Guide to Chapter One

Guide to Chapter One of Gareth Evans'
The Varieties of Reference

Rick Grush (Department of Philosophy, UC San Diego)
Copyright © Rick Grush 1999, 2000, 2001, 2002
Copyright holder grants permission to distribute, quote, and reproduce for any non-commercial purposes provided authorship and copyright are clearly indicated.

Chapter 1: Frege

Table of Contents:

- Section 1.1 Introductory
- Section 1.2 Meaning (Bedeutung)
- Section 1.3 Empty singular terms: preliminary remarks
- Section 1.4 Sense: Preliminary Remarks
- Section 1.5 Sense and thought
- Section 1.6 Empty singular terms: sense without Meaning?
- Section 1.7 Empty singular terms: sense without referent
- Section 1.8 Interpretational semantics and truth theories
- Section 1.9 Conclusions

Section 1.1 Introductory

This section is quite straight-forward. In it, Evans rehearses certain basic features of Frege's semantic theory, features which will be examined in much greater detail in the remainder of the chapter.

Evans initially draws attention to three aspects of Frege's semantic project.

First, the theory is constructed against the background of certain putatively antiseptic regions of language — regions devoid of complications like indexicals and demonstratives — in part because such regions are more well-behaved, and also in part because of Frege's larger project of securing the foundations of mathematics and the fact that the artificial languages of mathematics are similarly antiseptic. Frege though clearly intended the theory so constructed to have wider application to natural language.

Second, the theory was systematic, in that it analyzed the significance of complex expressions, including sentences, as dependent upon, or a function of, the significance of the constituents of the complex expression — a process which presumably bottoms out at lexical items such as words and morphemes. (Though Evans points this out, he is clearly sensitive to the fact that this account of how the meaning of specific complex expressions is a function of their components is compatible with the claim that the meaning assigned to these components may be a holistic function of their contribution to the meaning of these complex expressions.)

These two aspects of Frege's account come together in the following way. The primary sort of theoretically pristine complex expression with which Frege is concerned is the atomic sentence. Such a sentence is analyzed as a construct of two kinds of expression: i) one or more singular terms, and ii) a predicate of the appropriate degree (where the degree of a predicate is just the number of argument places the predicate has). Thus a sentence such as 'The cat is on the mat' would have two singular terms, 'The cat', and 'the mat', and a predicate of degree two, '... is on ...'. The sentence 'Patrick Stewart is bald' consists of one singular term, and a predicate of degree one.

Third, Evans claims that Frege's semantic theorizing underwent two phases. The first phase was what might be called a theory of meaning. This theory assigned to each significant linguistic expression of a given grammatical category (so far: atomic sentences, singular terms, and predicates) a semantic value of a type appropriate to that category — the expression's 'meaning'. Frege's recognition that this approach had important limitations led him to supplement it with a theory of sense, which supplemented this assignation of meanings with an assignation to each meaningful expression of a language a sense. The theory of the second phase maintained much the same theory of meaning as before, but now had another level added to it, so to speak. Now each significant expression was assigned two kinds of semantic value: a meaning, as well as a sense.

Section 1.2 Meaning (Bedeutung)

Evans in this section gives a brief introduction to relevant aspects of Frege's theory of Meaning, as applied to extensional sentences.

The section has four parts, roughly corresponding to the four paragraphs.

First, Evans discerns three general features of Frege's theory. These are: i) The conviction that the semantic significance of sentences consists in their being true or false; ii) The conviction that for any significant expression (sentence, name, predicate, whatever), its significance (or semantic power) is expressed in the ability of the expression to affect the truth value of the sentence in which it is embedded; and iii) This significance is afforded by the expression's being assigned an extra-linguistic entity (what this means and how it occurs is the subject of much of what follows). (Note: Regarding (ii), don't be confused by the fact that I listed sentences as types of expression that might effect the truth value of a sentence in which it is embedded. There are often sentences embedded in sentences, such as I believe that the cat is on the mat, where the subordinate sentence the cat is on the mat is embedded in the matrix sentence; or roses are red and violets are blue, of which the two sentences roses are red and violets are blue are components. The structure of the theory will be such that a sentence has the same semantic power — a truth value — whether it is free-standing or a component of a larger sentence.)

