Grice and Urmson

Speranza


J. O. Urmson (or "James Opie Urmson", as I prefer) set out to write a history of philosophical analysis in England since the beginning of the century.

But in fact, some critics will argue, he has written a history of logical atomism.

Moore’s analysis receives no treatment and logical positivism is not considered very thoroughly. Which is a pity, seeing that, as Austin said, "Some like Witters, but Moore's MY man".

The philosophers with whom Urmson (like Grice) is mainly concerned is Russell. Then Wittgenstein, and Wisdom.

Urmson writes in a beautiful (that some describe, wrongly as "awkward, pompous") style and his approach to the subject is indirect.

Urmson explains the various philosophical views and considers why they came to be held, but he never discusses whether they are true or not. But then, in vino veritas.

That is why the book is boring to a few, but VERY ENLIGHTENING to Griceians.

Nevertheless it provides a very sound background to a study of modern English philosophy and a number of interesting points are made.

Logical atomism, Urmson says, is a theory about the nature of language: All statements are either simple propositions of the subject-predicate form or else truth functional compounds of these.

An atomic proposition consists of a logically proper name, which stands for some object, and a

predicate word, which stands for a universal.

Logically proper names were thought to have several

properties which ordinary proper names lack.

Russell supposed that all ordinary proper names had

some descriptive function whilst logically proper names had none; and he supposed it to be a

consequence of this that only objects with which the speaker was acquainted could be named by

logically proper names. In use a name was thought to acquire a descriptive function and so to cease to

be logically proper.

Predicate words stood for universals, with which Russell supposed us to be

acquainted.

Relations, which are predicates of two subjects, triadic relations, which are predicates of

three, and all more complex relations, are universals which may be mentioned in an atomic

proposition.

All other propositions are built up by combining atomic propositions by means of

conjunctions, such as ‘and’, ‘if then . . .‘, ‘not’, ‘or’, to form molecular propositions which are truth-functions

of the components.

One proposition is a truth-function of a set of one or more others if its

truth is determined whenever the truths of the others are determined.

Urmson stresses that this was the

ideal of logical atomism this: in the

Principia Mathematica, which inspired the theory, general

conditionals are not truth-functions of atomic propositions. For example ‘All men are mortal’ or

‘Anything, if it is a man, then it is mortal’ is not a truth-function of such atomic propositions as

‘Socrates is a man’, ‘Socrates is moral’, ‘Homer is a man’, ‘Homer is mortal’. Russell considered himself

 

 

forced to allow other sorts of fact, for example general facts, to avoid such difficulties; though Ramsey

argued that general propositions were concealed truth functions of an infinite number of propositions.

The nature of language revealed, the logical atomists thought, the nature of reality.

The universe

consisted of just those sorts of things referred to in atomic propositions: the things, which the logically

proper names named, and the universals, for which the predicate words and relation words stood. The

atomic propositions, logically proper name and predicate word, pictured the world, which consisted of

things, properties, and relations.

The theory fulfilled three functions: it explained why the philosophical explanations of

mathematical terms in the

Principia Mathematica was a satisfactory explanation, because they were

given in a language which revealed the nature of the facts it described; it explained why classes, and

other abstract entities did not exist, because statements about them could be shown to be truth

functions of statements which did not mention them; and it explained how objects which could not he

experienced could yet be discussed, because statements about such objects could be shown to be truth

functions of statements which mentioned only objects which could be experienced.

Later in his brilliant essay, Urmson discusses why logical atomism was abandoned; but there is one

respect, which he does not mention, in which the truth functional analysis of language is unsatisfactory.

The atomists spoke as if a statement could be split up into the statements of which it was a truth

function, and these statements split up in their turn, until an end was reached and all the remaining

propositions were simple. But this is not correct, no end can be reached by this process. For example

any proposition

p is a truth function of the two propositions ‘p and q’ and ‘p and not q’; for ‘p is

equivalent to ‘

p and q.or.p and not q’. Now ‘p and q’ and ‘p and not q’ look more complicated than ‘p’,

but this is merely because of the way in which they are written. They could have been written using

single signs. The calculus of truth functions alone does not provide a way of recognising propositions

which are simple: unless it is defined syntactically there is no such thing as truth-functional simplicity.

