‘On Reference’ is a collection of 18 original articles. While united in their concern with reference, they deal with a large variety of topics, ranging from questions concerning the nature of reference, through the interaction of reference and cognition, to more specific questions about the semantics of particular referring expressions. The contributions are of high quality: thought provoking, insightful and engagingly written. Many have the potential to substantially advance the debate in their field.

In this review I will do two things. I will focus on a cluster of four essays (Chapters 10–13) that are concerned with a single topic: the view that proper names are predicates. Apart from illustrating the above-mentioned virtues, these essays are well suited to be discussed in conjunction due to their tightly connected subject matter. Before I get to this although, I will give mini-summaries of the remaining 14 articles, but space constraints prevent me from discussing them in the detail they deserve.

Part I. The nature of reference

Chapter 1: The book starts with somewhat of a counterpoint to its general agenda. In ‘The Illusion of Semantic Reference’ Christopher Gauker tries to advance the project of dispensing with a notion of semantic reference (as opposed to speaker’s reference). In particular, he tries to dismantle one reason for believing that such a project must be doomed from the start: the intuition that we have knowledge of what individual terms and predicates refer to. He proceeds in two steps. First, he argues that the intuition is really an intuition concerning our knowledge of the meaning of these expressions. Second, he develops an account of our knowledge of meaning that doesn’t require a notion of reference. Very roughly: in ascribing knowledge of meaning to someone we grant them a status; we recognize that there is no ‘need to interpret him or her by substituting some other words for those that he or she actually used’.

Chapter 2: Diego Marconi’s ‘Reference and Theories of Meaning as Use’ is also concerned with the tension between an intuition about reference and a project which, while not seeking to abandon the notion of reference altogether, still tries to downplay its importance. This time the project is Horwich’s theory of meaning-as-use with its deflationary account of truth and reference. And the recalcitrant intuition is the externalist intuition that expressions which are used the same way, such as ‘water’ and Twin-Earth ‘water’, can still have distinct extensions. Contra Horwich, Marconi argues that this tension is ‘irremediable’. Accordingly, we must either reject semantic externalism or refrain from a use-based account of meaning.

Chapter 3: In ‘Speaker’s Reference and Cross-Cultural Semantics’ Edouard Machery, Justin Sytsma and Max Deutsch also deal with a famous intuition but they do so from an experimental point of view. The main question is how stable Kripkean intuitions concerning his Gödel/Schmidt case are across subjects from different cultures. In 2004, Machery et al. presented experimental data suggesting that while the majority of American participants have Kripkean rather than descriptivist intuitions, this is reversed with Chinese participants.

A frequent criticism of that study is that it doesn’t sufficiently guard against the possibility of participants misunderstanding the questions posed as being concerned with speaker’s reference rather than semantic reference. This article presents several refined studies which employ different strategies in order to prevent participants from considering speaker’s reference. The authors take the results to support the original findings. In the study with most refinements in place, the results are ‘consistent with the claim that Chinese are significantly more likely to have descriptivist intuitions about proper names than Americans’. (However, it should be noted that in this study it was no longer a majority of Chinese participants that answered in accordance with descriptivism, but 43.9 percent, compared with 26.1 percent of descriptivist answers among Americans.)

Chapter 4: In her contribution, ‘Reference without Cognition’ Genoveva Martí leaps to the defence of an idea she takes to be at the heart of direct reference theories of names. The idea is that reference doesn’t require a ‘cognitive fix’: it is not what happens in a speaker’s mind that makes her use of a name designate its referent, but factors ‘outside of the cognitive sphere’ such as being part of an appropriate causal chain. Recently some direct referentialists (including Kaplan and Almog) have defected from this principle in taking referring to be ‘parasitic on having in mind’. Martí wants to bring the apostates back in line: according to her ‘having the referent in mind is neither necessary nor sufficient for a use of a name to refer to it’.

