THE ANALYTIC AND SYNTHETIC

Herbert E. Hendry
1959

 

ll
-. Thesis for ”10 Degree of M. A.
MICHIGAN STATE UNIVERSITY

IIIIIIIIIIIIIIIIIIIIIIIIIIII

3 1293 10598 5455 ‘

  
    
   

 
 

Lamar

Michigan Sun
Univm'sity

 

 

‘s

 

 

‘. 1w-

m;¢a-a\1~e
.’

330 034?
a 7

. 3.
F3 "‘ r "\

. J

4

1‘

..'_‘V

 

 

THE ANALYTIC AND SYNTHETIC

BY
HERBERT E. HENDRY

A THESIS
Submitted to the College of Science and Arts
Michigan State University of Agriculture and
Applied Science in partial fulfillment of
the requirements for the degree of

MASTER OF ARTS ,f, v.1/

Department of PhilOSOphy ‘ , 4" (:

1959

ACKNOWLEDGMENT

I should like to express my gratitude to
Professor William J. Callaghan for his help in the

preparation of this paper.

TABLE OF CONTENTS

page
I. ANALYTIC AND SYNTHETIC . . . . . . . . 1
II. AN HISTORICAL SKETCH . . . . . . . . 5
III. RECENT CRITICISNS . . . . . . . . . 25
IV. AN EXAMINATION OF THE CRITICISNS . . . 39
v. ANALYTICITY AND NATURAL LANGUAGE . . . 67
VI. SOME CONCLUDING REMARKS . . . . . . . 81

Bibliography . . . . . . . . . . . . . . . . 89

I ANALYTIC AND SYNTHETIC

An important and a frequently occurring concept
in modern philosophy, especially in modern empiricism,
has been that two distinct kinds of statements (proposi-
tions or judgements) can be distinguiShed and classified.
On the one hand there are said to be analytic statements;

on the other, synthetic statements. The difference be-

 

tween these two kinds of statements can be characterized,
though inadequately, by a description of the method by
which we come to determine their truth-values. For ex-
ample, consider the statement: "Generally speaking,
bachelors are more wealthy than married men." In order
to determine whether this statement is true, or false,
one should make an investigation into the respective in-
comes and savings of bachelors and married men; then by
comparing his findings, he should presumably arrive at a
conclusion. Statements of this kind are generally re-
garded as synthetic. On the other hand, if one considers
a statement like:

(1) All bachelors are unmarried males,

and if he understands what the words "bachelor," "unmar-

ried," and "male" mean, he knows that the statement is

true without making an empirical investigation into the

marital status of bachelors and married males. State-

ments of this kind are generally regarded as analytic.

But, still, this is not really enough to charac-
terize analytic statements. For we must also understand
the meanings of "all" and "are." Words like these are
usually called logical particles or logical constants.
Among others, expressions like "if," "then," "no," "non-"
and "un-" are also logical particles. These also play
an important part in the notion of analytic statements.
For example, we know that the statement:

(2) No non-banausiclites are banausiclites,

is true even though we might not understand the meaning
[of the word "banausiclite." Thus it is not enough, nor
is it in itself always necessary, to understand the mean-
ings of all the descriptive words (i.e., words which are
not logical particles) in order to determine the truth or
falsity of an analytic statement. But, generally speak-
ing, an analytic statement is one whose truth-value can
be determined by merely attending to the meanings of its
terms and to pertinent rules of syntax (i.e. those rules

which determine the correct usage of logical particles).

It may be noted, in passing, that the distinction
between logical particles and descriptive terms leads to

a distinction between two kinds of analytic statements.

First, there are logically true statements, i.e., state-
ments which are substitution instances of the truths of
logic. Second, there are statements which can be changed
into logical statements by substituting synonyms for 5127
231mg, Thus, statement (2) above represents an analytic
statement of the first kind. Statement (1) represents a
statement of the second kind. For, if we replace the ex-
pression "unmarried male" with its synonym "bachelor," we
have the statement "All bachelors are bachelors" which is
like statement (2). In this case, it represents a logical

truth.

Loosely speaking, it may be said that an analytic
statement is one whose truth-value can be determined sole-
ly by attending to the meanings of its terms, and a syn-
thetic statement is one whose truth-value can be determined
only by resorting to extralinguistic facts. Since analytic
statements have their truth or falsity determined by lin-
guistic convention (grammar, syntax), it is normally held
that they are void of empirical content, that is, that they
say nothing about empirical facts and that, in this sense,
they are uninformative. Synthetic statements, however, are
held to have empirical content; and in this sense, if they

are true, are held to be informative.

The making of this distinction has played an impor-

tant role in modern phIIOSOphy. It has appeared in the

writings of such authors as Hobbes, Locke, Berkeley, Hume,
Leibniz, Kant, and, on the contemporary scene, in the
works of the logical positivists, especially Ayer and Car-
nap, and in the works of many other empiricists. Recently,
however, the distinction has been severely criticized as
being ill-founded and untenable. With an end in view of
determining whether these criticisms are wholly justified,
it is the purpose of the following pages to examine some
of the arguments which have been advanced in favor of aban-
doning the distinction. It will be contended that though
the arguments advanced by the gradualists1 impose serious

problems, they do not justify a rejection of the distinction.

Before proceeding, however, it seems worth while
to present a brief historical sketch of the development of
this distinction in modern philosophy. This sketch, how-
ever, does not purport to be historical in the proper
sense; it will not trace the origin and growth of ideas.
Our chief concern is expository. This sketch will be use-
ful in two ways. First, it will afford a better charac-
terization of the distinction than the admittedly inade-
quate one offered above; and, second it will afford some
picture of the significance which has customarily been

associated with the distinction.

 

1The name "gradualist" is used to refer to those
who deny the tenability of the analytic-synthetic distinc-
tion. The name "dichotomist" is used to refer to those who
maintain the tenability of the distinction.

II AN HISTORICAL SKETCH

It is exceedingly difficult to find any place
in the history of phiIOSOphy to begin an account of the
analytic-synthetic distinction. The words "analytic"
and "synthetic" are Old and have many meanings. How-
ever, we are not here concerned with the words but with
what they mean in the sense described in the previous
section. The first philOSOpher to use the words in
this sense was probably Kant. But, clearly, Kant was
not the first to recognize the distinction (though he
thought to the contrary). The same, or at least a very
similar notion, is found in the works of Leibniz, Hume,
Locke and certainly in many lesser philosophers. The
following exposition begins with the views of John Locke.
However, the choice is not entirely arbitrary, for in
Locke we find the first modern philOSOpher who is
thoroughly concerned with language and its pervasive
importance in any epistemological inquiry. But, we
must be aware that Locke is not, in any sense, the
originator of the distinction; for we can go as far back

as the thirteenth century and find John Duns Scotus

concerned with the same dichotomy.l Indeed, we must
assume that men have always been aware that some state-
ments are true "just by definition," that some state-

ments are "trivially true," and in this sense Obvious.

For Locke, names are signs of ideas. Ideas, in
turn are signs of things which affect the mind. This
appears, today, as a rather naive and overly simple the-
ory of language. Indeed, Locke encounters immediate
difficulties in the absence Of the type-token distinc-
tion. It would seem that he needed the distinction not
only for names but also for ideas. This is especially

. . . 2 .
true once he begins to deal with universals. This,

 

lScotus argued that we should not admit a dis-
tinction between prOpositions which are analytic in them-
selves and propositions which are analytic "for us."
Thus, that a greater or lesser number Of persons do not
possess the requisite understanding of a prOposition's
component terms does not affect its analyticity.

2It is intimated that Locke was completely un-
aware of a need for distinguishing between the sign-
types and sign-tokens. This suggestion, however, is
not entirely correct as can be seen in the following
passage: "To return to general words, it is plain by
what has been said, that general and universal belong
not to the real existence of things; but are the inven-
tions and creatures Of the understanding, made by it for
its own use, and concern only signs, whether words or
ideas. Words are general, as has been said, when used
for signs of general ideas, and so are applicable indif-
ferently to many particular things; and ideas are gener-
al, when they are set up as the representatives of many
particular things: but universality belongs not to
things themselves, which are all of them particular in
their existence; even those words and ideas, which in
their signification are general." (John Locke, Ag Essay

however, is beside the point. Our interest is in Locke's
concern with language. Clearly, in this respect, he was
one Of the first philosOphers to recognize that the prob-
lems of language are inextricably bound up with those

of epistemology. And, he further realized that language
constitutes a major instrument for the acquisition and
extension of knowledge. In this respect one cannot fail
to appreciate his interest, acumen, and genuine contribu-

tions.

Locke classifies knowledge in several different
ways: intuitive, demonstrative, sensitive; actual or-
habitual; real or trifling. We are here concerned with
his distinction between real and trifling. Locke writes:

Before a man makes a prOposition, he is supposed
to understand the terms he uses in it, or else he
talks like a parrot, only making a noise by imita-
tion, . . . ; but not as a rational creature using
. . . fiverd? for signs Of ideas which he has in
his mind. he hearer also is supposed to understand
the terms as the speaker uses them... ... And there-
fore he trifles with words who makes such a proposi-
tion, which when it is made contains no more than
one of the terms does . . . ; v.g., "A triangle hath
three sides,". . . . And this is no further toler-
able than where a man goes to explain his terms to
one who is supposed or declares himself not to under-
stand them.

 

Concerning Human Understandin , ed. A. C. Fraser, Oxford:
CIarendon‘Press, n.d., VOI II, pp. 21-22.)

3Ibid., II, 298. NOte in this passage Locke's
insistence tfiat words must have some mental counterpart,
i.e., understanding or "idea." He is, I believe, quite
right in making this demand. This sort of thing suggests
the inadequacies of rigorous behaviorism. The problem of

He continues:

. . . We can know then the truth of two sorts
of prOpositions with perfect certainty: the one is,
Of those trifling propositions which have a certain-
ty in them, but it is only a verbal certainty, but
not instructive. And, secondly, we can know the
truth, and so may be certain in propositions which
affirm something of another, which is a necessary
consequence of its precise complex idea, but not con-
tained in it. . . .

For our present purposes, the important point to
note is Locke's attitude toward analytic (i.e., trifling)
propositions.5 The very terminology he uses suggests
that he sees neither phiIOSOphical significance nor prob-
lems which might be associated with the distinction. He
makes no attempts to offer a thoroughgoing justification
for making the distinction. Nor can we say that he makes
any serious attempt to define "trifling prOpositions" or
"real prOpositions." For Locke it is perfectly Obvious
that some propositions are trifling; they say nothing
about the world, and that is all there is to it. Their

only justifiable use is in informing the linguistically

uninformed.

 

adequately describing "mental counterpart," however, re-
mains unsolved.

thid.

5Also, it is worth noting that Locke emphasizes

the linguistic foundation of (at least some) "certain"
prOpositions more than, for example, Hume does.

In turning from Locke to the writings of Hume,
we find, on the other hand, great significance attached
to the analytic-synthetic distinction. Indeed, it forms
a pivotal part of his epistemology. We need not get in-
to the involved task of giving an exposition of Hume's
attempt to describe the difficult notion Of "relation."
The important point is that he makes a distinction be-
tween propositions expressing "matters of fact" and
propositions expressing "relations of ideas" and that
these notions correspond, roughly, to what have been
characterized as "analytic statements" and "synthetic

statements."

In the Enquiry Hume writes:

All the objects of human reason, or inquiry may
naturally be divided into two kinds, to wit, rela-
tions gf ideas, and matters 9; fact. Of the first
kind are the sciences of geometry, algebra, and
arithmetic; and in short, every affirmation which
is either intuitively or demonstratively certain.
That the square 2; the hypotenuse ii equal to the
sguares gig—the two mes, is a prOpOSItion GEEK"
expresses a relation between these figures. . . .
Propositions of this kind are discoverable by the
mere operation of thought, without dependence on
what is anywhere existent in the universe. Though
there never were a circle or triangle in nature, the
truths demonstrated by Euclid would for ever retain
their certainty and evidence.

Matters of fact, which are the second Objects
of human reason, are not ascertained in the same man-
ner; nor is our evidence of their truth, however
great, of a like nature with the foregoing. The con-
trary of every matter of fact is still possible; be-
cause it can never imply a contradiction, and is con-
ceived by the mind with the same facility and distinct-
ness, as if ever so conformable to reality. That‘thg
sun will not rise tomorrow is no less intelligiEIe
a proposition, and—Implies no more contradiction than

 

10

the affirmation, that it,will rise. We should in
vain, therefore, attempt to demonstrate its false-
hood. Were it demonstratively false, it would imply
a contradiction, and could never be distinctly con-
ceived by the mind.6

 

One cannot fail to recognize the striking paralled be-
tween Hume's notion of "relations of ideas" and analytic
propositions; and between his notion of "matters of
fact" and synthetic propositions. This passage, no
doubt, leaves much to be desired by way of clarity.
Just what is meant by "intuitively or demonstratively
certain" and "the mere Operation of thought" is, to say
the least, obscure. Whatever clarity the passage does
have, arises through his characterization of matters of
fact as being expressed by prOpositions which are nei-
ther themselves self-contradictory nor have as their
denials self-contradictory propositions. This notion
of self-contradiction, of course, was later to become

a customary way of marking the analytic-synthetic dis-

tinction.

