Martin asks for a compelling argument that intensions, e.g.

Martin asks for a compelling argument that intensions, e.g., propositions,
are required in a semantical or philosophical analysis of language. I would
not assert that the use of concepts like intension and Desi are necessary,
but their use seems to me convenient and natural for the reasons given
earlier (in §9 IV and in § 10 IV).
15. W. V. Quine on Logical Truth
A. The linguistic doctrine of logical truth. In the first part of his essay,
Quine offers an exposition and a detailed informal discussion of what he
calls "the linguistic doctrine of logical truth". I shall first comment on this
part, and later discuss his more specific criticisms of my conception of
logical truth.
The conception of the nature of logical truth, which was developed in the
Vienna Circle on the basis of Wittgenstein's ideas, and which I still
maintain in its essential points, was originally, before the construction of a
systematic L-semantics, formulated only in informal explanations. Among
the various formulations which Quine mentions and discusses critically,
there are some which today I would no longer regard as psychologically
helpful and would therefore avoid. One of them is the
-915characterization of logical truth as based on "linguistic fiat" or "linguistic
conventions". Neither does Quine's choice of the term "linguistic doctrine"
seem to be quite suitable for my conception. The term "linguistic
convention" is usually understood in the sense of a more or less arbitrary
decision concerning language, such as the choice of either centimeter or
inch as a unit of length. Sometimes the fact that a certain concept is
expressed in a certain language by a certain word is loosely said to be a
matter of convention. Although in this case there is no explicit agreement,
the term "convention" may nevertheless be employed to express the fact
that the usage might be changed by a decision, i.e., that a new word
might be chosen instead of the old one, without thereby changing any
essential characteristics of the given language. On the other hand, the
logical truth of the sentence "all black dogs are dogs" is not a matter of
convention even in the looser sense. Once the meanings of the individual
words in a sentence of this form are given (which may be regarded as a
matter of convention), then it is no longer a matter of convention or of
arbitrary choice whether or not to regard the sentence as true; the truth
of such a sentence is determined by the logical relations holding between
the given meanings. (In the present example, only the meanings of "all"
and "are" are relevant.)
A better informal characterization of logical truth mentioned also by Quine
describes it as truth based on meanings. To ascertain the truth of a given
sentence, it is necessary, first, to know the meaning of each part and
thereby that of the sentence as a whole; in other words, it is necessary to
understand the sentence. To ascertain the truth of some sentences, e.g.,
"Some dogs are black", it is further necessary to know certain facts of the
world. In the case of other sentences, e.g., "all black dogs are black", this
is not necessary; to understand them is a sufficient basis for the
determination of their truth. Such sentences are called "logically true" or
"analytic".
In the following discussion, I shall sometimes be compelled to discuss
Quine's views hypothetically, that is to say, on the basis of presumptions
about the meanings of his formulations, because I have not been able to
determine their meanings with sufficient clarity.
Quine says (in § II) that in a certain sense,
(1) elementary logic is obvious.
I presume that he does not understand the word "obvious" here in the
sense in which someone might say: "it is obvious that I have five fingers
on my right hand", but rather in the sense in which the word is used in: "it
is obvious that, if there is no righteous man in Sodom, then all men in
Sodom are non-righteous". In this case, one merely has to think in order
to recognize the truth; no observations of the men of Sodom are
-916needed. If Quine has this meaning in mind, we are in agreement.
However, Quine says later (in § III) of the linguistic doctrine of
elementary logical truth, which I shall call "LD" for short, that
(2) LD "seems to imply nothing that is not already implied by the fact that
elementary logic is obvious".
Again I agree. In other words:
(3) Whatever is implied by LD, is implied by (1).
Hence, since LD is implied by LD:
(4) LD is implied by (1).
Thus Quine, having accepted (1), must also accept LD. His argument
seems to be not a refutation of LD but rather a proof of it.
Indeed, I have the impression that Quine's critical argument (2) is not
meant as a refutation. He himself says soon afterwards: "I do not suggest
that the linguistic doctrine is false". I presume that he wants to say that
the doctrine is not false. (If so, I wish he had said sol) He nowhere says
that the doctrine is meaningless; this also would not accord with his
previous statement (2), nor with his remark that the doctrine plays a role
analogous to "0 = 0". Therefore we may presume that he regards the
doctrine as true. (If so, . .!) The main point of his criticism seems rather
to be that the doctrine is "empty" and "without experimental meaning".
