The Grice Club

Welcome

The Grice Club

The club for all those whose members have no (other) club.

Is Grice the greatest philosopher that ever lived?

Search This Blog

Thursday, August 30, 2012

"I regret that my poor choic of words caused some people to understand what I was saying"

Speranza



 Christopher Weywant, The New Yorker, Aug. 2012

R. Dale was drawing attention to this (elesewhere) -- and it may merit a Griceian analysis.

(i) I regret that my poor choice of words caused some people to understand what I was saying.

SELF-CONTRADICTORY.

Points:

Griceian:

distinction:

saying/meaning.

(ii) I regret that my poor choice of words caused some people to understand that I MEANT.

---

Strawson's point about 'understand' being a value-oriented concept.

(iii) "I want you to misunderstand me" -- as contradictory.

(iv) I regret that my poor choice of words caused some people to [OBVIOUSLY -- and the joke trades on this] MISunderstand what I was saying (or meant).

And so on.


        

Monday, August 27, 2012

Strawsonian presupposition and Griceian implicature -- or how to rob Peter to pay Paul

Speranza

We, as Grice and Strawson remarked, very commonly use expressions of certain kinds to
mention or refer to some individual person or single object
or particular event or place or process, in the course of
d'oing what we should normally describe as making a state-
ment about that person, object, place, event, or process. I
shall call this way of using expressions the t uniquely re-
ferring use'. The classes of expressions which are most
commonly used in this way are : singular demonstrative
pronouns ('this' and 'that'); proper names (e.g. 'Venice',
( Napoleon', 'John') ; singular personal and impersonal pro-
nouns ('he', 'she', T, 'you', 'it'); and phrases beginning
with the definite article followed by a noun, qualified or
unqualified, in the singular (e.g. 'the table', 'the old man',
' the king of France J ). Any expression of any of these classes
can occur as the subject of what would traditionally be
regarded as a singular subject-predicate sentence; and
would, so occurring, exemplify the use I wish to- discuss.

I do not want to say that expressions belonging to these
classes never have any other use than the one I want to
discuss. On the contrary, it is obvious that they do. It is
obvious that anyone who uttered the sentence, 'The whale
is a mammal', would be using the expression 'the whale'
in a way quite different from the way it would be used by
anyone who had occasion seriously to utter the sentence,
'The whale struck the ship'. In the first sentence one is
obviously not mentioning, and in the second sentence one
obviously is mentioning, a particular whale. Again if I

21



22 ESSAYS IN CONCEPTUAL ANALYSIS

said, 'Napoleon was the greatest French soldier', I should
be using the word 'Napoleon' to mention a certain indi-
vidual, but I should not be using the phrase, 'the greatest
French soldier', to mention an individual, but to say some-
thing about an individual I had already mentioned. It
would be natural to say that in using this sentence I was
talking about Napoleon and that what I was saying about
him was that he was the greatest French soldier. But of
course I could use the expression, 'the greatest French
soldier 3 , to mention an individual ; for example, by saying :
'The greatest French soldier died in exile'. So it is obvious
that at least some expressions belonging to the classes I
mentioned can have uses other than the use I am anxious to
discuss. Another thing I do not want to say is that in any
given sentence there is never more than one expression used
in the way I propose to discuss. On the contrary, it is
obvious that there may be more than one. For example, it
would be natural to say that, in seriously using the sentence,
'The whale struck the ship', I was saying something about
both a certain whale and a certain ship, that I was using
each of the expressions 'the whale' and 'the ship' to mention
a particular object ; or, in other words, that I was using each
of these expressions in the uniquely referring way. In
general, however, I shall confine my attention to cases where
an expression used in this way occurs as the grammatical
subject of a sentence.

I think it is true to say that Russell's Theory of Descrip-
tions, which is concerned with the last of the four classes of
expressions I mentioned above (i.e. with expressions of the
form 'the so-and-so'), is still widely accepted among logicians
as giving a correct account of the use of such expressions in
ordinary language. I want to show in the first place, that
this theory, so regarded, embodies some fundamental
mistakes.

What question or questions about phrases of the form
'the so-and-so' was the Theory of Descriptions designed to
answer ? I think that at least one of the questions may be



ON REFERRING 23

illustrated as follows. Suppose someone were now to utter
the sentence, 'The king of France is wise 5 . No one would
say that the sentence which had been uttered was meaning-
less. Everyone would agree that it was significant. But
everyone knows that there is not at present a king of France.
One of the questions the Theory of Descriptions was designed
to answer was the question : How can such a sentence as
'The king of France is wise* be significant even when there
is nothing which answers to the description It contains, i.e.,
in this case, nothing which answers to the description 'The
king of France ' ? And one of the reasons why Russell
thought it important to give a correct answer to this ques-
tion was that he thought it important to show that another
answer which might be given was wrong. The answer that
he thought was wrong, and to which he was anxious to supply
an alternative, might be exhibited as the conclusion of either
of the following two fallacious arguments. Let us call the
sentence 'The king of France is wise' the sentence S. Then
the first argument is as follows :

(i) The phrase, 'the king of France', is the subject of
the sentence S.

Therefore (2) if S is a significant sentence, S is a sentence
about the king of France.

But (3) if there in no sense exists a king of France, the
sentence is not about anything, and hence not about the
king of France.

Therefore (4) since S is significant, there must in some
sense (in some world) exist (or subsist) the king of France.

And the second argument is as follows :

(1) If S is significant, it is either true or false.

(2) S is true if the king of France is wise and false if the
king of France is not wise.

(3) But the statement that the king of France is wise and
the statement that the king of France is not wise are alike
true only if there is (in some sense, in some world) something
which is the king of France.



24 ESSAYS IN CONCEPTUAL ANALYSIS

Hence (4) since S Is significant, there follows the same
conclusion as before.

These are fairly obviously bad arguments, and, as we
should expect, Russell rejects them. The postulation of a
world of strange entities, to which the king of France belongs,
offends, he says, against 'that feeling for reality which ought
to be preserved even in the most abstract studies 1 . The fact
that Russell rejects these arguments is, however, less interest-
ing than the extent to which, in rejecting their conclusion,
he concedes the more important of their principles. Let me
refer to the phrase, 'the king of France', as the phrase D.
Then I think Russell's reasons for rejecting these two argu-
ments can be summarized as follows. The mistake arises,
he says, from thinking that D, which is certainly the gram-
matical subject of S, is also the logical subject of S. But D
is not the logical subject of S. In fact S, although gram-
matically it has a singular subject and a predicate, is not
logically a subject-predicate sentence at all. The proposi-
tion it expresses is a complex kind of existential proposition,
part of which might be described as a ' uniquely existential '
proposition. To exhibit the logical form of the proposition,
we should re-write the sentence in a logically appropriate
grammatical form ; in such a way that the deceptive similar-
ity of S to a sentence expressing a subject-predicate proposi-
tion would disappear, and we should be safeguarded against
arguments such as the bad ones I outlined above. Before
recalling the details of Russell's analysis of S, let us notice
what his answer, as I have so far given it, seems to imply.
His answer seems to imply that in the case of a sentence
which is similar to S in that (i) it is grammatically of the
subject-predicate form and (2) its grammatical subject does
not refer to anything, then the only alternative to its being
meaningless is that It should not really (i.e. logically) be of
the subject-predicate form at all, but of some quite different
form. And this in its turn seems to imply that if there are
any sentences which are genuinely of the subject-predicate
form, then the very fact of their being significant, having a



ON REFERRING 25

meaning, guarantees that there is something referred to by
the logical (and grammatical) subject. Moreover, Russell's
answer seems to imply that there are such sentences. For
if it is true that one may be misled by the grammatical simi-
larity of S to other sentences into thinking that it is logically
of the subject-predicate form, then surely there must be other
sentences grammatically similar to S, which are of the subject-
predicate form. To show not only that Russell's answer seems
to imply these conclusions, but that he accepted at least the
first two of them, it is enough to consider what he says about
a class of expressions which he calls ' logically proper names *
and contrasts with expressions, like D, which he calls 'de-
finite descriptions'. Of logically proper names Russell says
or implies the following things :

(1) That they and they alone can occur as subjects of
sentences which are genuinely of the subject-predicate form.

(2) That an expression intended to be a logically proper
name is meaningless unless there is some single object for
which it stands : for the meaning of such an expression just
is the individual object which the expression designates. To
be a name at all, therefore, it must designate something.

It is easy to see that if anyone believes these two pro-
positions, then the only way for him to save the significance
of the sentence S is to deny that it is a logically subject-
predicate sentence. Generally, we may say that Russell
recognizes only two ways in which sentences which seem,
from their grammatical structure, to be about some particular
person or individual object or event, can be significant :

(1) The first is that their grammatical form should be
misleading as to their logical form, and that they should be
analysable, like S, as a special kind of existential sentence.

(2) The second is that their grammatical subject should
be a logically proper name, of which the meaning is the
individual thing it designates.

I think that Russell is unquestionably wrong in this, and
that sentences which are significant, and which begin with



26 ESSAYS IN CONCEPTUAL ANALYSIS

an expression used in the uniquely referring way, fall into
neither of these two classes. Expressions used in the un iquely
referring way are never either logically proper names or
descriptions, if what is meant by calling them ' descriptions '
is that they are to be analysed in accordance with the model
provided by Russell's Theory of Descriptions.

There are no logically proper names and there are no
descriptions (in this sense).

Let us now consider the details of Russell's analysis.
According to Russell, anyone who asserted S would be assert-
ing that :

(1) There is a king of France.

(2) There is not more than one king of France.

(3) There is nothing which is king of France and is not

wise.

It is easy to see both how Russell arrived at this analysis,
and how it enables him to answer the question with which
we began, viz, the question : How can the sentence S be
significant when there is no king of France ? The way in
which he arrived at the analysis was clearly by asking him-
self what would be the circumstances in which we would say
that anyone who uttered the sentence S had made a true
assertion. And it does seem pretty clear, and I have no
wish to dispute, that the sentences (l)-(s) above do describe
circumstances which are at least necessary conditions of
anyone making a true assertion by uttering the sentence S.
But, as I hope to show, to say this is not at all the same
thing as to say that Russell has given a correct account of
the use of the sentence S or even that he has given an account
which, though incomplete, is correct as far as it goes ; and
is certainly not at all the same thing as to say that the model
translation provided is a correct model for all (or for any)
singular sentences beginning with a phrase of the form
1 the so-and-so ' .