One more remark concerning this passage deserves
to be made. In the Treatise Hume wrote and argued that
geometrical propositions express matters of fact, not
relations of ideas. It is generally agreed today that

the view expressed later in the Enguigy is more nearly

 

6

David Hume, Enguiries Concerning the Human
Understanding and Concerning the Principles ngNOrals,
ed. L. A. SEIby-Bigge, (2d ed.; Oxford: CIarendon Press,
1902), pp. 25-26.

11

correct. This is not the place for textual exegesis.

However, it may be said that there are probably two rea-

sons why Hume, in the Treatise, arrived at the conclu-
sion he did. First, he failed to distinguish between
"pure geometry" and "applied geometry." Second, he con-
fused the logical questions of geometry with the prob-
lems of genetic psychology. Had Hume seen these distinc-
tions more clearly, he probably would have contended

from the start of his work that geometrical propositions

express relations of ideas.

Whatever are the adequacies or inadequacies of
Hume's distinction, our chief concern here is with see-
ing what significance he attached to these notions. In
concluding his inquiry into human understanding, Hume
writes:

When we run over libraries, . . . , what havoc
must we make? If we take in our hand any volume;
of divinity or school metaphysics, for instance;
let us ask, Does it contain agy abstract reasoning
concerning quantity or number? No. Does it con-
tain an my experimentaI—reasoning concernin fig matter
of fact and existence? No. Commit it then to the
TIames: fOrIIt can contain nothing but SOphistry
and illusion.7

This passage is to some extent reminiscent of the early
searchings of logical positivism for a magical principle
of verification. Though we cannot call Hume a logical

positivist, the similarity is there.- Essentially they

7Ibid., 165.

12

are both concerned with the same problem. This.is seen
in A. J. Ayer's question concerning the above passage
from Hume. "What is this but a rhetorical version of
our own thesis that a sentence which does not express
either a formally true proposition or an empirical hy-

8

pothesis is devoid of literal significance?"

PrOpositions to be meaningful (or, in Hume's
language, not laden with "SOphistry and illusion") must
be either analytic or synthetic. That is they must ex-
press either relations of ideas or matters of fact. In
the latter case, they have empirical content, i.e.,
they refer to ideas derived from primordial sense impres-
sions. These propositions, however, are never completely
confirmed, and we can never be certain about their truth.
But, what then, about the truths of mathematics? These
are not only believed to be true, but necessarily true.
To say that they are synthetic is to say that they are
not necessary. To say that they do not refer to experi-
ence is to say that they are not empirical. The only
alternative for Hume was to say that they are analytic
(i.e. that they express relations of ideas). Thus they

are both meaningful and necessary.,

 

8A. J. Ayer, Language, Truth and Lo ic (2d ed.;
New York: Dover Publications, Inc., $947 , p. 54.

13

In turning to the philOSOphy of Leibniz, one
finds the highest of importance attached to the analytic-
synthetic distinction. (He makes the distinction one
between "truths of reason" and "truth of fact.") In ex-
pounding this view, however, there is some difficulty;
for, as Bertrand Russell has pointed out, Leibniz had
two systems of philosophy. One was exoteric and was

"based on theological optimism" while the other was eso-
teric and "based on two premises, the law of contradic-

"9

tion and the law of sufficient reason. We are con-
cerned here chiefly with Leibniz's esoteric philosophy.
This, of course, may not be entirely legitimate, for

the Russellian interpretation may not be correct. How-

ever, this is a frequent problem in the history of philos-

(3th 0

Without getting into a lengthy discussion of
metaphysics, we can briefly say that Leibniz was a kind
of metaphysical atomist. The world consists of an in-
definitely large number of infinitely small (i.e. unex-
tended) spiritual substances. These substances are
called monads. Each monad is "windowless." By "window-
less" it is meant that each monad is complete in itself

and unaffected from without (except, presumably, by God).

 

9Bertrand Russell, 5 History 9; Western Philoso-
phy: And Its Connection With olitical and Social gig:
cumstances From the Earliest Times to Egg Present‘Day_
(New York: Simon and schuster, 19437, pp. 531-596.

14

Thus, if each monad is unaffected from without, all of

its attributes are determined from within.

Russell succinctly states the relevance of the
two premises mentioned above. "Both depend on the no-
tion of an analytic prOposition. . . . The law of con-
tradiction states that all analytic prOpositions are
true. The law of sufficient reason (in the esoteric
system only) states that all true propositions are ana-

lytic."10

Now, because every attribute of any substance
(i.e., monad) is determined from within, it is necessary
that each monad has those attributes which it actually
does have; for if it did not, it would not be the same
substance. As a result, it would be self-contradictory
to deny, for example, that the Washington Monument is
555 feet and 1/8 inch tall. Thus, there are no truths
of fact. Every truth is a truth of reason. But the
distinction, however, still bears some significance.
For, man in his finitude never has complete knowledge

of any one substance; this alone would enable him to
deduce knowledge concerning all of its attributes.
Consequently, foreman, there is a difference between

truths of reason and truths of fact.

The foregoing remarks on Leibniz have neces-

sarily been sketchy and incomplete. Perhaps his most

 

lOIbid., 592.

l5

important contribution, in respect to the tOpic with

which we are concerned, is his definition of analytic
’propositions (i.e., truths of reason) as propositions
whose contradictories are self-contradictory. We have

already seen this suggested by Hume and we shall also

see it suggested by Kant.

One thing that we must notice is that Leibniz
attaches metaphysical as well as logical and epistemo-
logical importance to the analytic-synthetic distinc-
tion. And, we must further notice, that, although his
notion of analytic prOpositions is similar to that of
Hume and Kant, he is in radical disagreement with them

as to which propositions are analytic.

In Kant, a similar distinction is made. He

writes:

Either the predicate B belongs to the subject
A as something contained (though covertly) in the
concept A; or B flies outside the sphere of the con-
cept A, though somehow connected with it. In the
former case I call the judgement analytical in the
latter synthetical. Analytical judgements (affirma-
tive) are therefore those in which the connection

of the predicate with the subject is conceived through

identity, while others in which that connection is
conciived without identity, may be called syntheti-
cal.

 

11Immanuel Kant, A Criti ue of Pure Reason,
trans. Max Muller (2d ed. rev.; sew York: Macmillan
Company, 1927), p. 5.

16

This definition has often been criticized as being stated
in metaphorical language. However, there is good indica-
tion that Kant had in mind a notion very similar to that
of Leibniz.12 Even if this were not true, there is no
reason why the metaphor could not be given precision as,
for example, C. I. Lewis has given some degree of pre-

cision to "inclusion."13

It is difficult to say precisely what importance
Kant attached to this notion. A plausible answer might
be that it merely sets the groundwork for his argument
regarding the existence of synthetic g priori judgements.
Another reason might be that he felt it to be descrip-
tive of the mind in its act of judging. Some judgements,
he writes, are ampliative. Others are explicative. Ex-
plicative judgements add nothing to the concept of the
. predicates. Yet they do seem to cast concepts into order.
This presumably characterized the way the mind works

(ought to work?).

 

12For example, this is indicated when he writes:
"For as it was found that all mathematical conclusions
proceed according to the principle of contradiction
(which is required by the nature of all apodictic cer-
tainty), it was supposed that the fundamental princi-
ples of mathematics also rested on the authority of the
same principle of contradiction." (Ibid., 720) Many
other examples could have been cited. For, an argument
that Kant availed himself of two different principles,

see: Richard Robinson "Necessary PrOpositions " Mind
LXVII 1958), pp. 289-361..) ’ """'

13c. I. Lewis, "The Modes of Meaning," Philosophy
and Phenomenological Research, IV, (lth), p. 240.

17

Kant, unlike Hume, does not identify mathemati-
cal prOpositions with analytic prOpositions. Here,
then, is found an important clash of opinion. Analytic
propositions are true by virtue of their meanings; they
do not extend knowledge. Synthetic propositions are
not true by virtue of their meanings and do extend knowl-
edge. Yet there is disagreement as to which proposi-
tions are analytic and which are synthetic. This, in
conjunction with the above remarks on Leibniz, serves
as an indication that there might be something radical-
ly wrong with the distinction. That a lack of agree-
ment is possible, undoubtedly, is one of the reasons

which has occasioned recent criticisms of the notion.

Turning to Mill, we meet an altogether differ-
ent situation. As a description of verbal (analytic)
and real (synthetic) prOpositions he writes:

An essential proposition, . . . , is one which
is purely verbal, which asserts of a thing under a
particular name only what is asserted of it in the
fact of calling it by that name, and which, there-
fore, either gives no information, or gives it re-
specting the name, not the thing. Non-essential,
or accidental prOpositions, on the contrary, may
be called real prOpositions, in opposition to ver-
bal. They predicate of a thing some fact not in-
volved in the signification of the name by which
the preposition speaks of it, some attribute not
connoted by that name.

 

“John Stuart Mill, Philosophy of the Scientif-
;E_Method, ed. Ernest Nagel (Nedeork: -Hafner Publish-
1ng Company, 1950, p. 87.

18

Furthermore, real propositions, "if true, add to our
knowledge; they convey information not already involved

in the names employed."15

And he adds: "Accordingly,
the most useful, and in strictness the only useful kind
of essential prOpositions, are definitions which to be
complete, should unfold the whole of what is involved
in the meaning of the word defined, that is (when it is

16

a connotative word), the whole of what it connotes."

Mill is apparently in agreement with Locke in
regard to analytic propositions and the extension of
knowledge. Also, they appear to be in agreement over
the uses of analytic propositions. For both philoso-

phers, analytic propositions are pedagogical devices

or definitions, but no more.

Mill says explicitly that the distinction he

is drawing "corresponds to that which is drawn by Kant
and other metaphysicians between what they term analyt-
.ig and synthetic judgements."l7 The correspondence be-
tween the Kantian and the Millian notion also includes

a general agreement as to the extension of these terms.
This is true, at least, concerning the description of
mathematical prOpositions. For both hold that mathemati-

cal propositions are synthetic. The similarity ends

 

152.49." 88' 16M.) 850 l7Ibido, 88H.

19

here, however, since for Mill mathematical truths are
well established generalizations from experience; and

Kant, of course, takes them to be‘g priori truths.

On the contemporary scene the notions of ana-
lytic and synthetic statements have gained a great deal
of attention; with this attention they have, of course,
gained a greater degree of precision. A good example of
recent concern can be found in A. J. Ayer's Lan ua e,

Truth and Logic.

 

Ayer gives a cogent statement of his problem.
The empiricist, he says, encounters difficulties in con-
nection with the truths of formal logic and mathematics.

For whereas a scientific generalization is read-
ily admitted to be fallible, the truths of mathemat-
ics and logic appear to everyone to be necessary
and certain. But if empiricism is correct no propo-
sition which has a factual content can be necessary
or certain. Accordingly the empiricist must deal
with the truths of logic and mathematics in one of
the two following ways: he must say either that
they are not necessary truths, in which case he must
account for the universal conviction that they are;
or he must say that they have no factual content,
and then he must explain how a proposition which
is empty of all factual content can be true and use-
ful and surprising.

If neither of these courses proves satisfactory,
we shall be obliged to give way to rationalism. . . .
Or else we must accept the Kantian explanation which,
. . . only pushes the mystery a stage further
baCkol8

 

18Ayer, 72-73.

20

The problem Ayer is here concerned with is essentially
the same as Hume's problem. Ayer, of course, has bene-
fited from.the knowledge of new discoveries in geometry
and the work of Russell and Whitehead in logic and mathe-
matics. The contention that mathematical prOpositions
are analytic or tautologous became almost an irrevocable
dogma of empiricism in the early part of the present
century. This was largely due to Russell and Whitehead's
Principia Mathematics (1910-1913) and Wittgenstein's I

Tractatus Logico-PhiIOSOphicus (1922). The formerof

 

these works is an attempt to prove that pure mathematics
is deducible from logic. The latter work is concerned,
in part, with clarifying the nature of tautologies.
Although it is possible to question the complete suc-
cess of these two works, one can dispute neither their
significance, their contribution to the clarification

of logical and mathematical problems, nor their influ-
ence on the thinking of contemporary philosophers.
(White remarks that the only word to describe the achieve-
ment of the Principia is "monumental".)19 Consequent-
ly, it is quite natural that Ayer should be more fully

aware of the nature and gravity of the problem.