Chapter 5: Andrea Bianchi’s essay ‘Repetition and Reference’ is intended to be a step towards turning Kripke’s causal chain account of reference, famously put forward as a picture, into a systematic theory. Such a theory must have at least two components: an account of how names are introduced and an account of how they are passed on. Bianchi’s focus is on the latter which he interestingly connects to Kaplan’s work on the metaphysics of words. Reference is preserved from one speaker’s use of a name to another speaker’s use of the name if the two utterances count as repetitions of the same word in Kaplan’s sense of the term (which, controversially, Bianchi thinks can be understood in purely causal terms). Bianchi also offers a new take on Evans’s Madagascar case: while standardly construed as challenge for the passing-on component of a Kripkean theory, Bianchi thinks the solution should come from the introduction component: successful passing on of a name can be undermined by various events which include the (unwitting) introduction of a new name.

Chapter 6: The first part of the book is closed by Michael Devitt asking: ‘ShouldProper Names Still Seem So Problemtic?’. His answer is that they shouldn’t because he has already developed, in the 1970s and 1980s, the type of systematic theory which Bianchi thinks we are still lacking. The first half of his article contains a very useful and up to date presentation of this theory. Devitt also defends his account against the charge of not being a ‘full blown theory’. While acknowledging that several components ‘must look to future psycholinguistics’ for completion, he argues that the theory is as complete as any current theory of reference can reasonably be expected to be. The second half of the article is a renewed attack on what Devitt considers a common misconception: the idea that by rejecting descriptivism a causal theory of reference entails a ‘Millian’ conception of reference as direct. Against this Devitt insists that the name’s referent is but one aspect of its meaning; we also need to recognize a ‘mode of presentation’, albeit a causal not a descriptive one.

Part II Reference and cognition

Chapter 7: In ‘Thinking about an Individual’ Antonio Capuano contrasts to pictures of cognition. On the ‘inside-out’ picture a subject thinks about the world in virtue of a standing in some relation to an ‘intermediate entity’ which is ‘transparent’ to the subject in that it has some sort of immediate epistemological access to it. He ascribes this picture to Russell, Frege and more recently Tyler Burge and, citing considerations from Kripke and Donnellan pertaining to Paderewski cases, argues that it is inferior to an ‘outside-in’ picture of cognition. On this picture, ordinary objects ‘enter into the mind of a thinker rather than a thinker reaching out for them through a representation’.

Chapter 8: Marga Reimer’s ‘Drawing, Seeing, Referring: Reflections on Macbeth’s Dagger’ begins with an illuminating discussion of the notion of drawing an F. Reimer argues that this notion is ambiguous between an ontic reading, on which it entails the existence of an F, and a non-ontic reading, on which it doesn’t. She then argues that the same ambiguity applies to seeing an F and referring to an F and employs this idea to shed light on the semantics of ‘empty’ names in cases involving hallucination. On the non-ontic reading ‘Macbeth refers to (/sees) a dagger’ can be true even in the absence of a dagger. But on Reimer’s view, which is inspired by Salmon’s account of mythical names, the report still entails the existence of an object referred to, namely an abstract artifact unwittingly created by Macbeth in his failed attempt to refer to a concrete object.

Chapter 9: In his chapter, ‘The Cognitive Contribution of Names’, John Perry attempts to account for Frege’s Puzzle without accepting Frege’s (considered) conclusion in the form of sense. Rather, Perry’s solution bears some similarity to Frege’s Begriffschrift solution in that it regards the information conveyed by an utterance of ‘San Sebastian is Donostia’ to involve linguistic information: information about names and what they refer to. While the names ‘San Sebastian’ and ‘Donostia’ have the same semantic value, their referent, on Perry’s account part of what a hearer of the above utterance learns, is that the speaker is referring to an entity with two differently sounding names: ‘the direct cognitive contribution of a spoken name is how it sounds’. Perry goes on to defend this view against various problem cases (ambiguous names, empty names, indirect discourse).

