CHAPTER ONE

MEANING

PART ONE

Meaning and Reference: The Traditional Approach

When philosophers fight a raging battle over the question: ‘What should be a
theory of meaning?’, what exactly are they talking of? The words ‘meaning’ or
‘mean’ or any of their derivatives may be put to numerous uses. Surely the
philosophers cannot be concerned with all such uses. For example:

1. Tom meant to stop Jane from doing such a stupid thing.

2. The hike in the price of fuel means an overall hike in the price of all consumer
goods.
3. Life did not lose its meaning for her, even in such distress.
4. What is the meaning of this?
5. His sudden appearance in the present situation means trouble.
6. ‘Diction’ means manner of speaking or choice of words.
It is mostly the last use (i.e. use 6) of the word ‘mean’ which interests
philosophers. The main problems centering round this perhaps are: What does it take
for a word or a sentence to mean what it means? How do we understand what others
mean they make use of words, phrases or sentences? How do children learn the
meaning of words? How is the meaning of one term inter-related with meanings of
other terms within a language? How do we correctly decide what the meaning of any
particular word is? Do proper names have meaning? What are we saying about any
word when we say what it means?

A whole tradition has developed in philosophy, in the process of grappling

with these problems. As a result we now have quite a few theories of meaning.

Apparently the simplest of all such theories is the Referential Theory of

Meaning. This theory received support mostly of those who took proper names as the
typical unit of meaning. The meaning of a proper name is the individual whose name
it is. Thus the meaning of the name ‘Socrates’ is the designated individual, Socrates.

1
Philosophers have tried to construe the meaning of terms, other than proper names, in
the same way, saying that their meaning should be given in terms of the entities they
designate, i.e. in terms of their reference.

The major criticism that the theory faces is the one that Gottlob Frege put
forward. He says that tow expression may have the same reference yet they may differ
in meaning. (This I will discuss in detail later.) The other criticism that the theory
faces is that not all items of our language can be said to have any reference at all. But
that did not make them meaningless. Most referential theories have failed to explain
what connects expressions to their references and also how such a connexion comes
about without bringing in the role played by language users in establishing such a
connexion.

In view of these difficulties it seems that what is called the Ideational

Theory is a better and more adequate account of the connexion between words and
their referents. John Locke was an eminent advocate of the Ideational Theory among
classical writers.

II

Locke regarded language to be the means of communication. But what is it

that we can or do communicate to others? We want to communicate our ideas and
thoughts. Our ideas and thoughts, however, are objects of our mental world and there
is no way in which they may be publicly observable like our complexion. Ideas and
thoughts are private and we cannot possibly make others delve into our minds and
have a glance at them. So we need a tangible vehicle by means of which our inner
thoughts and ideas are made available to others. Locke took words as those tangible
marks. But words must be annexed to something with which both the hearer and the
speaker are directly acquainted. And that with which they can be acquainted
constitute the meaning of such words. What they can be acquainted with are ideas
given by experience. This empiricistic attitude made Locke understand meaning in
terms of ‘nominal essences’, i.e. the complex idea which a user of an expression
associates with the expression.

Here again we face a problem: nominal essence varies from person to person
and hence if meaning is given in terms of nominal essence it no longer remains

2
something invariant. And if meaning becomes variant it will be hard to find
something which Locke was looking for, namely something sharable which can serve
as the vehicle of communication. To make meaning objective, and public then, it
seems that meaning should be understood in terms of real essence. By ‘real essence’
Locke meant the inner constitution of things to which the words are to be applicable.
But real essences are not things with which one is acquainted. And being an
empiricist Locke could not accept real essences as the determinant of meaning. As an
empiricist Locke believed that we do and can annex words only to those properties
with which are acquainted. Yet he concedes that we nevertheless do want at the same
time to annex them to unknown internal real essences. For that might help us in
achieving something constant that will facilitate communication. So we find in Locke
a kind of tension as to whether one should associate the idea of nominal essence or the
idea of real essence with a word.

Locke was not unaware of this tension. He did make an effort to resolve this
by making some sort of compromise between his loyalty to empiricism and his
discomfort with subjectivity. His favourite example is the general name ‘gold’. The
idea ‘to which we have annexed’ the name ‘gold’ is the idea of having yellow colour,
a shinning look, great weight in proportion to size, malleability, fusibility. In other
words the idea of something that has all these characteristics, characteristics whose
joint presence helps us in identifying the object as ‘gold’. Locke does not identify the
nominal essence with these characteristics or properties but with the abstract idea of
such characteristics / properties, which are sometimes said to be in the mind. Real
essence, (in contrast with nominal essence) of gold, is the inner constitution of all
pieces of gold and it is this which determines the nominal essence being common to
all pieces of gold may be said to determine ultimately the kind of ideas that one
associates with the expression ‘gold’.

