Skip to main content

Full text of "An Analysis of the Concept of Meaning"

See other formats


Early Journal Content on JSTOR, Free to Anyone in the World 

This article is one of nearly 500,000 scholarly works digitized and made freely available to everyone in 
the world by JSTOR. 

Known as the Early Journal Content, this set of works include research articles, news, letters, and other 
writings published in more than 200 of the oldest leading academic journals. The works date from the 
mid-seventeenth to the early twentieth centuries. 

We encourage people to read and share the Early Journal Content openly and to tell others that this 
resource exists. People may post this content online or redistribute in any way for non-commercial 

Read more about Early Journal Content at 
journal-content . 

JSTOR is a digital library of academic journals, books, and primary source objects. JSTOR helps people 
discover, use, and build upon a wide range of content through a powerful research and teaching 
platform, and preserves this content for future generations. JSTOR is part of ITHAKA, a not-for-profit 
organization that also includes Ithaka S+R and Portico. For more information about JSTOR, please 


| ( T>R any theory of knowledge the concept of meaning will be 
-*- of fundamental importance. We find it interpreted in 
many different senses by various schools of thought, but in the 
midst of all such differences there is a substantial unanimity in 
admitting that only where meaning is present can there be truth 
and error. It is this which invests the notion with its great 
philosophical significance, for here we see that the truth problem 
itself must be attacked in terms of a theory of meaning. But 
while the concept of meaning is thus of central moment, we find 
comparatively little attention has been explicitly devoted to dis- 
cussing and accurately defining it. Every system of epistem- 
ology embodies a very definite theory of meaning, but by no 
means has every such system made that theory clear. The 
result has been inevitable. Innumerable misunderstandings 
have arisen. Wide spread and frequent failures sharply to de- 
fine issues and problems have resulted. And too often all hope 
of an adequate and satisfactory solution of the problems of 
epistemology has been banished by neglecting the need of ac- 
curacy on this fundamental point. From this situation the 
present discussion takes its rise. Its purpose will be to apply 
the logico-analytic method to the study of meaning. We shall 
see that it is possible to treat meaning, not as a simple ultimate, 
but as a complex of more fundamental entities. And the general 
philosophical interest and value of this analysis will be this, 
that it provides us with terms and concepts which enable us to 
deal with the epistemological questions at once more compre- 
hensively and more accurately than is otherwise possible. 

We may at the outset state our general thesis, which is that all 
meanings may be resolved into complexes of sense-data or 
'appearances.' That is to say, we propose an analysis of knowl- 
edge along lines often attempted without conspicuous success by 
Bertrand Russell and his school, lines which run parallel with his 
analysis of physical things into logical constructs of sense-data. 



In considering all such attempts to build up the world in which 
we live from the data of sense, it is essential at the outset to 
distinguish between what may be called the ontological and the 
epistemological points of view. Russell has succeeded in show- 
ing, at least in outline, how it is possible to regard the things of 
physics as constructs of sense-data. And such an exposition as 
is given in his article, "The Relation of Sense-Data to Physics," 
is essentially ontological in intent. For the entire interest 
centers upon the characterization and analysis of the objects in 
question. It is their ultimate constitution which is in question. 
The constitution and analysis of knowledge is simply assumed. 
What we have here is a discussion of the function of sense-data 
as constituents of the things of physics. But on the other hand 
there is quite another problem, namely the function of sense- 
data as elements in knowledge, and here our point of view calls 
us to treat the object not as existent, but as known. This is the 
epistemological point of view and problem. 

The distinction between the epistemological and ontological 
approach enables us to clear up a number of points in a manner 
which is of the greatest assistance to our whole discussion and 
which brings it into higher relief. In the first place, we see that 
from the ontological point of view it is a matter of indifference 
whether or no we have sense-data pure, so long as we have them 
somehow. It is often objected to the whole doctrine of sense- 
data that in actual experience we never encounter them in 
isolation, so that they are mere abstractions. But it can hardly 
be denied that such data do actually exist, at least in combina- 
tion, and this allows us to begin our analysis of physical reality 
in terms of appearances. In the second place, we see that from 
the epistemological point of view it does not matter whether the 
groups of data which constitute physical things are related to- 
gether in terms of congruence, as Russell sometimes maintains, in 
terms of cause, as Moore suggests, or in some other manner. 
It is enough that each such group or class is distinguished by a 
determinate pattern or order system, which gives us a criterion of 
individuation and enables us to distinguish between this thing 
and that. In the third place, it is absolutely essential that 


in maintaining our thesis that all meanings may be resolved into 
complexes of data, we frame a theory which shall take account of 
a consideration urged long ago by Berkeley, but too often ignored 
by the logico-analytic school, namely the heterogeneity of our 

