something like this is assumed in classifying Carnap and Quine as descriptivists (see Chapter 4 below).
- 30.
The contexts in question here are of course intralinguistic context s, i.e., other expressions surrounding a certain text, not to be confused with the extralinguistic context , made up of all the objects and states of affairs relevant for the comprehension of the text in question (L. Marti ’s terminology in her 2006).
- 31.
After Putnam 1975, Burge 1979 famously argued in favour of externalism by relying on the social dimension of language and since the 1970 s the literature on the topic has been growing.
- 32.
Following Armstrong and Bealer, I view properties and relations as universals . Some of course prefer a tropist ontology (see, e.g., Maurin 2002 and references therein) where they are understood as tropes, i.e. as particulars. However, the assumption of an ontology with universals should not be seen as an essential part of my present defence of descriptivism: a philosopher convinced that a tropist ontology is preferable should presumably be able to translate any talk of properties and relations understood as universals into a tropist way of speaking.
- 33.
An untensed proposition is basically a “proposition matrix”, in the terminology of Salmon 1986, p. 39. Richard 2003 speaks of “temporally neuter propositions” .
- 34.
Other temporal type II A-properties would be futurity and pastness. Temporal type II A-properties are typically contrasted with type II B-relations such as earlier and later (see, e.g., Q. Smith 1993 on this). Richard (2003, p. 30) notes that one can be a temporalist and still deny that there are A-properties. Richard attributes this view to Hugh Mellor . For present purposes we can confine ourselves to the view of the typical temporalist, according to whom temporalism and the acceptance of A-properties go hand in hand. Accordingly, I use “temporalism” in the following to mean typical temporalism.
- 35.
From a formal point of view, we can represent the property of being present by means of this lambda abstract : [λp ∃t(at(t, p) & A-present (t))]. Once the property of being present is represented in this way, (1a) becomes [λp ∃t(at(t, p) & A-present (t))]([λ B is# a president]). By lambda conversion , the latter is equivalent to ∃t(at(t, [λ B is# a president]) & A-present (t))), which is a more formal counterpart of (1b). Here (and elsewhere below), as “[λ B is# a president]” testifies, I have followed the common practice of representing propositions embedded in subject position in another proposition by means of a “vacuous” lambda abstract in which the lambda operator does not bind any variable.
- 36.
Salmon is not committed to taking the “at” of (1b) as tensely predicated and takes the time to be somehow part of the property conveyed by the predicate, “is a president” in this case.
- 37.
Richard (2003, p. 30) however notes that an eternalist could in principle accept A-properties, and takes Salmon to be an eternalist who admits them (note 9, p. 44).
- 38.
The issue arises whether the predicational link connecting |at| to |B is# president)| in |at(t, B is# president)| is tensed or not. It depends on whether we admit A-properties. If not, the link is tenseless. Otherwise, there is room for viewing it as tensed. The idea is that t is not atemporally such that Bush is a president with respect to t, as if the contingencies of becoming, or of the world history, had nothing to do with the properties that moments of time happen to have. Rather, t is presently a time with respect to which B is# a president, but at some point in the past (when t was not the time of the utterance in question, for such utterance did not exist), t was not a time with respect to which B is# a president, for at that point it was not yet determined that Bush would become a president. And if we are referentialists and we assume that |bush is# a president| is a proposition with Bush in flesh and blood as constituent, we might want to say that at some point, when Bush will pass away, t will lose the property, which now has, of being a time such that |bush is# president| is true at it.
- 39.
The property of being past can be represented by an appropriate lambda abstract just as we have done for the property of being present in note 35 above.
- 40.
In saying this, I presuppose the traditional tripartition of semiotics (understood as theory of signs, or semantics , theory of meaning, in a broad sense) into syntax, semantics (theory of meaning in a narrow sense) and pragmatics .
- 41.
Recall that I use words such as “expression”, “singular term”, “predicate” or “sentence” ambiguously to mean either a token or a type, unless more precision is required.
- 42.
For the referentialist the two propositions could be |at(t, h is# sunny)|, and |at(t', h is# sunny)|, where h is the place referred to by both the first and second “here” tokens and t and t' are two distinct times of utterance, corresponding to the two tokens of “it is sunny here”. For a descriptivist, on the other hand, the two propositions would involve descriptive contents determining h, t and t', rather than h, t and t' themselves.
- 43.
That there are sentences whose tokens are of this kind has long since been known. For example, Gale (in the introduction to § 4 of his 1968, p. 297) speaks of “pragmatically self-falsifying sentences” such as “I do not exist”.
- 44.
Predelli admits that Kaplan ’s arguments are not conclusive against the type-oriented approach (as is argued in Garcia-Carpintero 2000), but he provides these examples in order to offer a stronger support for it.
- 45.
At any rate, with appropriate adjustments, there might be a “translation” from one approach to the other and vice versa. If so, we need not worry too much in following the token-oriented approach here.
- 46.
See, e.g., Kaplan ’s Afterthoughts (1989a) and Bach 1994, p. 136. For a recent defence of this line, see Fodor and Lepore 2004.
- 47.
Note also that, alongside the notion of an ideal speaker (of a given language), we should also take for granted – as obviously presupposed in the former – that of a standard thinking subject, who is assumed to have normal mental capacities, i.e., she has neither superhuman, nor subhuman intelligence.
- 48.
I take a command to be a practition (Castañeda 1975), something akin to a proposition but devoid of truth value. This is of course inessential for the main topic of this book. For what matters here, someone unconvinced by Castañeda’s distinction between practitions and propositions may view a command as a proposition of sorts.
- 49.
To indicate that a token t has speaker or hearer meaning M for X, we may use rather obvious variants such as: X expresses M with t, t means M for X, X attributes M to t.
- 50.
Note that speaker, hearer and official meanings may have subconstituents, at least to the extent that a compositional view of meaning is assumed (as it is here). In this case, we may consider such subconstituents as in turn speaker, hearer or official (sub-) meanings.
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
Copyright information
© 2010 Springer Science+Business Media B.V.
About this chapter
Cite this chapter
Orilia, F. (2010). Introduction: Referentialism vs. Descriptivism. In: Singular Reference: A Descriptivist Perspective. Philosophical Studies Series, vol 113. Springer, Dordrecht. https://doi.org/10.1007/978-90-481-3312-3_1
Download citation
DOI: https://doi.org/10.1007/978-90-481-3312-3_1
Published:
Publisher Name: Springer, Dordrecht
Print ISBN: 978-90-481-3311-6
Online ISBN: 978-90-481-3312-3
eBook Packages: Humanities, Social Sciences and LawPhilosophy and Religion (R0)