Though Locke tried valiantly to resolve his tension, he was however not quite
successful. This is mainly due to his allegiance to empiricism, which finally lead him
to conclude that it is an abuse of words to be annexed to ‘real essences’ of which we
have no clear idea:

“. . . when a man says gold is malleable, he means . . . something

more than this, . . . viz., . . . what has the real essence of gold . . .
but . . . by this tacit reference to the real essence of the word gold . .

3
 . comes to have no signification at all, being put for somewhat
whereof we have no idea at all, and so can signify nothing at all . . .” 1

In spite of all these the Ideational Theory brought to our notice:

1. Reference is not all that there is to meaning:
2. Any theory of meaning to be acceptable must explain how the connexion
between a word and its referent is established.

These two things had a great impact on philosophers. We see that under the
influence of the Ideational theory many of those who put forward a theory of meaning
have differentiated meaning from reference, but they too have usually seen meaning
as that which determines reference. Precursors of the present day concept of meaning
are those of the mediaeval notion of intension, the Millian idea of connotation and the
Fregean notion of sense. But intension is precisely what determines extension,
connotation is precisely what determines denotation, and sense is precise is precisely
what determines reference.
So it seems that the prime motive behind the construction of a theory of
meaning has traditionally been that of giving an intelligible account of reference.
Reference being, so to speak, the paradigm of the connexion between language and
reality – both, in fact, the paradigm and the nodal point – a theory of meaning has
traditionally been conceived as that very theory which explains how our language
relates to reality. Therefore, in an evaluation of a theory of meaning in the traditional
mode, one of our prime consideration ought to be whether a theory of meaning can
serve the purpose of explaining reference or, what is the same thing, the language-
reality connexion.

The language-reality connexion is established, on the traditional view, via

meaning, i.e. intension or connotation or sense. On this traditional view, which is
usually given the names ‘description theory of meaning’ and ‘sense theory of
reference’, meaning or intension or connotation or sense is a conjunction of properties
or descriptions expressing this conjunction which is associated with a term (whose
meaning we are talking of): and it is this conjunction or the description of it which
determines what the extension, or denotation or reference of the term is. In other

1
This excerpt from Locke’s Essays III, ix and x is quoted from J.L. Mackie’s Problems from
Locke, Clarendon Press, Oxford, 1976, p. 95.

4
words it is the objects in which we find those properties to be present or the objects to
which the description of such properties fit, which are the objects constituting the
extension/denotation /reference of the term.

The ‘conjunction’ terminology is distinctively Lockean. John Locke advocated

this view and in doing so he was mainly giving a theory of meaning for general terms
(we do not find any specific theory of meaning for proper names in Locke). For Locke
each general term signifies a ‘sort of things’: and it signifies by being a sign of an
abstract idea in the mind of the user of the term. It is as though, an abstract idea (in
Locke’s terminology ‘nominal essence’) is associated with each and every expression
and it is this which constitutes the meaning of the expression and determines whether
or not we are correct in calling a particular object by that expression (or general name,
if you like).

In more recent years we find this Lockean theory in a slightly different

formulation in C.I. Lewis’ works. Lewis says that we can name an object by a term
‘T’ only if the object possesses a conjunction of attributes which constitute the
intension or connotation of the term ‘T’. Wittgenstein, who more or less totally agreed
with Locke and Lewis, said that the meaning/connotation/intention/sense of the term
is not constituted by the conjunction of all properties, but is constituted by a cluster of
enough features chosen from the conjunction. And it is this cluster of properties which
determines the extension of the term. Other traditionalists have spoken of concepts
which are associated with terms and each of which provides a criterion for belonging
to the extension of the corresponding terms. Though Locke had conceived this
description theory or conjunction theory of meaning for general names a tendency of
extending this theory to proper names also, has been found in later exponents of this
view. This extended theory was not welcomed by many and so it gave birth to a great
controversy among philosophers.

John Stuart Mill was perhaps the first to say that the traditional theory could
not be extended to proper names. Through it may be correct that associated with
definite descriptions there is always a set of properties which constitute the
connotation of the term/definite description, yet it is not correct that such a set of
properties is also associated with proper names. Proper names, according to Mill, are
non-connotative. It is not the case that we have associated with each proper name, a
set of properties which would decide which objects belonged to the denotation or

5
extension of the term. To take an example, the name ‘Dartmouth’ was given to the
city originally because the city was situated at the mouth of the river Dart. But once
the name was so given, it became totally independent of this reason because of which
the name was originally given. In fact, it is only when the name becomes so
independent that it starts functioning like a proper name. Moreover, the city is no
longer situated at the mouth of the river Dart, as the river has changed its course. But
this does not rob the city of its old name. Thus at least the reference of the proper
names were never determined by their connotation, because proper names were non-
connotative.