Part III Reference and semantics

Chapter 14: Marco Santambrogio’s ‘Empty Names, Propositions and Attitude Ascriptions’ tries to reconcile direct referentialism with the fact that some sentences involving empty names strike us as true (‘Le Verrier imagined Vulcan’), while others strike us as false (‘Vulcan is larger than Mercury’). He suggests that the problem can be mitigated once we realise that a direct reference theory is not committed to propositions with Russellian constituents.

Instead, he proposes to combine direct referentialism with an account of propositions whose name-constituents are equivalence classes of names: The contribution of name N to the proposition expressed by ‘N is a planet’ is the class of all names that have the same referent as N. In the special case where N is empty, the contribution is the class of all empty names. Santambrogio argues that this account of propositions is preferable to other accounts of propositions proposed by proponents of direct reference. It should be noted, however, that Santambrogio’s account is subject to a major limitation (as he admits): it entails that ‘Sherlock Holmes is a detective’ and ‘Harry Potter is a detective’ express the same proposition. Therefore, he cannot account for the fact we sometimes regard a sentence with an empty name as true and another as false, despite the fact that the only difference between them concerns which empty name they contain.

Chapter 15: Ángel Pinillos tackles another classic problem for direct referentialism in ‘Millianism, Relationism and Attitude Ascriptions’: to account for the intuitive difference in truth-values between (1) ‘Lois Lane believes Superman can fly’ and (2) ‘Lois Lane believes Clark Kent can fly’. Following Kit Fine (2007), Pinillos accepts the outlines of a relationist account of the difference between (3) ‘Lois Lane believes Clark Kent is Clark Kent’ and (4) ‘Lois Lane believes Clark Kent is Superman’. While each name’s individual contribution to the proposition expressed by the embedded sentence is just its referent, the names in the sentence embedded in (3), but not in that embedded in (4), are ‘coordinated’ and this makes for a difference in the proposition expressed. The challenge is to extend this account to (1) and (2) where the embedded sentences each contain only a single name.

Fine (2007, 2010) tried to solve the problem by appealing to inter-discourse coordination. Very roughly: coordination between the name in the sentence and a ‘position’ in the mental state of Lois Lane. Pinillos points out some problems for this approach and proposes an alternative which appeals to intra-discourse coordination instead. The key idea is to draw a distinction between the explicit content of the sentence (1) and extra content that is implicitly asserted in an utterance of (1). Pinillos then identifies coordination relations between the names in these explicit and implicit contents which can account for the intuitive difference between (1) and (2).

Chapter 16: The dilemma which Samuel Cumming tackles in ‘The Dilemma of Indefinites’ is this: a sentence of the form ‘an F is G’ has (i) a ‘singular content’ but at the same time it has (ii) an ‘existential truth condition’. By (i), Cumming means that that there are good reasons to treat ‘An F’ as a referring expression on par with names and demonstratives. By (ii), he means that the sentence is nonetheless true as soon as there is some F or another which is G. In tandem, the two features constitute a counterexample to a principle that might have appeared trivially true, namely that semantic content determines truth-condition. Cumming proposes to reject this principle and sketches a view that distinguishes with regard to a speaker who utters ‘an F is G’ between a ‘private commitment’ to a particular referent, which is meant to account for (i), and a weaker ‘public commitment’ to there being some F or another which is G, which is meant to account for (ii).

Chapter 17: In ‘A Unified Treatment of (Pro-)Nominals in Ordinary English’ Joseph Almog, Paul Nichols and Jessica Pepp are parting with a philosophico-linguistic paradigm. The paradigm concerns the analysis of sentences such as ‘No (every) pope admits that he is fallible’. Standardly, the noun phrase ‘No (every) pope’ is analyzed in terms of a quantifier and the pronoun ‘he’ in terms of a bound variable. Not so on the present account. Guided by their desire for maximal unification they proceed in two steps. First, they treat the noun phrase ‘every pope’ as referring to all the (contextually relevant) popes. Then they regard the ‘he’ as another referring expression, which co-refers with the noun phrase. In doing so, they reject the common tripartite distinction: deictic vs. anaphoric vs. bound pronouns. According to them, all pronouns have the common function of directly referring to the contextually most salient individual(s). That salience sometimes depends on extra-linguistic features (deictic pronouns) and sometimes on linguistic context alone (anaphoric and bound pronouns) doesn’t warrant a different semantics by their lights.