It is also easy to see how this analysis enables Russell to
answer the question of how the sentence S can be significant,
even when there is no king of France. For, if this analysis



ON REFERRING 27

is correct, anyone who utters the sentence S to-day would
be jointly asserting three propositions, one of which (viz.
that there is a king of France) would be false ; and since the
conjunction of three propositions, of which one is false, is
itself false, the assertion as a whole would be significant, but
false. So neither of the bad arguments for subsistent entities
would apply to such an assertion.

II

As a step towards showing that Russell's solution of his
problem is mistaken, and towards providing the correct
solution, I want now to draw certain distinctions. For this
purpose I shall, for the remainder of this section, refer to an
expression which has a uniquely referring use as ' an expres-
sion ' for short ; and to a sentence beginning with such an
expression as ' a sentence ' for short. The distinctions I shall
draw are rather rough and ready, and, no doubt, difficult
cases could be. produced which would call for their refine-
ment. But I think they will serve my purpose. The dis-
tinctions are between :

(Ai) a sentence,

(A2) a use of a sentence,

(A3) an utterance of a sentence,

and, correspondingly, between :

(Bi) an expression,

(62) a use of an expression,

(63) an utterance of an expression.

Consider again the sentence, * The king of France is wise '
It is easy to imagine that this sentence was uttered at various
times from, say, the beginning of the seventeenth century
onwards, during the reigns of each successive French
monarch ; and easy to imagine that it was also uttered
during the subsequent periods in which France was not a
monarchy. Notice that it was natural for me to speak of
'the sentence' or 'this sentence' being uttered at various



28 ESSAYS IN CONCEPTUAL ANALYSIS

times during this period ; or, in other words, that it would
be natural and correct to speak of one and the same sentence
being uttered on all these various occasions. It is in the sense
in which it would be correct to speak of one and the same
sentence being uttered on all these various occasions that I
want to use the expression (Ai) 'a sentence'. There are,
however, obvious differences between different occasions of
the use of this sentence. For instance, if one man uttered it
in the reign of Louis XIV and another man uttered it in the
reign of Louis XV, it would be natural to say (to assume)
that they were respectively talking about different people ;
and it might be held that the first man, in using the sentence,
made a true assertion, while the second man, in using the
same sentence, made a false assertion. If on the other hand
two different men simultaneously uttered the sentence (e.g.
if one wrote it and the other spoke it) during the reign of
Louis XIV, it would be natural to say (assume) that they
were both talking about the same person, and, in that case,
in using the sentence, they must either both have made a
true assertion or both have made a false assertion. And
this illustrates what I mean by a use of a sentence. The
two men who uttered the sentence, one in the reign of
Louis XV and one in the reign of Louis XIV, each made
a different use of the same sentence ; whereas the two men
who uttered the sentence simultaneously in the reign of
Louis XIV, made the same use l of the same sentence.
Obviously in the case of this sentence, and equally obviously
in the case of many others, we cannot talk of the sentence
being true or false, but only of its being used to make a true
or false assertion, or (if this is preferred) to express a true
or a false proposition. And equally obviously we cannot
talk of the sentence being about a particular person, for the
same sentence may be used at different times to talk about

1 This usage of 'use' is, of course, different from (a) the current usage in
which 'use 7 (of a particular word, phrase, sentence) = (roughly) * rules for
using ' = (roughly) 'meaning'; and from () my own usage in the phrase
'uniquely referring use of expressions' in which ' use ' = (roughly) 'way of
using'.



ON REFERRING 29

quite different particular persons, but only of a use of the
sentence to talk about a particular person. Finally it will
make sufficiently clear what I mean by an utterance of a
sentence if I say that the two men who simultaneously uttered
the sentence in the reign of Louis XIV made two different
utterances of the same sentence, though they made the same
use of the sentence.

If we now consider not the whole sentence, f The king
of France is wise ' , but that part of it which is the expression,
'the king of France', it is obvious that we can make ana-
logous, though not identical distinctions between (i) the
expression, (2) a use of the expression, and (3) an utterance
of the expression. The distinctions will not be identical;
we obviously cannot correctly talk of the expression 'the
king of France' being used to express a true or false pro-
position, since in general only sentences can be used truly
or falsely ; and similarly it is only by using a sentence and
not by using an expression alone, that you can talk about a
particular person. Instead, we shall say in this case that
you use the expression to mention or refer to a particular
person in the course of using the sentence to talk about him.
But obviously in this case, and a great many others, the
expression (Bi) cannot be said to mention, or refer to, any-
thing, any more than the sentence can be said to be true or
false. The same expression can have different mentioning-
uses, as the same sentence can be used to make statements
with different truth-values. 'Mentioning', or 'referring', is
not something an expression does ; it is something that
someone can use an expression to do. Mentioning, or
referring to, something is a characteristic of a use of an
expression, just as 'being about' something, and truth-or-
falsity, are characteristics of a use of a sentence.

A very different example may help to make these dis-
tinctions clearer. Consider another case of an expression
which has a uniquely referring use, viz. the expression 1 1 ' ;
and consider the sentence, 'I am hot'. Countless people
may use this same sentence ; but it is logically impossible



30 ESSAYS IN CONCEPTUAL ANALYSIS

for two different people to make the same use of this sentence :
or, if this is preferred, to use it to express the same proposi-
tion. The expression T inay correctly be used by (and
only by) any one of innumerable people to refer to himself.
To say 'this is to say something about the expression T :
it is, in a sense, to give its meaning. This is the sort of thing
that can be said about expressions. But it makes no sense
to say of the expression 'I' that it refers to a particular
person. This is the sort of thing that can be said only of a
particular use of the expression.

Let me use 'type 1 as an abbreviation for 'sentence or
expression'. Then I am not saying that there are sentences
and expressions (types), and uses of them, and utterances of
them, as there are ships and shoes and sealing-wax. I am
saying that we cannot say the same things about types, uses
of types, and utterances of types. And the fact is that we
do talk about types ; and that confusion is apt to result
from the failure to notice the differences between what we
can say about these and what we can say only about the
uses of types. We are apt to fancy we are talking about
sentences and expressions when we are talking about the
uses of sentences and expressions.

This is what Russell does. Generally, as against Russell,
I shall say this, Meaning (in at least one important sense)
is a function of the sentence or expression ; mentioning and
referring and truth or falsity, are functions of the use of the
sentence or expression. To give the meaning of an ex-
pression (in the sense in which I am using the word) is to
give general directions for its use to refer to or mention
particular objects or persons ; to give the meaning of a
sentence is to give general directions for its use in making
true or false assertions. It is not to talk about any particular
occasion of the use of the sentence or expression. The
meaning of an expression cannot be identified with the
object it is used, on a particular occasion, to refer to. The
meaning of a sentence cannot be identified with the assertion
it is used, on a particular occasion, to make. For to talk



ON REFERRING 31

about the meaning of an expression or sentence is not to
talk about its use on a particular occasion, but about the
rules, habits, conventions governing its correct use 5 on ail
occasions, to refer or to assert. So the question of whether
a sentence or expression is significant or not has nothing
whatever to do with the question of whether the sentence,
uttered on a particular occasion , is, on that occasion, being
used to make a true-or-false assertion or not, or of whether
the expression is, on that occasion, being used to refer to s or
mention, anything at all.

The source of Russell's mistake was that he thought that
referring or mentioning, if it occurred at all, must be mean-
ing. He did not distinguish Bi from 62; he confused
expressions with their use in a particular context ; and so
confused meaning with mentioning, with referring. If I
talk about my handkerchief, I can, perhaps, produce the
object I am referring to out of my pocket. I cannot pro-
duce the meaning of the expression, f my handkerchief ', out
of my pocket. Because Russell confused meaning with
mentioning, he thought that if there were any expressions
having a uniquely referring use, which were what they
seemed (i.e. logical subjects) and not something else in dis-
guise, their meaning must be the particular object which
they were used to refer to. Hence the troublesome mythology
of the logically proper name. But if someone asks me the
meaning of the expression 'this' once Russell's favourite
candidate for this status I do not hand him the object I
have just used the expression to refer to, adding at the same
time that the meaning of the word changes every time it is
used. Nor do I hand him all the objects it ever has been,
or might be, used to refer to. I explain and illustrate the
conventions governing the use of the expression. This is
giving the meaning of the expression. It is quite different
from giving (in any sense of giving) the object to which it
refers ; for the expression itself does not refer to anything ;
though it can be used, on different occasion, to refer to
innumerable things. Now as a matter of fact there is, in



32 ESSAYS IN CONCEPTUAL ANALYSIS

English, a sense of the word 'mean' in which this word
does approximate to 'Indicate, mention or refer to 5 ; e.g.
when somebody (unpleasantly) says, ' I mean you J ; or when
I point and say, 'That's the one I mean'. But the one I
meant is quite different from the meaning of the expression
I used to talk of it. In this special sense of 'mean', it is
people who mean, not expressions. People use expressions
to refer to particular things. But the meaning of an ex-
pression is not the set of things or the single thing it may
correctly be used to refer to : the meaning is the set of rules,
habits, conventions for its use in referring.

It is the same with sentences : even more obviously so.
Everyone knows that the sentence, 'The table is covered
with books', is significant, and everyone knows what it
means. But if I ask, 'What object is that sentence about ?'
I am asking an absurd question a question which cannot
be asked about the sentence, but only about some use of the
sentence : and in this case the sentence has not been used
to talk about something, it has only been taken as an example.
In knowing what it means, you are knowing how it could
correctly be used to talk about things: so knowing the
meaning has nothing to do with knowing about any par-
ticular use of the sentence to talk about anything. Similarly,
if I ask: 'Is the sentence true or false?' I am asking an
absurd question, which becomes no less absurd if I add,
*It must be one or the other since it is significant'. The
question is absurd, because the sentence is neither true nor
false any more than it is about some object. Of course the
fact that it is significant is the same as the fact that it can
correctly be used to talk about something and that, in so
using it, someone will be making a true or false assertion.
And I will add that it will be used to make a true or false
assertion only if the person using it is talking about some-
thing. If, when he utters it, he is not talking about any-
thing, then his use is not a genuine one, but a spurious or
pseudo-use : he is not making either a true or a false asser-
tion, though he may think he is. And this points the way



ON REFERRING 33

to the correct answer to the puzzle to which the Theory of
Descriptions gives a fatally incorrect answer. The Important
point is that the question of whether the sentence is significant
or not is quite independent of the question that can be raised
about a particular use of it, viz. the question whether it Is a
genuine or a spurious use, whether it is being used to talk
about something, or in make-believe, or as an example In
philosophy. The question whether the sentence is significant
or not is the question whether there exist such language
habits, conventions or rules that the sentence logically could
be used to talk about something ; and is hence quite inde-
pendent of the question whether it is being so used on a
particular occasion.

