Consequences of Reference Failure

Michael McKinsey

• Publisher: Routledge
• Year: 2019
• ISBN:
  9781000751772
• Edition: 1
• Topic: Philosophie
• Subtopic: Langue en philosophie

This book defends the Direct Reference (DR) thesis in philosophy of language regarding proper names and indexical pronouns. It uniquely draws out the significant consequences of DR when it is conjoined with the fact that these singular terms sometimes fail to refer.

Even though DR is widely endorsed by philosophers of language, many philosophically important and radically controversial consequences of the thesis have gone largely unexplored. This book makes an important contribution to the DR literature by explicitly addressing the consequences that follow from DR regarding failure of reference. Michael McKinsey argues that only a form of neutral free logic can capture a revised concept of logical truth that is consistent with the fact that any sentence of any form that contains a directly referring genuine term can fail to be either true or false on interpretations where that term fails to refer. He also explains how it is possible for there to be true (or false) sentences that contain non-referring names, even though this possibility seems inconsistent with DR.

Consequences of Reference Failure will be of interest to philosophers of language and logic and linguists working on Direct Reference.

1 Direct Reference and Descriptivism

History and Problems

According to the Direct Reference thesis, or DR for short, the proper names and indexical pronouns of natural language are what Bertrand Russell called “names in the logical sense” (Russell 1918, 201). Any term of this sort, which I will call a genuine term, is a term whose sole semantic contribution to the propositions expressed by sentences containing the term is simply the term’s semantic referent.¹ This idea, I take it, is equivalent to the idea that the proposition expressed by a sentence containing a genuine term is a strict function of the term’s semantic referent.² Following standard practice, I will call the propositions expressed by sentences containing genuine terms ‘singular propositions’.

Like many others, I endorse DR. There in fact seems to have been a consensus among philosophers of language for the past four decades or so that DR is true. In my own case – and I suspect in the case of many others – the main motivation for endorsing DR lies in the modal considerations first raised by John Searle (1958) and later clarified and forcefully applied by Saul Kripke (1972a), which show that proper names do not have the meanings of contingent definite descriptions. (A definite description in English is a term of the form ‘The F’, where F is a simple or complex predicate. A contingent definite description is one like ‘the inventor of bifocals’, which an object can satisfy in one possible world and yet fail to satisfy in another possible world.) Similar related points were raised by David Kaplan (1977), which persuasively support the conclusion that like names, indexical and demonstrative pronouns are also genuine terms.³

Historically, the main competitor of DR is the view that proper names and indexical pronouns have the meanings of or are abbreviations for definite descriptions. I’ll call this view ‘Descriptivism’. As we shall see, Kripke’s modal arguments don’t by themselves refute Descriptivism regarding proper names. Additional arguments are required. Moreover, even if in the end we think, as I believe we should, that Descriptivism is false, we should not assume that it follows that DR is true. Many philosophers, including Kripke, who believe that Descriptivism is false, are agnostic regarding the truth of DR. (This in some cases may be because they are agnostic about the existence of abstract singular propositions, to which DR is committed.)

I have two primary goals in this monograph. First, I wish to show in some detail that DR has important but rarely discussed consequences regarding the notions of logical truth and valid inference. These consequences follow from the fact that genuine terms (proper names and indexical pronouns) sometimes fail to refer, are ‘empty’. Second, I wish to suggest and defend an important qualification to DR, on which a small subclass of proper names that (following Gareth Evans’s (1979) terminology) I will call ‘descriptive names’, are allowed (by an idiomatic convention of natural languages) to be used as short for descriptions in a small range of sentential context-types. This suggestion will allow me to solve both the substitution problem for DR that Frege (1892a) discovered (see this chapter, Section 3) as well as the problem for DR emphasized by Russell (1918) that in some contexts, sentences containing empty names can nevertheless be true or false. (See this chapter, Section 5.)

In Chapter 2, I will argue that the DR-thesis implies that no sentence of natural language that contains a non-referring genuine term (used as a genuine term) can have a truth value, that is, can be either true or false. It follows that the only type of first-order logic that can be correctly applied to (the first-order fragments of) natural languages is a neutral free logic on which no sentence that contains any individual constant (proper name) is a logical truth.

In Chapter 3, I will describe and defend a formal semantics for neutral-free logic that is largely derived from the neutral free semantics proposed by Scott Lehmann (1994, 2001, 2002), but that differs from his semantics in its treatment of the quantifiers. Also in Chapter 3, I will propose a formal semantics for quantified modal logic that is based on my neutral free logic. It will turn out that many classical logical truths that contain individual constants (such as truth functional tautologies), while they cease to be logical truths on the modal semantics, re-emerge as a posteriori necessary truths under interpretations where the contained constants all refer.

In Chapter 4, I will defend my approach to reference failure by considering and criticizing a range of alternative approaches, some of which are proposed by other defenders of DR, while others are proposed by defenders of Meinongian semantics and defenders of substitutional quantification.

Finally in Chapter 5, I will propose and defend an account, consistent with a slightly qualified version of DR, of how there can be sentences of specific types that are either true or false, even though they contain non-referring names. The account will also provide part of a solution to the second major classical problem facing DR, the problem raised by the apparent fact that substitution of co-referring names in cognitive contexts fails to preserve truth value. The existing literature on these two problems for DR is truly huge, but in my view, none of the proposed solutions so far (including one of my own) succeed.

In the remainder of this chapter, I will discuss the history of the major literature concerning the semantics of proper names and indexical pronouns by defenders of both DR and Descriptivism.