III

When Frege evolved his notion of sense, however, there was a kind of
resurrection of the traditional theory which Mill had so bitterly criticized. Frege too
said that singular terms, i.e., both proper names and definite descriptions, did not
designate their references directly, but designated their reference via sense. The
traditionalists took Fregean sense the same as intension and connotation and used all
the arguments that Frege put forward to establish sense over and above reference, in
support of their own theory. Frege ascribed sense in addition to reference to proper
names for two reasons, viz.:

i) Unless we postulate sense we cannot explain the differences of two kinds of

identity statements.

ii) Unless we postulate sense we can never explain why the principle of
substitutivity fails in certain contexts.

Let us discuss the two reasons one by one:

i) Frege included both proper names and definite descriptions under names. The
reference of a name was that for which the name stood. If we consider the
following two identity statements.

A. The morning star = The morning star

B. The morning star = The evening star

Then we will notice that though the four terms involved refer to one and the
same planet and though both the sentences are statements of identity

6
 between the two terms involved in the statements, yet they have striking
dissimilarities in the following respects:

a) A is necessarily true, while B is merely contingently true.

b) A is analytic, while B is synthetic.

c) A is knowable a priori, while B is knowable a posteriori

d) While A is informative, B is uninformative.

Frege says, that which makes B so different from A is

the fact that though the two terms ‘the morning star’ and ‘the evening star’ has as their
reference the same planet Venus, yet the two terms are very different in respect of
sense. So names have both sense and reference.

Sentences too, for Frege, had both sense and reference Frege took sentences to
be kind of complex singular terms. All complex singular terms are formed out of one
or more simple singular terms by the help of a functional expression. Whole sentences
are no exception to this rule. As the complex singular term ‘the father of Satyajit Roy’
is formed out of ‘Satyajit Roy’ and ‘the father of . . .’ : the sentence ‘Socrates is wise’
is formed out of ‘Socrates’ and ‘. . . is wise’. The reference of a sentence (a complex
term of a sort, being determined uniquely by the references of the simple term (s) and
of the functional expression (s) out of which it is made of ) must be the truth-value of
the sentence. This is so because the reference of a complex term is that which remains
unaltered in and through all replacements of its constituents, by terms and functional
expressions having the same reference. And in the case of a sentence, it is nothing but
the truth-value of the sentence, which is as Frege prefers: either the True or the False
(both being objects). The sense of a sentence on the other hand is the thought
expressed by the sentence. The thought expressed by a sentence is the thought that the
conditions – the circumstances – which would make the sentence true are fulfilled: the
thought is the same as its truth-conditions.

ii) Now we may move onto discuss the second reason for admitting Frege’s sense.
We know that if we replace the constituent terms, functional expressions and
even sentences of a complex sentence by terms, functional expressions and
sentences with the same reference then the reference, i.e., the truth-value of the
whole sentence will remain unaltered. This is known as the principle of
substitutivity. In the case of truth functionally complex sentences this principle

7
 holds, but in the context of the ascription of belief and other propositional
attitudes this principle seems to fail. To take an example:

Though the term ‘Tully’ and ‘Cicero’ one co-referential yet it could well
be the case while the sentence “Tom believes that Cicero denounced
Catalina” is true, the sentence “Tom believes that Tully denounced
Catalina” is false. Again though the sentences “Socrates believed that if he
drank hemlock then he would die” is true, yet certainly the sentence
“Socrates believed that Bush was the President of USA in 1990” is false,
despite the fact that the sentences ‘If he drank hemlock he would die’ and
‘Bush was the President of USA in 1990’ are both coreferential, being
both true. This again provides us with a counter example of the principle
of substitutivity. In order that we may understand how Frege explained
this failure of the principle of substitutivity we have to keep in mind that
Frege also believed in, what is known as the Context Principle, which says
that a term has the meaning it has only in the context of sentence. His
main contention is that a term may change its reference with the change of
context. In the context of an ascription of a propositional attitude, which
he calls an oblique context, there is a kind of reference shift: and if we fail
to appreciate this then we will have a feeling that the principle of
substitutinity has failed. Frege says in an oblique context, the reference of
a term is not its customary reference but its customary sense. And in such
an oblique context if one replaces a singular term with a term having the
same sense then the truth-value of the whole sentence will remain
unaltered, i.e., there would be no failure of the principle of substitutivity.
It is in this way that Frege proved the need for postulating sense over and
above reference. Frege also said that it is the sense which determines the
reference.

