Mind & Language ISSN 0268-1064

Vol. 5 No. 2 Summer 1990 0 Basil Blackwell

Ar ticI e
Wittgenstein and Kripke on the Nature
of Meaning

PAUL HORWICH

Inspired by Wittgenstein’s Philosophical Investigations’, Saul Kripke2 has

recently presented an account of meaning arguing that, in some sense,
there is no such thing, and drawing from this conclusion some striking
consequences for linguistics and psychology. His line of thought in short
is, first, to show that attributions of meaning (for example, ’Jones means
addition by ”plus”’) do not describe ‘genuine facts’; second, to observe
that since such statements are nonetheless associated with communally
recognized conditions for appropriate use, then, despite their lack of factual
content, they are still perfectly significant; third, to infer from the character
of their assertibility conditions that a ’private language’ is impossible; and
fourth, to deduce that contemporary Chomskian linguistics is misguided
insofar as it attempts to explain the behavior of an individual by supposing
that he follows a personal system of rules. What I would like to do here
is to assess these ideas and comment on their relationship to Wittgenstein’s
philosophy.
It must be emphasized that Kripke does not himself endorse the reason-
ing that is suggested to him by reading Wittgenstein-though he does
admit to finding it somewhat compelling. So when in what follows I refer
to something as ’Kripke’s conclusion’, ‘Kripke’s argument‘, and so on, I
mean merely to be labelling the idea in question as one that Kripke
presents, without suggesting that he fully believes in it.

Ludwig Wittgenstein, Philosophical Investigations, Oxford: Blackwell 1953.

* Saul Kripke, Wittgenstein: On Rules andPrivate Language, Oxford: Blackwell, 1982.
106 Mind b Language
I. Construals of the Sceptical Thesis

I begin by asking what exactly Kripke's sceptical conclusion is. Just how
are we supposed to interpret the thesis that attributions of meaning do
not correspond to facts-or, as he puts it, that

There can be no fact as to what I mean by 'plus', or any other

word at any time (p. 21)?

The trouble is that this sort of statement is highly ambiguous. Let US

therefore go through some of the alternative ways of understanding it and
consider which might be the one that Kripke has in mind. I'll look at the
following three possibilities, to be clarified as we proceed:

(A) Words are meaningless.

(B) Meaning a certain thing by a word is not an intrinsic property of
a person-not a fact about him alone.
(C) Meaning attributions are not purely descriptive:-they are not
rendered true, or false, by any collection of natural, non-semantic
facts, but have normative, and therefore non-factual, content.

To start with, it is evident that Kripke's conclusion is not intended to

be the stunning, self-defeating claim that all language is senseless. This is
clear from the two-pronged structure of his discussion. He begins by
elaborating what he calb a 'sceptical paradox' whose conclusion is that
there are no facts about meaning. This argument consists in entertaining
a number of plausible candidates for what the facts of meaning might be,
together with demonstrations that none of these candidates meets certain
adequacy conditions that any account of meaning must allegedly satisfy.
The second part of Kripke's line of thought is what he calls a 'sceptical
solution' to the paradox. A straight, or normal, solution, he says, would
be the discovery of some mistake in the argument-for example, the calling
attention to some further kind of fact, not previously noticed, that would
satisfy the adequacy conditions. Kripke's sceptical solution, on the other
hand, is not intended to refute the sceptical conclusion or to suggest that
it can't be validly drawn, but rather to show that it is not as paradoxical
or radical as it might at first appear to be. Specifically, the idea is to show
that the sceptical conclusion does nor undermine our ordinary talk about
meaning. It need not induce us to reject claims of the form, 'Jones means
plus by "plus"'. It does not even compel us to deny 'It is a fact that Jones
means plus by "plus"' or 'It is true that Jones means plus by "plus"'-
provided that these are understood (as they are, in non-philosophical
contexts) as saying hardly more than 'Jones means plus by "plus"'.
Thus amongst our various construals of the thesis that meaning is non-
existent, there is one-evidently the most dramatic one-that Kripke quite
obviously does not have in mind. He is not claiming that meanings, like
 Wittgenstein and Kripke on the Nature of Meaning 107