These are the general presuppositions of our theory of meaning, 
and they determine our mode of approach, and the categories of 
our discussion. From the ontological point of view, the problem 
is to analyse the congeries of objects which we call the things of 
physics. This result is obtained by regarding them as classes of 
sense-data. Thus the fundamental notion for the ontological 
point of view is that of class occupancy. This relation is central 
in the whole ontological account. Over against this we have the 
epistemological account, whose problem is to analyse that con- 
geries of entities we call meanings, and which go to make up 
the body of discursive knowledge. There again the ultimate 
constituents are the same, namely sense-data. But the relation 
between them is radically different. It is now the relation of 
symbolism, by which one entity is said to 'stand for' another, 
that is of interest and importance. And our concern now is, 
given a sense datum, to determine how it can symbolise or stand 
for, another or others. Our task then, is to specify the properties 
of the relation of symbolism, or better, to analyse what we may 
call 'symbolic complexes,' that is to say, complexes of terms 
related by symbolism. And our object will be to show that a 
satisfactory account of knowledge can be given which takes its 
rise from the contention that each individual piece of meaning 
can be treated as a symbolic complex. 

The question at once suggests itself as to whether it is possible 
to speak of a general relation of symbolism which is always found 
and which remains the same in all essentials whenever and 
wherever one entity stands for another. There is a certain sense 
in which we may speak of a symbol without in any immediate 
way bringing in the notion of an entity which stands for another. 
It would be difficult to show that algebraical symbols stand for 
anything. This is a complication which will be further discussed 
later. It is, however, largely a matter of terminology, and when 


we explain that by a symbol we are to understand an entity 
which stands for another or others, we clear up this initial diffi- 
culty in admitting a general relation of symbolism. A more 
formidable obstacle appears when it is pointed out that in ex- 
perience we actually make use of two types of symbols, using 
the term in the sense in which we have defined it. The dis- 
tinction arises as follows. From the ontological point of view 
we may regard a physical thing as a class of sense-data. Now 
we may use one or more of these constituent data as symbols 
which stand for the thing itself, so that the patch of brown which 
I see here and now stands for all the other possible appearances 
whether visual, auditory or tactile, of the table before me. 
Here is one type of symbol. But on the other hand we may 
use the word ' table ' itself to stand for the class of heterogeneous 
sense-data which constitute the table. It is obvious that the 
word is not a constituent of or element in the object in the same 
sense as one of the constituent data. And yet it stands for or 
symbolises these data. Let us call these two types of symbolism 
respectively primary and secondary. This is obviously a very 
important distinction, and the question which is now made 
definite is whether between primary and secondary symbolic 
complexes there is any distinction so radical as to call for an 
entirely different treatment in the two cases, and to demand that 
we admit of two disparate kinds of knowledge. 

We begin by considering primary symbolism. Here we employ 
one or more of the constituent data of the thing to stand for the 
thing itself. This is what is ordinarily called perceptual knowl- 
edge. Now it is to be observed that even in perceptual knowledge 
we have something more than mere givenness. We have in 
fact a datum which functions as a symbol. An analysis of what 
is implied in ' functioning as ' a symbol will come later. At this 
point we merely observe that this is what actually occiirs. This 
is all in contradiction of Russell's analysis of perception which 
treats it as 'knowledge by acquaintance' or mere givenness. 
We have already remarked that from the ontological point of 
view it is a matter of indifference whether or no we ever have 
sense-data uncombined. This is far from being the case from the 