The second part consists of the application of these basics to the semantic power of singular terms. Since, according to Frege, expressions have their semantic power in virtue of being associated with some extra-linguistic entity, and since singular terms, such as names, typically have referents (e.g. the name 'Patrick Stewart' has, as its referent, a certain actor most famous for his portrayal of Captain Jean-Luc Picard on a recent television series), the identification of their referent as the extra-linguistic entity that imbues singular terms with their semantic power is natural (though as we shall see not without difficulties).

This identification of a singular term's referent as the extra-linguistic entity that imbues the term with semantic power is insulated from certain objections by Frege's exclusive concern (at this stage of theorizing) with extensional contexts. Two different singular terms with the same referent will have the same semantic power if semantic power is a function of referents. But it can seem intuitively that the names 'Clark Kent' and 'Superman' do not mean the same thing, even though they have the same referent. This intuition is heightened by placing the names in modal or intentional sentences: the sentence 'Louis Lane believes that Superman can fly' is true, whereas the sentence 'Louis Lane believes that Clark Kent can fly' is false. But extensional contexts just are those where such substitutions of co-referring expressions do not alter the substituted-in sentence's truth value.

Evans points out that a semantic theory might make other choices for the semantic power of singular terms other than their referents. One consideration in favor of appealing to something other than referent is the fact that problems remain for such a choice even when attention is constrained to extensional fragments of language. Many empty singular terms (noun phrases that refer to nothing, such as Santa Claus or the present king of France) would be meaningless — without semantic power — on this theory, yet they clearly seem not to be meaningless.

The third part consists in the application of the basics as articulated in the first part to the semantic value of sentences. In brief, the conviction that the significance of a sentence consists in its being either true or false, together with the conviction that an expression's semantic power lay in its association with an extra-linguistic entity, means that the Truth Value (either The True or The False, conceived as Platonic objects of some sort), was a natural construction for the 'meaning' of sentences.

The fourth and final part of this section concerns the kind of semantic power of, and extra-linguistic entities assigned to, expressions of other types. With the semantic powers and extra-linguistic assignations of expressions of the grammatical categories SENTENCE (S) and SINGULAR TERM (N) fixed, the semantic power of all other categories of meaningful expressions can be derived. Grammatical categories can be expressed in the following notation. The basic (underived) grammatical categories of SENTENCE and SINGULAR TERM are designated S and N respectively; other grammatical categories are derived from these two: a one-place predicate, designated S/N, is something that yields an S when combined with an N; a two-place predicate, designated S/N,N yields an S when combined with two Ns. In general, a derived grammatical category is one that yields an grammatical category of type a when combined with some specific number of instances of type b. An S/S,S would be an expression which yields a sentence when combined with two sentences. This would be a binary sentential connective, such as 'and' or 'or': 'The cat is on the mat and Patrick Stewart is bald.'

This is a taxonomy of grammatical categories which takes the categories SENTENCE and SINGULAR TERM as basic and derives others from them. The kind of extra-linguistic entity assigned to expressions of the derived categories as their semantic value is a reflection of their grammatical role. Just as the grammatical role of a one-place predicate (S/N) is to form a sentence (S) when combined with a singular term (N), the extra-linguistic entity assigned to one-place predicates are functions that map objects to truth values. More generally and perspicuously, the semantic value assigned to a grammatical category a/b will be a function that yields semantic values appropriate to as upon 'input' of a semantic value appropriate to bs. (This obviously generalizes to cases where the category calls for more than one instance of the type b.) Where a is S (to which are assigned Truth Values) and b is N (to which are assigned objects) — an S/N — the assigned extra-linguistic entity is a function from objects to truth values.

[Note: It should be apparent that the categorial descriptions N (singular terms) and S/(S/N) (quantifier expressions) are similar, in that both are grammatical categories whose semantic value is something that, in combination with an S/N (a predicate), provides a Truth Value. For instance, 'Patrick Stewart is bald' and 'Some actor is bald'. The difference is how they arrive at this semantic value.

In the first case, at the grammatical level we combine an N and an S/N. At the semantic level we are combining an object with a function from objects to truth values. This function produces a truth value — an extra-linguistic entity appropriate to Ss.