ghosts, phlogiston, and the lumiferous ether, are nothing but illusion. For
in that case we would be forced to abandon our ordinary practice of
meaning attribution. The moral might perhaps be put as follows. Strictly
speaking Kripke is not denying that statements attributing meaning may
be true or false, and that when they are true then there is a fact correspond-
ing to the statement. His view is rather that there are no facts of a
certain sort-we could call them ‘genuine facts’-regarding meaning; and
moreover that this is surprising and important. For we are very tempted
to think that the facts of meaning are of this special sort and, for this
reason, we are content to employ claims about meaning in linguistics and
psychology.
In accordance with this view of the matter, a second possible construal
of the sceptical conclusion is that it denies, not the existence of meaning,
but rather that meaning is a purely personal affair. This idea-that ’mean-
ings are not in the head’-is familiar from the work of Hilary Putnam3
and Tyler Burge4. It implies that if Jones means addition by ’plus’ then
there is no fact solely about Jones-no intrinsic characteristic of Jones
alone-in virtue of which this is the case. This construal of Kripke’s thesis
employs the old, but hard to explicate, distinction between intrinsic and
relational properties. Roughly speaking, the intrinsic properties of an
object are those, like ‘being spherical’ and ’being an electron’, that make
reference to no other object; whereas relational properties, like ’being one
mile from the Earth’s surface’ or ‘being taller than average’, do require for
their possession the existence of other objects.
Now philosophers have occasionally used ‘non-existent’ to mean ‘non-
intrinsic’. For example, it is sometimes said that there is no such property
as rednes-where the intention is to claim that a statement of the form
’X is red’ is not simply about X, the thing in itself, as one might at first
suppose, but is really about relations between the thing and people. For
it might be thought that ’X is red’ means something like ‘X would produce
such and such sensations in such and such circumstances’. Similarly, the
content of Kripke’s sceptical conclusion about meaning could be taken as
the claim that meaning a certain thing by a word is not an intrinsic
property of a person.
It is tempting to think that this is precisely what Kripke has in mind.
Many of his formulations suggest it: for example,
we must give up the attempt to find any fact about me in virtue
of which I mean ’plus’ . . . (p. 108)

Hilary Putnam, The Meaning of ‘Meaning’.In K. Gunderson (ed.), Language, Mind and
Knozuuledge: Minnesota Studies in the Philosophy of Science VIZ. Minneapolis: University
of Minnesota Press, 1975.
Tyler Burge, Individualism and the Mental. In P. French, T. Ueling and H. Wettstein
(eds.), Midwest Studies in Philosophy, Volume 4. Minneapolis: University of Minnesota
Press, 1979, 72-121.
108 Mind 6 Language
Wittgenstein’s sceptic argues that he knows of no fact about an
individual that could constitute his state of meaning plus . . . (p.
39)
the sceptic holds that no fact about m y past history-nothing that
was ever in my mind, or in m y external behavior establishes that I
meant plus . . . (p. 13)

So whatever ‘looking into m y mind’ may be, the sceptic asserts

that even if God were to do it, he still. could not determine that I
meant addition by ’plus’. (p. 14)

[emphasis added]

However, although Kripke evidently agrees that meaning is not a purely

personal property, it emerges that his sceptical conclusion is more radical
than this. For the mere fact that meaning is not intrinsic leaves it open
that when we say ’Jones means plus by ”plus”’ we are asserting the
existence of a certain natural relation between Jones and a linguistic com-
munity. In other words, the rejection of individualism regarding meaning
leaves open the following picture:-meaning plus is a property of the
public word ‘plus’, a naturalistic property that it has in virtue of a certain
communal use; and anyone uses ’plus’ with that meaning if his usage of
the word is in rough accordance with community standards. Here, the fact
that Jones means pIus by ‘plus’ would be partly a fact about Jones and
partly a fact about his linguistic community. No wonder God could not
determine what Jones means just by looking into Jones’s mind.
Although this interpretation is suggested, as I have indicated, by some
of Kripke’s remarks, it is quite clear on balance that it is not what he
intends. Most of Kripke’s formulations imply the much stronger claim that
there is simply no fact at all-not merely no fact about Jones-in virtue of
which Jones means plus. For example, he says:

no ’truth conditions’ or ‘corresponding facts’ in the world exist

that make a statement like ‘Jones, like many of us, means addition
by “+”’ true. (p. 86)

[Wittgenstein] does not give a ’straight’ solution, pointing out to

the silly sceptic a hidden fact he overlooked, a condition in the
world which constitutes my meaning addition by ‘plus’. In fact
he agrees with his own hypothetical sceptic that there is no such
fact in either the ’internal’ or the ‘external’ world. (p. 69)

It is important to realize that we are not looking for necessary and

sufficient conditions (truth conditions) for following a rule, or an
analysis of what such rule-following ‘consists in’. Indeed such
 Wittgenstein and Kripke on the Nature of Meaning 109
conditions would constitute a ’straight’ solution to the sceptical
problem, and have been rejected. (p. 87)

Moreover at the very end of his essay Kripke considers the idea that the
meaning of a word might be constituted by facts about its use within a
linguistic community, and he rejects it, complaining that

Such a theory would be a theory of the truth conditions of such

assertions as ’By ”plus” we mean such-and-such a function.‘ (p.
111)