Unsurprisingly, the authors have a harder time accounting for noun phrases such as ‘no F’ on the present model. Partly for this reason, they suggest an alternative on which it is the noun (‘pope’) which refers to the plurality of popes, while the determiner (‘every’/‘no’) indicates to which individuals belonging to this plurality the verb should be taken to apply to (all/none of them).

They close by situating their account within a more general causal theory of reference. They claim that our causal (in particular, perceptual) relation to the context of a pronoun is similar to the causal relation the user of a proper name has with the name’s bearer on the Kripkean picture of reference.

Chapter 18: Edward L. Keenan’s ‘Individuals Explained Away’ is concerned with the interaction of certain evaluative adjectives and agentive common noun phrases, as in ‘skilful surgeons’, ‘accomplished portrait painter’. These expressions are challenging for standard model theoretic treatments. The evaluative adjectives function similarly to non-evaluative adjectives (e.g. ‘female’) in that ‘Sarah is an accomplished portrait painter’ entails ‘Sarah is a portrait painter’. But unlike, e.g., ‘female’, they create a non-extensional context: Suppose the heart surgeons are identical to the portrait painters. It doesn’t follow that the skilful heart surgeons are the skilful portrait painters.

To deal with these constructions, Keenan first remodels standard extensional semantics in ‘purely Boolean terms’. On the resulting semantics individuals are ‘explained away’ in that ‘proper name interpretations are now derivative, defined in terms of properties and truth values’. In a second step, he generalizes his semantics so that it allows him to deal with non-extensional contexts and in particular to account for the challenging features of evaluative adjectives.

Discussion: predicativism about proper names

I will now turn to a more detailed discussion of the four chapters (10–13) not summarized above. These chapters concern predicativism about proper names and in particular the question of whether considerations pertaining to the uniformity of a semantic theory can serve as a motivation for predicativism. I will begin by describing the debate as it unfolds in Robin Jeshion’s and Delia Graf Fara’s contributions. I will then tentatively side with Jeshion, drawing on some additional considerations from Ernesto Napoli’s article. In doing so, I will focus on what I take the main line of dispute between these authors, while neglecting some subtleties and many interesting side issues.

Adopting Jeshion’s theoretically neutral taxology, we can say that ‘Alfred’ occurs in apparently referential position in

  • (1) Alfred is in Princeton

and in apparently predicative position in
  • (2) There are at least two Alfreds in Princeton.

Predicativists about proper names think that sentences such as (2) reveal an important insight into the semantics of proper names. The insight is that proper names are predicates. This claim can mean different things. On Fara’s version of predicativism it means that in all of their occurrences proper names have the same type of semantic value that bona fide predicates have.

While being silent about what exactly the semantic values of predicates (and hence of names) are, Fara is explicit about what the applicability conditions of names, qua predicates, are. According to her, ‘Alfred’ is true of an individual just in case that individual is called Alfred. More generally, she takes the following schema, dubbed the being-called condition, to characterize the applicability condition of those predicates which are proper names:

(BCC) A proper name ‘N’ is a predicate that is true of a thing iff it is called N1

According to Fara then, appearances are misleading when it comes to (1): while ‘Alfred’ doesn’t look like a predicate here, it still is one. In particular, Fara regards it as ‘the predicative part of a denuded definite description’, a definite description whose definite article is ‘unpronounced’ (Fara 2015: 60). On this view, the logical form of (1) is more perspicuously rendered as follows, where the square brackets indicate that the material they enclose is left unpronounced:

(1*) [The] Alfred is in Princeton.

The sentence is true iff the unique individual called Alfred is in Princeton. Worries concerning the uniqueness condition can be addressed by appealing to the familiar machinery of context induced implicit restriction of the predicate in question. But why insist that ‘Alfred’ is a predicate in (1) in the first place? Why go from the claim that ‘Alfred’ functions as a predicate in those sentences where this is in accordance with the sentence’s superficial syntax to the claim that it functions as a predicate in all sentences, including those where this creates a conflict with the sentence’s superficial syntax, a conflict which can only be resolved by the postulation of unpronounced material?