Urmson distinguishes two types of analysis which he believes Russell to have confused. The

first type, same-level analysis, has as an example the theory of descriptions by means of which a certain

class of expressions, such as ‘the King of France’, which seem to refer to objects which do not exist,

where shown not to be referential in character. ‘The King of France is bald’ was said to be equivalent to

‘There is a man who is King of France, there is only one man who is King of France, if any man is King

of France, then that man is bald’. This type of analysis, which shows the true logical form of

propositions in which definite descriptions occur (that is, shows that certain words which appear to be

proper names are not proper names), is to be contrasted with another type of analysis which shows that

certain types of entity, for example classes or physical objects, do not exist. Urmson points out that

 

 

showing ‘The King of England is old’ means ‘There is something which is King of England, etc.’ does

not show that the King of England does not exist. Similarly to show that expressions referring to classes

can be translated into expressions which do not refer to classes does not show that classes do not exist.

Nevertheless Russell seems to have supposed it did show this. What it does do is to destroy a powerful

argument that classes must exist.

Analysis with the purpose of showing that classes and physical objects

do not exist Urmson calls reductive analysis.

-- versus 'reductionist' analysis -- the distinction is Grice's.

Wittgenstein, according to Urmson, had an even stranger view than Russell about the nature

of logically proper names and in examining it Urmson gives an interesting explanation of why

Wittgenstein said that the world is the totality of facts, not the totality of things. It is this: things, for

Wittgenstein, were things of which various qualities could be predicated; it was not of a red pillar box

that the property ‘red’ could be predicated but of some substance which might possibly be red, or green,

or blue, but was none of them. This substance Urmson compares to Aristotle’s first substance.

Wittgenstein believed that these substances were abstractions, but necessary abstractions. Only the red

pillar box could he observed, not the colour less abstraction. The red pillar box was not a thing, but a

fact; and this is why the world divides, not into things, but into facts. Urmson:

“This and that are logically proper names of things which can be thus or thus, not of things

which are thus and thus. Otherwise ‘This is red’ would be a tautology.” This theory is quite untenable:

Of course ‘This is red’ is not a tautology even it I do refer, by ‘this’, to a red thing. When I do not refer

to a red thing the proposition is false; but the proposition is sometimes true, therefore I sometimes use

‘this’ to refer to a red thing. Nevertheless the theory may be what Wittgenstein intended. As Urmson

says ‘The difficulty which many find in interpreting the non-symbolic parts of the Tractatus is very

largely due to the fact that they do not realise how thoroughly metaphysical it is.’

Urmson has some interesting things to say about what Wittgenstein meant by his assertion

that sentences are pictures of the facts they are used to state. Wittgenstein, in accordance with his

doctrine that facts are groups or patterns of objects, held that sentences were facts, not objects. This

obscure doctrine is better put in this quotation, “must not say ‘The complex sign “aRb” says “a stands in

relation R to b’; but - must say ‘That “a” stands in a certain relation to “b” says that aRb’ (i.e. says that a

stands in a certain relation to b)”. The fact that verbal signs stood in certain spatial relations pictured,

and so stated, the fact that certain other things stood in certain other relations. This explanation does

avoid the obvious difficulties of understanding how a sentence can picture a fact. One thing can picture

another, but how can a thing picture a fact? Of course if this is a correct explanation, then the example

of a musical score picturing the music, and a gramophone record groove picturing the music are

misleading: for here we have one thing, the score, picturing another thing, the music. But the

 

 

explanation only leads into further difficulties. For what can be meant by saying that one fact pictures

another or that two facts are similar in structure? No doubt it was easy for Wittgenstein to see some

sense in this, believing as he did that facts were ‘combinations of objects’, but for those who do not hold

his view it is very obscure. Presumably part of what he meant is the same as what Russell meant when

he spoke of atomic propositions having an identical structure to atomic facts: name corresponding to

thing named, predicate word corresponding to universal. Urmson considers two criticisms of the

doctrine: one by Ryle, making the point that a fact cannot be said to be either like or unlike in structure

to a sentence or diagram: one by Ayer, making the point that though some languages employ signs

which picture the thing or action they name, others do not, and that this picturing is irrelevant to the