And this he takes to be incompatible with the sceptical conclusion. Thus

Kripke does not wish to assert merely that meaning is not a personal
matter.
These passages point us towards what I think is the correct interpretation
of Kripke’s sceptical conclusion. It is intended, I believe, to deny the
reducibility (in any sense) of meaning attributions to non-semantic state-
ments. It says that neither conceptual analysis nor empirical theory will
permit the determination of semantic facts by facts about, say, mental
processes, overt behavior or communal use. The thesis is, in other words,
that meaning attributions are not made true by any conglomoration of
non-semantic phenomena-so they can’t be incorporated into any grand
unified theory of the natural world.
This construal of Kripke’s conclusion is suggested in the first place by
his argument. For, as I’ve said, the strategy of the sceptical paradox is to
consider a series of non-semantic candidates for the facts in virtue of
which ’plus’ means what it does, and to argue against each of them,
concluding that there are no facts of the right sort to identify with facts
about meaning.
In addition, we can perhaps shed light on the character of Kripke’s
sceptical conclusion by comparing it to the emotivist position on ethics.
The general rationale for this point of view lies in the idea that there are
types of utterance-including expressions of value and (perhaps) first
person sensation reports-that, on the basis of their syntactic form, we
are tempted to assimilate to descriptions, expressing ’genuine facts’, but
that, in light of their role in the language, should properly be regarded as
entirely different speech acts. Specifically, it is claimed by emotivists that
’X is good‘ does not ascribe an objective property to X, but rather expresses
the speaker’s pro-attitude towards X. Similarly, it might be thought that,
syntactic appearances to the contrary, attributions of meaning are not used
in a way that is characteristic of descriptions, but have, in particular, a
normative function, and in that respect are non-factual. As Kripke points
out, when we say Jones means plus by ‘plus’ we imply not that Jones will
answer 125 when asked to add 68 and 57 but rather that he should answer
that way.
110 Mind 6 Language
II. Assessment of Kripke’s Argument

Let us adopt this interpretation and ask why Kripke thinks that attributions
of meaning are not rendered true or false by natural facts. In particular,
what is supposed to be wrong with the view (which, pace Kripke, would
seem to be Wittgenstein’s actual view) that the meaning of a word is its
communal use? Somewhat surprisingly, Kripke does not give this idea a
great deal of explicit a t t e n t i ~ nIt. ~comes up in a rather implausible form,
right at the end of his essay; and is dismissed with the claim that it can
be refuted by generalizing certain earlier arguments that were deployed
against a different theory:-namely that what an individual means by a
word consists in facts about how he alone is disposed to use it. Let us
review these arguments to see if they really do undermine the community-
use theory of meaning.
Kripke’s first objection to the dispositional account of meaning is, as he
puts it, that

As a candidate for a ’fact’ that determines what I mean, it fails to

satisfy the basic condition on such a candidate, . . . ,that it should
tell me what I ought to do in each new instance. (p.24)

The point is that we feel guided in our linguistic behavior by what we

mean, as if directed by a set of instructions; but nothing of this sort could
be accommodated by a brute disposition to say certain things and not
others.
Actually, it is rather odd that Kripke states this ‘guidance requirement’
so forcefully. For he himself comes to the view that there are in general
no such guiding facts of meaning-the sequence of rules to interpret rules
must come to an end somewhere. As he says,

ultimately we reach a level where we act without any reason in

terms of which we can justify our action. We act unhesitatingly
but blindly. (p. 87)

And if this is going to be eventually admitted, then it should not have been
insisted at the beginning that facts of meaning must have the character of
instructions. Therefore, if the dispositional theory is entirely adequate
except for failing to accommodate the intuition that meanings provide
guidance, then the proper conclusion is not that meanings are not dispo-
sitions, but a strengthened conviction that the naive intuition of guidance
should not be taken too literally.
But is the dispositional theory otherwise adequate? Kripke’s second

This criticism of Kripke is also made by Colin McGinn in his Wittgenstein on Meaning,
Oxford: Blackwell, 1984.
 Wittgenstein and Kripke on the Nature of Meaning 111
criticism is that such an account leaves no room for the possibility that
someone might use a word incorrectly. For if a person’s meaning consists
in what he is disposed to say, then what he does say (being the actualized
part of what he is disposed to say) must be perfectly in accordance with
his meaning. For example, if in normal circumstances someone is asked
to add 68 to 57 and gets 115 we want to judge that he has made a m i s t a k e
perhaps he has forgotten to carry. But according to the dispositional
account we must judge, it seems, that he has not made a mistake at all,
but rather that he means by the world ’plus’ some weird function whose
value for arguments 68 and 57 truly is 115.
Against this objection it is tempting to argue that the relevant dispo-
sitions must (and can) be characterized so as to make allowances for
mistakes. Provisos against the speaker being tired, distracted, or intoxi-
cated may be built in to the dispositional analysis:-viz. ’ S means plus by
“plus” if and only if S would say such and such, provided he is not tired,
drunk, . . . etc.’. Thus the possibility of error is accommodated. However,
it will be pointed out, in response to this move, that our normal ways of
identifying mistakes (and conditions likely to produce mistakes) involve
reference to community opinion: our criterion for ’going wrong’ is diver-
gence from what is generally said. Therefore the modified dispositional
analysis, once made complete and explicit, will involve facts about the
speaker’s linguistic surroundings.
We must allow that this response has considerable force against the
individualistic theory: namely, that what a person means is determined
solely by the dispositions of that person. It highlights the fact that
(ordinarily, and subject to certain conditions) individuals are said to mean
by a word whatever that word means in the linguistic community they
belong to-even when their own usage is to some extent improper. It
remains to be seen however whether an analogous objection will work
against the community-use theory.
Certain ways of using the word ‘plus’ are regarded by the community
as correct and certain ways incorrect. And individuals who are sufficiently
in accordance with these standards to qualify as speakers of the language
are said to be mistaken when their behavior violates the norms. Thus, the
identification of meaning with communal use has no difficulty in allowing
for the existence of individual errors. But is it going to be possible for the
community as a whole to be wrong? The answer, it seems to me, is that
this is indeed a problem for some use theories, but not for all. For example,
consider a theory that Kripke mentions:-that

for any m and n, the value of the function we mean by ’plus’ is

(by definition) the value that (nearly) all the linguistic community
would give as the answer. (p. 111)