With this remark I would certainly agree, and I am surprised that Quine
deems it necessary to support this view by detailed arguments. In line
with Wittgenstein's basic conception, we agreed in Vienna that one of the
main tasks of philosophy is clarification and explication. Usually, a
philosophical insight does not say anything about the world, but is merely
a clearer recognition of meanings or of meaning relations. If an insight of
this kind is expressed by a sentence, then this sentence is, although
meaningful (as we would maintain in contrast to Wittgenstein's view), not
factual but rather analytic. Thus I would interpret, e.g., the principle of
verifiability (or of confirmability), or the empiricist principle that there is
no synthetic a priori, as consisting of proposals for certain explications
(often not stated explicitly) and of certain assertions which, on the basis
of these explications, are analytic. Such philosophical principles or
doctrines are sometimes called theories; however, it might be better not
to use the term "theory" in this context, in order to avoid the
misunderstanding that such doctrines are similar to scientific, empirical
theories.
B. Two arguments of Quine's. Now I come to Quine's important
arguments (in §§ VIII-X) directed against my present view of the concept
of analyticity (or logical truth in the wider sense) as a semantical concept.
Quine's objections here should be considered in connection with those
-917in some earlier articles of his, which are reprinted in his book. 21 Some of
my arguments are closely related to those made by other authors. 22
I shall state two arguments by Quine, which were at first rather puzzling
to me and, it seems, to other authors. Then I shall explain what I
presume to be Quine's basic view, which underlies these two arguments
and would make them understandable. Finally, I shall indicate my position
with respect to this view. My proposals for the explication of analyticity
have always been given for a formalized (codified, constructed) language
L, i.e., a language for which explicit semantical rules are specified that
lead to the concept of truth (e.g., rules of the kinds A, B, C, and D in §
10). The explication is given by additional rules, essentially by a list of
meaning postulates (A-postulates) and, based upon them, a definition of
"A-true" (which I use as a technical term for the explicatum). The first of
Quine's critical arguments consists in the remark that the meaning
postulates are recognizable only by the label "meaning postulates" and
that the sense of this label is not clear (§ X); similarly he said in his book
(p. 33f.) that the semantical rules are recognizable only by the heading
"Semantical Rules" which itself is meaningless. I was puzzled by this
remark because neither Quine nor anybody else has previously criticized
the obvious fact that, e.g., the admitted forms of sentences of a
formalized language L are only recognizable by a label like "Sentence
Forms in L" preceding a list of forms of expressions, or the fact that the
axioms of a logical calculus are only recognizable by the label "Axioms".
Why should the same fact be objectionable in the case of meaning
postulates?
The second objection consists in Quine's statement that the concept of
analyticity is acceptable only if it is not merely explicated by rules in pure
semantics, but rather by an empirical criterion in behavioristic terms,
applicable to natural languages (§ IX); in his book (pp.56-64) he said the
same concerning synonymy. It seemed to me puzzling why for semantical
concepts like analyticity or synonymy the definition of a corresponding
empirical, pragmatical concept is required, while for other semantical
concepts like truth, the name-relation, and the like, a requirement of this
kind is not made.
My interpretation of Quine's intention is as follows, formulated in my
terminology. It seems to me that Quine's criticism is not directed against
the proposed semantical explicata. I believe that he would agree that,
____________________
21
W. V. Quine, From a Logical Point of View ( Cambridge, Mass., 1953). In the following,
the references to Quine citing page numbers refer to this book; those citing Roman section
numbers refer to Quine's essay in the present volume.
22
Benson Mates, "Analytic Sentences", Phil. Review, LX ( 1951), 525-34. Richard M. Martin
, "On 'Analytic'", Phil. Studies, III ( 1952), 42-47. H. P. Grice and P. F. Strawson, "In
Defense of a Dogma", Phil. Review, LXV ( 1956), 141-158.
-918e.g., my rules of the above mentioned kinds, leading to the definition of
"A-true", are in themselves exact and unobjectionable. His criticism is
rather that there is no clear explicandum, in other words, that the
customary pre-systematic explanations of analyticity are too vague and
ambiguous, and basically incomprehensible. This would make it
understandable why he requires for analyticity an empirical criterion, while
he does not require it for truth. In Quine's view, there is the following
basic difference. In the case of truth he recognizes a sufficiently clear
explicandum; i.e., before an explication had been given, the use of this
concept had been sufficiently clear, at least for practical purposes. On the
other hand, Quine sees no sufficiently clear, pre-systematic concept of
analyticity which could be taken as an explicandum. If an empirical
criterion for analyticity with respect to natural languages were given, then
this concept could serve as an explicandum for a reconstruction of a
purely semantical concept of A-truth. This seems to me to be Quine's real
motivation in demanding such a criterion. It would also explain Quine's
first objection. His remark on recognizability by labels only may be
understood, not as a criticism of the semantical rules themselves, but
rather as pointing out the lack of an explicandum. He says in his book (p.