The predicativists’ answer is that ‘we should try as much as possible to uphold a unified theory of proper names’ (Fara 2011, Unpublished Manuscript: 9) and that this is achieved by regarding proper names as predicates in all of their occurrences. In particular, predicativists think that their theory is more unified than, and hence preferable to, any view which regards ‘Alfred’ as a genuinely referring expression in (1). Such a referentialist view, predicativists maintain, will be forced to regard occurrences of ‘Alfred’ such as the one in (2) as somehow deviant or non-literal. In contrast, predicativists can treat ‘Alfred’ as having exactly the same semantic function, namely that of a predicate governed by (BCC), in both types of occurrences.

It is this argument from uniformity, advanced by Fara and other predicativists before her, with which Jeshion takes issue in her contribution ‘Names Not Predicates’ (Chapter 11). She begins by noting that predicativists themselves have to regard some apparently predicative occurrences of proper names as deviant in that they are not susceptible to their standard analysis, which proceeds in terms of (BCC). These occurrences include the following:

  • (3) George Wallace is a Napoleon.

  • (4) Dan Quayle is no Jack Kennedy.

  • (5) My mother thinks she is some kind of Martha Stewart.

The truth value of (3), for instance, predicativists admit, is not dependent on whether George Wallace is called Napoleon. Predicativists maintain, however, that their theory’s inapplicability to (3)–(5) does not undermine their theory’s uniformity, while the inapplicability of referentialism to (2) still shows that referentialism is less uniform than predicativism. Why? Because they deem the relevant uses of proper names in (3)–(5) non-literal, while regarding the uses of proper names in (1) and (2) as literal. A semantic theory of proper names should primarily account for literal uses, so the idea goes, and those uses predicativism, in contrast to referentialism, can handle uniformly.

Jeshion’s main line of attack on the uniformity argument now is to provide a host of sentences involving apparently predicative occurrences of proper names which, like (3)–(5), are recalcitrant to a treatment in terms of (BCC) but which, unlike (3)–(5), cannot easily be shrugged off as non-literal. Jeshion’s problem cases include the following.

Dynasty Cases: Names applied to individuals belonging to a certain family, where membership is matter of ‘bloodlines or marriage’ (pp. 237–8):

  • (6) Waldo Cox is a Romanov.

Contra (BCC), an utterance of (6) can be true if the person’s only name is ‘Waldo Cox’ while the person happens to be appropriately related to the Romanov dynasty.

Resemblance Cases: Names applied to individuals due to their having ‘certain salient characteristics of a certain individual’ (238):

  • (7) Two little Lenas just arrived.

Contra (BCC), an utterance of (7) can be true if two of Lena’s daughters have just arrived and these daughters, while not being named after their mother, resemble her in certain salient respects.

Artifact Cases: Names applied to objects ‘produced or designed by a certain individual’ (pp. 239–40):

  • (8) Two Stellas are inside the museum.

Contra (BCC), an utterance of (8) can be true if there are two unnamed paintings by Frank Stella in the museum.

Costume Cases: Names applied to individuals ‘fashioned as replicas, copies, models or some other variety of representation of a certain individual’ (240):

  • (9) Two Osama bin Ladens came to the Halloween Party

Contra (BCC), an utterance of (9) can be true if there are two individuals who are dressed as Osama bin Laden whose name they don’t share.

Jeshion stresses that she advances these cases to undercut a motivation for predicativism – the uniformity argument – rather than as a direct attack against predicativism. The cases are not intended to be counterexamples to the predicativists’ claim that all names are predicates. After all, they are all cases where names occur in apparently predicative position. Rather, they are meant to put pressure on the predicativists’ claim that by treating all names as predicates with an applicability condition given by (BCC) they have provided a theory which can uniformly account for names in all of their literal occurrences. By putting direct pressure on the uniformity argument in favour of predicativism, however, the cases of course also put indirect pressure on predicativism as such. After all, the uniformity of predicativism was supposed to be the main advantage over referentialism.