In the second case at the grammatical level we are combining an S/(S/N) — a quantifier — and an S/N — a predicate. At the semantic level, we are combining (i) a function from objects to truth values (the semantic value of a predicate), with (ii) a function from functions from objects to truth values, to truth values — that is, a higher level function that takes as input a lower-level function and produces as output a truth value.]

One of Evans' primary goals in this section is to distinguish, from within a broadly Fregean framework, the terms 'referent' and 'Meaning' (using the capital letter to mark the Fregean use of the term). The distinction, which will be discussed more in the following sections, is this: Meaning will be whatever extra-linguistic entity is assigned to an expression as its semantic value. Referent will be one possible choice for Meaning (for singular terms), functions might be another (e.g. for predicates). Evans will later argue that one can remain broadly Fregean by choosing, as the extra-linguistic Meaning for some singular terms, things other than their referents.

Section 1.3 Empty singular terms: preliminary remarks

This section is entirely straight-forward. In it, Evans recounts the problems that the sort of semantic theory outlined in the previous section faces when the singular terms lack a referent, aka empty singular terms. Evans then shows that Frege felt able to address these problems with his theory of sense, which was developed for independent reasons as a supplement to the theory of Meaning.

Evans begins by recounting how the sort of semantic theory outlined in the previous section is supposed to work for an atomic sentence, such as 'John is wise.' First, the singular term 'John' has as its semantic value an object: its referent, John. The predicate 'x is wise' has, as its Meaning, a function from objects to truth values, in this case a function which maps all and only wise objects on to the truth value True, and maps all others to False. This mapping determines the Meaning of the sentence as True (if in fact John is wise) or False (if John is not wise). In this way the meaning — the Truth value — of the sentence is secured.

There are two wrenches which might cause this semantic engine to misfire: first, the singular term may fail to be assigned an object (an empty singular term); and second, the function that is the semantic value of the predicate may not be defined over all objects (specifically, the object assigned to the singular term). In either case, the sentence in question will fail to have a Meaning secured. Evans says that before Frege developed his theory of Sense, he accepted this consequence. Thus Frege accepted that atomic sentences such as 'The table in the living room belongs to my grandmother' are strictly meaningless if there is nothing answering to the expression 'the table in the living room'. This view of matters is today more commonly associated with Russell, and Evans will discuss this more in the next section. For now, note the terminological point that a Russellian Singular Term will be a singular term whose significance depends on its having a referent — that is, a term such that, if empty, any extensional sentence using it as the grammatical subject will not have a Meaning (or: will be unfit for the expression of a thought).

Evans then briefly discusses Frege's theory of Sense, and claims that while the theoretical motivation for the supplementation of the theory of Meaning with the theory of Sense was not the problem posed empty singular terms, the theory seemed to Frege to be able to help with that problem. While the theory of Meaning was concerned with the Meaning of linguistic expressions, the theory of Sense was developed to account for features of language users' understanding of expressions of their language. Of particular interest was the potential informativeness of identity statements. According to the theory of Meaning outlined in the previous section, there is no difference of Meaning between the following two statements:

1) Cicero is Cicero.

2) Cicero is Tully.

Each sentence, according to the theory of Meaning, invokes the same object twice, and presents them to a function, the identity function, that returns the value True if the same object is presented to it by both singular terms, and returns False if different objects are presented to it. Thus at the level of Meaning (1) and (2) are identical. But clearly (2) can be informative, though (1) cannot be. Because the theory of Meaning fails to discern any difference between (1) and (2), Frege felt it necessary to supplement the theory of Meaning with a theory of Sense. The theory of Sense allowed for there to be a difference between expressions which had the same Meaning, by assigning to each expression not only a Meaning but also a Sense, such that two expressions with the same Meaning might nevertheless be assigned different Senses. In the case of sentences, their Meanings are Truth Values, but their Senses are thoughts. When the thought expressed by a sentence is understood to be a function of the Senses of its component expressions rather then merely their referents, theoretical room is thereby made for (1) and (2) to express different thoughts, even though strictly speaking they are identical at the level of Meaning. The notion of Sense thus makes available theoretical room to account for the different understandings of (1) and (2) that competent language users will have. Of course, a theory of Sense remains to be worked out.