33): "The difficulty is simply that the [semantical] rules contain the word
'analytic', which we do not understand!" He proposes to use instead the
untendentious term "K" for the class determined by the rules because the
use of the apparently familiar word "analytic" might give us the illusion
that we were in possession of a clear explicandum.
As I now understand Quine, I would agree with his basic idea, namely,
that a pragmatical concept, based upon an empirical criterion, might serve
as an explicandum for a purely semantical reconstruction, and that this
procedure may sometimes, and perhaps also in the present case, be a
useful way of specifying the explicandum. On the other hand, I would not
think that it is necessary in general to provide a pragmatical concept in
order to justify the introduction of a concept of pure semantics.
C. Empirical criteria for intension concepts. Since I agree with Quine's
basic idea, I have accepted his challenge to show that an empirical
criterion for intension concepts with respect to natural languages can be
given. I believe I have shown this in my paper on natural languages [
1955-3]. In the original draft of my reply to Quine's present essay, the
part dealing with natural languages grew finally beyond the space
available in this volume. Therefore, I elaborated this part and published it
separately as the above paper.
The basic ideas underlying my intensionalist thesis are simple. It seemed
rather plausible to me from the beginning that there should be an
empirical criterion for the concept of the meaning of a word or a
-919phrase, in view of the fact that linguists traditionally determine empirically
the meanings, meaning differences, and shifts of meanings of words, and
that with respect to these determinations they reach a measure of
agreement among themselves which is often considerably higher than that
reached for results in most of the other fields of the social sciences.
Quine's arguments to the effect that the lexicographers actually have no
criterion for their determinations did not seem at all convincing to me.
In my paper, I tried to show the possibility of giving operational rules for
testing hypotheses concerning the intensions of predicates of a natural
language, on the basis of responses by the users of this language. For the
sake of simplicity, I shall here take not "intension" but rather "analytic" as
an example. Let us suppose that two linguists study the natural language
L as used by the person X. Let us suppose that L consists of some English
words and English sentences, among them the following sentence:
(S
1
) "All ravens are black".
We assume that the two linguists agree on the basis of previous
experiments that X uses the words "all" and "are" in the ordinary sense,
and that X has repeatedly affirmed the sentence S 1 and hence
presumably regards it as true. Now the first linguist states the following
hypothesis:
(5) "The sentence S
1
is analytic in language L for person X".
The other linguist denies this hypothesis. In order to obtain evidence
relevant for (5), the linguists say to X: "Mr. Smith told us that he had
found a raven which is not black but white, and that he will show it to you
tomorrow. Will you then revoke your assertion of S 1 "? Let us consider the
following two of many possible responses by X:
(6) "I would never have believed that there are white ravens; and I still
do not believe it until I see one myself. In that case I shall, of course,
have to revoke my assertion".
(7) "There cannot be white ravens. If a bird is not black, then I just would
not call it a raven. If Mr. Smith says that his raven is not black, then
(assuming that he is not lying or joking) his use either of the word 'raven'
or of the word 'black' must be different from my use".
It seems obvious to me that a response like (6) would be disconfirming
evidence for hypothesis (5), while a response like (7) would be confirming
evidence for it. Thus it is clear that (5) is an empirical hypothesis which
can be tested by observations of the speaking behavior of X. If anyone is
still sceptical about this possibility, I should like to refer him to a recent
book by Arne Naess, 23 which shows by numerous examples how
____________________
23
A. Naess, Interpretation and Preciseness: A Contribution to the Theory of Communication.
Skrifter Norske Videnskaps-Akademi, Oslo, II. Hist.-Philos. Klasse ( 1953), No. 1.
-920hypotheses about the synonymy of expressions can be tested by empirical
procedures.
Furthermore, in my paper I have shown that it is also possible to
determine empirically the intension of a predicate of a language L for a
robot who makes observations and can receive and deliver messages in
the language L.