Ill

Consider again the sentence, ' The king of France is wise',
and the true and false things Russell says about it.

There are at least two true things which Russell would
say about the sentence :

(1) The first is that it is significant ; that if anyone were
now to utter it, he would be uttering a significant sentence.

(2) The second is that anyone now uttering the sentence
would be making a true assertion only if there in fact at
present existed one and only one king of France, and if he
were wise.

What are the false things which Russell would say about
the sentence ? They are :

(1) That anyone now uttering it would be making a true
assertion or a false assertion ;

(2) That part of what he would be asserting would be
that there at present existed one and only one king of France,

I have already given some reasons for thinking that these
two statements are incorrect. Now suppose someone were
in fact to say to you with a perfectly serious air : 'The king
of France is wise*. Would you say, 'That's untrue'? I



34 ESSAYS IN CONCEPTUAL ANALYSIS

think it is quite certain that you would not. But suppose he
went on to ask you whether you thought that what he had
just said was true, or was false ; whether you agreed or dis-
agreed with what he had just said. I think you would be
inclined, with some hesitation, to say that you did not do
either ; that the question of whether his statement was true
or false simply did not arise, because there was no such
person as the king of France. You might, if he were
obviously serious (had a dazed astray-in-the-centuries look),
say something like : 'I'm afraid you must be under a mis-
apprehension. France is not a monarchy. There is no king
of France.' And this brings out the point that if a rr an
seriously uttered the sentence, his uttering it would in some
sense be evidence that he believed that there was a king of
France. It would not be evidence for his believing this
simply in the way in which a man's reaching for his raincoat
is evidence for his believing that it is raining. But nor
would it be evidence for his believing this in the way in
which a man's saying, 'It's raining', is evidence for his be-
lieving that it is raining. We might put it as follows. To
say 'The king of France is wise' is, in some sense of 'imply',
to imply that there is a king of France. But this is a very
special and odd sense of 'imply'. 'Implies' in this sense is
certainly not equivalent to 'entails' (or 'logically implies').
And this comes out from the fact that when, in response to
his statement, we say (as we should) 'There is no king of
France 1 , we should certainly not say we were contradicting
the statement that the king of France is wise. We are cer-
tainly not saying that it is false. We are, rather, giving a
reason for saying that the question of whether it is true or
false simply does not arise.

And this is where the distinction I drew earlier can help
us. The sentence, 'The king of France is wise', is certainly
significant ; but this does not mean that any particular use
of it is true or false. We use it truly or falsely when we use
it to talk about someone; when, in using the expression,
'The king of France 1 , we are in fact mentioning someone.



ON REFERRING 35

The fact that the sentence and the expression, respectively,
are significant just is the fact that the sentence could be
used, in certain circumstances, to say something true or
false, that the expression could be used, in certain circum-
stances, to mention a particular person ; and to know their
meaning is to know what sort of circumstances these are.
So when we utter the sentence without in fact mentioning
anybody by the use of the phrase, 'The king of France 5 , the
sentence does not cease to be significant : we simply fail to
say anything true or false because we simply fail to mention
anybody by this particular use of that perfectly significant
phrase. It is, if you like, a spurious use of the sentence, and
a spurious use of the expression ; though we may (or may
not) mistakenly think it a genuine use.

And such spurious uses x are very familiar. Sophisticated
romancing, sophisticated fiction, 2 depend upon them. If I
began, 'The king of France is wise', and went on, 'and he
lives in a golden castle and has a hundred wives', and so on,
a hearer would understand me perfectly well, without sup-
posing either that I was talking about a particular person,
or that I was making a false statement to the effect that
there existed such a person as my words described. (It is
worth adding that where the use of sentences and expressions
is overtly fictional, the sense of the word 'about' may change.
As Moore said, it is perfectly natural and correct to say
that some of the statements in Pickwick Papers are about
Mr. Pickwick. But where the use of sentences and expres-
sions is not overtly fictional, this use of 'about' seems less
correct ; i.e. it would not in general be correct to say that a
statement was about Mr. X or the so-and-so, unless there
were such a person or thing. So it is where the romancing
is in danger of being taken seriously that we might answer
the question, 'Who is he talking about?' with 'He's not
talking about anybody' ; but, in saying this, we are not

1 The choice of the word 'spurious' now seems to me unfortunate, at least
for some non-standard uses, I should now prefer to call some of these * second-
ary' uses.

2 The unsophisticated kind begins : * Once upon time there was . . .*



3 6 ESSAYS IN CONCEPTUAL ANALYSIS

saying that what he is saying is either false or nonsense.)

Overtly fictional uses apart, however, I said just now that
to use such an expression as 'The king of France' at the
beginning of a sentence was, in some sense of 'imply', to
imply that there was a king of France. When a man uses
such an expression, he does not assert, nor does what he says
entail \ a uniquely existential proposition. But one of the
conventional functions of the definite article is to act as a
signal that a unique reference is being made a signal,
not a disguised assertion. When we begin a sentence with
'the such-and-such' the use of 'the' shows, but does not
state, that we are, or intend to be, referring to one particular
individual of the species 'such-and-such'. Which particular
individual is a matter to be determined from context, time,
place, and any other features of the situation of utterance.
Now, whenever a man uses any expression, the presumption
is that he thinks he is using it correctly : so when he uses
the expression, 'the such-and-such ', in a uniquely referring
way, the presumption is that he thinks both that there is
some individual of that species, and that the context of use
will sufficiently determine which one he has in mind. To
use the word 'the' in this way is then to imply (in the relevant
sense of ' imply ') that the existential conditions described by
Russell are fulfilled. But to use 'the' in this way is not to
state that those conditions are fulfilled. If I begin a sentence
with an expression of the form, 'the so-and-so', and then
am prevented from saying more, I have made no statement
of any kind ; but I may have succeeded in mentioning some-
one or something.

The uniquely existential assertion supposed by Russell
to be part of any assertion in which a uniquely referring use
is made of an expression of the form ' the so-and-so ' is, he
observes, a compound of two assertions. To say that there
is a is to say something compatible with there being several
<^s ; to say there is not more than one is to say something
compatible with there being none. To say there is one
and one only is to compound these two assertions, I have



ON REFERRING 37

so far been concerned mostly with the alleged assertion of
existence and less with the alleged assertion of uniqueness.
An example which throws the emphasis on to the latter will
serve to bring out more clearly the sense of 'Implied' in
which a uniquely existential assertion is implied, but not
entailed, by the use of expressions in the uniquely referring
way. Consider the sentence, 'The table is covered with
books'. It is quite certain that in any normal use of this
sentence, the expression 'the table 1 would be used to make a
unique reference, i.e. to refer to some one table. It is a
quite strict use of the definite article, in the sense In which
Russell talks on p. 30 of Principia Mathematics^ of using
the article 'strictly, so as to imply uniqueness'. On the
same page Russell says that a phrase of the form * the so-
and-so ', used strictly, 'will only have an application in the
event of there being one so-and-so and no more'. Now it is
obviously quite false that the phrase 'the table' In the sen-
tence 'the table is covered with books', used normally, will
'only have an application in the event of there being one
table and no more'. It is indeed tautologically true that, In
such a use, the phrase will have an application only in the
event of there being one table and no more which is being
referred to, and that it will be understood to have an appli-
cation only in the event of there being one table and no
more which it is understood as being used to refer to. To
use the sentence is not to assert, but it is (in the special sense
discussed) to imply, that there is only one thing which is
both of the kind specified (i.e. a table) and is being referred
to by the speaker. It is obviously not to assert this. To
refer is not to say you are referring. To say there is some
table or other to which you are referring is not the same as
referring to a particular table. We should have no use for
such phrases as 'the individual I referred to' unless there
were something which counted as referring. (It would make
no sense to say you had pointed if there were nothing which
counted as pointing.) So once more I draw the conclusion
that referring to or mentioning a particular thing cannot be



38 ESSAYS IN CONCEPTUAL ANALYSIS

dissolved into any kind of assertion. To refer is not to assert,
though you refer in order to go on to assert.

Let me now take an example of the uniquely referring
use of an expression not of the form, 'the so-and-so*. Sup-
pose I advance my hands, cautiously cupped, towards some-
one, saying, as I do so, 'This is a fine red one'. He, looking
into my hands and seeing nothing there, may say: 'What
is ? What are you talking about ? ' Or perhaps, ' But there's
nothing in your hands'. Of course it would be absurd to
say that, in saying 'But you've got nothing in your hands',
he was denying or contradicting what I said. So 'this' is
not a disguised description in Russell's sense. Nor is it a
logically proper name, For one must know what the sen-
tence means in order to react in that way to the utterance
of it. It is precisely because the significance of the word
'this 1 is independent of any particular reference it may be
used to make, though not independent of the way it may be
used to refer, that I can, as in this example, use it to pretend
to be referring to something.

The general moral of all this is that communication is
much less a matter of explicit or disguised assertion than
logicians used to suppose. The particular application of this
general moral in which I am interested is its application to
the case of making a unique reference. It is a part of the
significance of expressions of the kind I am discussing that
they can be used, in an immense variety of contexts, to make
unique references. It is no part of their significance to assert
that they are being so used or that the conditions of their
being so used are fulfilled. So the wholly important dis-
tinction we are required to draw is between

(1) using an expression to make a unique reference ; and

(2) asserting that there is one and only one individual
which has certain characteristics (e.g. is of a certain kind,
or stands in a certain relation to the speaker, or both).

This is, in other words, the distinction between

(i) sentences containing an expression used to indicate



ON REFERRING 39

or mention or refer to a particular person or thing ; and
(2) uniquely existential sentences.