Since every meaningful expression is to be assigned not only a Meaning but also a Sense, Evans points out that Frege felt able to address the problem of seemingly significant sentences with empty singular terms as their subject (or any of the predicate's arguments) by claiming that such singular terms, while lacking a referent, nevertheless might have a Sense, and this explains the (seeming?) significance of such sentences. Evans closes this section by foreshadowing the topic of section 1.5, which is an investigation into whether or not Frege's account allows expressions without a Meaning to have a Sense.

Section 1.4 Sense: preliminary remarks

This section is a brief introduction to what Fregean senses are, including both some remarks on how Frege tried to explain his notion of sense, as well as the beginnings of a somewhat cleaner proposal by Evans.

Frege used a number of metaphors to illustrate what a sense was. He talked of senses as a sort of particular viewpoint on the Meaning. He also talked about a sense as a way of presenting the referent, or as 'illuminating' different aspects of the referent. These 'ways of presenting' the Meaning would be objective, in that anyone who was sufficiently familiar with the language would not only know the Meaning of the expression, but would know the proprietary way in which a given expression 'presented' that Meaning.

Evans suggests that it will be more useful to analyze senses as ways of thinking of the Meanings. That is, given that when one thinks of something, one will be thinking of it in a particular way, these 'ways of thinking' of Meanings can be equated with senses — provided, of course, that these ways of thinking are conventionally associated with expressions in the language. So someone who thinks of the planet Venus as the object which shines most brightly in the morning sky, will be thinking of the same object, but in a different way, when thinking of it as the object which shines most brightly in the evening sky. The first way of thinking might be conventionally associated with some expressions, such as 'The Morning Star' or 'Phosphorus', while the second way of thinking of Venus might be conventionally associated with 'The Evening Star' or 'Hesperus'.

[Notice that I just cashed out senses as sorts of descriptions, such as the object which shines most brightly in the morning sky. This is one way to understand a sense on Evans' view — thinking of and object as the object which uniquely satisfies the description is a way of thinking of that object (though this is not a sense of a singular term, since descriptions are complex quantifiers and not singular terms). But Evans will, beginning in Chapter 6, explore other ways of thinking of objects.]

Evans then briefly explains that similar considerations apply not only to singular terms, but to any expression assigned a Meaning by the semantic theory. So for example, the function assigned to the predicate '... is a creature with a heart' might map exactly the same objects onto True as the function assigned to the predicate '... creature with kidneys', and so would be the same function (functions are fully determined by their mappings, and these functions have the same mapping, and hence are the same function) — thus these predicates have the same Meaning. In understanding either expression, one would be thinking of this function, but one would be thinking of it in different ways in the two cases. These would correspond to the different senses of the two predicates.

Evans closes with the following passage (p. 17):

"To summarize, then, I suggest we take Frege's ascription of a sense to a Proper Name to mean that not only must one think of an object — the referent of the term — in order to understand a sentence containing it, but also anyone who is to understand the sentence must think of the referent in the same particular way. It is therefore, for Frege, as much a public and objective property of a term that it imposes this requirement, as that is has such and such an object as its referent."

Section 1.5 Sense and thought

This section consists of three parts. The first introduces Frege's Intuitive Criterion of Difference, which is a constraint on the individuation of senses. The second part introduces Evans' own proposal for unpacking the notion of sense and answers an objection. The section concludes with a reconstruction of the Fregean model of the communication of thoughts via language that puts this notion of sense to use.

The first part of this section, running from the top of p. 18 to the bottom of p. 20, discusses Frege's one criterion concerning the identity of senses. Since Frege's purpose for the introduction of the notion of sense was to account for a cognitive phenomenon (the informativeness of identity statements), it is incumbent upon Frege to somehow link senses to what Evans calls propositional attitude psychology — that is, he must link senses to cognition, and in particular to the understanding that competent language users have of expressions of their language. Frege makes this connection directly, by making a cognitive/psychological test criterial for the individuation of senses. Evans calls this test the Intuitive Criterion of Difference. Evans states this criterion in the following way, which restricts its immediate application to one particular type of sense, thoughts (recall thoughts are the senses of sentences):