IV

Traditional theorists were impressed by Frege’s notion of sense and they used
it as a weapon against Mill’s idea that proper names were non-connotative. But
whether or not it is correct to give such a reading of the Fregean notion of sense is

8
something which should be carefully considered. This has become a matter of great
concern for Frege scholars in recent years, especially in the wake of the kind of
criticism that has been brought against the traditional notion of sense and sense theory
of reference, by philosophers like Saul Kripke,2 Keith Donnellan3 and Hilary
Putnam.4 These three are somewhat defending the Millian notion of proper names and
thus would opt for a new kind of Direct Reference Theory, rather than a sense theory
of reference.

We may however in passing say a few words about Russell’s notion of proper
names. Russell maintained that the names which Mill took to be proper names, were
all definite descriptions in disguise. ‘Socrates’ is not a proper proper name, it is the
truncated definite description “The snub nosed philosopher who taught Plato, married
Xanthephe, believed in the theory of Forms and drank hemlock . . .” . Thus such a
name could never be non-connotative. For Russell it is only the demonstratives which
are genuine proper names, or as he called them logical proper names. And
demonstratives alone were devoid of connotation.

The strength of Russell’s argument notwithstanding, we cannot say that it is

entirely correct. On his theory, for instance, the name ‘Bill Clinton’ was just a short
form of the definite description “President of USA in 1993”. But it could well be the
case that the outcome of the US-election of 1992 was very different and so it could
well be the case that not Bill Clinton but George Bush or Ross Pirot became the US-
president in 1993. So Bill Clinton might not have been the President of USA in 1993.
With a different outcome of the election the definite description ‘the President of
USA in 1993’ would no longer refer to Clinton. Such a situation where Bill Clinton
fails to be the President of USA in 1993 is possible. And hence there must be a way
in which we may be able to speak of this possibility about Bill Clinton. In other
words there must be a way in which we can speak of not only what is actual about
Bill Clinton but also what is possible about him. We can conceive of a possible world
where Bill Clinton is not the President of USA in 1993 but some one else is. In order
that such possibilities may be ascribed to Bill Clinton, and in order to be able to refer

2
See Saul Kripke, Naming and Necessity, Basil Blackwell , Oxford, 1980.
3
See Keith Donnellan, ‘Reference and Definite Descriptions’, Philosophical Review, LXXV, 1966:
reprinted in Stephen P. Schwartz (ed.), Naming, Necessity and Natural Kinds, Cornell University
Press, Ithaca and London, 1977 .
4
See Hilary Putnam, ‘Meaning and Reference’, The Journal of Philosophy, LXX (November 8,
1973), pp. 699-711: reprinted in Stephen P. Schwartz (ed.), Naming, Necessity and Natural Kinds.

9
to Bill Clinton in a possible world where he fails to be the US-Premier in 1993, we
cannot regard the name ‘Bill Clinton’ to be a disguised definite description ‘The
President of USA in 1993’. If the name was really the truncated form of the definite
description ‘The President of USA in 1993’ then in the possible world the name “Bill
Clinton’ would not refer to the man Bill Clinton, or, if you like, the man who in our
actual world was the Governor of Arkansas in 1992 or is the husband of Hilary
Clinton . . . . But there must be a way to refer to this same individual in all possible
world or speak of all kinds of possibilities about the same individual. In other words
we must have a way of securing trans-world identity of reference. If ‘Bill Clinton’
was nothing but the definite description ‘The President of USA in 1993’, then could
we meaningfully assert “Bill Clinton might not have been the President of USA in
1993”: for that would amount to saying the meaningless thing “The President of USA
in 1993 might not have been the President of USA in 1993”. All the description
which can be conceived to be applicable to a particular individual, may well not have
been applicable to that very person in some possible situation, or in some possible
world. Thus in order that we make a reference to the individual in a possible world or
in a possible situation where she fails to satisfy that particular description, we must
find a way of referring to her in all possible worlds, without involving any description
whatsoever. And it is the proper name like “Bill Clinton’ which serves this purpose.