What Russell does is progressively to assimilate more and
more sentences of class (i) to sentences of class (2), and con-
sequently to involve himself in insuperable difficulties about
logical subjects, and about values for individual variables
generally : difficulties which have led him finally to the
logically disastrous theory of names developed In the Enquiry
into Meaning and Truth and in Human Knowledge. That
view of the meaning of logical-subject-expressions which
provides the whole incentive to the Theory of Descriptions
at the same time precludes the possibility of Russell's ever
finding any satisfactory substitutes for those expressions
which, beginning with substantival phrases, he progressively
degrades from the status of logical subjects. 1 It is not
simply, as is sometimes said, the fascination of the relation
between a name and its bearer, that is the root of the trouble.
Not even names come up to the impossible standard set.
It Is rather the combination of two more radical misconcep-
tions : first, the failure to grasp the importance of the
distinction (section II above) between what may be said of
an expression and what may be said of a particular use of
it ; second, a failure to recognize the uniquely referring use
of expressions for the harmless, necessary thing it is, dis-
tinct from, but complementary to, the predicative or ascriptive
use of expressions. The expressions which can in fact occur
as singular logical subjects are expressions of the class I
listed at the outset (demonstratives, substantival phrases,
proper names, pronouns) : to say this is to say that these
expressions, together with context (in the widest sense), are
what one uses to make unique references. The point of the
conventions governing the uses of such expressions is, along
with the situation of utterance, to secure uniqueness of
reference. But to do this, enough is enough. We do not,
and we cannot, while referring, attain ifte point of complete

1 And this in spite of the danger-signal of that phrase, 'misleading gram-
matical form*.



40 ESSAYS IN CONCEPTUAL ANALYSIS

explicitness at which the referring function is no longer per-
formed. The actual unique reference made, if any, is a
matter of the particular use in the particular context ; the
significance of the expression used is the set of rules or con-
ventions which permit such references to be made. Hence
we can, using significant expressions, pretend to refer, in
make-believe or in fiction, or mistakenly think we are refer-
ring when we are not referring to anything. 1

This shows the need for distinguishing two kinds (among
many others) of linguistic conventions or rules : rules for
referring, and rules for attributing and ascribing ; and for
an investigation of the former. If we recognize this dis-
tinction of use for what it is, we are on the way to solving a
number of ancient logical and metaphysical puzzles.

My last two sections are concerned, but only in the barest
outline, with these questions.

IV

One of the main purposes for which we use language is
the purpose of stating facts about things and persons and
events. If we want to fulfil this purpose, we must have
some way of forestalling the question, 'What (who, which
one) are you talking about ? 5 as well as the question, 'What
are you saying about it (him, her) ?' The task of forestalling
the first question is the referring (or identifying) task. The
task of forestalling the second is the attributive (or descriptive
or classificatory or ascriptive) task. In the conventional
English sentence which is used to state, or to claim to state,
a fact about an individual thing or person or event, the
performance of these two tasks can be roughly and approxi-
mately assigned to separable expressions. 2 * And in such a

[* This sentence now seems to me objectionable in a number of ways, notably
because of an unexplicitly restrictive use of the word 'refer'. It could be more
exactly phrased as follows : * Hence we can, using significant expressions, refer
in secondary ways, as in make-believe or in fiction, or mistakenly think we are
referring to something in the primary way when we are not, in that way, referring
to anything'.]

2 I neglect relational sentences; for these require, not a modification in
the principle of what I say, but a complication of the detail.



ON REFERRING 41

sentence, this assigning of expressions to their separate roles
corresponds to the conventional grammatical classification
of subject and predicate. There is nothing sacrosanct about
the employment of separable expressions for these two tasks.
Other methods could be, and are, employed. There is, for
instance, the method of uttering a single word or attributive
phrase in the conspicuous presence of the object referred to ;
or that analogous method exemplified by, e.g., the painting
of the words ' unsafe for lorries' on a bridge, or the tying of
a label reading * first prize ' on a vegetable marrow. Or one
can imagine an elaborate game iri which one never used an
expression in the uniquely referring way at all, but uttered
only uniquely existential sentences, trying to enable the
hearer to identify what was being talked of by means of an
accumulation of relative clauses. (This description of the
purposes of the game shows in what sense it would be a
game : this is not the normal use we make of existential sen-
tences.) Two points require emphasis. The first is that the
necessity of performing these two tasks in order to state
particular facts requires no transcendental explanation : to
call attention to it is partly to elucidate the meaning of the
phrase, * stating a fact'. The second is that even this elucida-
tion is made in terms derivative from the grammar of the
conventional singular sentence ; that even the overtly
functional, linguistic distinction between the identifying and
attributive roles that words may play in language Is prompted
by the fact that ordinary speech offers us separable expressions
to which the different functions may be plausibly and approxi-
mately assigned. And this functional distinction has cast long
philosophical shadows. The distinctions between particular
and universal, between substance and quality, are such
pseudo-material shadows, cast by the grammar of the con-
ventional sentence, in which separable expressions play
distinguishable roles. 1

To use a separate expression to perform the first of these

[ J What is said or implied in the last two sentences of this paragraph no
longer seerns to me true, unless considerably qualified.]



42 ESSAYS IN CONCEPTUAL ANALYSIS

tasks Is to use an expression in the uniquely referring way.
I want now to say something in general about the conven-
tions of use for expressions used in this way, and to contrast
them with conventions of ascriptive use, I then proceed to
the brief illustration of these general remarks and to some
further applications of them.

What in general is required for making a unique refer-
ence is 5 obviously, some device, or devices, for showing both
that a unique reference is intended and what unique reference
it is ; some device requiring and enabling the hearer or reader
to identify what is being talked about. In securing this result,
the context of utterance is of an importance which it is almost
impossible to exaggerate ; and by 'context' I mean, at least,
the time, the place, the situation, the identity of the speaker,
the subjects which form the immediate focus of interest, and
the personal histories of both the speaker and those he is
addressing. Besides context, there is, of course, convention ;
linguistic convention. But, except in the case of genuine
proper names, of which I shall have more to say later, the
fulfilment of more or less precisely stateable contextual con-
ditions is conventionally (or, in a wide sense of the word,
logically] required for the correct referring use of expressions
in a sense in which this is not true of correct ascriptive uses.
The requirement for the correct application of an expression
in its ascriptive use to a certain thing is simply that the thing
should be of a certain kind, have certain characteristics. The
requirement for the correct application of an expression in
its referring use to a certain thing is something over and
above any requirement derived from such ascriptive meaning
as the expression may have ; it is, namely, the requirement
that the thing should be in a certain relation to the speaker
and to the context of utterance. Let me call this the con-
textual requirement. Thus, for example, in the limiting
case of the word ' T the contextual requirement is that the
thing should be identical with the speaker ; but in the case of
most expressions which have a referring use this requirement
cannot be so precisely specified. A further, and perfectly



ON REFERRING 43

general, difference between conventions for referring and
conventions for describing is one we have already encoun-
tered, viz. that the fulfilment of the conditions for a correct
ascriptive use of an expression is a part of what is stated by
such a use ; but the fulfilment of the conditions for a correct
referring use of an expression is never part of what is stated,
though it is (in the relevant sense of 'implied') implied by
such a use.

Conventions for referring have been neglected or mis-
interpreted by logicians. The reasons for this neglect are
not hard to see, though they are bard to state briefly. Two
of them are, roughly : (i) the preoccupation of most logicians
with definitions ; (2) the preoccupation of some logicians
with formal systems, (i) A definition, in the most familiar
sense, is a specification of the conditions of the correct
ascriptive or classificatory use of an expression. Definitions
take no account of contextual requirements. So that in so
far as the search for the meaning or the search for the
analysis of an expression is conceived as the search for a
definition, the neglect or misinterpretation of conventions
other than ascriptive is inevitable. Perhaps it would be better
to say (for I do not wish to legislate about 'meaning' or
' analysis ') that logicians have failed to notice that problems
of use are wider than problems of analysis and meaning.
(2) The influence of the preoccupation with mathematics and
formal logic is most clearly seen (to take no more recent
examples) in the cases of Leibniz and Russell. The con-
structor of calculuses, not concerned or required to make
factual statements, approaches applied logic with a pre-
judice. It is natural that he should assume that the types of
convention with whose adequacy in one field he is familiar
should be really adequate, if only one could see how, in a quite
different field that of statements of fact. Thus we have
Leibniz striving desperately to make the uniqueness of unique
references a matter of logic in the narrow sense, and Russell
striving desperately to do the same thing, in a different way,
both for the implication of uniqueness and for that of existence.



ON REFERRING 45

names like * The Round Table' substantival phrases
which have grown capital letters.

(3) Finally, they may be divided into the following two
classes : (i) those of which the correct referring use is
regulated by some general referring-cum-ascriptive
conventions ; (ii) those of which the correct referring
use is regulated by no general conventions, either of
the contextual or the ascriptive kind, but by conven-
tions which are ad hoc for each particular use (though
not for each particular utterance). To the first class
belong both pronouns (which have the least descrip-
tive meaning) and substantival phrases (which have
the most). To the second class belong, roughly
speaking, the most familiar kind of proper names.
Ignorance of a man's name is not ignorance of the
language. This is why we do not speak of the mean-
ing of proper names. (But it won't do to say they are
meaningless.) Again an intermediate position is
occupied by such phrases as 'The Old Pretender'.
Only an old pretender may be so referred to ; but to
know which old pretender is not to know a general,
but an ad hoc, convention.