"The thought associated with one sentence S as its sense must be different from the thought associated with another sentence S' as its sense, if it is possible for someone to understand both sentences at a given time while coherently taking different attitudes towards them, i.e. accepting (rejecting) one while rejecting (accepting), or being agnostic about, the other." p. 18-19

[Note: It is possible, though Evans does not explicitly do this here, to generalize this criterion to expressions of any grammatical category by exploiting the fact that the sense of a sentence is a function of the senses of the constituents, in a way exactly analogous to the way in which the Meaning of a sentence is a function of the Meaning of its constituents. The generalized version would be:

If two extensional sentences P and Q are identical except for the fact that P employs the expression 'a' where Q employs the expression 'b', and if it is possible for a rational and fluent speaker of the language in which P and Q are expressed to judge one to be true, while at the same time judging the other to be false or to be uncertain of the truth of the other, then 'a' and 'b' must have distinct senses.

The criterion Evans explicitly discusses is obviously a special case of the second, where 'a' is the entire sentence P and 'b' is the entire sentence Q.]

In the second part of this section, from the bottom half of p. 20 to the bottom of p. 21, Evans then turns to his own formulation of sense as a way of thinking of the referent. Evans' proposal is that a 'way of thinking of a' can be given by an account of what makes it the case that a subject S is thinking about the object a. This account will take a form such as "S is thinking about a in virtue of the fact that ... S ...". What follows the 'that' in this formulation will be some account which makes reference to both the subject S and the object a. Evans then says that another subject S' will be thinking of a in the same way as S if we get a correct account of why S' is thinking of a by simply replacing S with S' in the original formulation: i.e. "S' is thinking about a in virtue of the fact that ... S' ...".

[Note: One way of filling this out would be 'S is thinking about a in virtue of the fact that S is entertaining definite description D and predicating something of the object that uniquely satisfies that description, and object a is the object that in fact uniquely satisfies D'. But suppose we fill it out in the following way: 'S is thinking about a in virtue of the fact that S is entertaining a definite description and predicating something of the object that uniquely satisfies that description, and object a is the object that in fact uniquely satisfies the description S is entertaining'. In this case, on Evans' proposal, S and S' would be thinking of a in the same way, provided S' were simply thinking about it by description — even when S is using description D and S' is using a different description E. In a fuller account, we would want some way to eliminate this undesirable possibility. Evans touches on this sort of case at p. 20, but it is not clear that he manages to eliminate this difficulty.]

Evans then counters an objection to this proposal. The objection is:

1. Sense determines referent. (This is agreed on all sides.)

2. On your (Evans') account, 2 people at different locations thinking of their surroundings in the same way (as 'here') would be employing or grasping the same sense.

3. But then their senses would not determine a referent, because the same sense would lead to, so to speak, two different referents.

4. Therefore, your account cannot be correct.

Evans' answer is to deny (2), because on his proposal, one determines whether S and S' are thinking of a in the same way by substituting only occurrences of S with S', without substituting the names of the objects thought about (a' for a, in this case, the names for the two different locations). That is, in order to apply Evans' criterion to S and S' in the way required by the above objection, one would not only have to replace S with S', but also replace a (where S is) with a' (where S' is). But Evans' proposal does not allow one to replace the subjects AND the objects.

Note that because Evans' criterion requires that, in order for it to be applied, the formulation must retain everything except the substitution of S' for S, his criterion will have the result that sense determines referent.

Note also that while the Intuitive Criterion of Difference applies only to senses entertained by a single subject at a particular time, Evans criterion applies only to two different subjects, possibly at different times. Not also that Frege's criterion applies primarily to linguistic expressions, while Evans' applies to thoughts, regardless of their expression in language. It is not clear how Evans understands the relation between Frege's criterion and his own. They are not incompatible, and presumably he thinks each by itself is part of a complete account of sense, but that even together they are not the whole story.

In the final and third part, from the bottom of p. 21 to the end of the section, Evans closes with a recap of the Fregean model of communication which one arrives at with these ideas.

Section 1.6 Empty singular terms: sense without Meaning?

This section is rather unclear. The problem of significant or apparently significant atomic sentences employing empty singular terms is used to put pressure on Frege's semantic theory, and to suggest avenues of improvement.

