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

Tuesday, February 24, 2015

Norman Malcolm, Herbert Paul Grice, and Keith Sedgwick Donnellan

Speranza

Malcolm accused Moore of having misused the word “know” when he said that he knew that this was one human hand and that this was another human hand.

Malcolm claims that an essential part of the concept “know” is the implication that an inquiry is under way.

But surely that's merely a conversational implicature!

[Malcolm, "Defending common sense" -- Grice ("Moore and Philosopher's Paradoxes") also quotes from Malcolm, "Moore and ordinary language"]

Ordinary Language philosophy, as developed in Oxford in the first half of the 20th century, is widely thought to have been successful.

It was one element in what is now generally referred to as the ‘another linguistic turn of the screw’, which may be characterised roughly as the view that a focus on language is key to both the content and method proper to the discipline of philosophy.

Unfortunately, the received view regarding ‘linguistic philosophy’ as a whole, and Ordinary Language philosophy in particular, has become seriously over-simplified and misinterpreted.

 For example, Devitt and Sterelny quip, apropos of ‘linguistic’ philosophy, that when the naturalistic
philosopher points his finger at reality, the linguistic philosopher discusses the finger.

This captures an understandable rebellion against the idea that in philosophy, we are not to study reality or truth or morality etc., but the meaning of the words ‘reality’, ‘truth’, ‘moral’ etc.

But this understanding is simplistic and misleading.

We may challenge the received view that Ordinary Language philosophy is a failed project.

By correcting some misinterpretations and clarifying the application of the Ordinary Language method, we hope to show that such rebellion is misplaced.

We will focus on a very specific work within the Ordinary Language philosophical oeuvre, Cornell philosopher Norman Malcolm’s paper ‘Moore and Ordinary Language’. Our inspiration is of course, Herbert Paul Grice, "Moore and Philosopher's Paradoxes": a running commentary on Malcolm.

What we propose to do is reconstruct what I call the ‘Ordinary Language Argument’ from Malcolm's original essay, and defend it against both some initial and contemporary objections.

Our view is that a good deal of the foundational ideas of Ordinary Language philosophy may be based on just this argument.

Though we would hesitate to ascribe agreement with it on all details to all Ordinary Language philosophers past or present, such as Herbert Paul Grice, it does seem to capture the reasoning behind the primary concept, i.e. that the method most useful to philosophy is the observation and study of the ordinary uses of language.

What Grice, after Austin, called KEYWORD: LINGUISTIC Botany.

It is anyhow certainly a good case study: in this paper, much of what has come to be remembered as Ordinary Language philosophy is present, thus it makes a useful place to focus.

When Malcolm concluded in his essay that ‘ordinary language is correct language’ a battery of objections ensued.

Many of those objections were directed at what, we propose, was a serious misinterpretation of what Malcolm was arguing.

On the interpretation which we shall be rejecting, Malcolm was taken to have argued that any
philosophical proposition formulated such that it was not what an ‘ordinary person’
would assent to, or assert, was thereby false – and so any philosophical account of some
phenomenon which did not square with how that phenomenon would be described by
the ‘ordinary man’ (sic) was unacceptable (Benson Mates, p 66 -- Grice will go on to rely A LOT on Mates's "Logic" (Oxford) when he was designing his System Q -- later named System G by Myro).

It follows, on this interpretation, that anything formulated in ordinary language (that is, anything which is, or would be, ordinarily said in given situations), is thereby true, or (for referring terms)
truly refers.

That position is clearly absurd and, predictably, the immediate criticism was that what the ‘ordinary person’ would or would not say could not be taken to be philosophically illuminating: what is said in ordinary language is certainly unreliable, often not very interesting, and more often, quite simply false (e.g. Tennessen).

The more thoughtful version of this interpretation was the view that any philosophical theory which used some term or expression in such a way that it diverged from its ordinary meaning was unacceptable (or "inappropriate", or "misleading but true", as Grice prefers) and/or false.

The second immediate criticism was, again predictably – who is the Ordinary Language philosopher, or anyone else for that matter, to arbitrate on what is the ‘ordinary meaning’ of some term or expression?

Is there any such thing as the ‘ordinary’ meaning of terms and expressions, or ‘ordinary’ language anyway?

This interpretation of Malcolm’s argument persists today in the widespread view that the key argument of Ordinary Language philosophy is that we must reject any philosophical theory that is inconsistent with ordinary language, and we must forbid the introduction of new, revised, or technical uses of ordinary expressions (Dummett, p 683 -- "I intend to base my PhD dissertation on Dummett's "Frege". Have you read it?" Wrigley to Grice. Grice's response: "I haven't and I hope I won't").

The view is that according to Ordinary Language philosophy, any revision of ordinary language results in the ‘misuse’ of language, and must be prohibited (Soames, pp xiii - xiv).

We argue that these are all misinterpretations of Ordinary Language philosophy. Cfr. Travis, "Annals of Analysis".

Indeed, there is some irony in this interpretation.

It  has Ordinary Language philosophy proposing that linguistic meaning is utterly rigid and unchanging, which is precisely the view to be rejected (in fact, this is a view rather
more embraced by the Ideal Language philosophers, or 'formalists' or 'modernists' as Grice calls them).

In what follows, we shall reconstruct Malcolm’s argument in detail.

We suggest the argument has three elements:

-- a ‘linguistic’ argument.
-- a ‘modal’ argument, and
-- what has come to be known as the ‘paradigm case’ argument.

(Which we have discussed at length with R. Doyle in his free-will research and essay).

Throughout, we examine a representative sample of objections.

We argue that in each case, the objections either fail independently, or fail because they are aimed at a misinterpretation of what is being put forward.

Where the Ordinary Language Argument is genuinely flawed, we will suggest ways it can be improved.

Malcolm begins with a list of ‘philosophical’ propositions which, he says, share the property of being ‘paradoxical’ as they all ‘go against’ ordinary language.

The full list is as follows:

1. There are no material things
2. Time is unreal
3. Space is unreal
4. No-one ever perceives a material thing
5. No material thing exists unperceived
6. All that one ever sees when he looks at a thing is part of his own brain
7. There are no other minds – my sensations are the only sensations that exist
8. We do not know for certain that there are any other minds
9. We do not know for certain that the world was not created five minutes ago
10. We do not know for certain the truth of any statement about material things
11. All empirical statements are hypotheses
12. A priori statements are rules of grammar

Malcolm then casts Moore (unauthorised by Moore himself we should note) as refuting
these proposals by means of responding in the following manner:

"These claims are certainly wrong, for here’s one hand and here’s another; so there are at least two
material things”, and “This desk which both of us now see is most certainly not part of
my brain, and, in fact, I have never seen a part of my own brain”, and “The statement
that I had breakfast an hour ago is certainly an empirical statement, and it would be
ridiculous to call it an hypothesis”, and so on.

Malcolm claims that Moore actually refutes the philosophical propositions merely by
pointing out that they do ‘go against ordinary language’.

He says: "I hold that what Moore says in reply to the philosophical statements in our list is in each case perfectly true."

"Furthermore, I wish to maintain that what he says in each case is a good refutation, a refutation that
shows the falsity of the statement in question."

Now, Malcolm has focused, in his choice of ‘philosophical’ statements, on those expressing versions of scepticism.

Many objections to Malcolm’s arguments in this text latch on to this claim to refute certain versions of scepticism.

We shall say something more later as to whether Malcolm has indeed refuted (any version of)
scepticism.

But for the time being, we want to suggest that we focus not so much on the fact that the propositions chosen mainly reflect versions of scepticism, but on the fact that they have in common, as Malcolm actually points out, that they oppose (in some way to be explained) what is, or would be, ordinarily said in certain situations.

For example, we might say (though more must be said) that it would be a non-ordinary use
of the term ‘know’ to say (at least, without some further explanation of what one
means):

I do not know if there is a desk before me.

where one means by this, ‘not because I can’t see it, because I can see it clearly enough, indeed I can now sit right on it; but because no sensory information is certain, nor is it certain that any material
objects even exist’.

Regardless of whether what the philosophical claims express is true, it is clear that what they express is inconsistent with what would ordinarily be said in the situations in question. It is the nature of this inconsistency that we shall focus on.

It is this observation on the use of 'know' that Grice criticizes. Malcolm is just not distinguishing between what is strictly implied and merely conversationally implicated. Grice said that in Cornell, too!

Now, being merely inconsistent with the ways in which we ordinarily describe certain phenomena is not enough to distinguish the philosophical claims from any other claim in any significant way. Malcolm’s view is that the claims do not merely run up against common sense, or widely held views, thus challenging their truth.

He is not concerned with whether, for instance, there really are material objects.

He takes issue, not with the metaphysical or logical implications of the propositions, but with the semantic implications that Grice will re-label "conversational implicatures".

Malcolm's argument is that what would follow from the philosophical propositions, if they were true, is not that our ordinary ways of describing certain phenomena or situations turn out to be merely false; what would follow is that our ordinary statements would turn out to be ‘self-contradictory’ – they would both assert and deny something to be the case.

According to Malcolm, the philosophical propositions in question are not akin to and ought not be treated as scientific i.e. empirical, contingent hypotheses.