Can predicativists respond to Jeshion’s cases? The cases directly attack the claim that it is the applicability conditions yielded by the schema (BCC) which allow for uniform semantics for all occurrences of names. Can predicativists retract from this schema and replace it with a more general one which does allow for a uniform treatment not only of (1) and (2) but also of (6)–(9)? Perhaps, Jeshion suggests, names could be regarded as not only applying to their bearers but also to individuals who stand in some contextually specified relation to these bearers. Allowing this relation to be ‘x is a relative of y’ might account for dynasty cases, for instance, and letting it be ‘x is produced by y’ might account for artefact cases. While Jeshion points to some problems with this strategy, it doesn’t seem to be off the table. However, once predicativists renounce (BCC), the onus is on them to spell out the details of such an alternative account and then to argue that this alternative account is still more uniform than referentialism. In any case, concludes Jeshion, the uniformity argument does not support predicativism as it stands.

In her response, ‘Literal’ Uses of Proper Names’ (Chapter 12) Fara disagrees with this assessment. She wants to reserve the right to regard predicativism as providing a more unified semantic theory of names than referentialism while also sticking to her claim that it is (BCC), not some other schema, which accounts for this uniformity.

Fara’s defence of the uniformity argument against Jeshion’s cases is not itself uniform. Dynasty cases receive one treatment, artefact, resemblance and costume cases another. With regard to dynasty cases, the strategy Fara pursues in the present article is this. She accepts that when the expression ‘Romanov’ occurs in a dynasty case, it is used literally. But she maintains that it is not in fact used as a proper name. If this is right, then dynasty cases can’t embarrass (BCC) which predicativists intend to apply to proper names only. If ‘Romanov’ is not a proper name in (9), then what is it? According to Fara, it functions as a proper noun, in the same way as, says Fara, ‘Les Paul’ and ‘Flying V’ function as proper nouns, but not as proper names, in an utterance of (10) which can be true if the individuals involved play guitars of the appropriate models:

  • (10) I’m a Les Paul but I’m married to a Flying V.

Fara has since developed her treatment of dynasty cases elsewhere (Fara 2016, Unpublished Manuscript). I won’t go into this part of her general defence any further here. Instead I will focus on her treatment of artefact, resemblance and costume cases.

With regard to these, Fara points out that analogous cases can be produced which have bona fide predicates where Jeshion’s cases have names occurring in apparently predicative position:

Resemblance Cases: The following can be uttered truly if Lena’s children bear ‘a striking behavioural resemblance to kittens’ (266):

  • (11) Lena arrived with two little kittens in tow.

Artefact Cases (266): The following can be uttered truly if there are five paintings produced by gorillas and two produced by orang-utans at the primate art museum.

  • (12) There are five gorillas and two orang-utans at the museum.

Costume Cases (256): The following can be uttered truly if there were two guests dressed as ballerinas at the party.

  • (13) There were two ballerinas at the party.

Fara takes this to show that predicativists can unproblematically accept that (BCC) does not cover the original cases involving names. She reasons as follows: The sentence (12), for instance, contains a predicate that normally applies to an individual just in case it is a gorilla. But this applicability condition does not account for the particular occurrence of the predicate in (12). Yet this does not force us to revise the standard applicability condition of ‘gorilla’. But then the original case (9) should no more be taken to force predicativists to revise the application condition that (BCC) yields for ‘Stella’.