We can assert about an individual or a thing, by the help of a proper name, not
only what is actual about her or it but also what is possible about her/it, even if the
possibility is counter-factual. The proper name designates its referent on its own and
not via any description. It is as Saul Kripke, who has pointed out all that I have been
discussing in last paragraph, says, a ‘rigid designator’, which designates its reference
in all possible worlds; as opposed to a definite description which is a flexible
designator which changes its reference with the change of the possible world in which
we speak of it. Thus Kripke proves that Russell was wrong to consider proper names
to be disguised definite descriptions. And so Mill was right in saying that proper
names were non-connotative and that they rigidly designated their reference without
any assistance of connotation. And if Frege really meant connotation by sense then he
surely was wrong when the said that sense determines references.

Kripke was not only satisfied with treating proper names as rigid designators.
He went further, and treated general terms, or as they are called ‘natural kind terms’,

10
like ‘water’, ‘gold’ as rigid designators. A user of a natural kind term has his own
criterion by means of which he decides whether or not he should use a particular
natural kind term to designate a particular object. Now the question is whether or not
that criterion will determine the correctness of the use of the natural kind term. The
traditionalists would say that it does, while Kripke would say that it does not.
According to Kripke what it is to be included in the extension of a natural kind terms
like ‘water’ ‘gold’, etc. can never be analytically specified by some list of properties.
This he says because however much a thing may resemble those properties, the thing
may still fail to belong to the extension of the natural kind term. To take an example
the presence of the properties which we usually take to be the mark of an object’s
being gold may well present in a metal like fool’s gold, which is in fact not gold. But
then, is Kripke suggesting that there is not way in which we may decide whether or
not a particular metal is gold? Kripke is not making such an absurd suggestion. He
would say that present day scientific discoveries have told us that a metal with the
atomic number 79 is gold, or the designatum of the natural term ‘gold’. That which
determines whether or not the term ‘gold’ correctly applies to one particular metal is
its atomic structure. It is the inner constitution of the object which determines
whether or not a particular term correctly fits it. If and only if the stuff is gold or has
the atomic number 79 (as the present-day science tells us) that we can call it by the
natural kind term ‘gold’. Thus it is the inner constitution of the thing which is
determining the correctness of the use of the term ‘gold’, and not the criterion, we
ordinary people use while we make a use of the world. The correctness of the use is
determined by the ultimate constituents of the object, i.e. that which makes the object
the object it is. That ultimate constitution is not what the user has in her mind when
she uses the term. Thus what determines the correctness of the use of a natural kind
term is what scientific enterprizes are designed to reach. It could well be the case that
the inner constitution which present day science assigns to a particular object (like
atomic number 79 in the case of gold) may not be really the inner constitution and
there is always room for revision of scientific theories. But this does not in any way
alter Kripke’s main contention, which is: Whatever the inner constitution of gold be,
be it atomic number 79 or something else, it is that inner constitution and nothing else
which determines the correctness of the use of the natural kind term ‘gold’. Thus this
term has no such descriptions associated with it the applicability of which to a
particular object would make the object fall under the extension of the term. The

11
natural kind term too is like a rigid designator, and hence it too is devoid of
connotation.

Keith Donnellan who was another exponent of the theory of Direct Reference,
said not only proper names and natural kind terms, but definite descriptions too may
be regarded as rigid designators. Or to be more precise they may be put to one
particular use in which they function like proper names. We may distinguish between
two kinds of use of a definite description: an attributive use and a referential use. In
explaining this Donnellan takes the helps of an example to explain this distinction.
The story goes like this:

Smith is a very lovable person. He has been killed in a violent and gory way.
From the very gruesome nature of the murder one may make the statement “The
murderer of Smith is insane”. In making this use of the singular term ‘the murderer
of Smith’, Donnellan says, we are making an attributive use of it. In this case ‘the
murderer of Smith’ refers to any person if and only if the person satisfies the
conditions expressed in the open sentence ‘x murdered Smith’. But suppose
somebody has been arrested on the charge of murdering Smith and he is being tried in
the court of law. He behaves in a very peculiar way inside the court room. And seeing
his strange behavior I say “The murderer of Smith is insane”. Do I make an
attributive use of the singular term ‘the murderer of Smith’? Certainly not. It may
well be the case that I know that man who is being tried and who is behaving like a
lunatic while he stands at the dock is not the murderer of Smith. Yet I can refer to the
man by the definite description ‘The murderer of Smith’ though he perhaps does not
satisfy the descriptive conditions laid down by the definite description. That I can
succeed in referring to the man is simply because I am using the singular term like a
proper name, putting it to what Donnellan calls a referential use. So we see that when
a definite description is put to a referential use it is no longer connotative.