For example, the proposition:

The earth is spherical.

which is an empirical and contingent hypothesis, turned out to be true, even though it ran up against
the once current view that the earth is flat. But to assert

The earth is flat.

is simply false, not self-contradictory.

On the other hand, according to Malcolm, if it turns out that, say,

One never perceives a material object.

is true, then to assert

I perceive a material object.

is not merely to state a falsehood, but to state something like

I perceive something that I can not perceive”.

This seems an outrageous claim to make.

On what grounds does Malcolm make it?

Why could it not be the case that

One never perceives a material object.

is a perfectly empirical and contingent proposition, perhaps confirmable in the future by some
advanced physics and neuroscience?

To understand the basis on which Malcolm can
make such a claim, we must take a short detour through some of the theoretical
background that is not made explicit in the paper by Malcolm on Moore and ordinary language (so-called).

What Malcolm believes is that
he has reasons to assert that the philosophical claims in question are not what he called
‘empirical’ or ‘factual’, but rather ‘linguistic’.

A key element of ‘linguistic’
philosophy, we might recall, is a distinction between what was generally referred to as
‘factual’ or ‘empirical’ propositions, on the one hand, and ‘linguistic’ or ‘logical’ (or
sometimes ‘verbal’) propositions on the other.

This is a distinction between what we
might feel more comfortable calling ‘analytic’ and ‘synthetic’ propositions; since the
reference to ‘empirical’ tends to lead one to think that these are distinguished in virtue
of having ‘empirical verification’; and the reference to ‘factual’ leads to the thought
that the ‘linguistic’ propositions can’t be about ‘the facts’, and so must not really be
‘propositions’ at all (i.e. do not have truth-values).

But what Malcolm means is not
based on a kind of verificationism, or on the contention that any non-empirically
verifiable proposition must therefore be ‘logical’ or ‘linguistic’; nor was he committed
to the idea that ‘linguistic’ propositions are non-factual.

What Malcolm means, in
calling propositions ‘empirical’ is not so much that they are confirmable empirically,
but that they are asserted on the basis of empirical evidence.

What he means by
‘factual’ propositions are those that are about ordinary, contingent facts – as opposed
to those which purport to be about necessary, metaphysical facts.

By ‘linguistic’ propositions, he means those propositions which are regarded as, or purport to be,
necessary truths.

We contend that none of these views are unacceptable, on a proper
interpretation, which we now offer.

Malcolm and the early Ordinary Language philosophers, following the later
Wittgenstein on this issue, agreed to this distinction, essentially between analytic and
synthetic propositions, and it is certainly at work in arriving at the ‘outrageous’ claim
we are looking at.

And that is where they went wrong.

Some like Witters but Moore's MY man, Austin said -- and Grice followed suit.

For one, while Austin and Moore would RARELY consider utterer's implication ("I mean"), Witters never did!

Perhaps the most perennially disdained aspect of the distinction is
the contention that necessary, or analytic, propositions are ‘linguistic’ rather than
‘factual’ – otherwise known as the ‘linguistic doctrine of necessity’.

The terminology popular in the day is partly to blame here, as it sounds as though the claim is that
necessary propositions, because they are ‘linguistic’, are not to be understood as being
about the world and the way things are in it, but about words, or even about the ways
words are used.

This of course would have the result that necessary propositions turn
out to be contingent propositions about language use, which was correctly recognised to
be absurd (as noted in Malcolm, pp. 190-191).

Moreover, in distinguishing ‘linguistic’ propositions from ‘factual’ propositions, the thought was that the former are
therefore truth-valueless, or ‘non-cognitive’ or some such.

But this is not Malcolm’s,
or the Ordinary Language philosophers’ view at all.

Necessary propositions are not
(disguised) assertions about the uses of words, they are ‘about’ the world just as all
propositions are (and so are perfectly ‘cognitive’ bearers of truth-values).

But, on the
other hand, what makes them count as necessary, what justifies us in holding them to be
so, is not any special metaphysical fact; only the ordinary empirical fact that this is how
some of the propositions of language are used. (pp 192, see also Malcolm,
1942b).

On this view, it is through linguistic practice that we establish the distinction
between necessary and contingent propositions.

The distinction, then, comes down to
the different ways we behave with necessary or contingent propositions, and their
sources of justification, i.e. the justification we may have for holding them true. For
necessary propositions, that they have such a use in linguistic practice is what justifies
their necessity – and not the facts they describe, on this view.

On the other hand, the
source of justification for a contingent claim will always be the facts it purports to
describe - whether the facts are as the claim purports them to be.

The Ordinary Language treatment of a distinction between analytic and synthetic
propositions is not to be conflated with the distinction as held the Positivists, who were,
of course, famous for developing and holding it.

As Malcolm explains,  contrary to the view held by the Positivists, whether a proposition is considered necessary or contingent is a function of how we use the propositions, it is not a property
of propositions themselves.

And, unlike the Positivists, contingent propositions do not
count as contingent because they can be empirically verified, nor do necessary
propositions count as necessary because they are ‘self-evident’ in some sense.


Malcolm presages Quine in noting that this attempt at analysis of the distinction is
‘non-informative’.

But, contrary to Quine, Malcolm thinks that the
reason this analysis is non-informative is because it is an attempt to analyse the
properties of propositions themselves.

It is not a reason to reject the distinction
altogether, on Malcolm’s view, but a reason to see what makes a proposition necessary
or contingent is how it is used in the language. Since facts about how propositions are
used in a language are perfectly observable and empirical, such an ‘analysis’ involves
no circularity.

On Malcolm’s account, some propositions, we learn as we acquire
language, are treated differently than others, and are given different roles to play.
When we learn that “2+2=4”, we learn that we do not say “2+2=5”, or that if someone
does say that, we reject it, or correct it.

We learn that

2+2=5

is a miscalculation.

We learn that we do not say

Today is both Wednesday and not Wednesday” (in any literal,
non-metaphorical, poetic or comical sense), we learn that there is nothing that one can
say in the use of such a proposition.

We learn these things by observing how people
use words and expressions, what they accept and what they reject, how they react to a
certain form of words, what they infer and which propositions they use in inference and
so on.

Malcolm says “In order to see how misleading is the notion that the way we get
to know the truth of necessary propositions is by inspecting them, we must see that we
find out necessary truths in the same way that we find out the empirical truth that if you
suddenly jab a man with a pin, he will jump.”

We observe, and learn as
we acquire language, that some propositions cannot be negated without risking being
treated as not making any sense, i.e. those we call necessary.

We learn that, e.g. there is
nothing that we know of that can be said by a proposition of the form

“p and not p”.

Of
course, this is not to imply that all speakers are explicitly aware that there even is such a
thing as a “necessary proposition”.

But, for the most part if not the whole, even those
utterly without this concept will, usually, reject, perhaps as not making sense, the
negation of a necessary proposition (asserted literally, non-metaphorically, via conversational implicature, etc.).

We are now in a better position to understand why Malcolm and his contemporaries
sometimes called necessary propositions ‘linguistic’.

Necessary propositions, which
are used most generally in inference, or in argument, or in teaching and explaining
things --

A vixen is a fox.

and in logic and mathematics, may be used descriptively,
and hence ‘factually’ (“A vixen is a fox” is a description) just like contingent
propositions. But their effect, we might say, is to establish, clarify, confirm and to a
certain extent restrict the way certain terms are used.

The fact that some propositions
are used as necessary has consequences for the meanings and uses of the terms
involved. So, we might say that from the fact that we treat e.g.

All vixens are foxes.

as necessary, it follows that we do not use the term ‘vixen’ to apply to non-foxes
(literally, non-metaphorically etc.). And so, whilst necessary propositions do not assert
anything about the use of terms or expressions, the manner in which we use them, for
example, insofar as we reject their negation, has a certain upshot, call it, for the use, and
hence the meaning, of the terms involved.

Thus, if someone claimed to have found a
vixen that was not a fox, rather then treating this as an amazing discovery we would
more likely question whether that person properly understands the meaning of either
the term ‘vixen’ or ‘fox’, or wonder if they were using them in some new way.

On this view of the distinction between necessary and contingent propositions, which we
are attributing to Malcolm and the early Ordinary Language philosophers, and which is
based on the uses of propositions in a language, the line dividing them may shift, or
may be inconclusive.

Since it depends on the way expressions are used, and use can
change, no necessity is unrevisable. Malcolm went so far as to suggest the laws of
logic may well, one day, be different to what we accept now (1940, p 198), and that we
may well reject some necessary statements, should we find a use for their negation, or
for treating them as contingent ( p201ff).

Take, for example, the proposition

It is not the case that non-conscious objects think.

If and when a time comes that we
create, or decide on, an object that we count as ‘thinking’, but not as ‘conscious’, then
the original proposition would no longer be used as necessary.
Nevertheless, there is a feeling that some necessary propositions are necessary not
merely because of the way they function in language – that not all necessary
propositions are ‘linguistic’ or ‘logical’.

For example, propositions like

Nothing can be green and red all over.

This was one of Herbert Paul Grice's favourite utterances, and he would test it with the playmates of his two children: Karen and Timothy, when in Oxford.

 or “I cannot be in two places at once” do not appear to cadge
their necessity from the semantics of the terms, i.e. the meanings of the terms ‘red’ and
‘green’ are not obviously contrary in any way; nor are they in any obvious way logical
propositions.