It will be worth while to see how Ayer attempts

to give a solution to his dilemma.

 

19M. G. White, Toward Reunion in Philosophy,
(Cambridge: Harvard University‘Press,_l956), p. 290.

 

vy’e5'9.

Ir-l: 7*“-
“.U'Z‘I.

 

 

 

21

The principles of logic and mathematics are
true universally simply because we never allow
them to be anything else. And the reason for this
is that we cannot abandon them without contradic-
ting ourselves, without sinning against the rules
which govern the use of language, and so making our
utterances self-stultifying. In other words, the
truths of logic and mathematics are analytic propo-
sitions or tautologies.20
As for the surprising element in analytic statements, he
writes that "they call attention to linguistic usages,
of which we might otherwise not be conscious, and they
reveal unsuspected implications in our assertions and//"

beliefs."21

This view or very similar views have been quite
popular in contemporary empiricism and especially in the
movement of positivism. The only alternatives, it seems,
are a Kantian view of the g priori, which is eschewed by
logical positivism and really out of vogue for a great
many of the other more liberal empiricists; or the Mill-
ian view that mathematical statements are highly probable
empirical statements. The latter view is held to be

wrong because it confounds the contexts of discovery and

justification.22

If we turn to Ayer's definition of "analytic"
and"Synthetic," we see an attempt to give a sharper

clarification to these notions than, for example, the

 

21
ZOAYGI', 77. Ibido, 79-80.

22For example, ibid., pp. 73-75.

22

Kantian definition, which uses such terms as "concept"

and "judgement." These terms, needless to say, are

held suspect by many contemporary empiricists.

"A proposition," Ayer writes, "is analytic when\
its validity depends solely on the definitions of the
symbols it contains, and synthetic when its validity is

23 ,/

determined by the facts of experience." ///

This is an apparent improvement over the Kant-
ian definition in that it eschews the above mentioned
controversial notions. However, it still can stand im-
provement. For example, it ignores invalid prOpositions.
We then have a three-fold division: analytic prOposi-
tions; synthetic propositions; and invalid propositions.
But this is not adequate, for we also want to speak a
about invalid analytic and invalid synthetic proposi-
tions. For example, we want to say that "Not very many
bachelors are married" is an invalid analytic prOposi-
tion. Or, we could introduce a new term, for example,
"self-contradictory proposition." But, then what do we
call invalid synthetic prOpositions? Nor does Ayer seem
to be in accord with customary practice when he speaks
of invalid and valid propositions. Clearly, it would
be less confusing to speak of analytically true prOposi-

tions, analytically false propositions, and similarly

 

23Ibid., 78.

23

for synthetic propositions. This point has been dwelt
on because it seems to be an error into which many con-

temporary writers have fallen.

As has been said, the foregoing is admittedly
a cursory examination of the analytic-synthetic distinc-
tion and its vicissitudes in the history of phiIOSOphy.
Other figures could have been selected, perhaps, more
judiciously. The Opinions of rationalist philOSOphy,
with the exception of Leibniz's, have been conspicuous-
ly omitted. And certainly, the discussion could have

begun as far back as Aristotle and even further.

However, since we have not attempted to write
history proper but to select a reasonable cross section
of views, this method does not appear to be wholly un-
warranted. And, moreover, it gives some indication of
the relevance of the analytic-synthetic distinction in
modern empiricism. For example, we see that there has
by no means been a unanimity of opinion in connection
with this problem. Nor can we ascribe the relevance
associated with the distinction to a gradual histori-
cal develOpment, with later empiricists attaching more
significance to it than earlier empiricists. The simi-
larity in this respect between Locke and Mill vitiates
this view, unless, of course, we wish to count Hill as

an exception.

21.

Also, we see a wide divergence of opinion as to),
the extension of the word "analytic." PhiIOSOphers, as
different as Mill and Kant, hold mathematical prOposi-
tions to fall without the extension of "analytic", where-
as for Hume and Ayer, they do not. And, even between
Hume and Ayer there is a divergence of Opinion. In the
Treatise, as we have seen, Hume held that the proposi-
tions of geometry are synthetic; according to Ayer they
are analytic. And, according to Leibniz, all true propo-
sitions are analytic with the exception of those which .

.1"

predicate existence of their subjects. (/r

Often, when a doctrine in phiIOSOphy becomes ful-
1y established it is sometimes forgotten that the doc-
trine is by no means self-evident, but has merely become
accepted unquestioningly through frequent use. This is
perhaps the case with the analytic-synthetic distinction.
And, it is a welcome contribution of those who have called
the distinction into question. We must agree with this
fact whether or not we accept the conclusions which they
have drawn. In short, the thesis is not in any sense ob-

vious, and consequently, it needs to be defended.

With this in mind, we may turn to an examination
of the criticisms which have been advanced in favor of

abandoning the distinction.

III RECENT CRITICISMS

In recent years, since the initial criticisms
by W. V. Quine and M. G. White, there has been a bar-
rage of discussions and learned articles concerning
the traditional distinction between analytic and syn-
thetic statements. We have witnessed prOposals and
criticisms, counter-proposals and counter-criticisms
with such rapidity that the whole controversy is so
absurdly chaotic, that it is extremely difficult to
see just what the problem is. Moreover, there appears
to be little agreement, even within the two aligned
camps. Ostensibly, the problem is this: Is there an
absolute dichotomy between analytic statements and syn-
thetic statements? Or, on the other hand, is this a
hazy distinction with no clear line of demarcation;
that is, are some statements more or less analytic than
others, and are others more or less synthetic than these?
Is the distinction one of degree rather than kind?
Again, ostensibly, the question seems quite clear cut

and readily amenable to solution.

But this controversy has now been nurtured for

quite some time, and the prospect of any reasoned

26

agreement seems very dim. One is not repulsed by the
philosopher's inability to give a definitive answer

to the "big" questions usually associated with his pro-
fession. we stoically resign ourselves to this fact.
Though the recent controversy is not one of the "big"
problems, it is concerned with an important one and

it is closely related to "big" problems. For this rea-
son one impatiently awaits a satisfactory solution.

It is true that an inadequate distinction should be
abandoned. This no one denies. But, it is equally
true that a distinction which has proved as useful as
the analytic-synthetic distinction, if it can legiti—

mately be maintained, should not be abandoned.

The word "useful" demands some elaboration.
From the foregoing historical sketch, we have seen some
of the uses to which the distinction has been put.
All of the men we have considered use it as a basis
for a classification of knowledge. Locke, for example,
was enabled to make a partial identification of "cer-
tainty" and "trifling." Hume and Ayer were able to
identify mathematical propositions with analytic prOp-
ositions. More recently, the distinction has led to
a classification of the sciences into "formal" (logic,

pure geometry, etc.) and "non-formal" (physics, astronomy,

27

etc.).1 The distinction has also led to the identifi-
cation of logical propositions with analytic proposi-
tions. For example, "p v -p" is said to be analytic
or a tautology. Examples of these kinds, though they
do not justify the analytic-synthetic distinction, do

exhibit its usefulness.

The following section is devoted mainly to an
exposition of the nature of the arguments which have
been advanced to exhibit the untenability of maintain-
ing the distinction. The arguments are those given by
W. V. Quine in "Two Dogmas of Empiricism."2 These argu-
ments offer an excellent example of the misgivings of

other authors, for example, Nelson Goodman and M. G.

White.
Quine considers several definitions of "analytic":
(l) A statement is analytic if its denial is self-
contradictory.
(2) A statement is analytic if it is true by vir-

tue of its meanings and independently of fact.

 

1For a discussion of the distinction between for-
mal and nonformal science see the article "Formal and
Factual Science," by Rudolph Carnap in: Herbert Feigel
and May Brodbeck (eds.), Readin 3 1g the Philosophy of
Science (New York: Appleton-Century-Crofts,'Inc., 1953),
pp. I2§-128. Cf. Richard Rudner "Formal and Nonformal,"
Philosophy_g§,Science, XVI (19A9 , pp. Al-A8.

2W. V. Quine, "Two Dogmas of Empiricism," From a

Lo ical Point pf View: Nine Logico-PhiIOSOphical Essa 5
(Cambridge: Harvard University Press, 1953), pp. QO-h%.

28

(3) . A statement is analytic if it is a logical
truth or can be turned into a logical truth by
putting synonyms for synonyms.

(A) A statement is analytic if it is true in
every state-description.

(5) A statement is analytic if it can be reduced
to a logical truth by definition.

(6) A statement is analytic if and only if it

is true according to the semantical rules of

the language in which it appears.3

Definition (1) is of little value. For the
notion of self-contradiction, "in the quite broad sense
needed for . . . a definition of analyticity," Quine
writes, "stands in exactly the same need of clarifica-

tion as does the notion of analyticity itself. The

two notions are but one side of a single dubious coin.""

As for definition (2), it confounds meaning
with extension, and the sense of "meaning" presupposed
is specious. It will be well to dwell on this objection
for it has determined the direction of recent discus-

sions of the issue.

Following Frege, Quine distinguishes between

meaning and naming; or between meaning and extension.

 

3Cf. Benson Mates "Analytic Sentences," The
Philosophical Review, Li (1951), pp. 525-531.. Prafessor
Mates considers Quine's criticism of these and two other
definitions not listed above: "§_is analytic if and
only if‘g is true in all possible worlds" and "S is ana-
lytic if and only if §.could not possibly be faIse."

l"Quine, 20.

29

Quins writes that the class of all entities of which

a term is true is called its extension. This is, pre-
sumably, offered as a definition. To show that the
meaning of a term is not identical with its extension,
Quins mentions Frege's example of "Evening Star" and
"Morning Star," and several other examples showing that
this distinction can be extended to cover not only sin-
gular terms but general and abstract terms as well.
"Meaning," however, is left undefined. And we must be
satisfied with Quine's association of "meaning" with

"connotation" or "intension."

In a crucial passage he writes:

For the theory of meaning a conspicuous ques-
tion is the nature of its objects: what sort of
things are meanings? A felt need for meant entities
may derive from an earlier failure to appreciate
that meaning and reference are distinct. Once the
theory of meaning is sharply separated from the
theory of reference, it is a short step to recog-
nizing as the primary business of the theory of
meaning simply the synonymy of linguistic forms and
the analyticity of statements; meanings themselves,
as obsgure intermediary entities, may well be aban-
domed.

Here Quine is expressing sentiments to be found in vir-
tually all of the literature arguing the inadequacy of

the analytic-synthetic distinction.

For example, White writes:

Of all the "analyses" bequeathed to us by the
age of meanings, those which introduce queer entities

 

5Ibid., 22.

30

are the most suspicious. They are all modeled after
the reasonable analysis of what it means to say that
Jones is a father. To say this is to say that Jones
is a male and that there is someone whom Jones begat.
Here no new kind of entity has been introduced for
which we must invent a relation comparable to grasp-
ing a meaning or to whatever it is that we do to a
sake. The son is as seeable, as touchable, and as
smellable as the father, but meanings are unholy
ghosts. The inflationary introduction of queer en-
tities is best abandoned in philosophy as the early
moderns abandoned occult qualities in physics. In-
stead of trying to account for the epistemological
fact of understanding 5y inventing occult entities,
we had best take the fact of understanding as un-
analyzed or try to explain it, clarify it, illumi-
nate it from a more helpful point of view.

And Nelson Goodman writes of synonymy:

Under what circumstances do two names or predi-
cates in an ordinary language have the same mean-
ing? Many and widely varied answers have been given
to this question, but they have one feature in com-
mon: they are all unsatisfactory.

One of the earliest answers is to the effect
that two predicates have the same meaning if they
stand for the same real Essence or Platonic Idea;
but this does not seem to help very much unless we
know, as I am afraid we do not, how to find out
whether two terms stand for the same Platonic Idea.7

Along with Essences and Platonic Ideas, Goodman also
rejects "images," "concepts," and "possibilities."
Further examples which express these sentiments can be

found in most of the gradualist's literature.

 

6
M; G. White, Toward Reunion in Philosophy, (Cam-

bridge: Harvard University Press, I936), pp. 39-hO.

7Nelson Goodman, "On Likeness of Meaning" in:
Leonard Linsky (ed.), Semantics ggg the Philosophngf
Lan uave (Urbana: The University of_IIliHOis‘Press,
- 1952), p. 67.

31

From the above and the discussion in Section I,
it can readily be seen why so much space has been de-
voted to the notion of synonymy. And this shows the im-

portance of definition (3).