epistemological point of view. Here it is true to say that we 
never have mere data, but always data and meaning, which for 
us of course implies simply data which function as symbols. 
Thus the assertion that mere data are abstractions is not only 
true, as it is from the ontological point of view, but also important. 
Nor can we escape by the systematic device of insisting that in 
perceptual knowledge the datum is a symbol which stands for 
itself. Perhaps the only general formal property of the relation 
of symbolism on which it is worth while to insist is that it must 
be totally non-reflexive. This assumption constitutes the very 
life of discursive thought. The contrary assumption is precisely 
that of the mystic and is an explicit and deliberate abandonment 
of the entire scientific point of view. And the mere fact that 
we use a primary symbol by no means forbids us to assert that 
we are using a symbol. Now this at once raises the truth- 
problem; for Russell contends that the most important charac- 
teristic of knowledge by acquaintance is that it is neither true 
nor false. Here again we are brought face to face with issues 
which must be postponed until our analysis has gone further. 
For the moment we may be content with pointing out that the 
usage of ordinary language certainly suggests that there are 
truth values in perceptual knowledge. Russell would contend 
that if I have a perceptual hallucination, no truth value is in- 
volved, for the situation is exhaustively analysed when we have 
said that something is given. But a hallucination is constituted 
as such by an error in meaning, which for us is an error in sym- 
bolism. And there would seem no point in speaking of per- 
ceptual hallucinations at all unless meanings, or symbolic com- 
plexes, and truth and error arose at this level. Thus we must 
insist that perceptual knowledge like all other knowledge, is in 
essence symbolic, and that we must give an account of it, using 
as our fundamental and determining concept not knowledge by 
acquaintance, but primary symbolism, which is a genuine type 
of symbolism. 

In connection with our question as to whether we are bound 
to recognise two distinct and disparate kinds of symbolism, let 
us now consider secondary symbolism. A secondary symbol 


is any conventional sign whose content does not constitute an 
element of the object to which reference is made. By far the 
most important instance of such symbolism is to be found in 
language, and we may confine ourselves to this in the present 
discussion. And our immediate question is whether here we 
have a situation in essentials different from that which arises in 
the use of primary symbolism. Consideration of this question 
brings us face to face with a large number of other issues which 
must be met in the course of our discussion, but which may be 
postponed for the time being. Central among these, and indeed 
possessing an important bearing upon the question immediately 
before us, is the problem as to what elements in language con- 
stitute one single and unitary symbol. Shall we say, for instance, 
that only a substantive word or phrase can be treated as a symbol? 
And if so, what are we to do with verbs, among other elements of 
language? These are problems, which, as we shall find, involve 
us in very far-reaching considerations, but for the moment we 
may avoid them by introducing a convenient technicality, and 
refer to that word or congeries of words which functions as a 
symbol as a unit of language. Now the only apparent difference 
between a unit of language, that is to say a secondary symbol, 
and a primary symbol, is that with the former there is usually 
more emphasis on reference, while with the latter there is usually 
more emphasis on content. Bradley, in the opening chapter of 
his Logic, points out that in all symbolism there are present the 
two elements of content and meaning. In primary symbolism 
the immediate content usually bulks largest, though this is by 
no means always the case; for often a momentary sensory pre- 
sentation of an object has force and meaning only by virtue 
of standing for the object as a whole, as when the glimpse of a 
radiator summons us to spring aside to avoid an oncoming 
automobile. In secondary symbolism, however, the immediate 
content is comparatively unimportant, as is shown by the favor- 
ite and often used example of mistakes made in proof-reading. 
Beyond this there would seem to be no essential difference be- 
tween primary and secondary symbolic complexes, and we may 
conclude that it is justifiable to assume that there is a single 
relation of symbolism which is found in all knowledge. 


It is now time to turn to the problem which we raised in the 
above paragraph, and consider the general symbolic function 
of language, in which we have the best and clearest case of 
the working of secondary symbolism. This discussion will 
throw a flood of light on the nature of symbolism in general. 
The immediate difficulty which confronts any attempt to deal 
with language in terms of the notion of symbolism is that there 
seem to be many elements in language which can hardly be called 
symbols in any very obvious or immediate sense. Some terms, 
nouns substantive for instance, it seems at the first glance quite 
possible to regard as standing for some other physical entity. 
But when it comes to verbs, prepositions and conjunctions, it is 
obvious that the case is very different. For even though it 
might not be impossible to devise some ingenious interpretation 
which would endow them with an indirect reference to other 
entities, this would be a clear distortion of their natural force 
in the interests of a theory. The fact is that we shall find 
ourselves compelled to admit that the grammatical classifications 
and the analysis of language into separate words has very little 
bearing upon the dividing of it up into symbolic units. For here 
the principles and purport of the whole discussion are radically 
different. Bosanquet has well shown that in the particular case 
of the interpretation of the copula the grammatical analysis of 
the sentence is thoroughly misleading, and this is a consideration 
which is of general import and far reaching consequences. 