Many of us may feel that the necessity of these propositions is rooted in a
metaphysical reality.

And this, perhaps, gets us to the heart of the feeling of
dissatisfaction with the Ordinary Language view thus far presented; it might be
objected that it is obvious that we use some propositions as necessary truths; but that is
because they describe the necessary facts.

It is the existence of such facts that explains
why some propositions are necessary.

It is, for example, because of the (metaphysical)
nature of the colors that nothing can be both red and green all over.

Now, Malcolm would say of such an expressions, I think, that they have a sense of
necessity because we have no use for expressions such as “X is both red and green all
over”, or “I am right at this moment both in Sweden and Egypt” (literally, nonmetaphorically
etc), at this particular point in time in our language – that it is not clear
to us what we could say in their use.

Taking such propositions to be necessary is a
function, Malcolm would say, of how we (now) use those color and place terms – it is
because of how we use them, and thus of what they mean that we find a negation of the
expressions in question difficult to make sense of (e.g. the negation of “I cannot be in
two places at one”). The question whether there is some metaphysical fact that a
necessary proposition represents is, on this view, superfluous. Appeal to
metaphysically necessary facts, as an explanation of why we use some propositions as
necessary truths, does no work.

This is because we can simply observe the ordinary
facts about language use, or acquisition, and see that metaphysically necessary facts are
never brought into play – even when learning the necessity of a proposition. Learning
how to use the terms ‘red’, ‘green’ and ‘all over’ as we do, that is, learning their
meaning, is, at the same time to learn the necessary truth of “Nothing is red and green
all over”.

If one does not learn this necessary truth, then one has not learned how to use
the terms as others do (although, this is not to imply that one learns that “the
proposition (in question) is a necessary truth”, nor to imply that one even has to have
the concept of necessity, only that, in general, one would reject the negation of a
necessary proposition). We learn the necessity of necessary propositions, not be being
introduced or exposed to a reality which is distinct from the ordinary realm i.e. a realm
of ‘necessary facts’, but by being corrected for example, if we say something like “This
is both red and green all over” (and corrected in a different way, it should be said, to
how we might be corrected if we said “The earth is flat”).

It might be objected that how we learn to use necessary propositions is a distinct issue
from whether there are metaphysical necessities which those propositions describe, and
which make them necessary.

And it is a distinct issue, the Ordinary Language
philosopher would agree.

The Ordinary Language philosophy’s ‘anti-metaphysicalism’
we might call it, is not, we should notice, a claim that there are no metaphysical
necessities, only that whether or not there are is beside the point.

The existence of a
realm of metaphysical necessities is superfluous, since appeal to it cannot account for
why we use some propositions as necessary (we do so because they are so in the
language we speak), nor can appeal to it account for how we are justified in so using
them (that we speak the language so justifies us).

It cannot account for how we learn
from others, or acquire in the first place, the use of necessary propositions, because
there is very evidently no distinct process of e.g. intuiting the necessary facts and then
using the propositions that describe them as necessarily true.

Rather, observation easily
shows that, in fact, we learn necessary propositions as we learn how to use the terms
which are their components. For example, when we learn how to use the terms ‘vixen’
and ‘fox’ we at the same time learn the necessity of “All vixens are foxes”.

Certainly,
it does not seem correct to say that we learn, or somehow know, the necessary truth that
all vixens are foxes, without learning or knowing how to use the terms (in English in
this case) ‘vixen’ and ‘fox’ – or that these are distinct learning processes. But nor can
appeal to metaphysically necessary facts justify our use of some propositions as
necessarily true. If someone questioned your holding that “All apples are fruit” is a
necessary truth, and asked on what grounds you held it, it would not help much to
answer that it is a metaphysically necessary fact about apples, that they are fruit, that
makes the proposition necessary.

In fact, what justifies you is that you are a member of
a linguistic community that uses the terms ‘apple’ and ‘fruit’ in such a way that the
proposition “This apple is not a fruit” has no (literal, non-metaphorical etc) use, would
not be used to describe anything. You would be right, then, to simply answer “If you
know what the terms ‘apple’ and ‘fruit’ mean, you will see why all apples are fruits”.
So, to conclude the detour, it seems that the disdain and hostility towards the so-called
‘linguistic doctrine of necessity’ is misplaced, at least in the way it is understood by the
Ordinary Language philosophers.

The view is perfectly naturalistically respectable, and
merely agnostic vis. metaphysics rather than actively skeptical – on the interpretation I
have offered. There also seems little motivation to argue against the view that, for
necessary propositions but not contingent ones, negating one involves a change of
meaning of the component terms.

This does not prohibit such changes – it only points
out the fact necessary propositions are special in this respect, i.e. they have
consequences for our use of their component terms, and in this sense are indeed
‘linguistic’.
Now let’s return to Malcolm’s contention that if the philosophical propositions in
question turn out to be true, then our ordinary descriptions would turn out to be not
merely false, but self-contradictory.

The key premise is that philosophical propositions,
if true, must be necessary, and thus would trigger a revision of meaning for the terms
involved (whereas this is not the case if an ordinary contingent proposition turns out to
be true). Take the following proposition:

All vixens are foxes.

If we discover what seems to be a vixen that is not a fox, we have two options. Either
we reject the evidence that what we have found is really a vixen, i.e. we reject the
empirical evidence as faulty in some way, and maintain the truth of (A).

Or, we accept
it, and therefore reject the truth of (A).

Because we have, until now, treated (A) as
necessary (for whatever reasons), rejecting (A) will result in a revision of the meaning
of either ‘vixen’ or ‘fox’ or both. It seems clear that, if we accept that (A) has turned
out to be false in some way, we will simply not be able to continue using those terms as
we have in the past. Now, by contrast, consider the proposition:

All swans are white -- except some Australian ones which are black, and some Argentine ones, which are mainly white, but with a black neck.

The discovery of black swans did not manage to require a revision of our use of the
terms ‘swan’ or ‘black’ – because we did not use the proposition as necessary, even
though there was a time when no-one had encountered a black swan.

Now, according
to Malcolm’s view, the philosophical propositions in question are like (A), but not (B).


Thus, if the philosophical claims turn out to be true, we would have to revise the
meanings of the component terms, e.g. ‘perceive’, ‘know’, ‘certainty’ etc. Now, this is
not problematic in itself, since revision of meaning is not ruled out. The problem is
this: a revision of meaning in, say, the ‘vixen’ and ‘foxes’ case simply renders the
hitherto necessity “vixens are foxes” merely a contingency, after all. So, it remains
meaningful, if false, to claim that “All vixens are foxes”.

The situation is quite
different, however, if the philosophical propositions turn out to be true. If so, then a
proposition such as “One perceives independent material objects” is rendered not
merely contingently false, after all. If the philosophical proposition “One does not
perceive independent material objects” turns out to be true, and cannot be a contingent
thesis, as Malcolm will argue, then the proposition “One perceives independent
material objects” is rendered necessarily false, unlike the case involving foxes and
vixens.

Of course, the case has yet to be made that the theses in question must be
intended as necessary truths, rather than ordinary contingent hypotheses. But looking
ahead, we may note that if Malcolm is right – i.e. that the philosophical propositions, if
true, render ordinary uses of expressions self-contradictory, this presents a significant
obstruction to error-style theories about the meanings of certain terms, i.e. those that
have the properties we are here attributing to ‘philosophical’ statements.16
Malcolm’s paper continues thus: he imagines a dispute between Russell and Moore,
illustrated by the propositions noted above. Malcolm takes as an example Russell’s
assertion that “All that one ever sees when one looks at a thing is part of one’s own
brain”.

This is the sort of proposition that would follow from the philosophical thesis
(Russell’s version of it at any rate) that all we are acquainted with in perception is
sense-data, and that we cannot infer from this that there are independent material
objects external to it which are the causes of such perceptions17. Malcolm then asks
what sort of statement this is, and what sort is Moore’s opposing reply. We are asked
to notice that there is no disagreement, in Russell’s and Moore’s opposing propositions,
about the empirical facts of the matter.

That is, they (Russell and Moore) agree that the
phenomenon they are talking about is the one which is ordinarily described by saying
(for example) “I see the postman”. To gloss on Malcolm’s use of ‘empirical’ here,
what he means is that the problem is not, say, whether it really is a postman as opposed
to a milkman, or that the light is bad, or that Russell or Moore may be hallucinating or
dreaming. Malcolm cannot, or course, claim that no empirical facts whatsoever are
relevant to the situation, in particular the sort of facts, say, about the brain which may
be revealed by a future science.

We shall return to that, but it is key here to notice that
Malcolm is using ‘empirically’ in a contrastive sense here, i.e. to contrast the nature of
this disagreement with one in which a dispute about whether

I see the postman.

might
ordinarily be settled, i.e. by checking if it really was the postman, not the milkman;
checking that the subject making the claim is not visually impaired, or hallucinating.
So, the dispute is not, at least, ‘ordinary’ in this sense – getting a closer look at what
Moore claims to see will not help.

What is this sort of dispute really over then, if not about the (narrowly construed)
empirical facts of the matter, that is the facts that would normally be relevant to
denying that something is seen by someone?