1. Mill and Russell on Proper Names

The idea that proper names contribute precisely their referents to what is said by use of the sentences that contain the names could reasonably be taken as a piece of common sense. The idea has appeared in philosophical works as early as Plato’s Cratylus. In more modern times, John Stuart Mill, in his System of Logic (1843), provided a plausible and thorough account of the semantics of singular terms (proper names and definite descriptions) and general terms (simple and complex predicates). It’s clear I think that Mill would have endorsed DR as I’ve expressed it. He held that proper names have denotations (referents) but no connotations (semantic contents consisting of sets of attributes or properties), while general terms have both denotations and connotations. The denotation of a general term is the class of objects that satisfy the term’s connotation. Mill also held that definite descriptions have both denotations and connotations, as long as there is a unique object that satisfies the description’s connotation.

In his early work, Principles of Mathematics (1903), Russell discussed in some detail the semantics of the declarative sentences of natural languages as well as the constituents of such sentences, including singular terms (proper names and definite descriptions), predicates, and quantifier phrases. The fundamental concept used by Russell in this discussion is that of the proposition. Russell thought of propositions as abstract structured complexes that are the meanings of declarative sentences and whose structures and constituents correspond to the structures and constituents of the sentences that express the propositions. Each constituent of a proposition is also said to be the meaning of the corresponding constituent of a sentence that expresses the proposition.

Russell calls the meanings of proper names and definite descriptions terms, which are the objects or things that the propositions containing them are about. What counts as a term for Russell is anything that can be meaningfully mentioned, whether or not the thing exists. He held at this point of his career that whatever can be mentioned has being, even though it may not exist:

Numbers, the Homeric gods, relations, chimeras, and four-dimensional spaces all have being, for if they were not entities of a kind, we could make no propositions about them. Thus being is a general attribute of everything, and to mention anything is to show that it is.

(Russell 1903, 449)

The meanings of predicates Russell calls non-denoting concepts, where such a concept is a property or relation that is predicated or asserted of a term or terms by a given proposition containing the term or terms and the concept. Finally there are the meanings of quantifier phrases such as those of the forms every F, all F, any F, an F, some F, and the F, meanings which Russell calls denoting concepts.⁴ Denoting concepts have denotations, but those denotations are not always constituents of the propositions expressed by use of these quantifier phrases.⁵

It is clear, I think, that Russell’s semantic views in Principles logically imply the thesis of Direct Reference (DR). This may not be completely obvious since he calls the terms or constituents contributed to propositions by proper names, the names’ meanings. But note that these contributed terms are also always the things or objects that the names refer to, or denote (in Mill’s terminology). A similar issue arises concerning the properties and relations contributed to propositions by predicates. Russell says that such properties and relations are the meanings of the predicates. Would he also have allowed these properties and relations to be called the referents of the predicates? Perhaps, but perhaps not. I myself would prefer to say that predicates express or predicate properties and relations.

Russell’s (1903) concept of a proposition’s being an abstract complex entity that has the constituents and structure contributed by a corresponding sentence has had a revival among contemporary defenders of DR. This revival is largely due to the influence of David Kaplan, who endorsed the utility of Russell’s concept in his seminal work on the semantics of indexical and demonstrative pronouns. [See Kaplan (1977, 194–197). Other prominent defenders of Russell’s concept include Braun (1993), Salmon (1986), and Soames (1987)]. However, I should note that modern defenders of Russell’s model reject the metaphysical extravagance of The Principles (as Russell eventually came to do as well), and they require that propositions and their constituents must exist.

There’s no doubt that Russell’s concept of a proposition has heuristic value. However, the theoretical utility of the concept may be limited. Representing the unique structure and constituents of the proposition expressed by a simple atomic sentence such as ‘Socrates is wise’ seems straightforward. We could do it this way:

• <{Socrates}, [the property of being wise]>

Here, the curly brackets represent the proposition’s subject position, while the square brackets represent the proposition’s predicate position. We could then say that the proposition’s structure can be represented as: < {}, [] >. Here it seems that we’ve correctly represented the proposition as having a unique structure and a unique set of constituents.

However, as Frege (1892b, 49) pointed out, it often happens that two different sentences that contribute different structures or constituents can express the same thought (proposition). Frege mentions the active/passive distinction as an example. So compare:

(1a). John loves Mary.	< {*John, Mary*}, [the relation of loving] >
(1b). Mary is loved by John.	< {*Mary, John*}, [the relation of being loved by] >

(Here I use asterisks to represent ordered pairs.) Given that the names ‘John’ and ‘Mary’ contribute specific persons as referents, (1a) and (1b) clearly express the same proposition, even though the propositions are represented as having different constituents. The representations contain different ordered pairs in subject position and different relations in predicate position.

Here’s an example of two sentences that have different structures even though they express the same proposition:

1. (2a). John is bald, and John is tall.
   < CONJ(<{John}, [the property of being bald] >, <{John}, [the property of being tall] >) >
2. (2b). John is bald and tall.
   < {John}, [the property of being bald and tall] >

(Here, ‘CONJ’ represents conjunction.) (2a) and (2b) clearly have different structures. (2a) is a conjunction ...

Table of contents

• Cover
• Half Title
• Series
• Title
• Copyright
• Dedication
• Contents
• Acknowledgements
• 1 Direct Reference and Descriptivism: History and Problems
• 2 From Direct Reference to Free Logic
• 3 From Neutral Free Logic to a posteriori Necessity
• 4 Some Alternatives to Using Neutral Free Logic
• 5 Truths Containing Empty Names
• Bibliography
• Index