On this view the community is infallible. But other, more sophisticated,

dispositional theories are possible that would be able to accommodate the
112 Mind & Language
prospect of a community wide mistake. The communal disposition to use
a word in a particular way should not be regarded as simply the disposition
to treat certain sentences as definitely and permanently acceptable and
others not. In addition there are dispositions to sanction varying levels of
confidence (cashed out as ’betting behavior’) in the truth of certain
sentences-where the appropriate degrees of belief are a function of
observable circumstances. In other words there is the practice of revising
epistemological attitudes in specific ways in the light of new evidence,
and of acknowledging that earlier claims were wrong. Thus a ’communal
dispositional’ account of meaning is quite compatible with the recognition
of both individual and communal mistakes. To be sure, there are limits,
even on this account, to what mistakes are possible. But this is not
objectionable unless these limits are transcended by what we actually say;
and, in practice, the appreciation of communal error does seem to be
confined to the type of revision that is covered by the sort of dispositional
theory just mentioned.
A third and related argument of Kripke’s-one to which I alluded in the
previous section-is that the dispositionalist misconstrues the link
between meaning and verbal behavior, taking the relationship to be descrip-
tive when it really is normative. For the import of meaning plus by ‘+’ is
+
not that one will or would assent to ‘68 57 = 125’, but rather that one
ought to assent to it.
This point, while correct in itself, can easily lead to two mistakes. The
first, I have just addressed: it is simply not the case that a communal-use
theory implies infallibility. The second mistake is to suppose that the
normative implications of meaning cannot be reconciled with a naturalistic
reduction-to reason that since, given what Jones means by ‘plus’, ’68’,
+
and his other words, he ought to affirm ‘68 57 = 125’; and since it is
impossible for a purely factual antecedent to imply a normative conse-
quent; then it must be that the assumptions about Jones’s meaning are
not purely factual.
It seems clear that this argument is no good-specifically, that its final
premise is false. For there are numerous cases of obviously pure facts
implying obviously normative conclusions. Consider the fairly uncontro-
versial universal principles:-’Human beings should be treated with
respect’ and ’One should believe the truth’, which entail respectively ’If
Jones is a human being, then he ought to be treated with respect’ and ’If
+
it is true that 68 57 = 125, then one ought to believe it’. The general
point here is that in any normative realm-ethical, pragmatic, aesthetic, or
epistemological-there are bound to be principles specifying the normative
import of a range of non-normative circumstances. And such principles
will entail violations of the crucial premise.
Nor is it hard to see in particular why meaning-construed as communal
usageshould have normative implications. For knowledge is valuable
(both for its own sake and for its practical benefits); so evidently we ought
to strive for it: there are certain propositions-namely, true ones-that we
 Wittgenstein and Kripke on the Nature of Meaning 113
ought to believe. Moreover, given our language, we must express each of
our beliefs in a particular way. Therefore, within a given linguistic practice,
certain sentences ought to be affirmed and others not.
In other words, one ought (other things being equal) to assent to the
truth. In particular, one ought to assent to the proposition that
68 + 57 = 125. Therefore, one ought to assent to any sentence that one
understands to express that proposition. Now, given proficiency and par-
ticipation in a linguistic practice in which the constituents of the sentence
‘68 + 57 = 125‘ are used in certain ways, one will understand that sentence
to express the proposition that 68 + 57 = 125. Therefore, given partici-
pation in such a linguistic practice, one ought to assent to the sentence
‘68 + 57 = 125‘. Thus the communal-use theory has no difficulty in accom-
modating the normative implications of meaning.
Kripke’s fourth objection to the use theory of meaning is based on the
fact that the totality of our actual uses and dispositions to use the word
’plus’ is finite, whereas the set of truths involving the word is infinite. As
a consequence, there are sure to be sums involving numerals so huge that
we can have no disposition either to accept or reject them. Therefore,
given such an indigestible sentence, our dispositions do not settle whether
by ‘plus‘ we mean a function that would make it true, or a function
that would make it false. Moreover this problem arises for community
dispositional views just as sharply as it does for individualistic versions.
To see how we should respond to this point it is vital to work with an
explicit formulation of it. And it seems to me that the underlying reasoning
is as follows:-the meaning of a predicate determines its extension; but
our dispositions concerning its use do not; therefore the meaning cannot
be identified with those dispositions. Now this argument looks fine. For
its premises are plausible, and its form is simply Leibniz’s Law: X and Y
are not identical if X has a property that Y lacks. But further scrutiny
reveals a fatal dilemma, as follows. If ’determination of extension’ is taken
to imply that the extension may be discovered (in principle)-or specified
in certain epistemologically favourable terms-then the first premise is
false. There is simply no reason (except from the perspective of
verificationism) to agree that the meaning of a predicate incorporates a
method of nailing down exactly what it applies to. But if, alternatively,
‘determination of extension’ is taken to imply merely that some characteriz-
ation of the extension is fixed, then the second premise is false. The
dispositions of use that constitute the meaning of ‘F’ trivially determine
that its extension is the set of F’s. Thus there is no sense in which the
meaning of ’F’, but not its use, determines its extension.
Insofar as Kripke denies this, it seems that the basis of his argument-
what explains both its initial appeal and its ultimate failure-is straightfor-
ward verificationism. It is implicitly assumed that the meaning of a predicate
should comprise the sort of necessary and sufficient conditions for proper
application that would allow its extension to be discovered. And it is taken
for granted that when one is armed with an adequate account of the
114 Mind €3 Language
meaning of a predicate, one is in a position to explain why its extension
includes this thing, but not that, by showing how this, but not that,
satisfies the conditions.
There is ample evidence that Kripke does in fact make these assump-
tions. In the first place, the only dispositionalist accounts of the meaning
of ’plus’ that he considers seem designed to embody some way, at least
in principle, of establishing the extension of ’plus’. One example is the
crude communal-use theory that I quoted above in the context of respond-
ing to Kripke’s second argument; another example is the idea (p. 26) that
someone means plus by ‘+‘ if, when asked whether ’x + y = Z’ he answers
yes if and only if x +y = z (where ’x’, ’y’, and ‘z’ name the numbers x, y,
and z, and ‘=‘ means equals). In the second place, the main argument
against these accounts is that, on reflection, they fail to determine the
extension of ’plus’; they do not enable us to establish whether certain very
large, alleged sums are correct or not. Thus what Kripke’s argument really
shows is that a certain subset of dispositional accounts-namely, those
conforming to verificationist strictures-are inadequate.
But the use theory of meaning is by no means committed to verification-
ism, and should not be constrained by it. It is perfectly possible to identify
the meaning of a predicate with dispositions of use that are not sufficient,
even in principle, to enable its extension to be captured in epistemolog-
ically favourable terms.6 It is perfectly possible to reject the demand that
an account of the meaning of a predicate be capable of explaining why its
extension-characterized in epistemologically favourable terms-includes
what it does, rather than something else. Yet it is perfectly possible that
meanings nonetheless do determine extensions-in the sense that any
other predicate with the same meaning would have to have the same
extension. For it is a trivial fact that if a predicate means F-ness, then its
extension is the set of F’s. Consequently, whatever is the proper collection
of dispositions for the use of the predicate determines that its extension
is the set of F‘s. The explanation of why these dispositions determine that
extension is simply that these dispositions are identical with ‘meaning F-
ness’ (which, like other identities, is not susceptible of explanation) and
that whatever means F-ness must have that e ~ t e n s i o nThus,
.~ contrary to
Kripke’s assumptions, both the meaning and the use qf ’F’ determine