Malcolm asserts the following: “It
appears...that they disagree, not about any empirical facts, but about what language will
be used to describe those facts.” (p 9) Malcolm asserts, of the original propositions,
that “Both the philosophical statement, and Moore’s reply to it are disguised LINGUSTIC
statements.” (p 13 – our emphasis)

Now we can see that Malcolm is not claiming that
Russell’s proposition is about the words used.

What he means by a ‘linguistic’
statement is that it must be a claim to metaphysical necessity and thus will have
semantic consequences, namely that our hitherto use of the term ‘perceive’ has issued
in propositions that are necessarily false.
But it is still not clear what Malcolm’s grounds are for suggesting that the dispute
between Russell and Moore is not really about the phenomenon of visual perception,
but ‘really’ about the ‘correctness’ of the language used to describe such phenomena.
After all, surely if the facts are that we only ever see sense-data, then it is more
‘correct’ to say so. The problem, Malcolm is suggesting, is that the ‘facts’ Russell is
alluding to cannot be ordinary, contingent, empirical facts. That is, he argues that
Russell’s grounds for his proposal cannot be empirical.

He compares the suggestion
that a claim such as “No material-thing statement is ever known for certain” is enquiry
into the facts about ‘certainty’ with another ordinary (and properly empirically-based)
enquiry about a claim to certainty:

In ordinary life everyone of us has known of particular cases in which a
person has said that he knew for certain that some material-thing
statement was true, but that it has turned out that he was mistaken.
Someone may have said, for example, that he knew for certain by the
smell that it was carrots that were cooking on the stove. But you had
just previously lifted the cover and seen that it was turnips, not carrots.
You are able to say, on empirical grounds, that in this particular case
when the person said that he knew for certain that a material-thing
statement was true, he was mistaken.

Or you might have known it was
wrong of him to say that he knew for certain it was carrots not because
you had lifted the cover and seen that it was turnips, but because you
knew from past experience that cooking carrots smell like cooking
turnips, and so knew that he was not entitled to conclude from the smell
alone that it was certain that it was carrots.

It is an empirical fact that
sometimes when people use statements of the form: “I know for certain
that p”, where p is a material-thing statement, what they say is false. But
when the philosopher asserts that we never know for certain any
material-thing statements, he is not asserting this empirical fact…he is
asserting that always…when any person says a thing of that sort his
statement will be false. (p 11).
But even if a philosophical hypothesis is put forward not on the basis of empirical
evidence, that is not to say it may not be empirically confirmed by some future, more
complete science. Indeed, Russell, and many theorists since who have proposed
philosophical views which conflict with how certain phenomena are ordinarily
described have realised the importance of getting their philosophical propositions to
look as much like scientific hypotheses as possible.

This is important, crucial even,
because who can argue that a scientific hypothesis will never be confirmed by future
discovery? It is common practice even, to construct some radical theory about how
things really are, which contradicts the ordinary way of describing things, on the basis
of no empirical evidence whatsoever - but, it is argued, that is not to say that it won’t or
can’t be empirically confirmed in the future. The heart of the problem, for the
philosophical propositions, is, as we shall now see, not whether or not they are
empirically confirmable now, or at some future time.

The problem is, as Malcolm argues, that they cannot be contingent.

Malcolm has shown that the philosophical propositions in question are not, at least,
grounded on empirical evidence, but more is need to complete the argument that they
are really linguistic i.e. necessary. And so, perhaps Russell still has some options here:
he might argue that his thesis is merely contingent, and possibly confirmable by future
science, or he might argue that his thesis is necessary, but a posteriori confirmable by
the empirical facts – which would show it to be a genuine ‘factual’ dispute after all.
To take the latter option first, Russell could propose that his thesis is an a posteriori but
necessary truth, for example as Kripke (1980) has suggested is the case for propositions
such as

water = H2O

Isn’t the latter an empirical discovery that we made, and yet it
is necessarily true, since anything that is not H2O could not possibly be water? Even if
we accept Kripke’s point that there are such things as a posteriori discoverable
necessary truths, Malcolm’s argument still applies. The problem, according to
Malcolm is that necessary propositions have semantic consequences, consequences for
how we can and can not meaningfully use certain expressions, which ordinary
contingent proposition do not. If we accept that it is necessarily true that “We only ever
perceive sense-data”, whether it is confirmed empirically or not, the fact remains that
we would have to revise the meaning of the term ‘perceive’.

“To perceive x” can never
mean anything other than to examine one’s own sense-data. Moreover, it will follow
that this has always been the case, since it is a necessary truth. So, each time any one
of us has asserted that we perceived something other than sense-data, because
‘perceive’ has always ‘really’ meant nothing other than to examine one’s own sense
data, what one asserts is self-contradictory – one asserts that one is doing something,
but using an expression whose ‘real’ meaning entails that that thing cannot be done, e.g.
“I perceive something that I can not perceive” (p 11, also see fn 10, p 15).

On the other hand, it would seem that Russell does not have the option of avoiding
these ‘semantic consequences’ of his view, by retreating to the claim that his theory is
merely a contingent fact about the actual world, and that in some possible world, the
sense-data theory is false, i.e. a possible world in which one perceives not merely one’s
sense-data, but independent material objects.

Why not?

Firstly, to do so makes the
theory baseless – why suggest that although it is possible that we perceived material
objects, it just happens to be that we do not (of course, the sense-data theory says
nothing as to whether material objects actually exist or not) at this world. There
appears to be, not only no empirical evidence for such a view, but no other reason either
– for example, it cannot be grounded in a conceptual analysis of the concept ‘perceive’,
since the hypothesis now is that his view is contingent.

Thus, Malcolm establishes convincingly I think that Russell’s philosophical
proposition, and hence any philosophical proposition which shares the property of both
contradicting what would ordinarily be said, and claiming that this is the case
necessarily, will have the consequence that what would ordinarily be said of a given
situation is not merely factually wrong, but necessarily false. The essence of this
consequence is that what is ordinarily said is, and always has been, self-contradictory.
But why is this outcome so absurd? Perhaps some of our descriptions just are selfcontradictory
to assert, despite all appearances? It is important be clear that we must
not make the mistake of conflating ‘correct’ language with ‘true’ language. Perfectly
correct uses of language can, and often do, express falsehoods.

What Malcolm
contends is that ‘correct’ language is language that can be used to describe things –
truly or falsely. To use language ‘correctly’ is to use it meaningfully, even though what
is said may be false. Self-contradictory assertions, on the other hand, cannot be so
used, on Malcolm’s view, because they do not describe anything, truly or falsely
By an ‘ordinary expression’, I mean an expression which has an
ordinary use, i.e. which is ordinarily used to describe a certain sort of
situation.

By this we do not mean that the expression be one that is
frequently used.

It need only be an expression which would be used to
.
This seems prima facie true. Given the nature of our language, we do not, as a matter
of fact, use self-contradictory expressions to describe situations (literally, nonmetaphorically
etc). There is something deeply implausible about the suggestion that
for our uses of some terms, unbeknownst to us, we have been uttering selfcontradictory
expressions.

This cannot be the case, according to Malcolm, because, he
says, any expression which has a use therefore has an ordinary use and therefore has a
correct use. He says:
describe situations of a certain sort…To be an ordinary expression, it
must have a commonly accepted use; it need not be the case that it is
ever used. (ibid., p 14)

So, if Malcolm is right, it cannot turn out that some ordinary expression is, despite
appearances, necessarily false, for then we would be asserting self-contradictory
propositions. To do so would be to fail to use language correctly i.e. meaningfully, and
since this cannot be the case, then, ordinary language is correct.

Indeed Malcolm claims “The proposition that no ordinary expression is self-contradictory is a TAUTOLOGY” (p 16 – our emphasis).

This is a significant metaphilosophical result, one which has, as far
as I can tell, fairly much gone unheeded in subsequent philosophical work.
I contend that Malcolm’s conclusions should be restricted to claims about language,
since they do not show anything about ‘how things really are’ as such – particularly, he
has not shown that we really do see postmen and not sense-data, he has not shown that
we really can be certain of the truth of material-thing statements and so on, insofar as
these may turn out to be contingently the case, after all.

So, Malcolm has not refuted
epistemological or metaphysical scepticism.

Despite this, it seems to me that arriving
at this situation, the onus is on the author of the philosophical claim to argue that the
facts it claims do hold, either in the face of the untenable semantic consequences, or
argue that the semantic consequences are not untenable after all. But it does not seem
to be an option for the philosopher to argue that there are no semantic consequences of
her theory, or that the semantic consequences are irrelevant.

Any theory that says
something about the ‘way things really are’, particularly if what is said renders an
ordinary way of describing things necessarily false, will have consequences for what we
can and cannot say meaningfully – in just the same way as Malcolm has demonstrated
for Russell’s theory.


However, it does seem to me that Malcolm has taken his argument further than it
actually goes; in short, it looks like he is reading a metaphysics off language. But I
want to argue that it is not necessary to read Malcolm’s argument this way. My
suggestion is that we stick to using what Malcolm has strictly shown – some issues
about the use, logic and meaning of expressions – and that these conclusions need not
be read as producing metaphysical conclusions at all.

Let us take for example, an argument Malcolm elsewhere deals with.

He
suggests that it is not a ‘correct’ use of language to say

I feel hot, but I am not certain of it.