So far so good: but we now face the question that if the reference of names
and natural kind terms are not determined by descriptions, then how is it determined
at all? The answer that is provided by Putnam, Kripke and Donnellan is that the
reference is usually determined by causal chains. To explain this we may say that one
of the ways of deciding what the referent of a particular name was, we investigate
into the very event in which the name was given to a person in a ‘baptism’ or an
initial coinage of a term to refer to the present referent. Putnam extends the causal

12
theory of reference of proper names to natural kind terms as well. Putnam seems to
suggest that the same kind of baptism takes place when we make an initial use of a
natural kind term. What he says is that a paradigm example of some substance is
chosen and a natural kind term is given to it like a name. It may well be the case that
we at the time of baptism did not know what the inner constitution of the substance
was. As long as my use of the natural kind term is somehow casually linked up with
the initial act of baptism, I will succeed in referring to the substance, though it is the
inner constitution of the substance which determines the correctness of the use of the
term; and not what descriptions, attributes, properties I associate with the substance in
my mind. The meanings ‘ain’t in the head’ – it is the very stuff which constitutes the
substance which ensures the success of reference and there is nothing more to
meaning than reference.

Thus we find the Direct Reference Theory is seeking to forge a connexion

between expressions and things not through an extra-linguistic intermediary between
the two, but in a direct association between the two, possibly transmitted historically
from groups to groups of speakers. If such a theory of reference succeeds, a theory of
meaning in the traditional mould must be rejected altogether: a theory of meaning
will have no place in our understanding of the relation between words and objects –
between language and reality. In that case a theory of meaning would survive either
as a refinement of the theory of reference itself or a theory which has an altogether
different basis and purpose.

On the first alternative we can surely say that any study of language cannot
ignore the importance of convention. The reference to the event of baptism made by
the direct reference theorists, in explaining their causal theory of reference, itself
makes such a reference to convention. But if we want to exploit the importance of
convention to establish the objectivity of the connexion between words and things
then we also have to keep in mind that this convention must be something known to
the speaker himself. The speaker can hardly be expected to use her words according
to the existing linguistic convention and practice unless she actually knows what they
are. If this be so then we may say that perhaps Locke was right to associate nominal

13
essence with meaning rather than real essence. Moreover it still remains an unsolved
question as to whether it is right to consider Frege’s sense to be the same as Mill’s
connotation. Textual evidence from Frege and the new interpretation of Fregean sense
offered by Frege scholars like Michel Dummett, Gareth Evans and John Mc Dowell,
tell us that sense is not the Millian connotation, it is the mode of presentation of the
reference. I will not go into the discussion of sense as mode of presentation right
away, but I may just mention that this notion of sense is not without defect and there
is still room for doubting whether Frege really had this in mind when the spoke of
sense.

We may now pass on to an entirely different alternative. Under this alternative

a theory of meaning would no longer be tied up with the notion of reference, drawing
into justification as it were, from its ability to explain the latter. This radical
metamorphosis is what the truth-conditional theory and other molecular theories of
meaning have always tended to undergo. This is due to the fact that reference, or
extension, or denotation belongs primarily to terms – singular or general – which are
constituents of a sentence, and not to the sentence as a whole. As a result a meaning
theory which is designed to do a yeoman service to a theory of reference tends to be
atomistic in nature. It is as if meaning atoms relate to the sentence parts – the atomic
parts – and they join together to form the sentence meaning. With the demise of the
atomist theory, and with the dawning of the awareness that meaning belongs
primarily to sentence – at least to sentences if not to larger chunks of discourse, the
traditional motivation for having a theory of meaning of a certain type has weakened,
in fact almost evaporated.

PART TWO

From Atomism to Molecularism

This shift from atomistic theory to molecular theory of meaning has been
regarded by Quine as one of the ‘Five Milestones of empiricism’. 5 This shift is not
very sudden. Philosophers had started feeling uneasy about what the meaning of

5
W.V.O. Quine, ‘Five Milestones of Empiricism’, in Theories and Things, Belknap Press of
Harvard University Press, Cambridge, Massachusetts, 1981.

14
grammatical particles like ‘if’, ‘but’ could be, because these particles could not be
defined in the way other terms could be defined. Meaning of such terms could not be
grasped in isolation from the sentences in which they occurred. They could be defined
only in the context of such sentences. Thus for such terms philosophers offered, what
is called in philosophy, contextual definition. And it is this notion of contextual
definition which precipitated a revolution in semantics. Later on we find an extension
of this original insight and philosophers started considering other ordinary terms to
have the meaning they had only in the context of a sentence. The meaning of no term
could be understood in abstraction from the truth-conditions of sentences containing
the terms. The need for contextual definition of terms made us recognize the semantic
primacy of sentences and brought about what is known as the molecular theory of
meaning, as opposed to atomic theory of meaning. Quine says that the pioneer
molecularist was Jeremy Bentham who offered a contextual definition not only for
syncategorematic expressions but also categorematic expressions. It was Gottlob
Frege who generations after Bentham spoke of the semantic primacy of sentences and
evolved his famous Context Principle in The Foundations of Arithmatic.6 Russell too
was deeply influenced by the notion of contextual definition which paid rich
dividends in developing his technical logic.