The dispositions for the use of ‘plus’ that constitute its meaning are the basic ones-
those which, together with the meanings of other expressions, will provide a complete
account of our deployment of the word. The set of such basic uses will be somewhat
indeterminate: its boundaries will be unclear. But this simply mirrors the fact that it
often is unclear whether some new deployment of a word amounts to a change in its
meaning or not.
The present response to Kripke’s fourth argument involves a certain ‘deflationary’
theory of predicate satisfaction: namely, that the determination of extension by meaning
is fully characterized by the schema, ’x means F-ness + the extension of x is the set
of F’s’. This point of view (together with parallel accounts of truth and reference) is
elaborated and defended in my Truth, Oxford: Blackwell, 1990.
 Wittgenstein and Kripke on the Nature of Meaning 115
that it applies to the set of F’s; but neither need provide any alternative
specification of what it applies to.
A final argument against the use theory, apparently given some weight
by Kripke (pp. 774, is based on the idea that we should abandon the
traditional conception of the meaning of a sentence as consisting of its
truth conditions in favour of a conception of meaning as consisting of
assertibility conditions, i.e. circumstances of appropriate use. The implicit
line of thought appears to be that since the meaning of the sentence ’Jones
means plus by ”plus”’ is given by its assertibility conditions, we should
not expect it to have any non-trivial definitional analysis. That is to say,
we should not expect there to be any collection of more basic facts-e.g.
facts about usage-in virtue of which it is true or false.
The trouble with this reasoning is that it does not quite take us to the
supposed conclusion:-namely that there are no facts to which attributions
of meaning correspond. We may concede the superiority of ’assertibility
conditions‘ over ’truth conditions’ as the central notion in terms of which
meaning should be explicated. But all we are entitled to infer from this
concession is that attributions of meaning need not-as far as their own
meaningfulness is concerned-have non-trivial truth conditions. It does
not follow that they do not have such truth conditions. On the contrary,
excellent reason to believe that they do have them is provided by an
examination of the very assertibility conditions that Kripke suggests.