We might have two question here: why is this ‘incorrect’ language, and if it is,
what does this establish? Malcolm insists that his argument is not about the nature of
certainty, but the logic of expressions in which it features, or rather, the possible uses
such expressions may have.

He says:

We have never learned a usage for a sentence of the sort

I thought that I felt hot but it turned out that I was mistaken [that I felt hot].

In such
matters we do not call anything “turning out that I was mistaken”. If
someone were to insist that it is quite possible that I were mistaken when
I say that I feel hot, then I should say to him.

Give me a use for those
words! I am perfectly willing to utter them, provided that you tell me
under what conditions I should say that I was or was not mistaken.
Those words are of no use to me at present. I do not know what to do
with them…There is nothing we call “finding out whether I feel hot”.
This we could term either a fact of logic or a fact of language.


This highlights the fact that it is not part of Malcolm’s argument that ordinary language
cannot or must not be revised – that what has, for example, turned out to be an
‘incorrect’ use of language (e.g. “I only ever perceive my own sense-data”) cannot have
a use conceived for it, provided the conditions for its use are given.

So, we can accept
new or revised uses for ordinary expressions, only it is incumbent upon the philosopher
to explain the new use, to provide clues as to how it might be employed – but at the
same time, if this is undertaken, it needs to be made clear. It is illegitimate, for
example, to argue that one has a view about ‘the facts’ for which language is irrelevant
if the upshot is a revision of meaning for expressions ordinarily used


The Paradigm Case Argument plays an important role in this early phase of Ordinary
Language philosophy, indeed some think it constitutes Ordinary Language philosophy
(eg, Flew, 1966: p 273 acknowledges this tendency).

The PCA is, in my view, an
important part of the Ordinary Language Argument - but I think it too has been
misunderstood.
It has sometimes been thought that the PCA is an argument that shows that there must
be (at least some) true instances, in the actual and not merely possible world, of
anything that is said (referred to) in ordinary language.

For example, Roderick
Chisholm (1951) says

“There are words in ordinary language, Malcolm believes, whose
use implies that they have a denotation. That is to say, from the fact that they are used
in ordinary language, we may infer that there is something to which they truly apply.”
(p 180) If that followed, certain metaphysical truths, indeed empirical truths, could be
proven simply by the fact that we use a certain expression ordinarily. I do not think it is
necessary to interpret Malcolm, or the correct application of the PCA, as arguing this at
all.

That the PCA has been interpreted in the way I have just suggested is not surprising,
since the PCA has not been formulated, or argued for very well – it has indeed left itself
open to various misinterpretations. In what follows, we'll examine the interpretation
suggested – that the PCA is an argument designed to show that anything said in
ordinary language must have at least some true instances. We’ll then say why we do not
think this is Malcolm’s use of the argument, and what his use of it is.
Malcolm invokes the PCA by reference to one of the outcomes on the assumption that
the philosophical claims he is examining are true.

Now we can put to one side any
implicit modal claims and consider, say, the (actual) truth of the thesis that “No-one
ever knows for certain the truth of any material thing statement”. On that theory, it
turns out that an ordinary expression such as “I’m certain there is a chair in this room”
is never true, no matter how good our empirical evidence for the claim is. Malcolm
casts the ‘Moorean’ reply to such a view, that “[On the contrary] both of us know for
certain there are several chairs in this room, and how absurd it would be to suggest that
we do not know it, but only believe it, or that it is highly probable but not really
certain!” (Malcolm, 1942: p 12) Malcolm says,


What Moore’s reply does is give us a paradigm of absolute certainty,
just as in the case previously discussed his reply gave us a paradigm of
seeing something not a part of one’s brain. What his reply does is to
appeal to our language-sense; to make us feel how queer and wrong it
would be to say, when we sat in a room seeing and touching chairs, that
we believed there were chairs but did not know it for certain, or that it
was only highly probable that there were chairs. Just as in the previous
case his reply made us feel perfectly proper it is in certain cases to say
that one sees a desk or a pen, and how grossly improper it would be in
such cases to say that one sees part of one’s brain.

What Moore’s
replies to those philosophical statements remind us of is that there is
an ordinary use of the phrase “know for certain” in which it is applied to
empirical statements; and so shows us that Ayer is wrong when he says
that “The notion of certainty does not apply to propositions of this
kind.”…Moore’s refutation consists simply in pointing out that [the
expression “know for certain”] has an application to empirical
statements.

Now we can see something more illuminating about Malcolm’s claim that Moore has
refuted the philosophical claim: here he says nothing to the effect that Moore has
proven that it is true that there are in fact chairs and tables before us etc. All Malcolm
has claimed is that Moore has denied, indeed disproven, the suggestion that the term
‘certainty’ has no application to empirical statements. He has disproven that by
invoking cases where it is manifestly the case that the term ‘certainty’ has been, and is,
‘applied’. Nothing, notice, has been said as to whether there really are tables, chairs
and cheese before us - unless, that is, we confuse ‘having an application’ with ‘having
a reference’ or ‘having a true application’.

But we ought not commit this confusion,
because it is not necessary to understand the idea of a term or expression ‘having an
application’ in terms of ‘having a true application’. ‘Having an application’ means, as
Malcolm argues, having a use in a given situation. So, given situations, or contexts,
sanction the assertability of certain expressions, without guaranteeing their truth.
Malcolm’s example in the quote above is a ‘paradigm of absolute certainty’, but not,
notice, a proof of what it is that one is certain of. It is a ‘paradigm of absolute certainty’
because it is a prime example of the sort of situation, or context in which the term
‘certain’ applies – it is a paradigm of the term’s use: namely, in situations where we
have very good (though not infallible) empirical evidence, and no reason to think that
our evidence is not of the highest quality, that something is the case.

We understand Malcolm as suggesting here that a ‘paradigm case’ is an example of the
ordinary use of an expression. This is not to be confused with attempting to ‘analyse’
the meaning of some term, or to delineate necessary and sufficient conditions for its
application. Pointing out an example of something is not necessarily to point out an
exhaustive model of something, and equally, appealing to an example of what we
ordinarily mean when we say something is not to specify the meaning of a term. It is a
demonstration of how the term may be used, by showing one of its most classic uses.


The PCA method and point is easily misconstrued, and we'll explore at least two possible
alternative interpretations of what it is supposed to show.

Both the following
interpretations of the PCA understand that argument as (attempting) to show that there
must be at least some true instances of any ordinarily used term or expression. I want
to show that these interpretations of the PCA do, indeed, render the argument invalid,
but they are not the interpretation Malcolm was using, nor do they represent how
Ordinary Language philosophy used such an argument.

Firstly, the PCA may be interpreted as an ‘anti-realist’ argument, for example that our
‘counting’ a proposition true in virtue of the actual or possible available evidence we
have is all there is to the ‘truth’ of that proposition. On such an argument, one could
claim that since our counting X as true is all there is to X’s being true, then pointing out
any case (a paradigm case) where we do count X as true shows the proposal that there
are no true applications of X to be false (therefore, that there are at least some true
applications of X, where ‘true’ means ‘counted as true’).

A second possible way of interpreting the PCA is as an argument against the possibility
of the global falsity of the application of certain expressions, as opposed to the
possibility of local, particular false applications. On this use of the PCA, the argument
is that there must be at least some true applications of an expression for it to ‘make
sense’ that there be any false applications.

Notice that one could run the latter
interpretation of the PCA independently of the anti-realist understanding of ‘truth’ as
such. That is to say, one could run the ‘anti-global falsity’ argument on the premise
that there has to be at least some true applications of some expression even if we could
not recognise them (or ‘count’ them) as true applications.

To be sure, there may be mileage in running the PCA on either of these interpretations,
or on some combination of them. I am, however, unwilling to interpret Malcolm’s use
of the PCA in either of these ways, for the reason simply that both the anti-realist and
the (call it) ‘anti-global falsity’ interpretations of the PCA appeal to the notion of truth.
In Malcolm’s text, what we see is much more of an emphasis on meaning. It is evident
that Malcolm uses ‘paradigm cases’ as cases, or examples, of what we mean by the use
of a certain expression, and not as cases of the truth of the expression, or even of the
true application of the expression.

Let us take the second argument first: the ‘anti-global falsity’ argument, as we’ve called
it.

But now notice that appeal to a paradigm case is essentially superfluous to this
argument. In other words, the ‘anti-global falsity’ argument does not require appeal to
any actual paradigm cases: it simply postulates that unless there were at least some true
applications of an expression, there could be no false applications. In fact, what is at
the bottom of this argument is epistemological, that is, is an argument about concept
acquisition.

This version of the PCA goes something like this.

How would it be possible for us to
apply a term if all we have is false instances of that term? Indeed, there is something
attractive about this suggestion: how could we even acquire certain concepts if we have
no actual examples of such a thing – indeed if all we are ever presented with is false
examples of such a thing? This version of the PCA seems to be suggesting then that
there must be at least some actually true applications of some expression (not just
“counted as true”) to account for the expression featuring in language at all.
But, does this argument, as compelling as it may seem on the surface, actually go
through? After all, there are many terms and expressions we use ordinarily, technically,
and scientifically which have no true instances.