According to Fara, rather than taking (12) as a reason to revise our standard application condition for ‘gorilla’, we should regard this sentence as involving an instance of what has been called ‘deferred interpretation’ (see Nunberg 2004). Nunberg’s canonical example for deferred interpretation is this. A waiter can truly say

  • (14) The ham sandwich is sitting at table seven

when the person who ordered the sandwich is sitting at table seven. On Nunberg’s account, the predicative part of the definite description is subject to a deferred interpretation. On this interpretation it is not true of ham sandwiches but of persons who ordered ham sandwiches. In general, when a predicate receives a deferred interpretation there is a ‘salient function’ between the things satisfying the deferred applicability condition and those satisfying the normal applicability condition. In the ham-sandwich case, the function maps customers who ordered ham sandwiches to ham sandwiches. In the gorilla case, says Fara, the function maps paintings created by gorillas to gorillas. Fara’s point is that the predicativists should apply the same idea to the original case (9). The predicatively occurring name ‘Stella’ in (9), like the bona fide predicates in (12) and (13), is subject to a deferred interpretation. In the case of (9) the salient function maps artworks created by individuals called Stella to individuals called Stella. She thinks the same account applies to costume cases, and a similar one to resemblance cases.

In her ‘A rejoinder to Fara’ (Chapter 13) Jeshion, among other things, responds to Fara’s appeal to deferred interpretation as a means of dealing with the original problem cases. The gist of this response is that Fara’s appeal to deferred interpretation would make for an entirely satisfactory treatment of the problem cases if these had been put forward as mere counterexamples to (BCC). But this wasn’t quite meant to be the challenge. Rather, the challenge to the predicativists was meant to be this: account for the problem cases and convince me that the resulting account leaves your argument from uniformity intact. And this challenge, thinks Jeshion, hasn’t yet been fully met.

I am sympathetic to this assessment. In the remainder of this review, I would like to sketch why. While I take the following line of thought to be very much in the spirit of both Jeshion’s overall argument as well as Napoli’s contribution, ‘Names as Predicates?’ (Chapter 10), I don’t mean to ascribe it to either of them in exactly the form in which I will put it. Imagine someone objected to Fara’s appeal to deferred interpretation by refusing to distinguish between a normal applicability condition of ‘gorilla’ and a non-normal one which it receives in (12). Rather, the objector insists there is a single applicability condition for ‘gorilla’ and (12) shows that this condition is met by certain paintings. The objector might cite uniformity in favour of her view. She makes do with a uniform application condition, Fara needs normal application conditions and she additionally needs non-normal ones.

This objection seems unconvincing. Maybe the objector’s theory is more uniform. But this uniformity shouldn’t be regarded as a virtue. It handles cases uniformly which shouldn’t be handled uniformly. For there is a reasonably clear line to be drawn between normal usages of ‘gorilla’ (where it applies to gorillas) and non-normal usages (where it applies to paintings produced by a gorilla). An appropriate theory should reflect this difference.

If the objector claims not to see the difference the predicativist can explicate it as follows. When we distinguish between normal and non-normal usages of ‘gorilla’ we don’t mean to distinguish between statistically more and less prevalent usages. That distinction might be co-extensional with the one we are drawing (as it would seem to be in the present case), but it is not a statistical distinction we are trying to get at. What we mean is something like this. There is an asymmetry between the usages of ‘gorilla’ where it applies to gorillas and usages where it applies to paintings by a gorilla in that the former usages are, in some sense, prior to the latter.

One way to illustrate the asymmetry is this: the existence of the former usages (partly) explains the existence of the latter usages. It is because ‘gorilla’ can be used to apply to gorillas that it can also be used, in certain contexts, to apply to paintings made by a gorilla. The converse is false. It is not the case that we can use ‘gorilla’ to apply to gorillas because we can use it to apply to paintings made by a gorilla. A related way to get at the asymmetry is this: we can imagine a language community in which ‘gorilla’ is only ever applied to gorillas. It doesn’t seem absurd to suppose that ‘gorilla’ in that language means the same as ’gorilla’ means in English. We can also imagine a language community in which ‘gorilla’ is only ever applied to paintings made by a gorilla. But it does seem absurd to suppose that ‘gorilla’ means the same in that language as it does in English.