Though historically speaking the notion of contextual definition and the

Context Principle were in very close association with each other, yet one would land
up in great difficulty if one took the connexion between the two as more than
historical. Many have overdramatized the relation between the two and have taken
them to be quite inseperable. Those who have done this have taken Frege to have
suggested such a connexion between the notion of contextual definition and the
Context Principle. But we could assign such a view to Frege only if we took a ‘strong
interpretation’7 of Frege’s Context Principle. Under this strong interpretation, a word
has meaning only in the context of a sentence, i.e. a word or any substantial
expression has no meaning of its own. This interpretation makes the assertation that
each and every substantial expression can be defined contextually. Such an
interpretation is likely to do more harm than good to the principle itself. It may well

6
Gottlop Frege, The Foundations of Arithmatic, Translated by J. L. Austin, Basil Blackwell,
Oxford, 1950.
7
B.K. Matilal and P.K. Sen,and ‘The Context Principle and Some Indian Controversies Over
Meaning’, Mind, Vol. XCVII, No. 385, January, 1988.

15
be the case that there are certain expression which could be defined only
contextually. The classical example of the use of the contextual definition is Russell’s
theory of definite description.8 But Russell’s theory of definite description along with
the contextual definition he gives for definite descriptions, was not designed to give
the meaning of definite descriptions. What Russell intended to show is that we need
not suppose that such an expression has any meaning of its own. The logical analysis
of a sentence containing a definite description shows that it can be thoroughly
eliminated from the sentence. Russell through logical analysis shows that a definite
description (in fact, all descriptive phrases) are incomplete symbols. By incomplete
symbol Russell means: (i) that which has no meaning in isolation and by meaning
Russell means what is predominantly ‘denotation’; (ii) that which cannot be defined
in terms of some other expression which can be simply substituted for it without any
regard to any change in the context; and (iii) that which, as we have already said, can
only be contextually defined, i.e. by a translation of the whole sentence containing it
into an equivalent sentence which does not contain it.

From the above it is clear that the very notion of contextual definition was
evolved by Russell because he considered it as a means of defining incomplete
expressions. By conversational implicature it follows that there are complete
expressions and never Russell say that they too are contextually defined .Moreover in
the case of incomplete expression it is doubtful as to how far a contextual definition
provides us with the meaning of such as incomplete expression. It, in fact, follows
from the three characteristics of an incomplete expression that no such expression (in
this case no definite description) is a self-subsistent semantical unit at all. Moreover
the knowledge expressed by a sentence with a definite description is what Russell
calls Knowledge by Description, i.e. in which we know that there is a thing which
satisfies such and such description, without knowing which that thing is. Thus a
contextual definition of a definite description like  will not help us in answering the
question “What does  means?” The only meaning we can grasp at in the case of a
contextual definition is:

 means that by virtue of which S () means that 

Where S () is the sentence containing the definite description  and  is replaced by
a sentence semantically equivalent with S () but does not contain . But from this
8
Bertrand Russell, ‘On Denoting’ , Mind, 1957, pp. 479-493.

16
we hardly know what  means, we only know it means something by virtue of which
S () is semantically equivalent with . As a result, as an account of the meaning of
individual expressions contextual definition is very poor. And so if we take the
Context Principle, which is surely a semantical principle, to be too closely connected
with contextual definitions than the Context Principle will be of very little semantical
significance.

There is yet another reason why we should not adopt the strong interpretation
of the Contest Principle and should not overdramatise the relation between it and the
notion of contextual definition. The reason is that it would make the Context
Principle utterly incompatible with the Composition Principle. If there were no way
of knowing the meaning of the sub-sentential expression then it would not be possible
for us to decipher the meaning of a whole sentence by a kind of step-by-step
construction from meanings of words. And in such a case one will have to give up the
Composition Principle. But Frege held on to both the Context Principle and
Composition Principle. Perhaps it is because of his equal allegiance to both these
principles (along with the strong interpretation of the Context Principle), that we find
a kind of tension between Atomism and Molecularism in Frege.