From the middle of p. 22 to the bottom of p. 23, Evans points out that the problem posed by empty singular terms is not helped by the notion of Sense as Frege (or Evans for that matter) understands it. The first problem is that on both Frege's and Evans' criteria of sense identification, sense requires referent. Frege's metaphors of 'presenting the referent' in different ways, or 'illuminating' the referent differently, evidently require that there be a referent to be presented or illuminated, and Evans' formulation requires that there be an object thought about by the subjects S and S'.

Evans then recasts his objection slightly, pointing out that the core issue that renders empty singular terms problematic is Frege's assigning of extra-linguistic entities to expressions as their semantic value. The point here is somewhat obscure, and I think not expressed well. There is reason to think that Evans' real objection is not the assignation of extra-linguistic entities as semantic values per se, but of the monolithic nature of Frege's theory — monolithic in that it assumes that semantic categories are fully determined by grammatical categories, and in particular that the grammatical category of singular term (S/N) corresponds to a single semantic category such that all members of that grammatical category receive identical semantic analysis: assignation of the referent as the Meaning. The Varieties of Reference, as the title suggests and as the Introduction makes clear, will be concerned with the different kinds of semantic analysis that different kinds of singular term require.

The argument here is quite unclear for another reason. Evans expresses it as an argument against the assignation of Truth Values as the semantic values of sentences, and suggests that this might need to be revised. However, Evans nowhere in his book explicitly suggests another option here, and he also clearly endorses both the idea that the significance of sentences consists in their capacity to convey thoughts, and that a thought (expressed as a belief) must have some truth value. So the problem cannot be that significant sentences can fail to be directly connected to some truth value.

Rather, Evans' point seems to be this: First, Frege's semantic theory is systematic, in that the significance (Meaning or sense) of sentences is a function the significance (Meaning or sense) of its components. With this Evans will have no quarrel. And significant sentences are those fit for conveying thoughts, and thoughts (expressive of beliefs) must either be true or not true. Again, no quarrel so far.

But Frege's theory of Meaning assigns to Sentences a Truth Value as their Meaning, and it explains how a sentence comes to have a Truth Value as a function of the Meanings of its components, the predicates and singular terms. Empty singular terms undermine the process by which Sentences get a Truth Value on Frege's account. Thus such sentences should, strictly speaking, be insignificant, yet Frege seems to maintain that such sentences can be significant. This move cannot be sustained by appeal to the level of Sense, as Frege sometimes seems to, for two reasons. First, all of the accounts of Sense on the table so far explain Sense in terms of referent in one way or another. If a singular term is empty, then it is not clear that it can have a Sense at all, let alone one capable of supplying significance to the matrix sentence. Second, if such an account could be made to work, an account according to which singular terms lacking a Meaning could have a Sense capable of making a sufficient contribution to matrix sentences such that those sentences are also significant, then the Theory of Meaning becomes an otiose appendage to the theory of Sense.

Though he does not do this here, Evans will want to make the following positive moves (knowing where Evans wants to go will help in interpreting his arguments in this section). First, within the grammatical category of singular term (S/N) different semantic categories must be discerned, such as Proper Names, demonstratives, descriptive names, and perhaps others. Different accounts of the significance of expressions of each type might be necessary, and in particular some of them may be significant even in absence of a referent (e.g. descriptive names, discussed in Section 1.7), while others will not (e.g. demonstratives, discussed in Chapter 6). In all cases, though, a detailed examination of their significance and in what it consists will be necessary. For those expressions whose significance depends on their having a referent, a Russellian line must be taken: where there is no referent, the sentence employing the expression fails to express a thought, and is strictly meaningless. Nevertheless, such sentences cannot simply be assimilated to pure nonsense, but rather an account must be given of how such sentences can have the appearance of significance, or can convey 'apparent thoughts'.

Section 1.7 Empty singular terms: sense without referent

In this short section, Evans does two things. First, he introduces an example of a kind of singular term which can contribute to meaningful singular sentences even when lacking a referent — the descriptive name. Second, he summarizes the argument of the chapter to this point.

Evans begins with a gloss on what he takes to be right about Frege's overall semantic theory. First, it recognizes a level of Sense in addition to a level of Meaning. Second, it is Russellian in that strictly speaking it holds that atomic sentences employing singular terms which themselves lack a Meaning are not significant. Third, it points in a direction to look for an explanation of why such sentences sometimes appear to be significant: the conveyance of apparent thoughts.