What this shows is that the predicativists can justify their appeal to deferred interpretation, with its distinction between normal and non-normal application conditions, in their defence of (BCC) against Jeshion’s original cases. But in providing this justification they acknowledge that the more uniform semantic theory is not always the better. In particular, they acknowledge that when there is an asymmetry between two types of usages of an expression, in that one type of usage is prior to the other in the way described, then it is legitimate, and even desirable, for a semantic theory to reflect this asymmetry by treating these usages in a non-uniform fashion. In the hands of the predicativists, however, this concession now backfires when we reconsider the larger dialectical situation between predicativism and referentialism.

For recall that predicativists take their theory to be superior to referentialism because predicativism can, while referentialism cannot, handle apparently referential usages and apparently predicative usages uniformly. But referentialists can now respond to the predicativists in a way that is very similar to that in which we imagined predicativists to respond to the above objection. That is, referentialists can say that predicativism provides a uniform treatment of two types of usages of names which shouldn’t be treated uniformly, and that by the same standards that the predicativists appealed to in their justification of appealing to deferred interpretation. For referentialists can point out that there is an asymmetry between apparently referential and apparently predicative usages of names in that the former are, in some sense, prior to the latter.

One way to illustrate the asymmetry is this: we can imagine a language community in which ‘Alfred’ is only ever used in apparently referential position to talk about a particular Alfred. Provided that this is the same person that we actually talk about when using ‘Alfred’, it doesn’t seem absurd to suppose that ‘Alfred’ in that language means the same as ’Alfred’ means in our language. But we cannot say the same about apparently predicative usages. This time, it doesn’t even make much sense to suppose that there is a language community in which ‘Alfred’ is only ever used predicatively. At least not if predicativists are right that the application condition of ‘Alfred’ is provided by (BCC). As Napoli stresses in his article (213–6), whether an expression counts as a name depends on whether it used as a name. But for Alfred to acquire the name ‘Alfred’ mustn’t there be at least some utterances where ‘Alfred’ is used in apparently referential position to talk about Alfred? If so, then since these occurrences are banned from the language community in question, no one would have the name ‘Alfred’ (or any other name for that matter). But then, given (BCC), ‘Alfred’ (and any other name-predicate) would not apply to any individual. Accordingly, any assertion of a sentence involving an apparently predicative usage of ‘Alfred’ would be pointless: it would be either trivially true or trivially false (Napoli makes a point very similar to this, see (218)). In robbing a hypothetical language community of the ability to use names in an apparently referential way, we would also rob them of any reason to use them in an apparently predicative way. In contrast, robbing them of the ability to use names in an apparently predicative way does not undermine the point of using names in apparently referential position.

Relatedly, this consideration suggests that the existence of apparently referential usages of names (partly) explains the existence of apparently predicative usages of names, but not the other way around.

It thus seems that referentialists can make a plausible case that apparently referential usages of names are prior to apparently predicative usages in a way very similar to the way in which predicativists had to acknowledge normal uses of predicates to be prior to deferred interpretation usages. And predicativists themselves have to maintain that this priority justifies a non-uniform treatment. They cannot, then, hold the referentialists’ non-uniform treatment of names in apparently referential and apparently predicative positions against referentialism.

Of course, there are multiple ways in which predicativists can respond to this argument. Maybe they can justify their reliance on the deferred interpretation machinery in response to Jeshion’s problem cases in a way different from the one envisaged above. And perhaps such an alternative justification doesn’t undermine their uniformity argument against referentialism. Or maybe they can provide an alternative account of the application conditions of names which, unlike an account built around (BCC), allows them to deal with these cases without having to appeal to deferred interpretation in the first place. But either way, there is more work to be done before predicativists can safely claim their theory to be superior to referentialism because it provides a more uniform semantics for names.

1 For Fara’s reasons for dropping the quotation marks around the second occurrence of the schematic letter see her 2011a.


You can call me ‘stupid’ … just don’t call me stupid
Names are predicates
Philosophical Review
Semantic Relationism
Comments on Scott Soames’ ‘Coordination Problems
Philosophy and Phenomenological Research
Sematics: cross-cultural style
. The pragmatics of deferred interpretation. In
The Handbook of Pragmatics
 , eds.