Jones is entitled, subject to correction by others, provisionally to

say, ’I mean addition by “plus”‘, whenever he has the feeling of
confidence-*now I can go on!’-that he can give ‘correct’
responses in new cases; and he is entitled, again provisionally and
subject to correction by others, to judge a new response to be
’correct’ simply because it is the response he is inclined to give.
(P. 90)

Notice that the references to ‘correction by others’ are vital if Jones is

going to have the right to think that his word has the same meaning as
‘plus’ does in English. Therefore what Kripke is maintaining, in effect, is
that Jones is entitled to say ’I mean addition by “plus”’ whenever he is
entitled to believe that his responses more or less coincide with those of
others- i.e. that his dispositions to use the word match the communal
use. But the obvious explanation of this fact is that Jones may rightly infer
that he means addition by ‘plus’ from his belief that his use of the word
matches the community’s. Thus the ’communal-use’ truth conditions for
attributions of meaning provide a natural explanation of (and are certainly
not precluded by) the assertibility conditions that Kripke endorses.
Having examined all Kripke’s arguments, both explicit and implicit,
for the irreducibility of meaning, I conclude that his sceptical thesis is
unjustified. He has given no good reason to reject Wittgenstein’s view of
what the meaning of a word consists in-
116 Mind 6 Language
For a large class of cases-though not for all-in which we employ
the word ‘meaning‘ it can be defined thus: the meaning of a word
is its use in the language. (Philosophical Investigations, Section 43)

III. Implications for Linguistics and Psychology

Now I would like to turn to the question of what bearing Kripke’s line of
thought has on linguistics and psychology. He maintains that there is
something wrong with theories that would explain an individual’s
behavior by hypothesizing that the individual is following (perhaps
unconsciously) a certain set of rules. The alleged problem with such theor-
ies can be derived from Kripke’s sceptical conclusion about meaning. For
insofar as a rule is followed, it must exert some causal influence; so it must
have some sort of formulation. Therefore a rule is some sentence-like object
together with an interpretation. Therefore the rule one is following
depends on the meaning one gives to the formulation of the rule. Therefore,
if no facts correspond to meaning this rather than that by the formulation
of the rule, then no facts correspond to following this rule rather than that
one. But scientific explanations are restricted to statements that purport to
express facts. So if there are no facts in virtue of which a person follows
this or that rule, then attributions of definite rule following should not
appear in science.
In evaluating this position it is a good idea to consider it in the context
of the general question of whether a priori philosophical considerations
can legitimately constrain the development of a science. History is certainly
replete with philosophical complaints about specific scientific proposals-
complaints, for example, about action at a distance, atoms, curved space,
unconscious mental states, etc. But, as these examples indicate, such
attempts at interference do not succeed and eventually may come to seem
embarrassingly misguided. Typically the new theory uses some piece of
familiar terminology in a very unfamiliar way and contrary to principles
that are so extremely entrenched as to seem self evident. As a result there
is a tendency to allege that the new theory is absurd, self-evidently false,
perhaps even self-contradictory. It is forgotten that scientific theories are
evaluated solely on the basis of their observational adequacy, overall
simplicity, internal consistency and coherence with the rest of science.
Perhaps conformity with past usage counts for something, but major
revisions can be justified if they allow large enough gains in observational
adequacy and/or simplicity. In the face of any such radical revisions the
most that can be concluded is that the meanings of certain terms have
changed in the new theoretical context. But this is no objection to the new
theory. Thus a general mistake underlying much philosophical criticism
of science is to suppose that a word must preserve its old meaning in the
context of a very different theoretical employment.
 Wittgenstein and Kripke on the Nature of Meaning 117
As blatant as it is, when spelled out, this mistake has nevertheless been
extremely prevalent. In particular, it seems to infect Kripke’s criticism of
the way the concept of rule following is used in linguistics. For even if
rule following is ordinarily attributed in relation to a community, and not
to an individual considered in isolation, and even if its use were to have
connotations that are not entirely descriptive, there is no reason why the
notion cannot be stretched and refined in the context of a linguistic theory.
In fact it appears that such transformations have indeed occurred-varying
in extent from one theory to another. Take for example Noam Chomsky’s
recent account in Knowledge of Language.8 Although there is much talk of
‘rules’ and ‘rule following’, it is far from obvious that these terms are used
in such a way that the familiar distinction between ’following a rule’ and
‘conforming to a rule’ can be made. It seems, rather, that ’linguistic rules’
are simply laws of a certain low degree of generality that govern the activity
of one part of the mind/brain. Consider for example the Empty Category
Principle, which regulates the permissible ’distance’ between a moved
element and its trace. Nothing needs to be influenced by representations
of this law, let alone to understand it-any more than the planets need to
be guided by representations of General Relativity Theory. In that case,
Kripke’s problem cannot arise.
The notion of ’rule’ will not be the only ordinary concept to undergo
some modification in the context of linguistics. From the scientific point
of view it may be valuable to regard public languages, such as English, as
uninteresting entities that supervene on the ‘languages’ of individual
person^.^ Thus the term ’language’ will refer to an internalized idiolect.
Moreover, in accordance with this shift in perspective, one would want
an individualistic notion of ’meaning’. In this sense, what a person means
by ’plus’ would be some property of his mind that (together with other
factors) gives rise to his peculiar ways of using the word. Notice that this
would be a ’private language’ in Kripke’s sense-but certainly not in
Wittgenstein’s. Wittgenstein’s objection is to a language whose

individual words . . . are to refer to what can only be known to

the person speaking; to his immediate private sensations. So
another person cannot understand the language. (Philosophical
Investigations, Section 243)