In the case of phrases of the form 'the so-and-so' used
referringly, the use of * the J together with the position of the
phrase in the sentence (i.e. at the beginning, or following a
transitive verb or preposition) acts as a signal that a unique
reference is being made ; and the following noun, or noun
and adjective, together with the context of utterance, shows
what unique reference is being made. In general the
functional difference between common nouns and adjectives
is that the former are naturally and commonly used refer-
ringly, while the latter are not commonly, or so naturally,
used in this way, except as qualifying nouns ; though they
can be, and are, so used alone. And of course this functional
difference is not independent of the descriptive force peculiar
to each word. In general we should expect the descriptive
force of nouns to be such that they are more efficient tools



4 6 ESSAYS IN CONCEPTUAL ANALYSIS

for the job of showing what unique reference is intended
when such a reference is signalized ; and we should also
expect the descriptive force of the words we naturally and
commonly use to make unique references to mirror our
interest in the salient, relatively permanent and behavioural
characteristics of things. These two expectations are not
independent of one another ; and, if we look at the differences
between the commoner sort of common nouns and the com-
moner sort of adjectives, we find them both fulfilled. These
are differences of the kind that Locke quaintly reports, when
he speaks of our ideas of substances being collections of
simple ideas; when he says that 'powers make up a great
part of our ideas of substances' ; and when he goes on to
contrast the identity of real and nominal essence in the case
of simple ideas with their lack of identity and the shiftingness
of the nominal essence in the case of substances. 'Sub-
stance' itself is the troublesome tribute Locke pays to his
dim awareness of the difference in predominant linguistic
function that lingered even when the noun had been ex-
panded into a more or less indefinite string of adjectives.
Russell repeats Locke's mistake with a difference when,
admitting the inference from syntax to reality to the extent
of feeling that he can get rid of this metaphysical unknown
only if he can purify language of the referring function
altogether, he draws up his programme for ' abolishing par-
ticulars '; a programme, in fact, for abolishing the dis-
tinction of logical use which I am here at pains to emphasize.
The contextual requirement for the referring use of pro-
nouns may be stated with the greatest precision in some
cases (e.g. T and 'you') and only with the greatest vague-
ness in others ('it* and 'this')- I propose to say nothing
further about pronouns, except to point to an additional
symptom of the failure to recognize the uniquely referring
use for what it is ; the fact, namely, that certain logicians
have actually sought to elucidate the nature of a variable
by offering such sentences as 'he is sick', 'it is green', as
examples of something in ordinary speech like a sentential



ON REFERRING 47

function. Now of course it is true that the word c he* may
be used on different occasions to refer to different people or
different animals: so may the word 'John' and the phrase
'the cat'. What deters such logicians from treating these
two expressions as quasi-variables is, in the first case, the
lingering superstition that a name is logically tied to a single
individual, and, in the second case, the descriptive meaning
of the word 'cat'. But 'he', which has a wide range of
applications and minimal descriptive force, only acquires a
use as a referring word. It is this fact, together with the
failure to accord to expressions, used referringly, the place
in logic which belongs to them (the place held open for the
mythical logically proper name), that accounts for the mis-
leading attempt to elucidate the nature of the variable by
reference to such words as 'he*, 'she', 'it 1 .

Of ordinary proper names it is sometimes said that they
are essentially words each of which is used to refer to just
one individual. This is obviously false. Many ordinary
personal names names par excellence are correctly used
to refer to numbers of people. An ordinary personal name
is, roughly, a word, used referringly, of which the use is net
dictated by any descriptive meaning the word may have,
and is not prescribed by any such general rule for use as a
referring expression (or a part of a referring expression) as
we find in the case of such words as 'I 3 , 'this' and 'the', but
is governed by ad hoc conventions for each particular set of
applications of the word to a given person. The important
point is that the correctness of such applications does not
follow from any general rule or convention for the use of the
word as such. (The limit of absurdity and obvious circularity
is reached in the attempt to treat names as disguised 'descrip-
tion in Russell's sense ; for what is in the special sense implied,
but not entailed, by my now referring to someone by name
is simply the existence of someone, now being referred to,
who is conventionally referred to by that name ) Even this
feature of names, however, is only a symptom of the purpose
for which they are employed. At present our choice of names



48 ESSAYS IN CONCEPTUAL ANALYSIS

Is partly arbitrary, partly dependent on legal and social
observances. It would be perfectly possible to have a
thorough-going system of names, based e.g. on dates of
birth, or on a minute classification of physiological and
anatomical differences. But the success of any such system
would depend entirely on the convenience of the resulting
name-allotments for the purpose of making unique refer-
ences ; and this would depend on the multiplicity of the
classifications used and the degree to which they cut hap-
hazard across normal social groupings. Given a sufficient
degree of both, the selectivity supplied by context would do
the rest ; just as is the case with our present naming habits.
Had we such a system, we could use name-words descriptively
(as we do at present, to a limited extent and in a different way,
with some famous names) as well as referringly. But it is
by criteria derived from consideration of the requirements
of the referring task that we should assess the adequacy of
any system of naming. From the naming point of view, no
kind of classification would be better or worse than any other
simply because of the kind of classification natal or
anatomical that it was.

I have already mentioned the class of quasi-names, of
substantival phrases which grow capital letters, and of which
such phrases as 'the Glorious Revolution', 'the Great War',
1 the Annunciation ', ' the Round Table * are examples. While
the descriptive meaning of the words which follow the
definite article is still relevant to their referring role, the
capital letters are a sign of that extra-logical selectivity in
their referring use, which is characteristic of pure names.
Such phrases are found in print or in writing when one
member of some class of events or things is of quite out-
standing interest in a certain society. These phrases are
embryonic names. A phrase may, for obvious reasons, pass
into, and out of, this class (e.g. 'the Great War*).



ON REFERRING 49



V

I want to conclude by considering, all too briefly, three
further problems about referring uses.

(a) Indefinite references. Not all referring uses of
singular expressions forestall the question 'What (who,
which one) are you talking about ? ' There are some which
either invite this question, or disclaim the intention or ability
to answer it. Examples are such sentence-beginnings as 'A
man told me that . . .', ' Someone told me that * . .' The
orthodox (Russellian) doctrine is that such sentences are
existential, but not uniquely existential. This seems wrong
in several ways. It is ludicrous to suggest that part of what
is asserted is that the class of men or persons is not empty.
Certainly this is implied in the by now familiar sense of
implication ; but the implication is also as much an implica-
tion of the uniqueness of the particular object of reference
as when I begin a sentence with such a phrase as 'the table*.
The difference between the use of the definite and indefinite
articles is, very roughly* as follows. We use 'the* either
when a previous reference has been made, and when 'the*
signalizes that the same reference is being made ; or when,
in the absence of a previous indefinite reference, the context
(including the hearer's assumed knowledge) is expected to
enable the hearer to tell what reference is being made. We
use 'a* either when these conditions are not fulfilled, or when,
although a definite reference could be made, we wish to keep
dark the identity of the individual to whom, or to which,
we are referring. This is the arch use of such a phrase as
' a certain person' or 'someone' ; where it could be expanded,
not into 'someone, but you wouldn't (or I don't) know who'
but into 'someone, but Fm not telling you whoV

(ff) Identification statements. By this label I intend
statements like the following :

(id;) That is the man who swam the channel twice on
one day.



S o ESSAYS IN CONCEPTUAL ANALYSIS

(iia) Napoleon was the man who ordered the execution

of the Due d'Enghien,

The puzzle about these statements is that their grammatical
predicates do not seem to be used in a straightforwardly
ascriptive way as are the grammatical predicates of the
statements :

(iff) That man swam the channel twice in one day.

(ii) Napoleon ordered the execution of the Due

d'Enghien.

But if, in order to avoid blurring the difference between (ia)
and (iff) and (iia) and (iiff), one says that the phrases which
form the grammatical complements of (ia) and (iia) are being
used referringly, one becomes puzzled about what is being
said in these sentences. We seem then to be referring to the
same person twice over and either saying nothing about him
and thus making no statement, or identifying him with himself
and thus producing a trivial identity.

The bogy of triviality can be dismissed. This only arises
for those who think of the object referred to by the use of an
expression as its meaning, and thus think of the subject
and complement of these sentences as meaning the same
because they could be used to refer to the same person.

I think the differences between sentences in the (a) group
and sentences in the (ff) group can best be understood by
considering the differences between the circumstances in
which you would say (ia) and the circumstances in which
you would say (iff). You would say (ia) instead of (iff) if
you knew or believed that your hearer knew or believed that
someone had swum the channel twice in one day. You say
(ia) when you take your hearer to be in the position of one
who can ask: c Who swam the channel twice in one day?'
(And in asking this, he is not saying that anyone did, though
his asking it implies in the relevant sense that someone
did.) Such sentences are like answers to such questions.
They are better called 'identification-statements' than
'identities'. Sentence (ia) does not assert more or less than



ON REFERRING 5*

sentence (ii). It is just that you say (to) to a man whom
you take to know certain things that you take to be unknown
to the man to whom you say (iff).

This is, in the barest essentials, the solution to Russell's
puzzle about s denoting phrases 2 joined by f is' ; one of the
puzzles which he claims for the Theory of Descriptions the
merit of solving.

(*) The logic of subjects and predicates. Much of what
I have said of the uniquely referring use of expressions can
be extended, with suitable modifications, to the non-uniquely
referring use of expressions ; i.e. to some uses of expressions
consisting of 'the 1 , 'all the', 'all', 'jsome', 'some of the 1 , etc.
followed by a noun, qualified or unqualified, in the plural ;
to some uses of 'they', 'them', 'those', 'these'; and to
conjunctions of names. Expressions of the first kind have a
special interest. Roughly speaking, orthodox modern criti-
cism, inspired by mathematical logic, of such traditional
doctrines as that of the Square of Opposition and of some
of the forms of the syllogism traditionally recognized as valid,
rests on the familiar failure to recognize the special sense
in which existential assertions may be implied by the refer-
ring use of expressions. The universal propositions of the
fourfold schedule, it is said, must either be given a negatively
existential interpretation (e.g. for A, 'there are no Xs which
are not Ys') or they must be interpreted as conjunctions of
negatively and positively existential statements of, e.g., the
form (for A) 'there are no Xs which are not Ys, and there
are Xs'. The I and O forms are normally given a positively
existential interpretation. It is then seen that, whichever
of the above alternatives is selected, some of the traditional
laws have to be abandoned. The dilemma, however, is a
bogus one. If we interpret the propositions of the schedule
as neither positively, nor negatively, nor positively and nega-
tively, existential, but as sentences such that the question of
whether they are being used to make true or false assertions
does not arise except when the existential condition is fulfilled
for the subject term, then all the traditional laws hold good



52 ESSAYS IN CONCEPTUAL ANALYSIS

together. And this interpretation is far closer to the most
common uses of expressions beginning with 'all* and 'some'
than is any Russellian alternative. For these expressions
are most commonly used in the referring way. A literal-
minded and childless man asked whether all his children are
asleep will certainly not answer ' Yes ' on the ground that he
has none; but nor will he answer 'No' on this ground.
Since he has no children, the question does not arise. To
say this is not to say that I may not use the sentence, ' All my
children are asleep', with the intention of letting someone
know that I have children, or of deceiving him into thinking
that I have. Nor is it any weakening of my thesis to concede
that singular phrases of the form 'the so-and-so' may some-
times be used with a similar purpose. Neither Aristotelian
nor Russellian rules give the exact logic of any expression
of ordinary language ; for ordinary language has no exact
logic.