But, of definition (3) Quins writes, " we still
lack a proper characterization of . . . analytic state-
ments, . . . , inasmuch as we have had to lean on a
notion of 'synonymy' which is no less in need of clarifi-

cation than analyticity."8

Definition (A) is also inadequate. The notion
of "state-description" in this definition is Carnap's.
Quins generously observes that Carnap was under no illu-
sions about this point. For, as Quins notes;
. . . this version of analyticity serves its pur-
pose only if the atomic statements of the language
are, unlike 'John is a bachelor' and 'John is mar-
ried', mutually independent. Otherwise there would
be a-stats-description which assigned truth to 'John
is a bachelor' and to 'John is married', and conse-
quently 'No bachelors are married' would turn out
synthetic rather than analytic under the prOposed
criterion.9

Note in this passage that Quins speaks of "version of

analyticity." He admits that the notion makes sense

in languages whose atomic statements are mutually in-

dependent. But is not the notion, then, given an ade-

quate explication, at least for those languages?

 

8Quins, 23. 91bid.

32

Quins also finds difficulties concerning defini-
tion (5). For where, he asks, are we to find these
definitions? Who defined them thus, and when?

Are we to appeal to the nearest dictionary,
and accept the lexicographsr's formulation as law?
Clearly this would be to put the cart before the
horse. The lexicographer is an empirical scientist,
whose business is the recording of antecedent facts;
and if he glosses 'bachelor' as 'unmarrisd man' it
is because of his belief that there is a relation
of synonymy between those forms, implicit in general
or preferred usage prior to his own work.lO

There is essentially the same difficulty with the notion
of explication; for explication rests upon prs-existing
synonmies. Both definition and explication, consequently,

rest upon synonymy rather than explaining it.

However, Quins curiously adds:

There does, however, remain still an extreme
sort of definition which does not hark back to prior
synonymies at all: namely, the explicitly conven-
tional introduction of novel notations for purposes
of sheer abbreviation. Here the definiendum becomes
synonymous with the definisns simply because it has
been created expressly for the purpose of being
synonymous with the definisns. Here we have a really
transparent case of synonymy created by definition;
would that all species of synonymy were as intelli-
gible.ll

I believe that Quins has here misconstrued his own prob-
lem. The problem is not how terms become synonymous
(though this is an interesting and important one).

That is, that the logician may make synonyms by pure

fiat (this is not always the case, for even the logician

 

lOIbid., 21.. llibid” 25-26.

33

has certain strictures) is not of concern here. Nor

is the problem to be associated with the issue of whether
the lexicographer makes expressions synonymous or whether
(he merely discovers and records synonyms. These alter-
natives are not exclusive, and probably both contain

an element of truth. However, these questions are really
beside the point. For the lexicographer's dictionary

is certainly a language and is composed of definitions
which consist of synonymous expressions in their definisns
and definiendum. We might find good reasons for not"
employing "his language"; still, in relation to it,

the synonymy of expressions is quite explicit. The prob-
lem in the context we are considering, generally speak-
ing, is: What is it that the logician means when he
states, for example, that the expression "p :Jq" is
synonymous (or equipollent) with the expression "hp v q";
or what is it that the lexicographer means when he states
that the expression "sib" is synonymous with the expres-

sion "blood relation"?

Definition (6), according to Quins, is faulty
in depending upon the speciously clear notion of "seman-
tical rule." Let us suppose, he says, that we have an
artificial language L0 whose semantical rules have the
form explicitly of a specification, by recursion or

otherwise, of all analytic statements in L0. The rule

3h

tells which statements of L0 are analytic and which are
not. But the rules contain the word "analytic" which
we do not understand. We might "view the so-called rule
as a conventional definition of a new simple symbol
'analytic-for-Lo', which might better be written unten-
dentiously 'K' 36 as not to seem to throw light on the
interesting word 'analytic'."12 Here we have explained
'analytic-for-Lo' and not ianalytic for' or 'analytic'.
It would be better according to Quins if we "just stop

tugging at our bootstraps altogether,n13

It has been suggested, in the above exposition
of Quine's Criticisms, that the terms "logical truth,"
"self-contradiction," "synonymy," "analytic," and "neces-
sity," are but one family of terms intimately associated.
Indeed, Quins says that if he is given an intelligible
description of any of the last four notions, he will
find the others intelligible. It would then seem desir-
able to show just how these notions are related. For
example, the statement:

(7) It is not the case that if a woman is a
spinster she is a spinster,

is the contradictory of the statement "If a woman is a
spinster, she is a spinster" which is a substitution

instance of the logical truth:

(8) If p, then p.

 

121bid., 33. 13Ibid., 36.

35

Therefore, statement (6) is said to be self-contradic-
tory. But the statement:

(9) It is not the case that if a woman is a
spinster, she is an unmarried woman,

which is the contradictory of the statement "If a woman
is a spinster, she is an unmarried woman" is not a sub-
stitution instance of the logical truth (8). However,
because "spinster" means "unmarried woman," we want

to say that statement (9) is also self-contradictory.
Clearly there are two distinct notions of self-contra-
diction hsrs. This is what Quins is suggesting when

he says that "the notion of sslf-contradictoriness,

j._1_1_ Egg 9313.3 M _s_e_n_s_g needed for . . . a definition
of analyticity, stands in exactly the same need of clari-
fication as does the notion of analyticity itself."14

He is not at this point, at least, troubled by the self-
contradictory character of statement (7), but with the
self-contradictory character of statement (9). This
leads to the importance of synonymy. For if we substi-
tute in statement (9) the word "spinster" for the syn-
onymous expression, "unmarried woman," we change state-

ment (9) into statement (7).

If we define "analytically true statement" as

the converse of "logically contradictory statement," the

 

14Ihid., 20. Italics added.

36

same remarks hold, mutatis mutandis, for "analytic."v
Thus "analytic" and "self-contradictory" (in the sense
of statement (9) above) can be defined as each other's
denial. And, an adequate explanation of synonymy will
afford an adequate explanation of either analyticity

or the broad sense of self-contradictoriness.

Quins notes that an adequate explanation of
analyticity or self-contradictoriness will also afford
an adequate explanation of synonymy. For, to say that
"All and only spinsters are unmarried woman" is analyt-
ic, is to say that "spinster" and "unmarried women" are

synonymous expressions.

Moreover, "necessary" and "impossible" are of-
ten identified with "analytic" and "self-contradictory."
Consequently, these notions can be explained in terms

of each other.

Therefore, with the exception of "logical state-

ment" (logically true or logically self-contradictory
statement), all of these notions are interdefinable.
It is not certain whether Quins has similar misgivings
about the notion of logical truth as he does about the
notion of analyticity. His attempts to reduce analytic
truth to logical truth through the notion of synonymy

would seem to suggest that he does not. However, when

37

. . . . . 1
he remarks tnat "g2 statement is immune to reVlSlon" 5

it would seem that he does. If he does have misgivings
about logical truth, it might have been better if he

had directed his arguments toward this notion rather
than analyticity. One consequence of abandoning the
analytic character of logical statements is that the
statement "a=a" is theoretically subject to the same
corrigibility as "Snow is white." A further consequence
of this is that no matter how clear one makes the no-
tions of "synonymy," "meaning" or "necessity," he will
not satisfy Quins's demands. For the notion of logical
truth, and thereby the notion of analyticity, remains
obscure. Consequently, Quine's argument would have

been just as effective had he directed it toward the
notion of logical truth and it would have caused less
concern over analyticity and synonymity. However, had
he done this, we might not have benefited from his in-
sights into the genuine difficulties these other notions

do present.

In this section, we have seen the central argu-
ments which Quins brings to bear against the dichotom-
ist's thesis. The list is surely not complete. Quins
has many other interesting remarks which are concerned

with the.notions of "meaning," "synonymy" and "necessity."

 

l51bid., 43.

38

And we have ignored Quine‘s conclusion that the aban-
donment of the distinction leads to "another shift to-
ward pragmatism"?6 Quine's brand of pragmatism is a
peculiar one and will demand attention in later sections.
' We have also seen how the family of terms, "synonymy,"
"analyticity," "self-contradiction," "logical truth"

and "necessity" are interrelated.

 

lélbid., 20.

IV AN EXAHNATION OF THE CRITICISMS

The reaction to the criticisms advanced by Quine
and other similar criticisms has been violent. The
philosophical world, it seems, has been divided into
two camps; one arguing that the distinction is tenable
and the other arguing that it is not. As has been said,
the problem does not, at first glance, seem to be one
of the "big problems." Consequently, we are somewhat
ill at ease as we sojourn in this atmosphere of violent
disagreement. However, it is only partially true that
this is not a "big problem." The question is larger
than whether some statements are trivially true while

others say something about empirical facts.

That this is true can be seen if we note some
of the pivotal terms which constantly recur in discus-
sions of this problem. A statement is often said to
be analytically 2522, when by substituting synonyms
for synonyms, it becomes a logical Ragga. But synonymy
does not help. For synonymy is customarily defined as

a notion

sameness (identity or similarity?) of meaning,

which is equally vague. Or again, a statement is said

to be analytically true if it is true by virtue of its

#0

meanings or by definition; or, again, a statement is
analytically true when its denial is self-contradictory.
But it is said that we cannot understand "analytic"

for these notions are Viciously circular and are explained
only in terms of one another. Thus, it is said that we
need a more satisfactory criterion_in order to make

these notions understandable. That is, without a more
satisfactory explanation, the distinction cannot be
justifiably maintained. Thus, we are to regard the dis-
tinction as only a pseudo-distinction rather than an
absolute and £23; dichotomy (dualism?). And, if the
distinction is abandoned, we are led in "another shift
toward pragmatism." It has long been customary to iden-
tify necessity or certainty (two different problems)

with analyticity. Thus, it is said that there are no
necessary truths nor can we have certain knowledge.
Consequently, according to the gradualist, the statement
"If it is a clear day, then it is a clear day" is on

the same plane as the statement "If it is a clear day,
then it will snow tomorrow." Neither statement is neces-

sary. Indeed, no statement is necessary.

One can see that these notions have long been
chief concerns of philosophy. Some, of course, are of
more recent interest than others, but this is beside

the point. What is important is that they haVe long

#1

been of concern in philosophical debate. Aristotle went
to great length defending the law of self-contradiction.
But in the end, he says, it cannot be defended but must
be accepted as a first principle. And certainly Plato
was concerned with a crude form of pragmatism when he
pictured Socrates debating with Protagoras or Gorgias.
We can see that the analytic-synthetic problem converges
at some point with almost every central discipline in
philosophy. When it is asked whether "meanings" are to
be countenanced, a question of ontology and thereby a
question of metaphysics is being asked. When the prob-
lem is concerned with "ideas" as meanings or "images"

or with certainty, it is in the heart of epistemology.
And, surely, the whole question is concerned with the
making of decisions--decisions which may be either right
or wrong. For example, we must decide what are to be
taken as the requirements which must be satisfied for a
given congerie of events if it is to be called a lan-
guage. Some decisions will be arbitrary; others will

not. Here we are involved in normative questions.

Thus, the problem we are concerned with is not
really a peripheral or merely a technical one. Nor
should one be surprised that there can be such perva-
sive disagreement over a question which might seem,

prima facie, simple and unimportant.

#2

Since this problem is to a certain extent in-

volved in traditional problems of philosophy, it is

only reasonable to expect that it should take on their
character. A definitive solution like those of other
phiIOSOphical problems must wait, so to speak, until

all the work is done. This, of course, seems impossible.
And, here lies the difficulty of the problem. As one
cannot do logic, epistemology, or metaphysics in isola-
tion, neither can this problem be discussed wholly in
isolation. This is perhaps the dilemma of philOSOphy.
But this is not to say that the problem cannot be fruit-
fully discussed. The only alternative seems to be Pyr-

rhonism.

Consequently, our concern now is: If the prob-
lem we are presently considering is intimately associ-
ated with other "insoluble" (I use this word with re-
serve) problems, how can it be fruitfully discussed?
The answer to this, I believe, is that the analytic-
synthetic distinction must be discussed in a more lim-
ited way than has been customary. When one speaks of
"analytic" with the view of making it apply universally
to all statements of all languages, the notion is bound
to be somewhat fogg . Fruitful discussion must wait
until the problem is made more specific. This is done

by relating it to specified languages, for example, as

43

Carnap has done with the notion of state-description.
The notion of state-description is quite senseless when
it is applied to, say, "plain English." Though the no-
tion cannot be applied with cogency to gll_languages,

it can be applied to specific languages. Similarly, I
believei that one must realize that any one definition
of "analytic" will not apply to all languages-~though
there might be some common core of meaning which is to
be found in all ("some" would be strong enough) specif-
ic notions of analyticity. This approach is akin to
what White has labeled "finitism." Instead of viewing
"analytic" as a term which is peculiarly sacrosanct and
standing outside all others, we must regard it as apply-
ing to specific languages with, perhaps, different mean-
ings in each of these languages. We may, as Quine sug-
gests, abandon the term "analytic" for one less tenden-
tious, for example, "true in all state-descriptions."
The apprOpriate term will, of course, vary from language
to language. This is the only way to fruitfully speak
about the notion (or rather the notions), and it is
senseless to attempt to find some meaning of "analytic"
which will apply to all languages. This is quite as
senseless as saying that "thing" means the Same thing

in a "phenomenalistic language" as it does in a "thing"
language." Though the words are the same, they do not

have the same meaning. The following section will be

hh

concerned with exhibiting the futility of attempting to

have "analytic" apply univocally to all languages.