What is wrong about Frege's theory is that since he took it to be applicable to singular terms generally, and not just some subset of them, his semantic theory as it stood was not viable. This was because there are classes of referring expressions which, even if empty, contribute to meaningful singular sentences and thoughts. It might be thought that the best example here would be definite descriptions, but recall that definite descriptions are complex quantified expressions, and hence are grammatically of the form S/(S/N), and not S/N, that is they are, grammatically speaking, quantifiers rather than singular terms. Evans provides as an example of an expression of the type S/N which is significant even when lacking a referent the descriptive name (which will be treated in greater detail in Chapter 2). These are names whose referent, if it has one, is fixed by a descriptive reference-fixing stipulation, such as

3) Let us call whoever invented the zip Julius.

Evans claims that given a stipulation such as (3), sentences such as

4) Julius was an Englishman.

are perfectly meaningful, and can convey genuine thoughts even if, in fact, the name is empty (because, e.g. nobody invented the zip, or because it was discovered by accident, or devised by a computer program, or whatever). In Chapter Two Evans will argue at length that descriptive names are referring expressions (are of the form S/N), and also that even when empty atomic sentences employing them are fit to convey genuine thoughts, not merely apparent thoughts.

Evans summarizes the point he has been arguing for in this chapter with the following crucial passage from page 32:

"I have not argued that the entire structure of a Fregean semantic theory would break down if one insisted upon ascribing a Fregean sense to an empty singular term. What I have claimed is that dire consequences follow from the ascription of a Fregean sense to any expression that has no Meaning or semantic value; hence the consequences ensue for Frege because he selected, as appropriate semantic values for singular terms, the objects that would be regarded as their referents. But the equation between semantic value and referent is by no means mandatory. In fact it has to be given up if empty singular terms are to be ascribed a sense."

This is the key passage of this chapter for purposes of interpreting Evans' proposal. Clearly, Evans takes it that a broadly Fregean semantic theory is still viable, provided that, at least for some categories of singular term if not all, their semantic value is not identified with their referent. Singular terms which can be empty, and yet contribute to meaningful sentences, must have as their semantic values something other than their (purported) referents.

Evans closes by providing one example of such an alternate proposal: one can assign to singular terms not objects (referents) but sets, such that the set assigned to a singular term either contains exactly one object (the referent) or is empty (the empty set). On this proposal, all singular terms will have a semantic value, an extra-linguistic Meaning. It should be stressed that this is not a proposal Evans wants to endorse, but is merely an example of how one can choose entities other than referents to be the Meanings of singular terms. Evans' own proposal for various kinds of singular term will form the bulk of the book.

Section 1.8 Interpretational semantics and truth theories

This section is probably the most difficult of the chapter, and for many readers it will also be the most dispensable. The entire purpose is to head off an objection to Evans' enterprise to the effect that his treatment of the difference between Russellian and non-Russellian singular terms will be of interest only to theorists who adopt an interpretational semantic theory, but is not relevant to those interested in Davidsonian truth-theoretic semantic theories. Evans argues that even from a truth-theoretic approach, the difference between Russellian and non-Russellian singular terms is important. Those uninterested in this issue can safely skip this section — it will be of no consequence for the remainder of the book.

Evans begins by showing that the Fregean semantic theory he has been discussing is a species of interpretational semantic theory. Such a theory assigns to expressions of various semantic categories entities of some sort as their semantic value: both an assignation of a kind of entity as an appropriate interpretation for expressions of a particular semantic category, as well as the particular assignation of a given entity of this kind to a given expression of the semantic category. Evans then introduces another sort of semantic theory, Davidsonian truth theories. Evans' reason for doing this is not that he thinks such truth theories are preferable. Indeed, he argues that interpretational theories are needed even for those who engage in the truth-theoretic enterprise. Rather, he introduces truth theories here in order to show that even from within such theories, the distinctions Evans wants to draw between Russellian and non-Russellian terms needs to be drawn. The following passage makes Evans' intentions on this matter clear (p. 34):

"... the rather cumbersome theories in which entities are assigned to expressions as their semantic values seem ultimately unavoidable... However, it may help to increase the clarity and appeal of the points I have been making if I express them in terms provided by one of these more familiar and less cumbersome theories."