GRICE and WARNOCK on the implicatures of metaphysics and logic

Speranza

As Grice and Warnock remarked, one is not accustomed in philosophy nowadays to the
assumption that one is either a Platonist or a Nominalist.

These venerable names, with their deceptive suggestion of
clear and sharp opposition, are no longer regarded as profit-
able banners under which to attack philosophical problems
and opponents, nor as party titles exhausting the possibilities
of disagreement. However, disputes couched in exactly
these terms are still keenly waged among mathematical
logicians. It is said that there is, attached to the study of
mathematical logic, a different and in some ways more
important enquiry t called Ontology : and that leading ques-
tions in this enquiry are, for example ; what abstract entities
there are in addition to the concrete objects with which we
are all familiar ; or whether, on a more radical view, there
may be no abstract entities at all. The central ontological
question is, it seems, the question whether there are abstract
entities. It is commonly supposed that there is no difference of
principle (though certainly there are very many differences of
some sort) between properties, relations, concepts, numbers,
classes ; that all of them are in some way reducible to
classes, and so that the admission of these lets in all the rest.
And accordingly as one does or does not make this admission,
one is a Platonist or a Nominalist.

1 This is an extensively revised version of a paper originally published in
the Proceedings of the Aristotelian Society for 1950-51. That paper was very
defective, being confused at many points, in some passages irrelevant, and
written also in a rather disagreeable polemical tone. The present version,
though still directed at the same targets, is milder, shorter, and, I hope, much
clearer. I am most grateful to the editor for allowing me the opportunity to
make these changes.

75



y6 ESSAYS IN CONCEPTUAL ANALYSIS

It is, as one would expect, exceedingly difficult to come
to grips with this debate, since the doctrines between which
one is to choose are so curiously worded. Do we believe
that there is a 'reality behind linguistic forms' ? Surely we
do. But Professor Quine regards this as the thin end of the
Platonic wedge, 1 which we must be prepared to extrude if
we wish to 'renounce abstract entities'. Are we willing, or
not, to make this renunciation ? But we do not know what
it is that we are invited to renounce. It would be both
arrogant and rash to assume that these queer-looking dis-
putes are quite without substance, but it does not appear at
first sight that any sensible choice could be made between
such alternatives.

Professor Quine has made numerous highly expert
attempts to sharpen the ontological issue for us. However,
in this paper I shall seek to show that the apparatus which
he brings to bear does not clearly or naturally apply to some
at least of the fields in which he has advocated its employ-
ment. I believe that some scrutiny of the logical symbolism
and logicians 1 devices which he uses, and some comparison
of these with certain features of ordinary language, will
reveal that the logician's apparent sharpening of the issue
involves in part the manufacturing of unnecessary problems,
and In part a distortion of what may be quite serious pro-
blems ; and I shall suggest that this may come about through
an insufficient sense of the perils involved in imposing the
neat simplicities of logic upon the troublesome complexities
of language. It is hardly necessary to say that 1 pick no
quarrel with mathematical logic itself, but only with some
of the peripheral uses to which its weapons are sometimes
put. The particular weapon which is, as I shall suggest,
importantly misused in the present case is the existential
quantifier.

A preliminary distinction should be made at once. The
central ontological question is, as I said above, the question
whether there are abstract entities. But this question is in
1 *On Universals', Journal of Symbolic Logic, September 1947.



METAPHYSICS IN LOGIC 77

an important sense secondary to the question whether some
given system of discourse implies that there are abstract
entities. The initial problem is said to be that of detecting
the ' ontologica! commitments J of a language or some depart-
ment of a language ; thereafter the different question can be
raised what language, and hence what commitments, one is
to adopt. It is not suggested that any strictly logical tests
will serve to answer the latter question, which seems to be
regarded as 'pragmatic' ; it is, however, claimed that the
question of ontological commitment falls within the purview
of the logician. As Quine puts it, 'perhaps we can reach no
absolute decision as to which words have designata and
which have none, but at least we can say whether or not a
given pattern of linguistic behaviour construes a word W as
having a designatum' ; I and one is held to be ontologically
committed to the existence of such entities as must be
designated by those expressions of one's language which
one takes to be designating expressions. A nominalistic
language will be such that all expressions in it which are
taken to have designative uses designate only concrete
objects ; a platonistic language will be such that it contains
expressions, construed as having designative uses, the desig-
nata of which must be abstract entities. The first pro-
blem, then, is that of deciding whether a given language or
part of a language is platonistic or nominalistic; and if
satisfaction were obtained on this point, it would be possible
to proceed to the further question, which is the proper sort
of language to use.

(l) 'WHAT IS THERE?'

Are there classes ? Do numbers exist ? Are there such
things as abstract entities ? Quine has on more than one
occasion boiled down such typical ontologists* questions to
the simple and uncompromising formula, 'What is there ?'

It needs no argument to show that this way of posing

1 'Designation and Existence *> Journal of Philosophy > 1939.
G



7 8 ESSAYS IN CONCEPTUAL ANALYSIS

the question, perhaps never meant to be taken seriously, is
unprofitable; It appears to invite a quite indefinite and
possibly endless range and variety of answers. There is a
pen in my hand ; there is a pain in my ankle ; there is a
virtue in necessity ; there is general confidence in the dollar.
These are all correct expressions, and what they state may
well be true. It cannot be supposed, however, that to add
to such truths at random is the proper way to solve the
ontological problem. We are really more interested in the
question what kinds of things there are are there abstract
as well as concrete entities ? and even this question is, as
has been pointed out, strictly secondary to the question what
kinds of things we are committed to believing that there
are. So let us try to approach this latter question more firmly.

(2) DESIGNATION

One method of approach to the problem begins with
the unexceptionable assumption that, if a given expression
designates something, then there is something which it
designates ; or more cautiously, that if an expression has a
designative use, there is something which in that use it
designates. If, for example, 'Tito* has a designative use,
then there is such a person as Tito. It is taken for granted
that there are concrete objects, such presumably as Tito,
which may be designated ; it is a question whether ex-
pressions which, if they had designative uses, would desig-
nate abstract entities, are in fact taken to have designative
uses. If they are so taken, then we must hold that there are
abstract entities ; if not, not or at any rate not unless
they turn up in some other way.

We require, then, tests by which to decide what ex-
pressions are taken to have designative uses, and hence
what we must hold that there is to be designated. Quine
has more than once described two such tests, admitting
that they are not absolutely conclusive. I shall seek to show
that the case is worse than this.



METAPHYSICS IN LOGIC 79

The most important of these tests consists In an operation
called 'existential generalization'. Suppose I say

(i) Leeds is a City

Then, since there is in fact a city of which 'Leeds' is the
name, I am presumably entitled to state that there is some-
thing of which my statement is true. That is, I can safely
assert

(la) Something is a city. Or

(i) There is something which is a city. Or even

(ic) There is an x such that x is a city.
If on the other hand I had said

(2) Valhalla is mythological,

I would certainly wish to convey that there is actually no
such place ; and it is assumed that I must object to the
inference that there is something which is mythological. I
must not allow

(20) There is an x such that x is mythological,
since my point was that there is not a place called ' Valhalla'.
It is then argued that this difference between the logical
behaviour of 'Leeds' and 'Valhalla 1 can be attributed to the
fact that ' Leeds' is, and 'Valhalla* is not, taken to designate,
name, or refer to an actual place. It might seem, then, that
we have in this device a method of deciding in general
whether or not a given expression is regarded as having a
designative use, or (to put the point more insidiously) as
designating something.

Suppose then that we try to apply this test of existential
generalization to disputed cases say to 'appendicitis', or
'17', expressions which, it is held, must designate abstract
entities if they designate anything at all. We might say,
for instance,

(3) Appendicitis is painful. Or

(4) 1 7 is a prime number.

Now is there something of which each of these statements is
true ? Can we infer that something is painful, that some-
thing is a prime number ?



go ESSAYS IN CONCEPTUAL ANALYSIS

But here we encounter a curious difficulty. How can we
possibly decide whether or not to tolerate these inferences ?
The trouble is that

(30) There is something which is painful, and

Something is a prime number,



are wholly odd and mystifying sentences, for which it is
difficult to imagine plausible contexts of utterance. And
for this reason it seems impossible to pronounce generally on
the question of their admissibility.

But let us see what can be done. Suppose someone says,
'He is suifering from appendicitis'. I might, if I were un-
certain about this diagnosis, reply, ' Perhaps he is ; he is
certainly suffering from something 7 ; and it might turn out
in the end to be appendicitis. Or suppose I have worked
out a sum, and found the solution of it to be 17; but I forget
this, and later when I try to re-work the sum I get stuck.
In such a case I might well say, with an air of dogged be-
wilderment, 'Well, something was the right answer* per-
haps in order to insist that the sum does work out somehow,
is not insoluble. There are thus some cases at least in which
one might use 'something' where, if one had more or better
evidence, or exact knowledge, one would instead have said
'appendicitis' or ' 17' ; and so perhaps one would have no
reason to object, though one might be extremely puzzled,
If one were invited to reverse the usual procedure and to
replace 'appendicitis' or '17' by 'something'. One decides,
let us say, to accept existential generalization in these cases.

What does this prove ? It is supposed to prove that one
is thereby recognizing appendicitis and 17 as ' somethings',
entities, and, furthermore, as abstract, Platonic entities. But
surely it does not prove anything like this ; for at this point
one begins to encounter the invaluable non-simplicity of
ordinary speech. The difficulty is that * something' does not
behave in the way required in the logician's argument. For
if I inform a bored or inattentive listener that Valhalla is
mythological, it would be perfectly in order for him, if



METAPHYSICS IN LOGIC 81

questioned about our conversation, to say, 'He was telling
me that something or other was mythological 1 ; and this
use of 'something* would not be taken as evidence that he
thought there really was such a place, nor would his report
be condemned as self-contradictory. And if I say, secret-
ively, that I am imagining something, I do not thereby evince
belief in the actual existence of what I imagine. If one were
to use the queer-sounding sentences, * There is something
which is mythological 1 , or "There is something which I am
imagining', one would certainly perplex one's hearers ; but
the use even of these odd sentences cannot be said to be flatly
ruled out merely because the mythological does not, and the
imagined may not, actually exist. Still less (indeed in no
way) conclusive is the mere use of ' something *, without
1 there is', in sentences of a quite different construction.