With this by way of justification, we now turn
to an examination of recent criticisms. A few prelim-
inary remarks seem in order so as to prevent later con-
fusion. There are several senses in which all statements
can be said to be synthetic. But, as I shall attempt to
show, these are not concerned with the tenability of

maintaining the analytic-synthetic distinction.

It is a basic tenet of empiricism that all knowl-
edge is occasioned by experience. This would be one be-
lief which is embraced by all of the divergent schools
of empiricism. But, without experience, statements would
not be possible, for there would be nothing to express.
However, if from this we were to conclude that all state-
ments are synthetic, it would be to miss the point. For
this would be to confound origin with validation. The
question is not whether all knowledge is occasioned by
experience but whether, given experience, some statements
are self-justifying; that is, whether some statements,
though occasioned by experience, are true independently
of experience. This is similar to Kant's problem. But
it is different in the respect that he argued that truth
can be Obtained independently of experience and yet say
something about experience. Analytic statements, however,

purport to say nothing about experience.

45

Another, similar error would be to say that all
statements are synthetic because we can never be certain
that we perceive (hear, see, or feel) them veridically.
Every statement has as its vehicle of communication some
physical-~sensib1e--thing. Thus, when one reads a book,
he is perceiving physical configurations of ink. And
when one listens to a lecture, radio, television or what
have you, he is perceiving (hearing) certain sounds which
are transmitted in waves. Or when a blind person is
reading braille, he is perceiving certain configurations
of physical bumps. Thus, there is always the possibility
that what one perceives is not veridical. For example,
one might read the statement "Misery and suffering are
prerequisites of salvation" as "Misery and suffering
are perquisites of salvation." However, to offer evidence
such as this for the synthetic character of all statements
would be to misconstrue the problem. This, again, would
be to confuse origin and validation. But there is an
additional error in that this objection confounds the
psychological notion of certainty with the logical and
epistemological notion of analyticity. One can feel cer-
tainty with respect to any statement, even with respect
to self-contradictory ones, just as one may doubt the
truth of analytic statements. Thus it must be presup-
posed or established that there has been veridical per-

ception of a sentence-token before one can proceed to

46

argue that it is analytically (or synthetically) ground-
ed. Once this is seen, we see that the only sense in
which certainty can fruitfully be brought into the pres-
ent question is through identifying it with the logical
notion of necessity. And, then to say that analytic
statements can be doubted (i.e., that they are not cer-
tain) is self-contradictory. In view of this anomalous
consequence, it would not seem advisable to identify
certainty with necessity, a term which is confused enough
as it is. Or, if "certainty" is being used in a psy-
chological sense, the central question, as we have seen,

is being ignored.

These points are obvious enough. I believe
that they would be subscribed to by most empiricists.
However, the presuppositions necessary for making the
analytic-synthetic distinction are not limited to those
mentioned above. This point will be made clearer in

what follows.

It appears that Quine has fallen into a similar
error--though his reasons are more subtle-~when he says
that "no statement is immune from revision." His rea-
sons for making this remark are, in part, identical with
his reasons for rejecting the analytic-synthetic dichot-
omy. As further evidence, he writes, "revision even of

the law of excluded middle has been prOposed as a means

#7

of simplifying quantum mechanics."1 This is true. But
the decision appears to be one of convenience. And, it
is quite aside from the problem of the analytic. The

problem, in recent contexts has presupposed the law of
excluded middle. This presupposition seems entirely

warranted. Moreover, we have no reason to believe that
there would not be analytic statements if the law were

abandoned.

For these reasons, I believe, reference to "im-
munity from revision" has no important place in discuss-
ing the tenability of the analytic-synthetic dichotomy.
That no statement is immune from revision is not a par-
ticularly alarming truth. The reasons for this, to
repeat, are due to the possibility of non-veridical per-
ception, or the fact that the law of excluded middle
might possibly be abandoned. In the latter case, how-
ever, it is not individual statements which will be
abandoned, but whole bodies of statements. What should
still be maintained is that if the law of excluded mid-
dle were true, then such and such statements should
have been analytic. Consequently,I believe, it would

be better not to speak of "certain statements" or

 

1W. V. Quine, "Two Dogmas of Empiricism," From‘g
Lo ical Point 2f_View: Nine Lo ico-Philoso hical Essays
(Cambridge: Harvard University Press, 1953 , p. #3.

48

"statements immune from revision" in conjunction with

the present problem.

The arguments advanced by the gradualists are
not so easily put aside as those which we have just
been considering. Some of the gradualist's arguments
are, of course, more biting than others. But, they do
not, I believe, offer sufficient evidence for reject-
ing the analytic-synthetic dichotomy. There are, it
seems, several pyimgflfggig reasons why the distinction

should not be abandoned too hastily.

First, the distinction has enjoyed being a part
of a long phiIOSOphical tradition. When men of such
intellectual stature as Locke, Hume, Kant, Ayer and
Carnap defend a thesis, I believe, one should think
twice before abandoning it. It is true that some philos-
ophers have regarded it with importance while others
have attributed it with little significance. Most of
them, however, have recognized a difference. This is

the important point.

Second, there has been a wide concordance of
Opinion as to the respective extensions of the terms
"analytic" and "synthetic." There are, of course, dif-
ferences of Opinion as we have seen. For example, Kant

regards mathematical statements as synthetic whereas

1+9

Hume and Ayer do not. But if this is true, it is also
true that there is wide agreement. For example, it is
generally agreed that statements like "All bachelors are
males" represent an analytic truth; and it is agreed
that statements like "The Washington Monument is over
thirty feet tall" are synthetic. Thus, it seems that

we not only have to account for disagreement about the
extension of "analytic" but also for agreement. There
must be something which leads us so consistently towards
this agreement. If it is not that one kind of state-
ment says nothing about empirical facts while another
kind does, then what is it? If the distinction is

abandoned, this question must be answered.

Third, there is an intuitive feeling that there
are some statements which are trivially true and say
nothing about facts of experience. Even Locke and Mill,
who attached little significance to these notions, felt

a need to take cognizance of their existence.

Finally, the notion has proved to have a high
heuristic value. This can be seen in its use as con-

ceived by Hume, Ayer and a great many other empiricists.

The above remarks do not in any way offer a complete
justification of maintaining the distinction. That is,
the analytic-synthetic dichotomy might be completely

illusory, and this fact would not be incompatible with

50

the above facts. What these remarks do show, however,
is that the distinction should not be abandoned too

hastily.

One cannot fail to be impressed by the frequent
recurrence of the word "understand" in the writing of
Quine. He says that he does not understand the terms
"analytic," "self-contradiction," "necessity" and "syn-
onymy." These terms, he says, are interdependent and
each is in exactly the same need of clarification as
the others. Many who would argue in favor of the grad-
ualist doctrine hold, on the contrary, precisely the
opposite, that is, that they understand these notions
perfectly well and that there is no need of radical
clarification in order to make their position tenable.
M. G. White anticipates such a rejoinder and writes:

We begin by saying that we do not understand.

But our opponents may counter with Dr. Johnson that
they can give us arguments but not understanding.
And so it ought to be said that the objection is a
little less meek; the implication is that many who
thigk they understand really don't either.2
Volumes have been written on the human understanding.
And it would be unfortunate if one had to wait for a
definitive solution of these difficulties before he

could tackle the present problem. Unless the word

understand" is to be a mere subterfuge, it seems highly

 

2
M. G. White, "The Analytic and the Synthetic: An

Untenable Dualism," in Sidney Hook (ed.), John Dewe :
Philosther a: Science and Freedom (New York: fiiaI Press,
’ p. C

5l

desirable to see just what Quine and White do understand.
However, the two philosophers differ as to what sort of
explanation they would take as establishing satisfactory
understanding. Quine wants a definition of "analytic"
which would permit his "making sense of the idiom 'S' is
analytic for (language) 'L', with a variable '8' and
'L'."3 Or in terms of synonymy, presumably, we must be
able to make sense of the idiOm 'A' in 'L1' is synony-
mous with 'B' in 'LZ', with a variable 'A', 'B', 'L1'
and 'Lz'. This, of course, does not preclude 'L1' and
'L2' being the same language. Moreover, he prefers that
the definiens of the definition be stated in behavioris-

tic terms.

White's demands are more lenient. He will be
satisfied with a term which is extensionally equivalent
with "synonymous with." That is, he will be satisfied
with a criterion. The difficulty here, however, is that

he must be able to‘gnderstand the criterion better than

"synonymy."h

One is reminded in these discussions of the
long debates over the question of whether we are justi-
fied in believing that the future will resemble the

past. Very few of these discussions bothered to consider

 

3Quine, 33.
"White, Analytic and Synthetic, 320.

52

what sort of justification would be taken as adequate.
The next question, of course, is what sort of criteria
are we to take as justifying our criteria for believing
that the future will resemble the past. Clearly, this
could continue forever without being fruitful. But
there does seem some such legitimacy in asking the first
question, i.e., what would justify our believing that

the future will resemble the past.

Similarly, it is fruitful to ask what sort of
evidence would be adequate to support the contention
that the analytic-synthetic dichotomy is tenable. There
seems to be some evidence that Quine's demands are too
severe. Many would claim that we do not need such a
definition to justify maintaining the distinction. Many
might hold that "natural language" is not the sort of
thing which can be subjected to the rigor of absolute
criteria. They might hold that this is not necessary.
And, all that is necessary is that we have some sort of
idea of "synonymy" as it occurs in artificial language.
Indeed, there is some evidence that the entire contro-
versy is not concerned so much with the intelligibility
of the old definitions as with the implications which

are to be drawn from the distinction.

A clear formulation of a problem, is essential

to any adequate discussion of the problem. This seems

53

trivially true. Yet it is precisely this important
element which has most conspicuously been absent.from
discussion in the recent controversy. And, this, per-
haps, has precluded even a tentative solution. That
this is the case can be seen from a remark made by White:
After presenting [gradualist] views like these
I frequently find philos0phers agreeing with me.
Too often they are the very phiIOSOphers whose views
I had supposed I was criticizing. Too often, I.
find, the criticisms I have leveled are treated as
arguments £2; what I had supposed I was opposing.5
Clearer evidence could not be given that something is
amiss. It is indeed odd that trained logicians should
construe the same arguments to support incompatible
conclusions. This would not be a fruitful supposition.
But it is not so strange that one and the same argument
can be viewed as leading to different but compatible
conclusions. In the recent discussions, this latter,
I think, might very well be the case, and it exhibits
the necessity of a precise statement of the problem.
Not only would a precise statement tend to preclude the
possibility of similar arguments supporting supposedly
incompatible conclusions, but it would also make the
problem more manageable and amenable to solution. We
cannot say that the whole problem is merely verbal and

that there is really no controversy at all. This would

certainly be a paradoxical conclusion.

 

51bid., 328-329.

51+

A great part of the difficulty in reaching a
reasonable agreement can be attributed to the notion I
of language. "Language," in itself, is an elusive term.
We know pretty well what an artificial language is.
This is due to the very fact that it is artificial. We
do not, however, understand the nature of natural lan-
guage as well. Is natural language to be identified with
the actual physical phonemes, words or statements uttered?
Or must meaning of some sort or another be brought into I
the picture? Is there one natural language or are there
many? If there are many, must we also call each dialect
of each language a language? Must we speak of communica-
tion between animals as a language? Does each species
have a different language? If so, do they also have
different dialects? Must we go further and distinguish
private from public language? Embarrassing questions
like these can be created at will and considerably ex-
tended. It is quite evident that "language" is not as
intelligible as might at first be believed. And it is
equally evident that to make this concept understandable
in any rigorous sense, one must make certain arbitrary
decisions. These decisions, however, are not always
entirely arbitrary, for certain heuristic considerations
are bound to be of some significance. The probable rea-
son for the need to make these decisions is that there

is no single language which is referred to by "language."

55

It is not meant that the term "language" should be pre-
served only for rigorously defined concepts. For, I
believe, we must retain non-rigorous senses of "language."
There must be some systematic ambiguity. For example,

I should like to think of dogs as communicating. And,

I should like to think that there is a dog-language.

But, I hardly suspect that dogs make analytic utterances.
What is necessary is that when one asks for an explica-
tion of a semantic concept, he should specify what he
means by language, in this context, and give some indi-
cation of what language(s) he is concerned with. That
all of these difficulties have been resolved in some
sense has been an unwarranted presupposition in recent
discussions by both the gradualists and the dichotomists.
And this, it would appear, has made fruitful communica-
tion virtually impossible. This has important consea
quences. For it Shows that some decision must be reached
as to the range of applicability of the predicate "lan-

guage."