Indeed, Malcolm urges us not to
understand the PCA in this way – his own example to the contrary is the term ‘ghost’.
Any assertion of the proposition, for example “There is a ghost” is false; there are only
(in the actual world) false instances of this expression. And yet, we know perfectly
well what the terms means, and it features in many other ordinary, perfectly meaningful
expressions as well (e.g. “the ghost walked through the wall”). So, it simply does not
follow that because we have paradigm cases, or any cases for that matter, of the
ordinary use of expressions, such that there must be at least some true applications of
those expressions.

Perhaps, it might be suggested, what matters is that there be at least some applications
which we count as true. But, the argument just given is not altered by exchanging the
concept of ‘truth’ for ‘what is counted as truth’. It remains the case that there are many
propositions for which we have no true instances, nor any instances that we count as
true, which are nevertheless perfectly meaningful.

Our suggestion is that Malcolm’s use of the PCA argument was not to argue that there
must be at least some true applications of ordinary expressions at all (i.e. whether we
mean true-even-if-unknowable or counted as true). Indeed, Malcolm nowhere mentions
that Moore has proven that there are ‘true’ applications of a proposition, only that that
there is an application.

Many expressions in our language have, or have had, or could
have, a use, but only false instances: “I’ve seen a ghost”, “The earth is flat”, “A tonne
of feathers will fall to the earth more slowly that a tonne of bricks”, “My chances of
winning the lottery depend on how many people buy tickets”. When an expression ‘has
an application’, a paradigm case of that application is an example of how the expression
is used in the language, according to its meaning.

The point of appealing to paradigm cases, then, is not to guarantee the truth or ordinary
propositions, but to demonstrate that they have a use in the language. But, one might
think, so what? That an expression has a use is no more philosophically illuminating
than to confirm that we believe our uses of expressions are meaningful. What we
really want to know is: are they true? But this misses the point. The PCA is an
effective move against any view which implies that some ordinary expression does not
have a use, or application. It is effective because it demonstrates such implications to
plainly be false. But I think it does more, and this is why it is really an interesting
philosophical tool. It is also part of the point, of appealing to paradigm cases of the
application of some term or expression, to demonstrate that any suggestion that we are
misapplying it would be absurd. This suggestion raises the question; what else could
the expressions of our language mean, other than, precisely, what we use them to mean?
It is we, language users, who determine through the use of expressions, what it is they
could be said to ‘mean’, and so it is impossible that we have not gotten the meaning
‘right’. That is as impossible as it would be for someone to suggest that the rules of
chess are incorrect, and therefore that what we play is not really ‘chess’.

Correcting somewhat Malcolm’s own stance, the Ordinary Language Argument, as
we’ve seen, is not an argument that has metaphysical results, i.e. anti-sceptical, nor
ontological. It is strictly an argument with semantic results, that is, it shows that
expressions are meaningful because they have a use. Let us now take away the
magnifying glass from Malcolm’s argument – that ordinary language is correct
language - and see what the wider implications are. The immediate outcome is a
generally applicable constraint on philosophical theories: it should not be an
implication of any theory that the ordinary use of some expression turns out to be, not
merely, but necessarily, false – unless it can be shown that is no problem with the idea
that some of our expressions are, despite appearances, self-contradictory. A theorist
who suggests something that is radically inconsistent with what would ordinarily be
said must make clear whether she takes her theory to be a merely contingent
hypothesis, or rather a claim to necessary truth. Unfortunately for many theorists, this
creates a dilemma: if purported to be contingent, it is difficult to ground, or motivate
the view, in the absence of empirical, or otherwise testable evidence for the hypothesis
or even just a good reason to think it is so. On the other hand, if the thesis is proposed
to be necessarily true, it will have the unfortunate consequence that what is ordinarily
said turns out to be self-contradictory – a difficult (impossible according to Malcolm)
position to defend.

But how does this help solve philosophical problems?

Many have thought that the
Ordinary Language view supposes that philosophical problems will go away once we
see language aright. But it does not seem in any way compulsory to think that Ordinary
Language philosophy is a kind of philosophical quietism. The result that ‘ordinary
language is correct language’ should be used as a safeguard against baseless
speculation, but it is not a cure, as it were, for all philosophical problems. Sometimes,
attention to the use, or possible use, of philosophically problematic terms may help to
resolve some issues, in the sense that they may not seem problematic anymore in quite
the same ways. But this is not the end of the matter, it is just the beginning.
Does this answer the charge, voiced by Devitt and Sterelney, that ‘linguistic
philosophy’ fails the traditional philosophical project of helping us to understand
reality, what there is, the nature of truth etc, insofar as it focuses on meaning? The
answer is that we do not take our eyes off reality when we focus on language and
meaning. Understanding the nature of phenomena and understanding how and why we
describe it in the ways we do amount to the same thing.

It is not the case that the
Ordinary Language philosopher is interested in linguistic meaning independently of an
interest in the reality language is supposed to describe. The point is to understand the
phenomena of reality, but equally, the thought is that there is no method of doing so
independently of studying how we use language to describe it – as there is in the
empirical sciences. We understand phenomena through understanding, for example,
what we would say of it and what not, what we might be prepared to say and what we
would reject, whether we might consider saying something else of it and what follows
from any of these observations, as we come to know the phenomena better through
observation etc.

But observing and attending to the ordinary uses of language does not solve
philosophical problems, because there is no end-point to arrive at. Philosophical
problems are not like mathematical problems in this sense. How we use expressions,
what we are prepared to count as ‘good’ or ‘right’ or ‘justified’ or ‘conscious’ or
‘certain’ etc, is in constant evolution, and so the exploration of meaning is an ongoing
project.

Recognising that what we now refer to as ‘Ordinary Language’ philosophy was never a unified
movement, or program, as such, but rather a collection of views amongst a variety of thinkers - amongst whom we may count the ‘Wittgensteinians’: Malcolm, Ambrose, Wisdom and Lazerowitz, to name a few; and the later ‘Oxford’ philosophers, Ryle, Austin, Herbert Paul Grice, and later still, Grice's student, Strawson. Somewhat more contemporary Ordinary Language sympathisers include Stanley Cavell and Oswald Hanfling and some members of the Grice Club.

It is true that the views of these thinkers do not always agree, and the distillation of what I present as a unified argument behind Ordinary Language philosophy would not be accepted in all detail by each and every one.

J. L. Austin, for example, perhaps the most well-known Ordinary Language philosopher, refrained from
explicitly formulating a meta-philosophical argument, unless Grice identified it!

He urged the method of attending to the nuances
of our ordinary uses of language (he was himself a master of this), but maintained that this method was a
beginning point, not a final arbiter, in the resolution of philosophical problems. (1956, p 182)
Nevertheless, I do think that the argument we present, via Malcolm, captures some key aspects of the view
that holds these thinkers together in a single category.

The term (as "Grice the futilitarian") was coined by Bergmann in 1953 (in Rorty, 1992, p 64).

The other element is the interestingly
connected, but primarily opposing stance of Logical Positivism, or ‘Ideal Language’ philosophy, alla Carnap.

The
history of these two views, i.e. Ordinary Language philosophy and Logical Positivism, is deeply
intertwined and the two distinct lines of thought are not always carefully distinguished in the literature.
This is evidenced by a habit of generalising about ‘linguistic’ philosophy, rather than addressing the quite
different stances taken on this ‘focus on language’.

The first misunderstanding here is to imagine ‘ordinary’ language to be distinguished from, say,
‘technical’ language – i.e. the language used in specialised disciplines such as medicine or physics. Nor is it to be conflated with the language spoken, or used, by one group (e.g. lay-persons) as opposed to another (e.g. the ‘educated’), within the same linguistic community. Words and expressions are not in and of themselves ‘ordinary’ (or not), rather, the focus is on their use (and as will become apparent, some uses count as ordinary and others do not).

The second misunderstanding is that what is at issue is whether certain uses of expressions express something that is true or false. This is a more complex misunderstanding than might appear, and is addressed more fully in the text. Suffice it to say here that the fact that a good deal of what we say ordinarily may well be false is a fact that needs to be distinguished from the issue as to whether certain uses of expressions count as ordinary or not.

That is, we must not conflate the (Ordinary Language philosophy) concept of ‘ordinary use’ with ‘expressing something true (or false)’, nor the concept of ‘non-ordinary use’, or even ‘misuse’ with ‘expressing something false (or true)’.

The two issues, ‘use’ and ‘truth’, though obviously connected, are
nevertheless logically distinct, since there is nothing contradictory in saying that a perfectly ordinary use
of some expression may express a falsehood, nor in saying that a non-ordinary use may state a truth. Of
course, we need to know more about what, on this view, does count as an ordinary or non-ordinary use,
and that will become clearer as we progress through Malcolm’s argument. But it does need to be
emphasised that reading into the claim that some uses of expressions are ordinary uses, that what they
express is (thereby) true, is the root of a good deal of misunderstanding of the Ordinary Language view.

Thus it is necessary to point out the distinction right from the beginning, and explain it more fully in due
course.

Or rather, with what would ordinarily be said, of some phenomenon or situation.

We sometimes use
(lower case) ‘ordinary language’ to stand in for this, i.e. for ‘what would ordinarily be said’.

Moreover,
the concern is with propositions, and not ‘sentences’.

We have for the most part put syntactical
considerations to one side in this argument as they do not, as far as we can see, play any major role in it.