One might say that the kind of contextual definition of definite description
Russell speaks of is not the kind of contextual definition of definite description and
other terms Frege speaks of. This is simply because Frege’s theory regarding definite
description is very different from Russell’s theory. Frege does not take a definite
description to be an incomplete symbol which cannot occupy the subject position of a
proposition. He takes it to be like a proper name/singular terms, which can and does
occupy the position of a subject, which is a self-subsistent semantical unit, and which
stands for an object. On a Fregean analysis a contextual definition of a term would
tell us that the expression has sense and reference. But it is doubtful as to whether
such a contextual definition would tell us what the reference and the sense is. So we
are again facing the same problem here again the contextual definition and the strong
interpretation of the Context Principle becomes semantically poor.

In the face of all such difficulties, some have given a rather weak
interpretation of the Context Principle. This principle ways that it is not the case that
sub-sentential expressions have no meaning of their own. But the best way of
ascertaining the meaning of sub-sentential expressions is by observing how such

17
expressions are used in sentences to say something. But this is something no one
would deny and it seems that Frege has something more definitive in mind when he
spoke of the Context Principle. This becomes evident from the development of
Frege’s theory of number where he says that what matters is not that we should
identify something to be the meaning of a subsentential expression; but it is sufficient
if we can say what contribution it makes in the determination of the meaning of the
sentence in which it occurs. And it is this what is expressed in what may be called the
moderate interpretation of the Context Principle. In defence of this interpretation of
the Context Principle Dummett says: the main function of language is saying things
by the use of it: but one can hardly say anything without the use of whole sentences.
It is only by the use of whole sentences that we can make linguistic moves. Hence
sentences become the semantic unit in relation to which we may try to find out the
meaning of subsentential expressions. We succeed in saying things by subsentential
expressions only because they contribute to the significance of the sentence of which
they form parts. Thus the meaning of the sub-sentential expressions consists in the
contribution which they make to the meaning of the sentences.

Dummett goes on further and says that if this be the Context Principle, then
there is no reason why Frege could not adhere to it along with the Composition
Principle. Dummett says that the two principles give answer to two entirely different
kinds of question and there is no reason whatsoever to take them as two opposing
principles. If we have to find out what the meaning of the sentence is then we have to
first find out the meaning of the parts of the sentence. We can never know or
understand what the meaning of the whole sentence is unless we know or understand
in advance what the meaning of the words that compose it is. So far we are concerned
with the recognition of the meaning, the word is primary and in that case it is the
Composition Principle which helps us. But when we ask: ‘What kind of a thing is the
meaning of an expression?’ or ‘What kind of function does a particular expression
perform?’, then we can only answer such question by taking the meaning of the
sentence as primary. Thus it is the Context Principle which gives us a general
explanation of what it is for any expression to have a meaning. Hence the meaning of
a word or any sub-sentential expression can be understood only as consisting in the
contribution which it makes to the determination of the meaning of the sentence. We
can never understand what function a word or any sub-sentential expression performs

18
unless we know what linguistic act is performed by the sentence in which it occurs,
because it is only by means of a sentence that we can say something.

Thus we see that while the Context Principle is answering the more general
kind of question ‘What is meaning?’ the composition Principle is answering the more
specific kind of question ‘What is the meaning of “water”?’. And so the apparent
conflict between the Context Principle and the Composition Principle is resolved by
Dummett. This solution however somewhat weighted in favour of the Context
Principle, because in philosophy it is the more general questions which are more
significant. And perhaps there is nothing wrong in the solution being weighted in
favour of the Context Principle: because it is in this principle that we find the
molecularist turn. A turn which is welcome in the face of the difficulties we found the
atomistic theory facing.

We may however at this point take notice of the fact that Frege himself did
not explicitly say or claim the Context Principle to be a semantic principal. In his
works, never does Frege himself claim that he was giving a theory of meaning. Frege
scholars however have tried to trace a distinctive theory of meaning in Frege. This
they have done, as we have said in the very beginning of the chapter, by identifying
Fregean sense with meaning or by identifying a theory of meaning with a theory of
reference, or by advocating a kind amalgamation of Frege’s theory of sense and
theory of reference as a theory of meaning. And undoubtably it is in Frege’s Context
Principle that we can trace the modern Molecular and Holistic Theories of meaning as
advocated by Davidson, Dummett, McDowell and the like. But all these remarks
would have very little significance unless we know what we really understand by the
term ‘theory of meaning’ in contemporary philosophy of language. And we should
also have a kind of pre-theoretic understanding of the concept of meaning. I will start
the next chapter by addressing myself with these two things.

19