If we consider these difficulties in connection
with the demands made by Quine, the problem becomes one
of determining what sort of things fall within the vari-
able ‘L'.6 Even, if this problem can be given a satis-

factory solution, we have no right to assume that there

 

61t is interesting to speculate whether (or how)
Quine would quantify over L's.

56

will not be some language in which the concept of analyti-

 

city has no meaning. Nor can we assume that the concept
will apply in every language. That is to say, there very
well may be some "languages" in which there are no state-
ments which are analytic. There are also problems of
translatability here, for it is possible that there could
be a language in which the word "analytic" cannot be
translated, i.e., the notion may be ineffable in some
languages. This, however, is not to say that there might
not be a meta-language in which statements in the object
language could be properly identified as analytic (or
synthetic). When it is said that these matters are poss-
ible, it is meant that there is no logical contradiction
involved. It may very well be that such language never
occurs, but this is beside the point. This seems to be
the question: Is a notion intelligible only if the same

explication of it can be used for all "languages"?

Failure to take account of the great variety of
referents assignable to the term "language" is not the
only ground for the controversy under review. The con-
troversy seems also to have been assimilated to the
ancient nominalist-realist feud. Thus, Quine, White,
and Goodman are antecedently hostile to any proposed
solution which features hypostatized meanings and propo-
sitions. Ruling out "obscure intermediary entities,"

Quine takes the primary object of the theory of meaning

57

to be the synonymy of linguistic forms and the analyti-
city of statements. This attitude, as has been suggested,

characterizes the gradualist's position.

surely, though, the frequent use in the litera-
ture of the terms "meaning" and "proposition" suggest
that they have certain heuristic value, even when employ-
ed in an "hypostatizing" way. And, how would one justify
a sweeping injunction against hypostatization? Hypothet-
ical entities do respectable work in other disciplines
and it is entirely conceivable that they might be of use
in the theory of language. Nor does it seem pure obscur-
antism to declare that these entities exist "after a
fashion" or "in a certain sense." There are difficulties
with these phrases, but, then so are there difficulties
with the word "exist." And, it is not really consistent
with finitistic principles (White's, for example) to in-
sist upon a univocal interpretation of "exist."7 Nor
does it seem consistent with pragxatism to ignore the
heuristic aspects of a platonistic use of "meaning" and

"prOposition."

For by hypothesizing "meanings," one gains a
certain degree of explanatory ability just as one gains

a degree of explanatory ability by hypothesizing atoms.

 

7n. G. White, "A Finitistic Approach to PhilOSOph-
icgl Theses," The Philosophical Review, LX (1951), pp. 299-
31 .

58

latter hypothesis admittedly has proved more fruitful.
But this is perhaps because the former has not been taken
more seriously. Both hypotheses seem fully consonant
with pragmatism. I do not know whether meanings or atoms
exist, however, I do know that the latter enables scien-
tists to make certain predictions (and explanations),
and, I can see no reason why the former should not prove
equally fruitful. Quine writes that a conspicuous ques-
tion for semantics is the nature of its objects. Carry-
ing this query further, I believe that it may be asked,
in the same sense, what is the nature of atoms. The
truth of the matter is that one knows atoms and meanings
only in an indirect way. To be sure, such hypotheses
have many difficulties. It would be useless to deny
this. But no one could reasonably reject explanations

of this general kind as intrinsically wrong-headed.

And it does seem that the investigation of the possibili-
ties in the use of meanings and propositions as hypothet-
ical entities has been prematurely abandoned by many

thinkers.

Quine desires to have "significance" (i.e., "hav-
ing meaning"), "synonymy" and "analyticity" explained in
terms of behavior. Yet, he also says that "we are not
concerned with synonymy in the sense of complete identity

in psychological associations or poetical quality; indeed,

59

no two e: {pressions are synonymous in such a sense. We
are concerned only with what may be called cognitive
synonymy."8 He does admit that he does not know just
what cognitive synonymy is. Yet it does seem to be a
little gratuitous to assume that such notions (i.e.,
"analyticity," etc.) can be explained entirely in behav-
ioral terms, that is, to assume that such notions can

be defined without the assistance of some concepts which
cannot be defined in behavioral terms. That this is
presumptuous can be seen by noting the difficulty, and
most likely the impossibility, of maintaining a distinc-
tion between cognitive and non-cognitive synonymy on
purely behavioral grounds. The argument against behav-
iorism sometimes takes the form of Stating that intro-
spection often yields evidence or reveals data inaccess-
ible to an outside observer.9 This argument, I believe,

is entirely tenable.

However, even if this evidence is too subjective,
Quine's behaviorism founders on other rocks. It is non-
finitistic. I do not mean to suggest that Quine subscribes

to White's finitism. However, it is against such views

 

8Quine, 28.

9For example, see: Bertrand Russell, Human Knowl-
edge: Its Scoge and Limits, (New York: Simon and Schuster,
19h

). pp. h5- -h7, 50- 51.

60

as behaviorism that finitism is most biting. And here

White's argument applies most cogently.

Quine, as we have seen, makes a distinction be-
tween what he calls "theory of reference" and "theory
of meaning." "Meaning," he says, "is not to be identi-
fied with naming." To illustrate his point he considers
as examples '9' and 'the number of the planets'. Now,
both of these terms denote the same abstract entity.
But, they cannot both be said to have the same meaning;
for observation of the planets was needed and not mere

reflection on meanings, to determine this sameness.

Now, I do not wish to deny that there is a dis-
tinction here; nor do I want to suggest that it is not
necessary that we make the distinction. But, I should
say that Quine's choice of words is misleading. What
he calls meaning is no doubt not the same as naming.

But, both of these are modes or kinds of what we would
normally call meaning. This suggests that there is more
than one mode or kind of synonymy. So long as "synonymy"
is defined as sameness of meaning, we will have as many
kinds of synonymy as we have meaning. For example, we
might say that "Morning Star" is extensionally synonymous
with "Evening Star"; or we might say that "bachelor" is

intensionally (and extensionally) synonymous with

61

"unmarried male."lo To be sure this will not always

be in accordance with our practice in the use of natural
language, and we will come up with some weird specimens
of synonymy. Distinguishing between modes of synonymy
certainly does not lessen the burden of giving Quins a

general (if not universal) criterion for determining

synonymy.

Similarly, it is difficult to understand how

Quine and others can distinguish between the theory of
reference and the theory of meaning on purely behavioral
grounds. They can indicate that meaning is not the same
sort of thing as naming. But can a behavioristic crite-
rion be given which will distinguish the two? For ex-
ample, can one (attending to overt behavior only) sharply
distinguish the occasions when someone is using, in the

"naming way," expressions like "nine" or "the number of

 

10I do not say that these are the only kinds of
synonymy. For example, one might speak of two words
(expressions or sentences) as being synonymous: (a) if
they evoke the same (or similar to a high degree) behav-
ioral responses; (b) if they are verified in the same
(or similar)ways; (c) if they have the same deductive
power in conjunction with certain other sentences. The
qualifications noted in (a) and (b) above are necessary.
For unless we allow some leniency along these lines, we
are in the ludicrous position of not only failing to un-
derstand what it means to say "A is synonymous with B"
but also failing to understand what it means to say "A
has the same meaning as A." This applies especially to
those who with Quine would demand that the general crite-
rion be behavioral.

The kinds of synonymy suggested above, of course,
do not purport to be exhaustive.

62

the planets" from those other occasions when he is
attending to "meaning" proper? Perhaps this difficulty
vanishes when spoken and written utterances are lumped
under overt behavior, but any pragmatic basis for dis-
tinguishing behavioristic procedures vanishes at the
same time. Considerations like these may account for
Quine's dissatisfaction with the notions of analytic

and synthetic and the related one of synonymy.

With these points in mind, we may turn to a con-
sideration of one attempt to offer a satisfactory solu-
tion to the problems of analyticity and synonymity.
Carnap has recently prOposed a procedure for empirically
testing hypotheses concerning the intension ("meaning"
in Quine's sense) of a term. This procedure, he claims,
enables us to determine empirically "the general con-
cept of the intension of any predicate in any language
for any person at any time," and that these intensions
have a clear, empirically testable sense. Once he has
outlined this procedure, he is able to define our trouble-

some terms.

"Two expressions are synonymous in the language L
for §.at time 2 if they have the same intension in.L

for X at t.

63

"A sentence is analytic in.§ for §.at t.if its
intension (or range of truth-conditions) in.L for §_at

. ll
‘3 comprehends all p0531ble cases."

The procedure and definitions are explicitly
said to comprise a refutation of Quine's position.
The following is a brief resume of the salient features
of Carnap's outlined procedure: It is agreed that the
determination and analysis of extension for a natural
language is methodologically a sound scientific proce-
dure. The problem is: If a linguist can determine the
extension of a given predicate, how can he go beyond
this and also determine its intension? For example,
suppose that two linguists have reached complete agree-
ment on the extension of a given predicate (say "Pferd")
in a given language (say German) for a given person
(say Karl), and the one writes in his lexicon:
(l) Pferd, horse,
and the other writes:
(2) Pferd, horse or unicorn.
Since (1) and (2) have identical extensions, what empir-
ical data can we consider to determine its intension?
None of Karl's reactions when he is confronted with a

Pferd will assist the linguists in their endeavor. The

 

llRudolph Carnap, "Meaning and Synon y in Natural

Languages," PhilOSOphical Studies, VII (1955 , p. #5.

64

answer is that they must takejn to account not only ac-
tual cases but also possible cases. This is done either
by drawing a picture of a unicorn or describing one

(in German) and then asking Karl if he is willing to
apply the word "Pferd" to a thing of this kind. Using

a similar procedure we can determine the intension of

any term.

This outline, of course, suffers from over-
simplification and a great deal of expatiation would
be necessary to make it a completely accurate statement
of Carnap's prOposal. But it is sufficient and accurate
enough to show that Quine's requirements for an adequate

criterion of analyticity are not satisfied.

To begin with, there is some difficulty concern-
ing what sort of things the variable L is to refer to.
This is true of both Quine's formulation of the problem
and Carnap's prOposed definition. Thus, there is some
possibility that Quine and Carnap do not mean the same
thing by "language." This is suggested by Carnap defin-
ing analyticity not merely for a given language, but for
a given language £9;- 3 m pierson _a_t_ 3 22119.11 .t_i_r_n_e;.

On the whole this procedure seems preferable. For it
allows us to accountifor the changing of meaning and

the possibility of a statement being analytic for one

person but synthetic for another. Quine's formulation

65

of the problem, which does not take note of variable
"persons" or variable "times," however, apparently
makes the implicit assumption that words do not change
in meaning; that they mean the same thing for every-
body; and that every statement which is analytic for
one person is analytic for everyone else. This seems

to me to be obviously false.

However, if these difficulties are put aside
(illegitimately, of course), it can readily be seen
that Carnap's proposals do not satisfy the requirements
stipulated by Quins. The most obvious indication that
this is true is Quine's general hostility towards the
notions of "intension" and "prOposition." I can see
no reason why he should not find Carnap's use of these
terms equally objectionable, for, the ontological ques-
tion still remains unanswered. But, a more important
objection for our present concern is that the above
definitions do not give "analyticity" and "synonymy"
the universality which is demanded. Obviously, there
are some languages, namely, purely extensional languages,
which do not make use of intensional concepts. In these
languages Carnap's definition finds no application.
Thus, even though Carnap has defined these concepts for
SOme languages (the adequacy of his definition is not
of immediate concern), he has not defined them for all

languages.

66

So far it has been argued that the demand to
make the analytic-synthetic distinction apply to all
languages is too stringent. Our main presumption has
been that the wide variety of things which might be
taken as possible referents of the term "language" make
these demands vague, unreasonable, and in fact, imposs-
ible. It is suggested,.however, that "analytic" can
fruitfully be applied to languages of a restricted
range. The difficulty Quins has with this view, as we
have seen, is that he does not understand "rule for L."
Though "rule for L" does impose problems (those pointed
to by Quins), I do not believe his criticism can be
taken as sufficient reason to reject "analytic" as it
applies to some languages. This can be seen from the
very fact that if the evidence is taken as sufficient,
we are not only rejecting "analytic" but alsolgl; arti-
ficial languages. For, certainly, artificial languages
are governed solely by their rules. Heuristic consider-
ations, alone, make this alternative undesirable, or

more correctly, absurd.

In the remaining section it will be argued that,
though "analytic" is not a perfectly precise notion as
it is applied to natural languages, it is not entirely
senseless to speak of analytic statements in natural
languages. By natural languages we intend languages

like English, French, etc.