making of an assertion. Urmson criticizes this point of Ayer’s on the grounds that Wittgenstein did not

intend picturing to be taken so literally: any similarity which would enable someone knowing the rules

to build up the object from the picture (as the music may be constructed from the score) counts as

picturing. But Ayer’s criticism applies equally no matter how liberally the term ‘picturing’ is taken. The

important points seem to be: signs may picture the things they name or the properties they ascribe, but

they need not; sentences are things and cannot picture facts; to say that the fact that a sentence is

written down, or that certain words have been arranged in a certain order, pictures any other fact

snakes no sense. Urmson makes these points but in a rather roundabout fashion.

Given the doctrine that all propositions are truth-functions of atomic propositions the antimetaphysical

doctrines of Wittgenstein, leading him to the conclusion that there are no philosophical

propositions (Philosophy is an activity not a doctrine), must follow. It was this belief, that the

statements of science and logic were the only significant statements, that characterized logical

positivism. Urmson’s treatment of logical positivism is much less satisfactory than his treatment of

logical atomism. He gives a justification of the Verification Principle which is in fact a justification, not

of the theory that we know the meaning of a statement when we know how to find out whether or not it

is true, but of the theory that we know the meaning of a statement when we know what it is for the

statement to be true. This latter theory is acceptable enough, but empty. The Verification Principle may

be unacceptable, but it is interesting. This confusion has often arisen and is fostered by talking of the

situations which verify propositions. What makes ‘Snow is white’ a true sentence in English is that snow

is white; but this is not a description of how to test the sentence. The verificationists supposed that the

description of how to test it would have to be concerned with the sense-data that someone trying to test

it would have; so they were led to phenomenalism.

Urmson emphasizes that the logical positivists, like the logical atomists, considered the main

business of philosophy to be analysis, though of language, not of facts. They believed that there were no

 

 

philosophical but only logical propositions. Urmson says that ‘Ayer maintained that we must be

allowed to talk about the agreement of propositions with reality. Yet this is quite impossible on the

general account of the nature of philosophy given by Ayer in

Language, Truth and Logic.’ Unfortunately

Urmson does not explain exactly what it is about the philosophy of

Language, Truth and Logic that

makes it impossible to speak of the relations between language and reality.

He seems to think that it is

because philosophy is to be restricted to logical statements. But such a statement as ‘A proposition is

true if it agrees with reality, false if it does not’, however uninformative or misleading it may be, would

certainly have been supposed to be a logically true statement by a logical positivist, such as Moritz

Schlick, who asserted it. Perhaps Urmson is led to this view by his mistaken insistence that the logical

positivists held that all philosophical statements must be tautologies. Now if ‘tautology’ is taken in the

strict sense of ‘statement that is logically true and whose logical truth arises from its truth-functional

structure’, then, although Wittgenstein did believe this, the logical positivists did not.


Urmson seems to make

the same mistake in his discussion of statements that one proposition logically implies another. There is

a difficulty, he says, in supposing that these statements are tautologies. Of course there is and this fact

does provide an objection to Wittgenstein’s views, but it does not provide one to the views of the logical

positivists, for it may be that statements about entailments are logically true without being tautologies,

or it may even be that they are empirical statements.

Urmson gives as one of the reasons for the rejection of logical atomism the invention of

systems of logic other than that described in the Principia Mathematica. These other systems, he says,

had as much right to be regarded as perfect languages mirroring the structure of the facts as had that of

the Principia, yet they differed in structure from it. Hence the Principia could no longer be regarded as

a perfect language the structure of whose sentences corresponded to the structure of the facts. Now I do

not knew exactly what other systems Urmson has in mind, but I do not think that any systems of logic

have been devised which cover the same inferences as the Principia does and yet have a radically

different structure. Are there any, for example, which discuss the predicate calculus and yet do not use

name variables and predicate variables? Urmson is exaggerating.