A typical truth theoretic clause of the type which might be used to derive truth conditions for a singular sentence would be:

(5) The referent of 'Aphla' is Aphla.

But a clause such as (6) will work as well for the derivation of the truth conditions for embedding sentences (i.e., nothing from within the Davidsonian semantic theory will favor (5) over (6)):

(6) The referent of 'Aphla' is Ateb.

Some of the truth-theoretic semantic theories will also be able to serve as theories of the sense of the terms, namely those using clauses such as (5). Thus for Russellian singular terms (singular terms that are significant only when they have a referent), such truth theories are capable of serving as theories of sense, and thus able to play a role in the understanding of the contribution of such expressions to sentences expressive of thought and belief. But given non-Russellian singular terms, axioms such as (5) and (6) will not work, simply because there will be nothing which follows the 'is' in the clauses.

So we need clauses such as (8) for descriptive names:

(8) (x) (the referent of 'a' = x iff x is j).

[as in the text, underlining indicates the scope, so to speak, of uniqueness, but unlike the text, I use bold to indicate which expression is the one for which uniqueness is being asserted.]

If it is allowed that there can be singular terms which lack a referent, yet are significant (non-Russellian singular terms), then it will be the case that complex sentences in which such terms are embedded may in fact be true. Therefore, it will be incumbent upon a theory of the application of the predicate 'true' (such as a Davidsonian theory) to accommodate such terms with clauses of some sort, possibly such as (8). With appropriate modifications to classical logic, non-Russellian singular terms can be used to express their own semantic contribution to embedding sentences in the way required by such truth theories, via clauses of a form like (8).

Thus Evans argues that a distinction between Russellian and non-Russellian singular terms is both required (to account for the truth of complex sentences embedding empty non-Russellian singular terms), and possible (given modification to the logic (e.g. free logic) and clauses such as (8).

Section 1.9 Conclusions

In this section Evans presents and clarifies the main results he wants to draw from this chapter.

Evans closes with the following conclusions:

1) Frege's theory of sense and reference forced him to be a Russellian about singular terms, at least most of them, whether or not he liked or recognized this fact.

2) Even if (1) is wrong, there was nothing to prevent Frege from recognizing Russellian singular terms.

3) Even granting (2), there was nothing to prevent Frege from recognizing non-Russellian singular terms.

Therefore there is the possibility of an eclectic Fregean theory, recognizing both Russellian and non-Russellian singular terms provided. This of course requires giving up a monolithic semantic theory for singular terms, and recognizing different semantic categories within that grammatical category. But Evans warns that as yet we have no idea what a Russellian thought could be, how a way of thinking of something could be such that in the absence of the object, there is no such thought to be had. (The move from Russellian singular term to Russellian thought will be clarified later.)

Evans then claims that even such an eclectic Fregean theory will not be able to handle all singular terms. Specifically, proper names will not fit into a broadly Fregean semantic theory. The reason is that proper names work, i.e. subserve communication, without it being the case that the interlocutors share a sense — that is, they can both understand the proper name correctly, even though they do not think of the referent in the same way, and thus could have thoughts distinguishable by the Intuitive Criterion of Difference. [Note that Evans here is making appeal to Frege's Intuitive Criterion of Difference, which is defined only for a single speaker, to a case involving two speakers. Presumably he has in mind a situation in which a single speaker entertains the two thoughts that are being entertained by the two speakers in the situation under discussion.]

So as examples we have (at least) these three semantic categories within the broader grammatical category of singular term, each requiring its own semantic theory:

a) Russellian singular terms (e.g. demonstratives and indexicals: this, that, here, there, I); Require a referent in order to be significant; theory of Sense is applicable.

b) Non-Russellian singular terms (e.g. descriptive names such as Julius); Do not require a referent in order to be significant; theory of Sense is applicable.

c) Proper names (e.g. Joey Buttafucco); require a referent in order to be significant; theory of Sense is not applicable.

[End of Guide to Chapter One]

Copyright holder grants permission to distribute, quote, and reproduce for any non-commercial purposes provided authorship and copyright are clearly indicated.