The concern is more specifically with the semantics of propositions (which can be treated as classically
truth-conditional).

Historically, the Logical Positivists under the early work of Rudolf Carnap, and their contemporary
heirs, now known as ‘minimalist’ or ‘invariantist’ semanticists. Go to the Carnap Corner, hosted by R. B. Jones.  (Or pay a visit to the City of Eternal Truth).

It’s necessary here to say something as to the existence and method of identification of ‘ordinary uses’ of terms and expressions.

It has been objected that it is not clear, and perhaps not possible, for anyone
(let alone an armchair philosopher) to say what would or would not ‘ordinarily be said’ in given
situations. We are, of course, interested in what would be said of given situations, in particular, i.e. in
what a given situation or phenomenon would ordinarily be said to be a case of.

Though this warrants some explanation, the complaint is much less genuinely problematic
than is thought.

It seems clear that we can identify non-ordinary uses of expressions fairly readily. Take
the case of the use of the term ‘know’ in

I do not know if there is a desk before me.

As used in its
‘sceptical’ sense (i.e. where it is used to mean, not that e.g. the light is dim and I’m not sure if it is a desk,
or some other sort of table, but to mean that that no sensory information can be ‘known’), simply
pointing out that further explanation is required to make this use clear (if one wants to make it clear)
shows that we can see and agree quite easily that this use (i.e. call it the ‘sceptical use’) is ‘non-ordinary’.

This is not to say that the sceptical use is to be prohibited, or avoided or is somehow faulty; much less is
it to show that what is expressed on this use is false. Nevertheless, it seems hard to deny the claim that
the sceptical use of the term ‘know’, in this case, is not the ordinary use – whatever otherwise its merits
may be.

But why label the sceptical use as the non-ordinary use?

Why can it not count as the ordinary
use?

After all, it is a quite common use amongst philosophers, and in such contexts often doesn’t even
require ‘further explanation’ to make the sense with which it is used clear.

The answer is that the distinct
uses of the expression bear an asymmetrical dependence relation.

What this means is that the meaning of
a non-ordinary use of some expression depends on the meaning of the ordinary use; or rather, the
possibility of making a non-ordinary use clear depends on the recognition and acceptance, or assumption
of the ordinary meaning; but not vice-versa.

This makes it possible to test, of two (or more) ways a term
may be used, which, if any, is a non-ordinary use.

For instance, in the case we are considering, the
sceptical use of the term ‘know’ depends on the ordinary use, because it would be hard to make sense of
the sceptical use if we did not already know that there is some other use (of ‘know’), a use which must at
some level be recognised, if one is to recognise what the sceptical use is denying (i.e., that ‘knowledge’ applies to beliefs about an independent material realm).

Grice speaks of platonic knowledge. Knowledge, strictly applies to 2 + 2 - 4. But we still can be "deemed" to know other things, like that the pillar box is red.

If the ordinary use of the term ‘know’, were not recognised (i.e. its use to apply to propositions about an independent material realm), or assumed, the
sceptical use would not make sense: because the term ‘know’ would then mark a completely different
distinction (essentially, that between valid and invalid mathematical or logical propositions, rather than
as between beliefs which have been found to be true, and those which have been found false). The term,
on the supposition that its meaning depends on no other use, would not even apply to non-logical or nonmathematical
claims at all.

So,

a) the idea that ‘ordinary’ language may not even exist is rebuffed, since
non-ordinary uses can readily be detected, and they count as non-ordinary insofar as their use depends on
ordinary uses, and not vice-versa. But also

b) regarding the philosopher’s or anybody’s justification for a
claim as to ‘what would be said’ in a given situation: whilst no-one is an absolute authority on all of the
uses of expressions of a language, anyone who is a member of a linguistic community has an equal right
to make claims as to what would or what would not be ordinarily said (as argued by Cavell).

The
ultimate ‘authority’ can only be the linguistic community as a whole – certainly, nothing outside of it
could be.


Or, we should say, our version of Malcolm, which is a somewhat revised and corrected interpretation of
his argument.
We do not stray from the main points he maintains, nor the overall rationale and conclusion,
but some of what I attribute to Malcolm is an improvement on what is said in the actual text. For
example, Malcolm claims that his (Moore’s) responses to the philosophical claims refute them – but
Malcolm equivocates on whether what has been refuted is the metaphysical picture the claims suggest, or rather the linguistic (or semantic) picture that they entail. I say more about this below, however, we ignore
this equivocation in order to present the argument as it should be read, which is as an argument which
has no metaphysical consequences whatever, but only semantic consequences, i.e. consequences as to the
meaning of terms and expressions, and how they may be used.


We understand ‘contingent’ to mean ‘possibly true’ or ‘true in some possible world’ and
‘necessary’ to mean ‘necessarily true’ or ‘true in all possible worlds’.


He explicitly wrote, on a couple of occasions (1942, 1940), that this view was not a version of
‘verificationism’, i.e. that the distinction between ‘factual’ and ‘linguistic’ propositions was not based on
epistemic considerations, such as how we may come to know or confirm them.

In fact, Malcolm urges
that the content of all propositions, even if we do distinguish between what we call ‘factual’ and ‘logical’
(synthetic and analytic, contingent and necessary), is nevertheless learned, confirmed and understood
equally ‘empirically’ (i.e. through observing others, practicing, and being corrected when we go wrong
etc).

On this view, our knowledge and understanding of, say, the propositions of mathematics and logic,
are not acquired through intuition of a distinct metaphysical realm, but in the same way we acquire
knowledge and understanding of propositions about material objects.

That is, working out whether a
proposition is, say, necessary or contingent is “…a matter to be settled by the eyes and ears (Wisdom’s
phrase) and not by some mysterious intuitive act of the intellect.” (Malcolm, 1940, p 195) In particular,
according to Malcolm, the necessity of necessary propositions is learned through observation –
observation of how the propositions are used in the language. And, what is observed, i.e. how we use
language, what we accept as following from certain propositions, what we reject, what we do and do not
say and so on, is as ‘factual’ as anything else we observe.
Wherever it is claimed that we do not say something of the form “p and not p”, or that such
propositions ‘have no use’, it is always intended that this is the case for a literal, non-metaphorical, nonpoetic, non-comical etc uses.

The use is more important than the proposition itself, in this sense, because
we sometimes we use contradictions very ordinarily.

A: Is your work going well?
B: Well, it is and it isn’t.

But we wouldn’t call this a case of contradicting oneself.

Similarly, sometimes necessary propositions are negated in a perfectly meaningful way.

He’s a bachelor in all but the fact that he’s married.

The necessity of arithmetic and logic is thus, on Malcolm’s view, justified in the use of language.

This is in accord with the later Wittgensteinian views about the necessity of arithmetic and logic, in our view.

This position has little to say about the ‘reality’ or ‘nature’ of numbers.

We do not think it is correct to
interpret it as a rejection of the possibility of the existence of a metaphysically necessary realm of facts,
but rather that their existence cannot help us to answer any of our fundamental, philosophical questions
concerning arithmetic or logic. It was, we think, an insistence that the question whether there is or is not a metaphysical realm which is described by the language of arithmetic and logic is beside the point.

The point is that if we stick to observing how we actually do and do not use the expressions of arithmetic and logic, how we learn them and how we teach them, then what is incontrovertible is that we are trained, for example, not to assert

2+2=5

we are trained to respond to ‘continue this series: 2, 4, 6, 8….’ in a
certain way.

But none of these observations alone ought to be counted as good reasons to think that we
are describing, or conforming to, a mathematical reality that exists independently of our actual practices
of counting, calculating, continuing series etc.

This is neither to accept or reject such a reality, but to
insist that it is how we make use of the propositions of arithmetic and logic that justifies their necessity.

Appeal to any further facts, over and above the fact that we do so use a proposition in this way is
superfluous if we wanted to understand, as an example, why 2+2=4, and why 2+2 cannot = 5.

One may,
if one wishes to, speculate that the answers to those questions are to be found in the nature of
mathematical reality.

But appeal to metaphysical facts cannot justify our use of the proposition “2+2=4”
as necessary – since all we can do is speculate about their existence. We do not call on such a realm to teach arithmetic, for instance, nor to carry out proofs and calculations –i.e. we do not suggest that one
simply ‘consider mathematical reality’, for example, as a method of learning mathematics, or as a way to
find a proof or calculation. What does justify our taking the proposition to be necessary is that we do
not, as things stand, have a use for the expression “2+2=5” in arithmetic.


Therefore, Quine’s complaint that we have no way of analysing the analytic/synthetic distinction
without circularity bites less here – as Grice well knew (see his defense of a dogma), the Ordinary Language philosopher would agree that there is no
‘analysis’ of the distinction, insofar as we could provide necessary and sufficient conditions on what
counts as a ‘necessary’ or ‘contingent’ proposition (or analytic or synthetic), and moreover, that the
boundaries are not hard and fast.

Since whether a proposition counts as necessary or contingent is a
function of how we use it, there is no reason why propositions cannot change category, or fall on the
borderline between categories.

Perhaps “Nothing can be both red and green all over” and “I cannot be in
two places at once” are such borderline cases – it seems we may be willing to allow them to be negated
should, say, physics show us that there is, after all, a use for e.g. the proposition

“There is a special
material which, under special circumstances, looks both red and green all over”

or “Certain quantum
particles can be in two places at once”.