It was fundamental to both logical atomism and logical positivism that all significant

propositions could be analysed into atomic propositions or basic propositions. Urmson discusses at

some length the impossibility of this: the impossibility of translating statements about nations into

statements about people, or statements about physical objects into statements about sense-data. Faced

with this impossibility, Urmson says, philosophers had two alternatives; one, to argue that the failure is

due to the vagueness of ordinary languages and so to attempt to construct artificial languages in which

translation is possible; the other, to insist that it is ordinary language that requires investigation and to

 

 

abandon any philosophical method that calls for analysis into basic propositions. But to argue that it is

ordinary language that ought to be investigated is not to provide an alternative explanation of why the

attempts at analysis have failed, nor to provide an explanation of why these analyses, which the logical

positivists and atomists supposed must be necessary, are not necessary after all. The investigation of

language is one of the methods of philosophy, not its purpose. Its purpose is the solution of

philosophical problems. Therefore only if the investigation of ordinary language is capable of solving

the problems which philosophers believed could be solved by the analysis of statements about physical

objects into statements about sense-data can it be said to provide an alternative to this analysis. Oxford

philosophers often believe that the investigation of ordinary language is an end in itself.

Finally Urmson has a chapter in which he describes contemporary views of the nature of

philosophy and of the correct method of doing philosophy. The idea that a philosophical analysis

revealed the structure of a fact was discarded in favour of the view that an analysis showed how the

form in which a proposition was expressed could mislead people into supposing it to have

consequences which it did not have. Philosophical problems were now thought of as puzzles, produced

by language, which could be dissolved by an investigation of language, rather than as questions to

which a yes or no answer could be given. Similarly, philosophical pronoucements were thought of less

as truths, than as drawing attention to distinctions and similarities in language. The Verification

Principle, for example, distinguishes between the use of descriptive sentences, such as ‘The moon will

be full tonight’ and that of evaluative sentences such as ‘That was a good action’. In this way, some

philosophers continued the method of analysis without believing that it showed the structure of facts or

that it showed that certain types of object, such as numbers or classes, did not exist; whilst others

abandoned the method of analysis and addressed themselves to explaining the functions of words in

other ways.

It is the second method that has Urmson’s support. But it is a method which has grave dangers

for the continuation of philosophy as a serious study, because it solves problems in toe easy a fashion.

For example, Urmson quotes a paper by G. A. Paul called ‘Is there a problem about Sense-data?’ in

which Paul says: “My intention has not been to deny that there are sense-data, if by that is meant that

(1) we can understand, to some extent at least, how people wish to use the word ‘sense-datum’ who

have introduced it in philosophy, and that (2) sometimes statements of a certain form containing the

word ‘sense-datum’ are true, e.g. ‘I am seeing an elliptical sense-datum’ of ‘a round penny’. Nor do I

wish to deny that the introduction of this terminology may be useful in helping to solve some

philosophical problem about perception; but I do wish to deny that there is any sense in which this

terminology is nearer to reality than any other which may be used to express the same facts; in

 

 

particular I wish to deny that in order to give a complete and accurate account of any perceptual

situation it is necessary to use a noun in the way in which ‘sense- datum’ is used, for this leads to the

notion that there are entities of a curious sort over and above physical objects which can ‘have’ sensible

properties but cannot ‘appear to have’ sensible properties which they have nut got.” Now the important

part of the solution of the problem of sense-data is the discussion of whether such ‘curious things’,

which cannot but appear as they are, can possibly exist. If they can, then perhaps they do, and a

language in which it was possible to discuss the existence of sense-data would be nearer to reality than

one in which this was not possible; in the simple sense that there would be facts expressible in the

former which were not expressible in the latter.

Again, Urmson argues that the view that propositions can be analysed into propositions

stating how to verify them is a metaphysical principle which must be abandoned. But he appears to

accept the legitimacy of such analyses of descriptive propositions so long as it is done under the name

of showing how the proposition is used’ rather than under that of ‘finding an analysis’. In this way a

careful investigation of the consequences and justification of the Verification Principle is avoided. Any

paradoxes to which it leads are put down to the mistake of looking for exact translations. It may be that

these paradoxes arise from the Principle itself.

In short, then, Urmson’s criticisms of logical atomism and logical positivism are those of a

contemporary Oxford philosopher of ordinary language. Some of his criticisms will not appear

important to other philosophers, who would rather emphasize points which he does not. But that's the way the world goes, no?

Cheers for Urmson, and Grice!