There are two senses in which it is possible to regard language 
as a system of symbols. The first of these has no immediate 
interest for us, and need be considered only as clearing the 
way for that interpretation which concerns our present purpose. 
Language may be regarded as symbolic in the same sense as 
algebra. It may in fact be treated as an algebraic scheme by 
means of which certain relations, or better, certain types of order, 
can conveniently be analysed and exhibited. Here again it is 
important to observe that the grammatical analysis of language 
is irrelevant. This is clearly demonstrated by Russell's theory of 
'incomplete symbols.' The point of this theory seems to be to 
show that a large number of language elements which at first 


sight appear to possess symbolic force in the sense of ' standing 
for' some other or 'external' entity, have, in fact, significance 
only or at least primarily as elements in a language-algebra. 
Only by such a treatment is it possible to avoid the contradic- 
tions pointed out by Meinong. And, further, this theory makes 
it evident that the subdivisions of language which are employed 
by grammar, its subdivision, that is to say, into separate words 
as the ultimate units, cannot hold water for epistemology. 

The consideration that a correct treatment of language as an 
algebra gives us a principle of division quite other than that of 
grammar leads at once to a study of the second sense in which 
it is possible to regard language as symbolic. It is evident here, 
that as was remarked in another connection, our definition of 
a symbol as that which stands for something else, implies that 
to speak of an algebraical symbol is to use the term in a different 
and for us relatively unimportant sense. We now come to a 
study of language as symbolic in the sense in which we have 
defined the term, that is, as 'standing for' something else. For 
this treatment the positive value of noting that language may, 
and for some purposes must be regarded as an algebra is, first, 
that it clearly shows how misleading is the attempt to use the 
grammatical categories in arriving at the ultimate symbolic 
elements in language, and second, that it gives us a reliable clue 
to the discovery of those elements. Let us proceed by means of 
a particular instance. What is the symbolic force of the sentence, 
' the light is burning ' ? What we have said at once forbids the 
seductively simple solution of saying that here we have two ele- 
ments, 'the light' and 'burning', related by a copula having some 
such force as the logical 'times.' For both 'the light' and 
' burning ' must be regarded as incomplete symbols which cannot 
immediately stand for anything. Moreover we cannot provide 
a solution by introducing the notion of propositions or objectives, 
and saying that the secondary symbolic unit is ' that the light is 
burning ' or ' the burning of the light.' For as Russell has pointed 
out, propositions themselves must be regarded as incomplete 
symbols. We seem therefore definitely forced to the conclusion 
maintained by Bosanquet, that the symbolic unit of language 


is the sentence itself, that is, the asserted proposition. We 
cannot further subdivide language in the interests of a symbolic 
scheme, but must simply say that the unit which 'stands for' 
something will be in this case the sentence, 'the light is burning.' 
It is valuable to note the congruence of this conclusion with the 
obvious facts of primary symbolism. When I use a primary 
symbol for reference to and knowledge of the light, I do not 
take the factual complex to pieces. I do not have before me a 
sense-datum of light apart from burning, or a sense-datum of 
burning apart from light. I have a sense-datum of a burning 