V ANALYTICITY AND NATURAL LANGUAGE

The arguments advanced by Quins and other grad-
ualist authors have consisted largely of destructive
criticism. They have been directed at showing the un-
tenability of maintaining a sharp distinction between
analytic and synthetic statements. If these arguments

re taken to be successful, abandonment of the distinc-
tion leaves a vacuum which has previously been filled
with an explanatory tool. It was observed in earlier
sections that the notion of analyticity had been gener-
ally taken to throw light on such terms as "necessary,"
"contradiction," and "logical truth." If the analytic-
synthetic distinction is abandoned, these notions also
must be abandoned, or at least, given radically new
meaning. For the purpose of the present section, atten-
tion will be directed toward a single lacuna which,

if the distinction is abandoned, calls for especial

consideration.

As we have seen, statements which occur in a
natural language like:

(1) A rose is a rose,

69

the same time, attending to it's positive side. That

is, it might be argued that the two are identical.
This, however, does not follow. For all things which
are not dichotomous certainly do not, necessarily, dif-
fer in degree. Think of two different specimens from
one lot of telephone poles; we would not speak here of
two kinds of telephone poles, but neither would we speak
of them as representing different degrees of telephone-
polensss. It is true that in some cases differences,
which are not differences of kind, are differences of
degree. This is particularly noticeable in cases where
measurement is applicable, for example, temperature.
One hundred degrees Fahrenheit is certainly different
from thirty degrees Fahrenheit. But we do not say that
this represents a difference of kind. We say that it
represents a difference of degree, namely, warmer than.
We might notice, however, that this interpretation is
not necessary; and that to say there is not a differ-
ence of kind is not to say that there is a difference
of degree, for, clearly, there may be no difference at

all (other than numerical).

Yet, the gradualists do not deny that there is
a difference between statements like, for example, (1)
and (2). It does not, then, seem unreasonable to re-

quest an explanation or an account of this difference.

70

And it is precisely such an explanation (or justifica-
tion) which is conspicuously lacking. However, to this
end there seem to be several possibilities which should

be examined.

For example, it might be said that statement
(1) is more analytic than statement (2). But this is
of very little help. For it is exactly the notion of
analyticity which has so severely been called into ques-
tion. We cannot understand what it means to say that
one statement is more necessary than another, or that
one expression is more synonymous with a second expres-
sion than with a third expression, unless we understand
what it means to say that a statement is analytic, or
that a statement is necessary, or that one expression
is synonymous with another. Nor is the difficulty here
to be associated with the word "understanding"; for
were we seeking a definition or criterion of "more ana-
lytic," we should be confronted with the same situation.
There are, however, certain conditions which limit the

truth of this statement.

Admittedly, if we take "more analytic" as the
fundamental concept in the field (in the way in which
some ethical speculation takes "better than" instead
of "good" as its basic notion) the above objections

would lose force. And something very like this is

71

involved in proposals that we substitute "nondescrip-
tivéffor the more familiar "descriptive" accounts of

analyticity.

Thus, White suggests that perhaps our criterion
of self-contradiction should not be conceived in its
usual "question begging manner." He remarks that all:

. . . it self—contradiction has to do is to pro-
ducs a csr ain feeling of horror or queerness on
the part of psOpls who use the language. They be-
have as if they had seen someone eat peas with a
knife. Such an approach is very plausible and I
would be satisfied with an account of the kind of
horror or queer feelings which peOpls are supposed
to have in the presence of the denials of analytic
statements.
And horror is pretty obviously a matter of degree. That
Quins is sometimes thinking along similar lines can be
seen in his talk about centrality 13,3 conceptual scheme
. . . . 2 . . . .
and 1mmun1ty from reV131on. The p0551bil1t1es of these
suggestions are developed more fully in Bernard Psach's
"A Nondsscriptivs Theory of Analytic."3 Indeed, he
seems to be the first to make the distinction between

descriptive and nondsscriptivs theories of analyticity

 

in. G. White, "The Analytic and Synthetic: An
Untenable Dualism," in Sidney Hook (ed.), John Dewe :
Philosopher of Science and Freedom (New Yor : BiaI Press,

E50): pp. BEE-3750

2w. V. Quins, "Two Dogmas of Empiricism," From g

Lo ical Point g£_Visw: Nine Logico-Philosophical Essa 5

(Cambridge: Harvard UniversityPress, 1953), pp. 52-46.
3Bernard Peach, "A Nondsscriptive Theory of Ana-

lytic," The Philosophical Review, LXI (1952), pp. 349-367.

 

 

72

explicit. The following is a loose synopsis of what
Peach takes to be the character of the distinction be-

twsen these two kinds of theories.

In general, he writes, a descriptive theory of
the analytic is one that claims to describe the nature
of analyticity, or to reveal the essence of analyticity.
It puts primary emphasis on classification, assuming
that analyticity is understood, when on the basis of
knowing that a statement has a nature N, it is known
to belong to a category 5. Descriptive theories empha-
size inspection and understanding of a statement. Non-
descriptive theories,,howsver, emphasize actions or atti-
tudes that are correlated with, but go beyond "mere"
inspection and understanding. They do not emphasize
classification. A nondescriptive theory will emphasize
non-logical correlations between statements and the
attitudes or actions of people, who are concerned with
them within some context, rather than syntactical char-
acteristics and logical relations between statements.

It is based on the notion of willingness to act or main-

tain an attitude in accordance with a conviction.

I find it difficult to take such a proposal
seriously. Some of the reasons for this are listed by
White. Who, he asks, is supposed to feel the horror

in the presence of the denial of an analytic statement?

73

And, further we must distinguish between firmly believed
synthetic statements and analytic statements. For ex-
ample, there are persons who are not one bit horror-
stricken by the denial of "It is either raining outdoors
or not," whereas they could not possibly bear the thought
that there might not be tables and chairs. They can
readily think up a third alternative for the first state-
ment, for example, that it is snowing. On the other
hand, to say that there are not chairs is clearly ab-
surd, they hold, for what do we sit in every day unless
they be chairs? To say that they do not understand what
is meant by "not raining" in this instance is clearly

to beg the question. It is not a very good explication
of analyticity which makes the notion deviate so far
from its common understanding. The flaws in this no-
tion are obvious and they can readily be multiplied.
Perhaps the most important is a pragmatic one. What

good is it? What purpose does it serve? Why not say
"more firmly believed" rather than "more analytic"?
Clearly, analyticity loses philOSOphical significance

when it becomes nondescriptive.

There remains one further proposal which is to
be examined. This is presented in Nelson Goodman's

"On Likeness of Meaning."A There are two aspects of

 

hNelson Goodman, "On Likeness of Meaning" in:
Leonard Linsky (ed.), Semantics and the PhilOSOphy 2f

7h

his paper; one is negative and the other is positive.
The negative side is characterized by arguments against
the use of such notions as Essence, Platonic Ideas, im-
ages, concepts or possibilities in explaining the notion
of synonymity. The positive side is found in his sugges-
tion that "we might try the very different and radical
theory that two predicates have the same meaning if and
only if they apply to exactly the same things--or in
other words, have the same extension."5 Here we shall
be concerned with the positive side of Goodman's thesis.
Its essence is suggested in the following quotation:
. . . although two words have the same extension,
certain predicates composed by making identical
additions to these two words may have different
extensions. It is then perhaps the case that for
every two words that differ in meaning either their
extensions or the extensions of some corresponding
compounds of them are different. If so, difference
of meaning among extensionally identical predicates
can be explained as digference in extension of cer-
tain other predicates.

For example, the predicates "centaur" and "uni-
corn" have the same extension, namely the null exten-
sion. This Goodman calls its primary extension. But
it is not the case that, for example, "unicorn-picture"

(i.e., a picture of a unicorn) has the same extension

as "centaur-picture." These are what Goodman calls

 

Language (Urbana: The University of Illinois Press,
2 ’ p. 67-74.

5Ibid., 69. 61bid., 71.

75

secondary extensions (of "unicorn" and "centaur"). It

is important to note that Goodman does not intend to

limit secondary extensions to pictures:
. . . for the secondary extensions . . . consist
also of the extensions of "Q-diagram," "Q-symbol,"
and any number of other such compound terms. In-
deed actual word-inscriptions are as genuine physi-
cal objects as anything else; and so if there is
such an actual physical inscription that is a P-
description and is not a Q-description, or vice
versa, then "P" and "Q" differ in their secondary
extensions and thus in meaning.

Unfortunately, a consequence which follows from
this procedure is that no two words have the same mean-
ing, for there is always a phrase like "P-description
which is not a Q—description" to which applies "P-descrip-
tion" but not "Q-description." Further, it has been
shown by Richard Rudner in "A Note On Likeness of Mean-

8 .
ing" that this program does not allow for any two tokens
of the same type to have the same meaning. For instance,
consider statement (I) in conjunction with the predicate
"a rose description which occurs in the fifth place in
(1)." This predicate is applicable to the fifth word
in (l) but not to the second. Consequently, the second

and fifth words in (l) are not synonymous, nor is the

statement analytic. I believe that this line of thought

 

7Ibid., 72.

 

Richard Rudner, "A Note on Likeness of Meaning,"
Anal sis, X (1950), pp. 115-118.

76

can be carried even further and that it can be shown
that a given token is not even synonymous with itself.
Consider, for example, the predicate "a rose descrip-
tion which occurs in the fifth place in (l) at time to."
Now, this predicate applies to the fifth word in (l)

at time tO but not at t1. Therefore we can assume, if
Goodman's analysis is accepted, that not only do two
word-types have the same meaning, but also that two
tokens of the sggg type do not have the same meaning,
and further that the same word-token does not have the
same meaning at different times. (Perhaps it would be
desirable to so define token that we cannot legitimate-
ly speak of the life-span of a given token. In other
words, "token" could be defined in such a way as to

m ke the first instance of the word "rose" in (l) a dif-
ferent word-token now than it was when it was first

written.)

That these facts might compose a reductio gg_
absurdum of Goodman's thesis need not at the moment de-
lay us, for we are here concerned in seeing whether a
case can be made out for synonymy which can be inter-

preted as a matter of degree.

This, is, indeed, the conclusion which Goodman
draws. ". . . we should do better never to say that two

predicates have the same meaning but rather that they

77

have a greater or lesser, or one or another kind, of
likeness of meaning."9 And, also the most we can say
about statements is that they are more or less nearly
analytic. And, similarly for "necessary" as it applies
to statements. It is not facetious to ask how are we
to tell when an expression §_is more synonymous with

an expression'g than an expression‘g. That is, we need
some sort of workable criterion. What we are in need
of here is not a definition of "synonymous with"; Good-
man has provided us with this (two expressions are syn-
onymous when they have the same primary and secondary
extensions). What we do need is a definition of the
expression "is more (or less) synonymous with." Good-
man's solution of this problem is that it would be best
to construe degree of synonymy as degree of interre-
placeability (in non-intensional contexts). I am not
too sure of what is meant by degree of interreplace-
ability. I believe it must be something like this:

An expression‘g has a higher degree of interreplace-
ability than an expression §_for an occurence of an ex-
pression 9, if and only if, when gland g are substituted
forig, those contexts in which‘gpyields the same truth-
value as Q_are more numerous than those contexts in which
§,yields the same truth-value as Q. How can we speak

of interreplaceability at all? We cannot literally pick

 

9Goodman, 73.

78

up the fifth word-token in statement (3) and replace it
with the second word-token, for in this very process
the token is changing its meaning (i.e., its location
is different and the process presumably takes time).
Should token be redefined as was suggested above, we
could not speak of interreplaceability at all. I find
the notion of interreplaceability or substitution a
mysterious enough process without the additional prob-
lems which it accumulates from Goodman's analysis. Simi-
larly, we find troubles with the distinction between
sentence-token and'sentence-type. I believe we must
admit that this analysis allows of no substitution, for
there is really nothing to substitute, nor, if there

were, any place to substitute it.

It might plausibly be said that this argument
is merely eristic. I cannot decide myself whether it
is or not. Yet, if it is, I can find no line where con-
structive criticism ends and "logic chOpping" begins.
It is conceivable that certain rules might be advanced
with regard to identity of sign-tokens or the legitimacy
of considering time factors. But, I believe such pro-
cedures would be wholly arbitrary. Furthermore, this
procedure begs the question. For we must choose those
rules which will determine what we had presystematically

believed about synonymy.

79

But, even if all of these problems are disposed
of, another fact demands consideration. Quine and White
have denied that they (or anyone else) really understand
analyticity. But here we have a perfectly understand-
able notion as it is drawn out by Goodman. As was in-
dicated above, it is not without serious problems, but
nevertheless "we get distinctions that are as fine as
anyone could ask."10 Now, the important point to be
noted here is that in order for Goodman to define "more
synonymous with," he had to define "synonymous with."
Synonymity is consequently a well "understood" notion.
But the fact is that there are no synonymous expres-
sions. (And, as I have attempted to argue, there can
be no expression more synonymous with another than that
other is with a third.) And I believe that we can
assume that White cannot legitimately presume "that a
suitable criterion is likely to make the distinction
between analytic and synthetic a matter of degree."11

For he cannot presume that a suitable criterion which

 

10Ibid., 73. The distinctions are fine. How-
ever, we might ask whether a standard workable crite-
rion for determining secondary extensions could be giv-
en. How, for example, do we know whether a given den
scription is a description of a unicorn? The extension
of a term is often taken to be determined by its inten-
sion. In Goodman's analysis, though, there are no in-
tensions. Thus, does the primary extension of a term
determine its secondary extension; or is it the other
way'aroundZ

llWhite, Analytic and Synthetic, p. 330.