Moreover, necessary truths, on the Ordinary Language view,
must always be kept within the perspective of how we can, do or might use expressions – i.e. they are not
to be taken as dictating absolutely inflexible strictures on linguistic use.

For example, we might consider
it a necessary truth that

If one knows x, then one believes x to be true.

However, there is a use of these
terms, e.g.

I know it is true, but I simply don’t believe it.

which is not a violation of the necessity of the
first proposition. (Malcolm, 1940, p 194)

And finally, no necessary proposition is unrevisable on this
view – since our usage can always change.

For example, Malcolm claims “We can quite well imagine
our practice [in respect of our use of negation] should gradually alter until everybody used double
negative expressions, such as

Ain’t nobody here.

And if this happened we should say then that

~(~p) = ~p

was a law of logic.

Cfr. L. J. Cohen's misreading of Grice (Cohen is playing a bit of a Cockney).

Thus, this view has no problem with the idea that a
proposition which has been discovered to be true a posteriori, and thus counts as a genuinely ‘factual’
empirical claim, comes to be used as a necessary proposition e.g. “Water is H2O”.

It may also be found in, for example Wisdom, 1936-7, Lazerowitz, 1955 and Ambrose 1952.

As a
general rationale behind a ‘use’-based theory of meaning, we think it may be held by many contemporary
theorists as well, including Grice.


But there are very few who would explicitly agree that they maintain a ‘linguistic
doctrine of necessity’.

So, it is not, paradoxically, the claim that there are no metaphysical facts – which is, of course, a
metaphysical claim (or, if this is sometimes intimated by Malcolm and the Ordinary Language
philosophers, it ought not to be).

It might be argued that a metaphysical realm of necessary facts does explain why we use some
propositions as necessary.

This may be so, but the fact is it does not explain why we use a particular

expression as necessary, e.g.

All triangles are three-sided.

A given person uses such an expression as
necessary because he or she has learned the meanings of the terms ‘triangle’ and ‘three-sided’ in his or
her language, and not because he or she has been exposed, in some way, to a special fact, i.e. to the
metaphysical necessity of the three-sidedness of triangles.

Once a person has sufficiently mastered the
use of the term ‘triangle’, nothing further is required to explain why she holds “All triangles are three-sided” to be necessarily true.

For example, one might suppose it is open to a theorist to propose that he or she has discovered the
proper meaning of ‘perceive’, i.e. that it doesn’t apply to a relation between perceivers and independent
material objects after all.

And so, we have simply been in error in applying it the way we have.

The
problem for such a theory is that if it is suggested that our prior uses of some term are, in fact, necessarily
false, then the onus is on the theorist to explain how we have come to be uttering such self-contradictory
expressions, regarding some phenomenon or other.

This is quite a distinct theoretical project from that
involving claiming our propositions have simply been false all along. This is, for example, a distinctly
semantic project – i.e. one involving showing how some expression, which has evolved to have some use
in the language, and is used in just such a way, does not have this use after all. It is not clear what
grounds one could have for making such a suggestion. On the other hand, if a theorist were to suggest
not that we have been expressing necessary falsities all along, but have just been plain wrong, then one would assume that the grounds for such a claim would be ordinary and empirical – so it would be a scientific hypothesis.

Making this kind of claim is not to make a claim about the meaning of the terms
we use, however. This would simply be to make an empirical hypothesis which, if proven correct, would not have the result that we have hitherto been uttering necessary falsehoods. This latter would not amount to an error-theory about meaning. See Russell, 1927.

So self-contradictions are ‘incorrect’ uses of language, but not all so-called ‘incorrect’ uses of
language, on this view, are necessarily meaningless. For example, the ‘sceptical’ use of the term ‘know’ is classified, on Malcolm’s view, as a ‘misuse’ of the term, but only because it is not the ordinary use.

Such a use is, of course, allowed, and obviously it is not meaningless. But using the term this way
requires some further explanation, or indication to the effect that it is not being used in its ordinary sense.
So, contrary to some detractors, on the Ordinary Language view, not all ‘incorrect’ uses, or even
‘misuses’ of expressions are thereby false – nor are they prohibited. So, it is not prohibited to use
expressions in ‘philosophical’ or other novel ways – but one must make reference back to the ordinary use in order to make the new use clear. If one decides to use the term ‘certainty’ only to apply to mathematical or logical proofs, but not to empirical observations, then that is allowed, as it were – but the user must make this way of using the term clear before doing so. Malcolm’s view is that certain philosophical theses introduce just such new uses of expressions – non-ordinary uses – but fail to acknowledge it, claiming, instead, to have ‘analysed’ or ‘given an account of’ the expression or term in question.

Once again it is worth noting that, for Malcolm, it is not the proposition itself which may be said to be ‘correct’ or ‘incorrect’, but how it is used. It is a correct use of the proposition

I see part of my brain.

for instance, in certain circumstances such as, e.g. if one were to view a photograph taken during one’s brain surgery.

Used in this way, i.e. to make a contingent claim about a particular situation, there is
nothing contradictory about it. But to say the same thing of every perceptual experience would not make
sense, given what we do mean by ‘perceive’. Moreover, not all apparent contradictions are incorrect uses
of language – if they are not used as, say, literal contradictions. He says “We do not call an expression
which has a descriptive use a self-contradictory expression. For example, the expression “It is and it
isn’t” looks like a self-contradictory expression. But it has a descriptive use. If, for example, a very light mist is falling…and someone, asking for information, asked whether it was raining we might reply

Well,
it is and it isn’t.

We should not say that the phrase, used in this connection, is self-contradictory.” (p 16)

Many articles written in response to Malcolm’s paper reject the claim that he refutes sceptical theses.
For example, see Chisholm (1951: pp 175 – 182).

It has been noted, for example in Flew (1966: pp 268 – 269) that this changes the argument from the
‘nature’ of certainty, to one about the standards of empirical evidence we should use as criteria for its
application. And so, it is legitimate (and not a ‘misuse’ of language) to ask whether our standards we
ordinarily use for the application of the term are high enough. But as Flew notes, the moment the
argument about standards of empirical evidence veers into the suggestion that no standards are high
enough, then it is veering into the suggestion that any application of the term ‘certain’ to empirical
statements is self-contradictory (since it would be to assert that something which is necessarily not certain, is certain). And so on.


REFERENCES

Ambrose, A. ‘Linguistic Approaches to Philosophical Problems’, Journal of Philosophy, XLIX, pp 289-301.

Austin, J L.  ‘A Plea for Excuses’, in ed. Chappell (1964), pp 41-63.

Bergmann, Gustav ‘Logical Positivism, Language, and the Reconstruction of Metaphysics’ (in part), reprinted in ed. Rorty, pp 63-71.

-- Grice as an English Futilitarian.

Cavell, Stanley  ‘Must We Mean What We Say?’, reprinted in ed. Chappell, pp 75-112.

Chappell, V C., Ordinary Language, (Prentice-Hall).

Chisholm, Roderick, ‘Philosophers and Ordinary  Language’, reprinted in ed.Rorty, pp 175 – 182.

Devitt, Michael and Sterelny, Kim, Language and Reality: An Introduction to the Philosophy of Language, Second Edition, (Blackwell).

Dummett, Michael. Frege: Philosophy of Language, (Duckworth).

--- Wrigley to Grice: I will base my dissertation on Dummett's "Frege". Have you read it. Grice: "I haven't, and I hope I won't."

Feyerabend, Paul and Maxwell, Grover, eds., Mind, Matter and Method (University of Minnesota Press).

Flew, Antony, ‘Again the Paradigm’, in eds. Feyerabend and Maxwell,  pp 268 – 269.

Kripke, Saul, Naming and Necessity, (OUP).

Lazerowitz, Morris, The Structure of Metaphysics, (Routledge and Kegan Paul).

Malcolm, N. ‘Are Necessary Propositions Really Verbal?’, Mind, XLIX, pp 189-203.

--  ‘Moore and Ordinary Language’, reprinted in ed., Chappell, V C, pp 5-23.

— ‘Certainty and Empirical Statements’, Mind, LI, pp 18-46.

— ‘Philosophy and Ordinary Language’, Philosophical Review, LX, pp 329- 340.

Mates, Benson, ‘On the Verification of Ordinary Language’, in ed. Chappell, pp 64-74.

Quine, W V O, Word and Object, (MIT Press).

Rorty, Richard, ed., The Linguistic Turn, (The University of Chicago Press).

Russell, Bertrand, The Analysis of Matter, (George Allen and Unwin).

Ryle, Gilbert. ‘Systematically Misleading Expressions’, reprinted in ed. Rorty pp 85-100.

Soames, Scott Philosophical Analysis in the Twentieth Century, Volume 2, The Age of Meaning, (Princeton University Press).

Tennessen, Herman. ‘Ordinary Language in Memoriam’, Inquiry, 8, pp 225- 248.

Wisdom, John. ‘Philosophical Perplexity’, Proc. Arist. Soc., XXXVII, pp71-88.

Wittgenstein, Ludwig (1953), Philosophical Investigations, trans. G E M Anscombe, (Blackwell).

No comments:

Post a Comment