We are now in a position to turn to our final task of dealing 
with the truth-problem by means of the categories of our previous 
discussion. We have been driven to the conclusion that only 
sentences can be regarded as the ultimate symbolic elements of 
language. Let us now look a little more closely into the con- 
sequences of this position. What is obviously involved is that 
certain word-structures will, and others will not, possess symbolic 
force, or 'stand for' something else. To return to our former 
example, the word-structure 'that the light is burning' is not 
a secondary symbol at all. It is, to use Russell's terminology, 
an incomplete symbol. But on the other hand, 'the light is 
burning' is a complete secondary symbol, a unit of knowledge, in 
that it stands for something else. How then are we to account for 
this important difference? It would seem very paradoxical to 
attempt to do so in terms merely of the formal differences of the 
word-structures involved, though such an attempt might be 
made without the certainty of failure. But the formal difference 
in the two cases can only with great difficulty and by dint of great 
ingenuity be treated as the sufficient ground of the immense and 
vital distinction of function. But this formal difference is 
symptomatic of the true and sufficient ground of that distinction. 
In order to understand it we must look beyond the set of symbols 
as mere physical entities which correspond with other physical 
entities, and take into account a factor which in a sense does not 
appear within the symbols themselves, though the word-struc- 
ture usually indicates its presence. This is the factor of assertion, 


which is indeed essential to the very existence of a symbolic 
complex. And we reach the conclusion that we cannot have 
symbols of any kind in the absence of assertion. A mere form of 
words does not in and of itself constitute a secondary symbol. 
In and of itself it is no more than a physical fact among other 
physical facts. But when we assert this form of words, then it 
takes on the unique function of symbolism, in that it is used to 
stand for other facts. 

Now it is clear that in assertion something has emerged of 
which our analysis has not as yet taken account, but which 
is nevertheless essential to its completion. Moreover it is evident 
that we have now come into the presence of the truth problem, 
for only an assertion can be true or false. And the question at 
once presents itself as to whether we can deal with this notion in a 
way which will not be vitally inconsistent with the entire analysis 
of knowledge which has been presented. This question is cogent 
for the reason that the so called coherence theory of truth, be- 
ginning with an analysis of assertion, substantially ignores the 
element of symbolic correspondence as insignificant, though 
admitting that it exists. We, however, in beginning with the 
element of symbolism, have substantially implied that it is by no 
means insignificant. But our analysis of symbolism has driven 
us to admit that the element of assertion is essential. Can we 
then do justice to the factor of assertion without violating the 
rights of the factor of symbolism, and to the factor of symbolism 
without violating the rights of assertion? 

Before we attempt to answer the question it will be well to 
remind ourselves of the dialectical impossibility of accounting 
for truth and error in terms of symbolic correspondence alone, 
without any reference to assertion. It may be correct to say that 
truth arises where a symbol actually corresponds to something 
other than itself. But this, though a necessary, is not a sufficient 
condition. Its insufficiency is best seen when we ask how in these 
terms we are to account for error. In error our symbol will not 
stand for any existing entity, though it may or may not seem to 
do so. Shall we then say that we have a symbol which symbolizes 
nothing? This would seem to be a contradiction in terms, for of 


course the 'nothing' in this case is not the null class. Or shall 
we say that our symbol stands for something which it does not 
mean? This we cannot do without admitting a factor other than 
that of symbolic correspondence, for at once we have a new ele- 
ment, the meaning or intention of the symbol. Thus the notion 
of symbolic correspondence breaks down in dealing with error. 
And it is in fact quite as inadequate in dealing with truth. A 
true judgment is certainly a case of successful symbolic corre- 
spondence, but in order to define this correspondence we need 
some such element as the intention of the symbol, for a true judg- 
ment is not a symbol which corresponds to anything, but to 
some one selected thing. And it is obvious that if we introduce 
the notion of the meaning or intention of a symbol we immedi- 
ately stultify our whole discussion, the point of which is to arrive 
at a definition of meaning. Thus the assumption of any such 
element would be a petitio principii. 

To return after this parenthesis to our question, which has 
now defined itself more in detail, our problem is to determine 
how to account for knowledge by means of the notions of sym- 
bolism and assertion, giving each their rights, and to avoid the 
difficulties attendant upon introducing the notion of the meaning 
of symbols. So far assertion has remained a mysterious and 
unexplained tertium quid, over against symbol and object. It is 
now time to ask what actually happens when I make an assertion 
such as, 'the light is burning.' This sentence is a symbol in our 
own sense, which means that it uniquely corresponds to another 
physical fact. And it is evidently the question as to how this 
other fact, to which our symbol uniquely corresponds, is selected 
in such a way as to constitute its denotation. This is accom- 
plished by assertion, and in order best to understand how this 
is done, let us look at the process of verification. When I doubt 
the assertion that the light is burning, I look to see whether in 
fact the light is burning. Or if I am not interested in verification 
as such, I may go and press a switch ,to turn the light out, or may 
sit content if my desire is that the light burn. Always the point 
is that I commit myself to some course of action which has refer- 
ence to the specific objective situation to which the symbol 
corresponds. Thus the whole significance of the symbol re- 