D. Analyticity and change of language. Quine shows (in his book, pp.
42-46) that a scientist, who discovers a conflict between his observations
and his theory and who is therefore compelled to make a readjustment
somewhere in the total system of science, has much latitude with respect
to the place where a change is to be made. In this procedure, no
statement is immune to revision, not even the statements of logic and of
mathematics. There are only practical differences, and these are
differences in degree, inasmuch as a scientist is usually less willing to
abandon a previously accepted general empirical law than a single
observation sentence, and still less willing to abandon a law of logic or of
mathematics. With all this I am entirely in agreement. But I cannot follow
Quine when he infers from this fact that it becomes folly to seek a
boundary between synthetic and analytic statements. I agree that "any
statement can be held true come what may". But the concept of an
analytic statement which I take as an explicandum is not adequately
characterized as "held true come what may". First of all, I should make a
distinction between two kinds of readjustment in the case of a conflict
with experience, namely, between a change in the language, and a mere
change in or addition of, a truth-value ascribed to an indeterminate
statement, (i.e., a statement whose truth value it not fixed by the rules of
language, say by the postulates of logic, mathematics, and physics). A
change of the first kind constitutes a radical alteration, sometimes a
revolution, and it occurs only at certain historically decisive points in the
development of science. On the other hand, changes of the second kind
occur every minute. A change of the first kind constitutes, strictly
speaking, a transition from a language L n to a new language L n+1. My
concept of analyticity as an explicandum has nothing to do with such a
transition. It refers in each case to just one language; "analytic in L n "
and "analytic in L n+1 " are two different concepts. That a certain sentence
S is analytic in L n means only something about the status of S within the
language L n ; as has often been said, it means that the truth of S in L n is
based on the meanings in L n of the terms occurring in S. To be sure, this
status has certain consequences in case of changes of the second kind,
namely, that analytic sentences cannot change their truth-value. But this
characteristic is not restricted to analytic sentences; it holds also for
certain synthetic sentences, e.g., physical postulates and their logical
consequences.
-921E. The concept of analyticity in philosophy. I believe that the
distinction between analytic and synthetic statements, expressed in
whatever terms, is practically indispensable for methodological and
philosophical discussions. This is also indicated by the fact that this
distinction is made by a large majority of philosophers, including some of
those who do not explicitly acknowledge the distinction in these terms or
even reject it. As an example, let me refer to a philosopher whose work I
esteem very highly, although I cannot agree in all points with his views.
This philosopher once undertook to destroy a certain doctrine, propounded
by some other philosophers. He did not mean to assert that the doctrine
was false; presumably he regarded it as true. But his criticism concerned
its particular kind of truth, namely that the truth of the doctrine was of the
analytic kind. To be sure, he did not use the word "analytic", which he did
not seem to like very much. Instead, he used other expressions which,
nonetheless, clearly seem to have essentially the same meaning as
"analytic". What he showed was that various attempts to assign an
experimental, empirical meaning to this doctrine remained without
success. Finally he came to the conclusion that the doctrine, even though
not false, is "empty" and "without experimental significance".
16. Herbert G. Bohnert on Definitions and Analyticity
I am in agreement with most points in Bohnert's discussions, in particular,
with his arguments against Quine's objections to the concept of
analyticity; and I find many of his explanations illuminating. Therefore I
shall restrict myself to only a few comments.
I agree with Bohnert's remark (in § II) that today an approach to
semiotics by way of a behavioristic sociologism, analogous to the
approach of the earlier psychologism, poses a threat to the drawing of
precise distinctions in logic. Bohnert believes that Quine's requirement of
an empirical criterion for synonymy is an example of this kind of
sociologism. This would be the case if, as Bohnert believes, Quine actually
had the intention of founding logic upon empirical concepts. Quine's
formulations on this point admit of a variety of interpretations, including,
perhaps, Bohnert's as well. In my reply (§ 15) I have given Quine the
benefit of the doubt and have suggested a hypothesis about the
motivation of Quine's requirement in such a way that I would be able to
agree with it. It remains for Quine to make clear which of the
interpretations he has in mind.
Bohnert's discussion of what he calls "recipe terms" is clarifying and
useful. He is certainly right that in general syntax, and even more in
general semantics, some concepts cannot be adequately introduced by
exact general definitions for all languages, but only by different recipe
definitions for different classes of languages. However, it is possible to
-922give general exact definitions both for A-truth (analyticity) and for truth
(see § 10 I, D and G) provided that other suitable concepts occurring in
these general definitions are introduced by recipe definitions.
Bohnert points out correctly (in § III) that the term "definition" must be
introduced as a recipe term, newly defined for each language or class of
languages by an enumeration of forms. The situation is similar for the
term "meaning postulate". Nevertheless, it is possible, as Bohnert
indicates, to give general directives, though not in the form of exact rules,