The ’ostensive‘conception of meaning involved here goes well beyond the

idea of meaning as individualistic. In particular, there is no reason to
suppose that individualistic meanings cannot, in general, be grasped by
others. Consequently, it is quite possible to attack and reject the ostensive

Noam Chomsky, Knowledge of Lnnguage, New York: Praeger, 1986.

9McCinn (op. cit.) shows that Wittgenstein did not try to argue that ‘meaning’ and
’rule-following’ are essentially communal.
118 Mind 6 Language
model-as Wittgenstein does-without impugning other individualistic
conceptions such as the one just mentioned.
In assessing such linguistic theories, the real issue of acceptability hangs
on whether they possess the usual theoretical virtues familiar from other
sciences: namely, predictive success, explanatory power, simplicity, etc. If
those virtues are present, then no further defence of the terms ’rule follow-
ing’, ’language’, and ’meaning‘ is called for. Moreover, the question
whether the new use of such terms is sufficiently similar to their use in
ordinary language to say that their meanings have not been changed is a
question which, even if answerable at all, has no importance whatsoever
for linguistics.
It is ironic that Kripke should saddle Wittgenstein, of all people, with
responsibility for a critical attitude towards the use of the notion of rule
following in linguistics. Quite on the contrary, Wittgenstein would regard
such an outlook as the height of philosophical folly. His central idea was
that philosophy should merely describe what is done in order to facilitate
the elimination of philosophical confusion, and that it cannot interfere
with common sense or with uncorrupted scientific practice.

ZV. Wittgenstein’s Point of View

One of the reasons that Wittgenstein’s late philosophy is so often said to

be obscure is the persistent attempt to extract from his writings metaphys-
ical theses when there are none to be found. Never mind that he repeatedly
emphasises his opposition to philosophical theorizing and says that such
projects are based on confusion. These remarks are ignored or treated as
rhetorical exaggerations. Perhaps it seems incomprehensible that a great
philosopher could write an important philosophy book without opinions
on the traditional questions. Thus it is widely accepted that Wittgenstein’s
views are exceptionally elusive-and the issue becomes whether or not
this is his fault. On the sympathetic side, there has arisen a tradition of
Wittgenstein interpretation according to which his ideas, though profound
and important, are essentially unstatable in clear terms.
Kripke endorses this interpretive stance, offering the following expla-
nation of why it is so difficult, and perhaps impossible, to give a precise
formulation of Wittgenstein’s views. Consider claims to the effect that
there are numbers, mental states, facts about meaning, etc. Each such
thesis, Kripke says, has both an ordinary construal and a special philo-
sophical construal. In their ordinary senses, Wittgenstein has no quarrel
with any of these theses. Of course there are numbers, and so on. It is in
their philosophical senses that Wittgenstein disagrees with them. How-
ever, it turns out that there are insuperable difficulties in formulating the
philosophers’ theses. Part of the problem is that whatever is said by way
of explanation may itself be construed either ordinarily (which would be
uncontroversial and add nothing) or philosophically (which would merely
 Wittgenstein and Kripke on the Nature of Meaning 119
call for further elucidation). For example, the mathematical Platonist, who
says ’Numbers exist’, tries to amplify his view with the assertion that
‘Numbers are a special type of entity outside space and time and indepen-
dent of human thought’. Taken straightforwardly, this is obviously true,
but does not help to explain the Platonists’ initial assertion. Taken philo-
sophically, it is just as obscure as the claim it is designed to clarify. Thus
the philosophers’ theses cannot be captured. Therefore, Wittgenstein’s
views-the denials of these theses-cannot be captured either. That is
why, when Kripke attributes to Wittgenstein the view that there are no
facts about what we mean, he warns the reader that Wittgenstein would
not have been content with that characterization of his position, but that
this is the best that can be done.
I disagree with Kripke, and with the general idea that there is something
inevitably unstatable in Wittgenstein’s philosophy. I think that the appar-
ent difficulty in formulating Wittgenstein’s views derives from the fact
that they do not take the form of metaphysical theses-including denials
of the theses of others. Instead his central claim is that the traditional
puzzles and questions are illfounded, stemming from confusion rather
than ignorance. Kripke is right to say that Wittgenstein has no objection
to the existence of numbers, pains, meanings, etc. He goes wrong, in my
view, when he looks for a different way-a phiIosophica1 way-of constru-
ing these claims, such that Wittgenstein would deny them. In fact what
Wittgenstein objects to is the puzzlement itself-the queerness, the peculi-
arity, that philosophers have found in commonplace facts. This sense of
oddness can give rise to philosophical theories designed to ‘solve’ the
’problems’-behaviorism and intuitionism, for example. But these are
theories that, Wittgenstein would say, are the product of muddled think-
ing, and so they should not be taken seriously. Note however that this is
not to argue that they are false or that it will never be rational to accept
them; for who knows what science will bring. But empirical reasons of a
kind quite different from those given by the confused philosopher would
have to be provided- and that can never be ruled out. Notice also that
there is no conflict between Wittgenstein’s metaphilosophy-his rejection
of philosophical theory-and the fact that he does make various claims-
for example, that meaning is use, and that philosophical problems arise
in a characteristic way. Wittgenstein’s objection is to the traditional con-
ception of philosophy as a field capable of penetrating deep mysteries and
producing a priori knowledge of the most profound aspects of reality. He
has no objection to the psychological and linguistic observations that will
help us to diagnose and cure such ways of thinking. Thus Wittgenstein’s
main aims are to point out the illegitimacy of certain problems and sol-
utions. His conclusions seem elusive only if one is looking for something
more.
In the case of meaning, Wittgenstein’s method is, first, to exhibit-
almost to wallow in-the way that we feel meaning to be an extraordinarily
bewildering phenomenon; second, to expose the general philosophical
120 Mind 6 Language
dispositions and particular linguistic structures that engender that feeling;
and third, to neutralize these sources of bewilderment by bringing them
to light.
There is not the space here to describe how these things are actually
done; but let me at least try to convey some of the flavour of his project,
in order to contrast it with Kripke’s. Here is how Wittgenstein proceeds
in the Blue Book’O, beginning with a characteristically acute description of
our puzzled state of mind:

