Professional Documents
Culture Documents
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms
Wiley is collaborating with JSTOR to digitize, preserve and extend access to Noûs
161
These ar
of Venn
when we
possible
But cont
for this.
The sub
or unpro
area hav
between
the trad
contensive relations) and relations of intent (or: intensive rela-
tions). I shall begin by explaining this distinction.
The content of a predicate is the set of predicates which are,
by the definition of F, true of a thing of which F is true. Thus
G belongs to the content of F (G is contensively included in F) if
the conventional definiens of F is a conjunction of G and some other
predicates. My content is what Keynes called the conventional in-
tension.2 It also seems to fit fairly well what Bradley, Bosanquet,
and Lewis meant by intension, Kant by Inhalt and the Port-Royal
Logic by comprehension.3
To explain the notion of intent, it may be easier to begin
with a listing of traditional characterizations. The intent of a predi-
cate has been characterized as:
cepts. But this does not rule out the continued demand for an inter-
pretation of these concepts: in that case, however, answers which
go outside semantics proper must be acceptable. Such a demand
for interpretation can only be construed as a request that we show
the concepts in question to have a significant role.
There is in fact an obvious field which may provide such a
grounding for semantic concepts, namely, pragmatics. And I pro-
pose accordingly that we construe the demand for an interpreta-
tion of semantic concepts as answerable by the exhibition of a clear
pragmatic counterpart. For example, the semantic statement "The
tenn W denotes the object 0" has as pragmatic counterpart "The
person X uses the term W to refer to the object 0 (at time t)."
The exhibition of a pragmatic counterpart does not define the se-
mantic concept. It is perfectly legitimate to use the semantic con-
cept of denotation, even if no term is used to refer to the same
t-hing by all persons or even by any person at all times. The use of
the semantic concept signals a certain level of abstraction, which
involves disregarding the vagaries of contextual factors, idiosyn-
cratic usage, changes in usage with time, and so on.14 But had the
concept of denotation no clear pragmatic counterpart, its use would
have no clear relevance to the study of language at all.
So now our task is to provide similar pragmatic counterparts
to relations of intent. For brevity, we shall look only at identity of
intent, which we regard as one plausible explication of synonymy.
The pragmatic counterpart to "F is intensively identical with G" is
"F and G are synonymous for X (at time t)." But what is meant
by the latter? Notwithstanding the well-known puzzles and doubts
which have been raised regarding synonymy, it seems to me that
the literature contains sufficient grounds for regarding "F and G
are synonymous for X" as in no way more problematic than "X uses
W to refer to 0'. We may briefly recount the relevant points here.
An early, but very thorough discussion, is Carnap's "Meaning
and Synonymny in Natural Languages"'J5 In particular he holds that
proper interrogation of the person X will provide factual evidence
for or against the assertion that two terms are synonymous for X.
Thus he describes how a linguist might determine the intention of
"Pferd" for a German Karl.
(i) Have you ever (to the best of your knowledge) come across
anything which you call an A but of which you said that it was
not a B?
Faced with this kind of question, the subject may express a long-
standing commitment ("Of course notl"), or a new intention ("I
have in the past, but now I think that is wrong."). Or he may
have to make a decision, because prior to the questionnaire, the
question may never have arisen. The reasons for this commitment,
intention, or decision can be of various sorts. It may be that he
believes the usage in question to be "incorrect", "not standard",
"not the Queen's English", and that he wishes to conform to norms
which (he believes to) exist in his society. Or he may be concerned
with the emotive effect on others. Or again his grounds may be
largely inductive: he may adhere to a theory so strongly that he
wishes to replace at least part of his prescientific framework with
t-his theory.
The difference between questions (i) and (ii) is that the
former inquires about factual belief, and the latter about commit-
ment or intention. It is certainly not a difference of retrodiction
All the languages in the class C(S*) have this syntax. The logical
space must in each case be some nonempty set H, and the inter-
pretation function f must assign to each predicate F some subset
f(F) of H. Now we turn to the notions of model, truth, and val-
idity. We define these for an arbitrary member L of C(S*), whose
logical space and interpretation function we denote as H and f.
Truth: Eve
h(Fa) = f(F)
h(A & B)=h(A)rnh(B)
h(Q---A) =H - h(A)
h(A- B) = H if h(A) c h(B)
A A, the null set, otherwise
The sentence A is true in the model < loc,X > for L
only if loc(X) E h(A).
This discussion does not throw new light on the nature of S5, which
has long been known to admit of such interpretations. But it places
the relation of intensive inclusion in a familiar frame of discussion.
all these logicians agree that (/x) has at least the properties of the
ordinary quantifier and either state or conjecture that the proper-
ties of (/x) are exactly the properties of (x). We shall see below
that this is demonstrably so for the present interpretation.
Let us take a concrete example: a language which has predi-
cates F,G, . . ., of various degrees, variables x,y, . . ., the quan-
tifier (/x), and so on. The sentence (/x)(FxDGx) has as in-
tended interpretation:
The predicate F is intensively included in the predicate G.
It is read as:
It is read as:
"? Note that we have listed El among the logical signs instead of with
the predicates; this simplifies the presentation somewhat.
This is the precise form of the assertion that (/x) has the same
formal properties as (x). Note that any given member L of C(S)
will in general have valid statements which are not quantifica-
tional theorems. These residual valid statements reflect the mean-
ing relations among the predicates represented by means of H and
f. They are essentially what Carnap called the "meaning postu-
lates". This term is misleading here, however, because the mean-
ing structure may be so complex that no formal postulate system
can capture all these residual valid statements (the theory of the
subject may be essentially incomplete, by Godel's theorem). In this
sense our method of logical spaces is essentially stronger than
Carnap's method of meaning postulates.