garded as that which stands for something other than itself would 
seem to be that it forms an element in a context of action directed 
towards that for which it stands. The only apparent exception 
to this generalization appears to arise when I have no practical 
interest whatever, not even that minimum of practical interest 
which is exhausted by the process of verification, but merely 
contemplate the sentence and its implications. But here it 
ceases to be a symbol in our sense, and is treated merely as an 
algebraical equation . 

Thus the assertion of a symbol may be taken to mean its 
selection as an element in behavior directed towards some 
other object, so that the symbol stands for this object. It is 
behavior that sets up the unique correlation of symbol and 
object which is required and which is the basis for a theory of 
truth. The introduction of this concept of behavior fills in all 
the lacunae which a dialectical analysis reveals in the system of 
epistemological symbolism which has been presented. It is not 
our purpose here to attempt any thoroughgoing account of the 
detailed functioning of behavior in linking up symbol and object, 
but only to point to this function as providing the solution of our 
difficulties. In general, an epistemology which is merely dualistic 
breaks down. It cannot account for truth and error, and in these 
terms it can never arrive at an adequate analysis of symbolic 
complexes. A symbolic complex involves something more than 
mere word-structures on the one hand, and factual situations on 
the other. It is a complex of unique structure and constitution, 
and is found only where consciousness, objectively interpreted as 
a functioning organism, in the course of a train of behavior, 
makes use of one set of physical facts as conveniences in dealing 
with another, so that the two sets are correlated, and the first 
'stands for' the second. Only on these terms is it possible to 
have symbols at all. 

The whole of this discussion may now be gathered up and 
brought to a head by passing over several more or less unrelated 
points, (a) It is to be observed that this theory of meaning 
is humanistic and involves the complete abandonment of the so- 
called subjective point of view. We are not dealing with ideas 
or purposes or other mystical and mysterious entities. We are 


dealing with entities whose undoubted reality is obvious to the 
most tough-minded common sense. Meaning becomes for us a 
unique and complex ordering of physical entities set up by the 
introduction of an organism capable of specific response. The 
entire truth problem ceases to possess any cosmic breadth, and 
becomes centered round man. 

(b) We have obtained a point of view which seems very suc- 
cessfully to avoid the peculiar complexities which emerge in all 
theories dealing in propositions or objectives, and to avoid them 
by putting the discussion on quite another basis. A successful 
analysis of knowledge in terms of behavior, symbol and object 
demonstrates the uselessness of objectives and kindred concep- 
tions, by filling their places with more manageable categories. 
And surely there is something repugnant to common sense in the 
very notion of propositions or objectives, isolated pieces of mean- 
ing vaguely floating in some undiscovered region, and, like the 
ether, privileged against all the laws of logic. Our own analysis 
has at least this advantage that it deals only in the ponderable 
and valid elements whose presence in the knowledge-situation 
is not open to doubt. 

(c) The interpretation which sees in the assertion of a secondary 
symbol simply its use by consciousness objectively understood 
is obviously and immediately applicable in the case of primary 
symbolism. That sense-datum which stands for the totality of 
sense-data which constitute the object in question is simply that 
which is selected by the organism in question. With secondary 
and primary symbolism alike the moment of assertion is the same. 
In the latter as in the former case the presence of a sentient being 
is the necessary condition for setting up any symbolic complexes. 

(d) All this suggests that the province of the science of episte- 
mology is to fill in the outlines of the general theory, and to show 
in specific cases how consciousness brings together symbol and 
object. It is this painstaking and detailed analysis of the various 
types of symbolic complex, a study which obviously cannot even 
be begun here, which must go to make up the body of the theory 
of knowledge. 

James L. Mursell. 

White Plains, New York.