We are tempted to think that the action of language consists of

two parts; an inorganic part, the handling of signs, and an organic
part, which we might call understanding these signs, meaning
them, interpreting them, thinking. These latter activities seem to
take place in a queer kind of medium, the mind; and the mechan-
ism of the mind, the nature of which, it seems, we don‘t quite
understand, can bring about effects which no material mechanism
could. Thus e.g. a thought (which is such a mental process) can
agree or disagree with reality; I am able to think of a man who
isn’t present; I am able to imagine him, ‘mean him’ in a remark
I make about him, even if he is thousands of miles away or dead
’What a queer mechanism’ one might say ‘the mechanism of
wishing must be if 1can wish that which will never happen’. (Blue
Book, p. 3)

Wittgenstein goes on to note that the meaning of a sign is its use. He

attributes the mystery of meaning to our failure to recognize this fact and
then attempts to explain this failure. Thus he says:

The mistake we are liable to make could be expressed thus: We

are looking for the use of a sign, but we look for it as though it
were an object co-existing with the sign (One of the reasons for
this mistake is again that we are looking for a ’thing corresponding
to a substantive’). (Blue Book, p. 5)

In other words, expressions such as ’The meaning of ”table”’, ’He doesn’t

mean what he says’, etc. deceive us into thinking of a meaning as an entity
that we associate with a word and that has the amazing capacity to reach
into the world to pick out its object and determine what would be correct
usage of the word.
What remains is to undermine the attractiveness of this way of thinking.
This is done mainly by exposing its origins and showing that they consist
in false analogies, crude overgeneralizations, and other characteristically
philosophical bad habits of thought-habits which begin to loose their

Ludwig Wittgenstein, The Blue and Brown Books, Oxford: Blackwell, 1958.
 Wittgenstein and Kripke on the Nature of Meaning 121
grip as soon as they are brought out into the open.
To summarize, I have mentioned five respects in which I think Kripke’s
interpretation gives a distorted picture of Wittgenstein’s philosophy. First,
Wittgenstein does not deny that attributions of meaning are made true or
false by natural facts. On the contrary, he says that they concern facts
about the use of words. Second, Wittgenstein would oppose the idea that
philosophy can govern the direction of scientific research. In particular,
he was quite critical of the philosopher’s tendency to be mesmerized by
the ordinary use of a term and then to reject as incoherent some scientific
theory which involves a different use. (See his discussion of ‘unconscious
pain’ in the Blue Book.) Third, Wittgenstein does not argue against the
possibility of an ’individualistic language’. The target of his private langu-
age argument is, more specifically, the prospect of words that mean nothing
to others because they are defined ostensively by the speaker to refer to
incommunicable aspects of his experience. Fourth, Wittgenstein’s ideas
are not hard, let alone impossible, to formulate. The impression of obscur-
ity comes from not taking seriously enough his radically anti-theoretical
metaphilosophical outlook. Finally, Wittgenstein does not invent a paradox
and then develop a sophisticated theory of meaning to resolve it. Rather
he addresses the fact that we do regard meaning as mysterious, and his
aim is to articulate the sources of this sense of mystery, and to show
thereby that they are illegitimate and avoidable.”

Department of Linguistics and Philosophy

Massachusetts Institute of Technology
Cambridge, M A 02139

I would like to thank Ned Block, Warren Goldfarb, Itziar Laka, Charles Marks, Gabriel
Segal and Meredith Williams for their helpful comments on an earlier draft of this
paper.