It is in fact pretty obvious that one's readiness or reluct-
ance to use 'something' in the cases mentioned has really
no sort of connexion with the question whether or not one
supposes that diseases, numbers, etc., are abstract entities,
possible designata of abstract expressions. The word
'something' has an entirely different function. One is
ordinarily disposed to use the word 'something' in cases
where one does not know what in particular, or where for
some reason one does not wish to specify ; and there is no
sharp restriction upon the sorts of expressions which in such
cases one cannot or does not wish to use, so that one has
recourse to the use of 'something'. Hesitation in admitting
such sentences as (30) and (40) is indeed justified not,
however, because their admission would entail acceptance of
any philosophical doctrine, but because it would be very hard
to see when or why one might wish to say such things, or what
one could possibly be getting at if one did.

The failure of existential generalization to do for us what
is required can be explained, in part, briefly as follows. In
manipulating the symbolism of logic, if I have the expression
' Fa', I am undoubtedly entitled to write down the expression
'(3x)Fx'; and in so far as the rules for the use of this



82 ESSAYS IN CONCEPTUAL ANALYSIS

expression are fixed, there is no uncertainty as to what is
meant. But if I come across the expressions ' 17 is a prime
number' or ' Valhalla is mythological', I cannot be sure
that I am right if I say, * Something- is a prime number',
nor can anyone else be sure that I am wrong if I say, ' Some-
thing is mythological'. For the former sentence is doubt-
fully admissible as English, the latter might be intelligible
and true in suitable contexts. In any case it would be quite
impossible to say, simply on the basis of someone's readiness
to employ these sentences, that he was a Platonist or self-
inconsistent ; for in the ordinary language in which they
purport to be expressed, they simply would not have the
implications thus imputed to them.

But perhaps a yet more important consideration is this.
The test of existential generalization is most simply em-
ployed as a device for revealing how names of actual persons,
cities, etc,, may be made to function differently in some con-
texts from story-tellers' names for mythical or fictitious
persons and cities. But it is further supposed that the very
same device can be applied at once to the job of detecting
the existence or non-existenee of abstract entities. This
assumption appears to embody the supposition that the
question whether there are or are not abstract entities is
just like the question whether there is or is not a city called
4 Leeds ' ; that, if there are no abstract entities, then appendi-
citis, etc., belong in the same list as Pegasus, Apollo, Mr.
Pickwick; that, if '17' does not designate anything, it fails
to do so in the same way as 'Cerberus' fails.

Now here again it is surely in point to draw a contrast
between logic and language. If we have a form of discourse
already reduced to the pattern of quantificational logic, then
doubtless we can draw a simple distinction between ex-
pressions allowed to be ' substituends ' for bound variables,
and expressions debarred from such employment. But there
is no warrant for the belief that expressions in ordinary
language can be dichotomized in a similarly simple manner.
It seems almost too obvious that no one device could force



METAPHYSICS IN LOGIC 83

'Pegasus', '23', ' intelligence ', 'redness 1 , and 'republican-
ism' into a single bag. No doubt none of these designated
a concrete object, but they fail to do so If indeed they can
be said even to fail in ways that are utterly diverse.
' Pegasus J designates no concrete object, and it is true that
there is no such thing as Pegasus ; ' republicanism ' designates
no concrete object, but that there is no such thing as re-
publicanism is, of course, straightforwardly false ; ' 23 ' does
not designate a concrete object, but that there is no such
thing as 23 is so queer a remark that, unless further explained,
it must defy the assignment of any truth-value.

At this point a protest might also be entered against the
alleged dichotomy between concrete objects and abstract
entities. It is manifest that, unless this distinction is clear,
we do not clearly know what Platonism or Nominalism is,
and also that, unless it is exhaustive, we do not know that
these are necessary alternatives. But consider such a list as
the following: (i) 'gravitational field'; (2) 'the North
Pole' ; (3) 'the Heaviside layer' ; (4) 'the Common Law' ;
(5) 'shadows'; (6) 'rainbows'; (7) 'the Third Republic'.
Any of these expressions may occur in true or false state-
ments not in fiction or myth. There is such a thing as the
Common Law ; there are such things as rainbows ; there
was such a thing as the Third Republic, etc. None of these
things could be called a Universal; none has 'instances';
some require the definite article ; yet none would naturally
be called ' concrete ' ; and it is at least uncertain which, if
any, should be labelled 'particular'. What is referred to by
(2) or by (3) has a definite position ; shadows and rainbows
have dimensions ; and the Third Republic had a definite
duration. But shadows and rainbows, though visible, cannot
be touched, heard, or smelt; the Common Law cannot be
seen, and also has no position, shape, or size ; the Heaviside
layer can move, but cannot be seen or heard or felt to be
moving. And so on. Again, it may very well be that in the
symbolism of logic some clear distinction can be made
corresponding to that alleged between the abstract and the



84 ESSAYS IN CONCEPTUAL ANALYSIS

concrete ; but that this is so, if it is so, settles nothing when
we return to ordinary words. The distinctions here are per-
haps not useless, but they are certainly neither precise nor
exhaustive. It surely follows that outside logic no
definite sense can be attached to the supposed assertions and
denials of the Nominalist, just as no definite results could be
obtained by the device of existential generalization.

The second test by which it has been hoped to identify
expressions having designative uses is the converse of existen-
tial generalization, and is called ' application'. We have
so far attempted to decide whether or not (say) 23 is an
entity, or is thought to be so, by asking whether what is true
of 23 is thereby true, or thought to be so, of an x, a some-
thing. It is now proposed that we take some formula which
we know to be true of all x's, and ask whether it is thereby
true of (say) appendicitis. Suppose we agree that, for all
values of x, x=x ; can we proceed to infer that appendicitis
= appendicitis ?

This is not much help. In addition to the over-simplifica-
tions already noted, there is here the further defect that the
conclusion to be drawn (or rejected) by application seems
merely fantastic. Why should such an expression as ' ap-
pendicitis = appendicitis' ever be written down, uttered,
asserted, or denied ? It says nothing whatever about any-
thing ; it is not a mathematical equation ; it does not look
like any sort of logicians 1 theorem. And if we were for any
reason persuaded to allow this sort of expression, it would
be hard indeed for the speaker of plain English to see why
any version of it should be, or indeed how it could be, denied.
Pegasus = Pegasus' looks odd, but not deniable ; there seems
to be nothing wrong with 'pink=pink', nor yet with 'if=if '.
If this is a test for designative use, then every expression
designates.

Let me say again, at some risk of being tedious, what I
think it is that goes wrong with, the logico-ontologist's argu-
ment. It is supposed that because, in the symbolism of logic,
certain distinctions can be clearly drawn and certain infer-



METAPHYSICS IN LOGIC 85

ences made, the same should be true of discourse in general ;
that since we can be clear what sorts of logicians 8 expressions
may be used, and how they may be used, in contexts of
existential quantification, it should be possible similarly to
discover the 'existential commitments* of ordinary talk.
This is, however, not so. For the expressions supposed
to correspond to the existential quantifier ('There is . . .',
c . . . exists 5 , "Something . . .', 'There is something
which . . .') are too diverse and intricate in their uses to
yield the necessary results; and the supposed distinction
between abstract and concrete entities is too wavering and
non-inclusive. The Nominalist, launched with this inappro-
priate equipment upon the field of ordinary discourse, is
obliged ('There is no such thing as appendicitis') to conduct
his campaign in a manner so exceedingly awkward that
doctors and other philosophical non-combatants must in-
evitably be assailed along with the Platonic army.

(3) FRAGMENTATION OF SENTENCES

I would next like to enter, consistently with my general
thesis, a mild protest against another logicians' practice,
doubtless innocent enough in many contexts, but liable to
cause much perplexity in the present case. Consider the
straightforward statement

(5) There is a prime number between 13 and 19.

This might indeed be called though for reasons given
below only with due caution an existential statement, so
that it should be a fair case for the use of the existential
quantifier. But even here the conventions of logicians are
fraught with some peril. For in their hands such a sentence
is apt to become, by accepted translation of the symbolism,

(50) There is something which is a prime number
and is between 13 and 19.

It is important to notice that into this transformed version



86 ESSAYS IN CONCEPTUAL ANALYSIS

an 'and ' has mysteriously entered, so that the whole sentence
now appears to contain as a proper part the sentence, * There
is something which is a prime number', or 'There is a prime
number'. But how is this surprising appearance generated ?
It is of course an accepted rule of logic that from (gx)(Fx.Gx)
we may without qualms derive (3x)Fx ; there is no doubt
that the latter expression is well formed, or that it is entailed
by the former. From this, however, it does not follow that
the first half of (5) or (5*) is by itself an impeccable and
intelligible sentence in English ; on the contrary, it is clearly
not so. For if someone were to say, ' There is a prime number ' ,
and then stop, one would wait expectantly for the rest of his
observation; one would not suppose that he had already
come to the end of it. One does not assert bare Being. ' Go
on J , one might say, ' what about it ? ? And if he were merely
to repeat, ' There is a prime number', this being the whole of
his contribution, one would be left in bewilderment. What
can he be getting at ? Can he suppose that anyone has ever
said that there is not a prime number ? And even if someone
had ever said this, in what would the disagreement .have
consisted ? If one takes it for granted that the baffling frag-
ment, * There is a prime number' is really a proper part of a
conjunctive sentence and so could stand alone, it may seem
necessary since it has no ordinary use to invent some
curious sense for it, to interpret it as meaning something odd
that there is an object called a prime number, an Entity,
one of the things that are. Here indeed we seem to be tread-
ing Platonic ground ; but it is easy to see that by this path
at least we would never have got there, if the original plain
statement had not been broken in two. In general : Sen-
tences of English cannot usually be taken to pieces in the
way in which their corresponding formulae can be.

(4) EXISTENTIAL QUANTIFICATION

Let us now make a final and more head-on attack upon
the logico-ontologist's apparatus. It is sometimes said, with



METAPHYSICS IN LOGIC 87

a view to clarifying the issue, simply that we must admit
into our accepted ' universe of entities' ail those things which
we allow to be values of the bound variables of quantification.
We may discourse of classes, as Boole does, or of proposi-
tions, as in the prepositional calculus, without thereby com-
mitting ourselves to Platonism ; for we can discourse in these
ways without taking the fateful step of * quantifying over J a
class, or a proposition. If we take this crucial step, however,
we fall into the ontological grip of the existential quantifier.