80

will make the distinction a matter of degree, will not

also (and more appropriately) make it a dichotomy.

There is an important point here which should
be noted. One can always ask, concerning a definition
or criterion, to what extent it applies to the actual
universe. For example, Goodman's criterion makes the
distinction a sharp one. But as it turns out, his
criterion finds no application. Or, better it finds
application but makes all statements synthetic and
merely contingent. This statement is obvious, but it
is also often overlooked. The geometrician deals with
the admittedly idealized notions of "point" and "straight
line." Though points and straight lines have never
been found to be manifest in the empirical world, these
notions have proved fruitful and even necessary not
only in carpentry and other banausic arts, but, also
in surveying, astronomy, and a myriad of other empiri-
cal sciences. In the light of these facts, I do not
believe that the logician, semanticist, or linguist
should be overly chagrined if he fails to find that
the semantical notions of "analyticity" and "synonym-
ity" do not apply with absolute precision to natural
languages. The justification of applying notions in
natural language is largely a pragmatic one. And,
even if there were no synonymous terms or analytic

statements (which I doubt), the necessary use of these

81

semantical notions is determined largely by pragmatic
considerations. The ubiquitous employment of diction-

aries is sufficient evidence for this statement.

However, should one still insist that the
analytic-synthetic distinction be abandoned, it should
perhaps be best if it were abandoned wholeheartedly.
And it should be better to quit talk of degrees of
(more or less) analyticity, etc. This kind of talk
hurts the ear and seems to vitiate against the mean-

ing of the terms in question.

VI SOME CONCLUDING REMARKS

No single conclusion is to be drawn from the
discussion of the foregoing sections. Although the
general tone has been to suggest that the distinction
drawn between analytic and synthetic is a sound method-
ological procedure, it cannot be said that this belief
has been adequately defended or proved to be "right."
The reasons for this should now be evident. The fore-
most of these reasons is the extreme complexity of the
problem. We have attempted to exhibit this by cursori-
ly discussing some of the attitudes of different men
in the history of philosophy toward this and related
concepts. It has been shown that the distinction has,
engendered disagreement in the past. The disagreement
has taken essentially two forms. First, there has been
a division of opinion as to the importance of the dis-
tinction and, second, there has been general disagree-
ment as to the extensions of the two terms, "analytic"
and "synthetic." These facts have been taken as evidence
to support the contention that the distinction is by no
means self-evident or as obvious as might sometimes be

believed. We have attempted to show that the distinction

83

is important due to its frequent recurrence in the writ-
ings of modern philosophers. This suggests that the

problem is not one so simple or isolated that it can

be posed as "Is it true that every statement is analytic
or synthetic but not both?" The very complexity of the
problem seems to demand that it be given a clear formu-
lation. This, however, is conspicuously lacking. As a
result one cannot make up his mind as to who is defend-

ing what, and just what is the disagreement.

That the problem is not an isolated one suggests
that an adequate defense can be given only when it is
incorporated into a full-grown theory of metaphysics and
epistemology. If this is true, then it is likely that
different solutions are determined to some extent by
temperament and even the philosopherfs social environ-
ment. How much influence this has cannot be said. But
it would be futile were anyone to deny it. For one can-
not deny that these factors play an important part in
the formation of one's metaphysic. It is true that the
phiIOSOpher attempts to diminish the influence of these
factors. But it is equally true that he never wholly
succeeds. This can be seen to be true merely by gazing
our eyes upon the divergent schools of contemporary

philosophy 0

8h

More specifically it has been suggested that
the trouble is not so much with the terms "analytic"
and "synthetic" as it is with the term "language." This
view is not particularly original. It has been argued
by Benson Matesl and Richard Martin.2 The important
point to realize, in this respect, is that there is no mus
right analysis of language. The relationship between
one analysis and another is not that the one might be
right while the other is wrong, but that one is better
than the other. I do not know what criterion we should
use to determine which of any two proposed analyses of
language is the better. But certainly there will be

some pragmatic factors involved.

0
The general sentiments that there is no single
right analysis of language are expressed by both Carnap
and Lewis. Lewis writes in "The Modes of Meaning":

I should like to express my conviction that if
there be any one analysis of meaning in general which
is correct, then any number of other analyses will
be possible which are equally correct: for much the
same reasons that if any set of primitive ideas and
primitive prOpositions are sufficient for a mathe-
matical system, then there will be any number of
alternative sets of primitive ideas and propositions
which likewise are sufficient.3

 

1
Benson Mates, "Analytic Sentences," Philosophi-
cal Review, LX (1951), pp. 525-34.

2Richard Martin, "On 'analytic'," Philosophical

 

3c. I. Lewis, "The Modes of Meaning," PhilOSOphy
and Phenomenological Research, IV (l94h), p. 236.

85

And Carnap writes in "Empiricism, Semantics, and Ontology":

For those who want to develop or use semanti-

cal methods, the decisive question is not the alleged
ontological question of the existence of abstract
entities but rather the question whether the use of
abstract linguistic forms or, in technical terms,

the use of variables beyond those for things (or
phenomenal data), is expedient and fruitful for the
purposes for which semantical analyses are made, viz.
the analysis, interpretation, clarification, or con-
struction of languages of communication, especially
languages of science... . . [The question is not

a question simply of yes or no, but a matter of
degree.

We have also argued that difficulties of aban-
doning the distinction are as great as those of main-
taining it. It is usually agreed that there is a dif-
ference between statements like "Fido is Fido" and "Light
travels at a speed of approximately 186,000 miles per
second." If it is maintained that the difference is
not one of kind, the difference still remains to be ex-

plained. To say that the difference is one of degree
does not help the issue. For we must ask the question,
degree of what? And this problem seems to be as perplex-
ing as the other. Such a proposition must be defended.
Nor does it seem adequate to describe the difference
merely as different feelings people have about state-

ments.

 

hRudolph Carnap, "Empiricism, Semantics, and On-
tology," Revue Internationale gg_Philosophie, IV (1950),
p. 39. ‘

86

We must recognize, too, that abandoning the
distinction has consequences in fields outside linguis-
tic and logical studies. Suppose we agree that it is
an "ill-founded dogma" and discard it. Will it still
be possible to give the phrase "empirical philosophy"
any halfway determinate meaning? Quine has given us
a well-known metaphor describing human knowledge. Knowl-
edge, he writes, is one gigantic conceptual scheme,

". . . a field of force whose boundary conditions are
experience. A conflict with experience at the periphery
occasions readjustments in the interior of the field.
_Truth values have to be redistributed over some of our
statements. Reevaluation of some statements entails
reevaluation of others, because of their logical inter-
connections--the logical laws being in turn simply cer-
tain further statements of the system, certain further
elements of the field."5 This sort of holism is some-
times convincing. And the metaphor may be apt. But

the fact remains that it is a metaphor and needs detailed
defense and explanation (as does the term "analytic").
_Nothing in this passage suggests that its elaboration
would result in a theory of knowledge which could be

called "empirical" rather than, say, "idealistic."

 

5W. V. Quine, "Two Dogmas of Empiricism," From

Lo ical Point 9£_View: Nine Logico-Philosophical Essay
(Cambridge: Harvard University Press,‘I953), p. h .

mlm

 

 

87

Rather the contrary, what Quine says would work quite
smoothly into the context of a Hegelian or neo-Hegelian
epistemology, with a coherence-theory for estimating
the amount of truth achieved at a certain moment in a

dialectical process.

Abandonment of the distinction is said to lead
to a more thoroughgoing pragmatism. In the words of
Peirce we may ask, what are its practical consequences?
Is not Quine, for example, still teaching logic, as
is, for example, Carnap? And would not Carnap's students
do as well on Quine's examinations as Quine's Students
would do on Carnap's examinations? At a lower level
still, Quine's recommendations would leave the evidence-
gathering situation unaffected. In the teeth of the
difficulties he adduces, we would validate "A rose is
a rose" by analytic methods involving reference to syn-

onymity.

In the light of these facts it is hard to take
seriously the prOposal that the distinction should be
abandoned. Yet this is not to say that the gradualist
critique has been both senseless and futile. Nothing
could be further from the truth. For certainly the
gradualist arguments have brought into view genuine
difficulties. Perhaps the most important consequence

of their criticism is its exhibition that one cannot,

88

in clear conscience, speak indiscriminately of th§_
analytic-synthetic dichotomy. The gradualist con-
tribution is one of bringing a central tenet of em-
piricism (a tenet which had, in fact, become dogmatic)
into the Open, not for rejection, but rather for re-
examination, reevaluation and polishing. And thus, due
to the wide variety of possible referents of the term
"language," it can be seen that one ought to be more
specific when he speaks of "analytic" (or "synthetic").
The distinction, appropriately, applies with greater
facility to artificial languages than to natural lan-
guages. This was to be expected. But, as we have seen,
the distinction is not purely senseless even when applied
to the flexible and somewhat erratic field of natural

language.

BIBLIOGRAPHY

Ayer A. J. Langflage Truth, and Logic. 2nd edition.
’ New Yor : over PuBIications, nc., 19h7.

Carnap, Rudolph. "Empiricism, Semantics, and Ontology,"
Revue Internationale de Philosophie, IV(l950 ,
pp . 20-4.0.

. "Meaning and Synonymy in Natural Languages,"
‘Philosophical Studies, VII(1955), pp. 33-47.

 

Feigl, Herbert, and Brodbeck, May (eds.). Readin 8.3g
the PhilOSOphy_of Science. New York: AppIeton-
Century-Crofts,_Inc., I953.

 

Goodman, Nelson. "On Likeness of Meaning," Anal sis,
X(1949), pp.

Hume, David. Enquiries Concerning the Human Under-
standin and Cbncerning tHeIPrincipIes of-
horaIs, edited byL.A. SEIBy- Bigge. 233
editio . Oxford: The Clarendon Press, 1936.

Kant, Immanuel. A Criti ue of Pure Reason, translated
by Max Nuller. and Ed. revised.‘INew York:
Macmillan Company, 1927. -

Lewis, G. I. "The Modes of Meaning," Philoso h and
Phsnomenological Research, IV(I9EE), pp. 236-
2A .

Linsky, Leonard (ed.). Semantics and the Philosophy
of Lan a e. Urbana: The University of
IIlinois Eress, 1952.

Locke, John. An Essa Concerning Human Understandin ,
edited-by . . Fraser. Oxford: The Clarendon
Press, 189A.

martin, Richard. "On 'analytic'," Philosophical
Studies, III(l952), pp. 42-h7.

Mates, Benson. "Analytic Sentences," The Philosophical
Review, LX(1951), pp. 525-53h.

90

Mill, John Stuart. PhilOSOphy of the Scientific
Method, editedIby Ernest—Nagel. NewfiYork:
Hafner Publishing Co., 1950.

Peach, Bernard. "A Nondescriptive Theory of Analytic,"
The Philosophical Review, LXI(l952), pp. 349-
357.

Robinson, Richard. "Necessary Propositions," Mind,
LXVII(1958)’ pp. 289-30h.

Rudner, Richard. "A Note on Likeness of Meaning,"
Analysis, X(l950), pp. 115-118.

"Formal and Nonformal," Philosophy of
Science, XVI(l9h9), pp. Al-LS.

Russell, Bertrand. A,Histo Of Western Philosophy:
And Its Connection wltEIPoIitical and Social
Circumstances‘from the Earliest Times pg 35g
Present‘Day. New”York: Simon and’Schuster,

1945.

. Human Knowledge: Its Scope and Limits.
New York:fiSimon and Schuster, I9 8.

 

Quine, W. V. From g_Logical Point o§.View: Nine
Logico-PhilOSOphicaI ESsa s. Gambridge:
Harvard University—Press,_ 953.

White, M. G. "The Analytic and Synthetic: an Untenable
Dualism," in Sidney Hook (ed.). John Dewe :
Philosopher of Science and Freedom. New York:
Dial Press, I950, pp. 3IEEBB .

. Toward Reunion in Philosophy. Cambridge:
arvardeniversiterress, 1956.

. "A Finitistic Approach to Philosophical
ITheses," The Philosophical Review, LX(1951),
PP. 299’3160

ROOM USE ONLY

 

 

MICHIGAN STQTE UNIV. LIBRRRIES

31293105985455