Let us begin at a point where all seems reasonably clear.
We may, for example, wish to indicate that some algebraic
formula holds for all values, or for at least one value, of the
variable occurring in it ; and here ' values J has the familiar
sense of ' numerical values '. Dealing with integers we might
say, 'For all values of x, 2x is even', or 'For at least one
value of x, x=7~3'. And the first of these expressions
states that whatever integer we choose, if we multiply it by
2 we have an even number ; the second that there is at least
one integer equal to 7 minus 3.

But a statement of this latter kind has been thought to
raise a peculiar difficulty. It might be described as an
existential statement, and if so, do we not by making it
commit ourselves to the important view that a certain
number (in this case 4) exists ; and so, since a number is
presumably an abstract entity, to the acceptance of Plato-
nism ? Are we not thus trapped by the existential quantifier ?

To this there appear at once to be the following objections.
First, the existential statement in question, whether true or
false, can be shown to be true or false by purely mathematical
operations. In fact every schoolboy knows quite well that
it is true, even if he has never so much as heard of Plato,
and there could be no serious argument about it. Further,
even if one were to call for a full demonstration, one would
be offered no contentious philosophical arguments, but
only a bit of mathematics. The whole matter is entirely
remote from the arena both of Platonic and of anti-Platonic
philosophizing.



88 ESSAYS IN CONCEPTUAL ANALYSIS

And in any case, second, the statement does not state
that the number 4 exists, but only that there Is an integer
equal to 7 minus 3 a very different matter. To say that
there is a number of a certain sort is not at all the same as
to mention a number and then assert that it exists. The
question whether there is an integer equal to 7 minus 3 is
closed, once we have said and if necessary shown that 4 is
such a number. Whether the number 4 itself exists is, if
there could be any such question at all, a quite different
question a different sort of question; and certainly we
do not answer it in the affirmative merely by answering
affirmatively the other question.

However, it seems, I suppose, tempting to argue that, if
there is an integer equal to 7 minus 3, and 4 is such a number,
then at least there must be such a number as 4. This I take
to be odd, rather trivial, but presumably true. But even
this does not either require or license us to say that the
number 4 exists \ for the phrases 'There is . . .', 'There is
such a thing as . . .', and '. . . exists' are not, as they are
so often assumed to be, synonymous and interchangeable.
Consider

(6) There are tigers in Africa.

(?) Tigers still exist.

(8) There are such things as tigers.

First, clearly (6) would sound odd and incomplete if we
omitted from it the words 'in Africa'. It would then be, in
fact, a mere fragment of a sentence, just as * There is a prime
number* was a fragment of a sentence. Sentence (7) is per-
fectly natural, in as much as it contains the word 'still', and
thus would be understood as conveying the information that
tigers are not extinct. f Tigers exist ' would be by compari-
son queer ; it is not easy to see why anyone should want to
say such a thing, though perhaps it might be intelligible
enough in some suitable context. I think that (8) would
usually be understood as a denial of the idea that tigers are
fictitious or mythological beasts, employed to distinguish



METAPHYSICS IN LOGIC Bg

tigers from phoenixes and unicorns. It might have other
uses, but this would be typical enough.

Now it is no doubt the case that the seriously made
statement 'There are tigers in Africa* would not be true
unless (a) tigers still existed, were not extinct, and (3) there
were such things as tigers, non-fictitious , non-mythological.
But it is equally clear that to say that there are tigers in
Africa is not to say that tigers still exist, nor is it to say that
there really are such animals. Of course there are con-
nexions, but there are also marked differences, between these
three statements ; the situations, questions, counter-asser-
tions, etc., which would naturally call for their utterance are
quite distinct.

Consider next

(9) There are shadows on the moon.

There is no doubt that this is both intelligible and true. But
in this case there seems to be no plausible counterpart to
sentence (7). What could be meant by saying that shadows
(or, the shadows) exist ? There is no question of shadows
being or not being extinct of their still, or perhaps no
longer, existing. Certainly we should not say that the
shadows on the moon do not exist ; this would be too much
like saying that they are not really there (but are really seas,
or due to defects of eyesight, etc.). But if we say that the
shadows on the moon exist, or in general that shadows
exist, might we not appear to be suggesting that shadows
lie about on things as sheets do that perhaps they could
be taken up and erected for shelter ? After all, a shadow
is in many ways more like the absence of something than the
presence of anything ; in a way, there is nothing there
when we say there's a shadow. And so, if someone were to
ask whether shadows exist, we should not know what he had
in mind, we should feel reluctant to answer either yes or no.
We do not in fact use the word * exist' in talk about shadows.
What then of * There are such things as shadows* ? The
most plausible use that I can think of for this is an ironic



90 ESSAYS IN CONCEPTUAL ANALYSIS

one, calling attention to the obvious addressed, for in-
stance, to a painter who always leaves the shadows out of
his pictures. Similarly, one might say 'There are such
things as tigers* as an ironically phrased reason for not
spending the night in the open. These remarks are clearly
quite unlike the enigmatic 'Shadows exist ', or ' Tigers exist'.

In this connexion numbers, classes, properties, etc. re-
semble shadows rather than tigers. We can say * There is
a number which, multiplied by 3, gives 21'; but we feel
wholly baffled by the question whether this number is,
whether it exists, whether numbers exist. No one surely
supposes that numbers might be extinct, or that they figure
only in legend and fiction ; and every one knows that there
is such a subject as arithmetic, that in this sense there are
such things as numbers. Of course one does not wish to
deny that numbers exist one does not use the word
'exist' at all, in talk about numbers.

The expressions I have been considering are of course
familiar ones, and it is really pretty obvious that they have
different uses. However, employment of the existential
quantifier is liable to blind the logician's eye to just such
points as these. To this one device is given the job of sym-
bolizing such phrases as 'There is . . .', 'There is such a
thing as . . .', '. . . exists', and even 'Some . . .', 'At
least one . . .', and 'There is something which . , .'. Be-
cause all these phrases are ordinarily dealt with by the use
of the existential quantifier, it is easy to assume that they
are interchangeable, all really the same; it may even be
naively supposed that logic has somehow proved that they
are really the same, and that one must be wrong if neverthe-
less one insists that they are different. (Too often the boot
gets on the wrong foot in this way as if a map-maker
should complain that the mountains were inaccurate.) It is
of course possible that/0r some purposes perhaps for most
of the purposes of logicians the phrases in question are
not relevantly different, and so that a single symbolic device
may be more or less workable. If there is, say, a green



METAPHYSICS IN LOGIC 91

book on my table, then it is at any rate true, for what It Is
worth, that there is something of which I could say i It is a
green book J ; that there is at least one book which is green ;
that there is such a thing as a green book ; even, perhaps,
by stretching matters a little, that some books are green and
that a green book exists. But for many other purposes the
differences between these locutions will remain of vital
importance. It will often be highly important to remember
that where, for example, it is intelligible and true to say
'There is such a thing as x', to say that there is x or that x
exists may be unintelligible, and even if not unintelligible
will usually be different. Those who use expressions of the
sorts supposed to correspond to the existential quantifier are
not all saying the same kind of thing, nor can there be any
single philosophical position to which they are committed
by their readiness to use such expressions. The proper
understanding of these expressions- is not assisted, but on
the contrary rendered almost impossible, by the lavish intro-
duction of quantifiers. For thus the harmless will constantly
be transformed into the peculiar ; progress will be held
up by unnecessary questions, needless scruples, and false
dilemmas ; and in the obscurity Plato's ghost will seem
to be lurking.

I have been arguing in this paper that the efforts of the
logician to clarify problems of ontology fail, since the devices
employed all turn on notions of quantificational logic, par-
ticularly on the use of bound variables and the existential
quantifier ; and that this apparatus has little or no clear
application to the ordinary words and idioms in which the
problems are initially expressed. I need finally to defend
this argument against the charge of irrelevance. Professor
Quine has recently written l that t the philosophical devotees of
ordinary language are right in doubting the final adequacy
of any criterion of the ontological presuppositions of ordinary
language', since 'the idiomatic use of " there is" in ordinary

1 From a Logical Point of View, p. 1 06.



92 ESSAYS IN CONCEPTUAL ANALYSIS

language knows no bounds comparable to those that might
reasonably be adhered to in scientific discourse painstakingly
formulated in quantificational terms '. However, he observes
that this is a minor affair, since the enquiry into ontological
commitments is properly concerned, not with ordinary lan-
guage at all, but with 'one or another real or imagined
logical schematization of one or another part or all of
science J . If so, it would of course follow that my argument,
though within its own limits possibly sound, has been sub-
stantially beside the point. Now one way in which I might
seek to ward off this charge would be by maintaining, first,
that most of the last ten years' literature of ontology embodies
no such awareness as Quine now expresses of the limitations
of symbolism in application to ordinary language; and
second, that there is also little evidence in that literature of
any particular concern with science, except for one or two
brief statements that some unspecified science is the subject
ultimately in view. However, to adopt either of these
courses would involve much rather acrimonious citation of
texts, with much risk of distortion, misconstruction, and mis-
understanding ; it would be a mere exercise in post mortem
polemics. In any case I think that it is more to the point
to make a counter-charge of irrelevance. If it is true that
ontology in its modern dress can get firmly to grips only
with scientific discourse really or imaginedly schematized,
then certainly it can have little relevance, if any, to philo-
sophical problems about universals, concepts, etc. abstract
entities in general, the supposed subject-matter of the
enquiry. For these problems arose from, can be posed,
clarified, discussed, and (in their way) settled in, quite
ordinary, unregenerate language ; they can neither be con-
fined to, nor settled in, language of any specially regimented
pattern. If one cannot deal with philosophical problems of
ontology upon the field of discourse in general, one cannot
deal with them at all ; one can only pass them by. This
itself, if admitted, would be a useful conclusion to reach.
At least it would be clear that there are problems which we



METAPHYSICS IN LOGIC 93

cannot look to the logician to settle for us, and that the old
problems of ontology remain among them.

Perhaps after all this is really to say no more than that
they are philosophical and not logical problems. But in
uttering this highly charged platitude, I suspect that one can
only be pretending to have arrived at an absolutely uncon-
troverslal terminus.