Is Meaning Normative?
ANANDI HATTIANGADI
Abstract: Many people claim that semantic content is normative, and that thereforenaturalistic theories of content face a potentially insuperable difficulty. The normativityof content allegedly undermines naturalism by introducing a gap between semantic‘ought’s and the explanatory resources of naturalism. I argue here that this problem isnot ultimately pressing for naturalists. The normativity thesis, I maintain, is ambiguous;it could mean either that the content of a term prescribes a pattern of use, or that itmerely determines which pattern of use can be described as ‘correct’. For the anti-naturalist argument to go forward, content must be prescriptive. I argue, however, thatit is not. Moreover, the thesis that content supplies standards for correct use isinsufficient to supply a similar, a priori objection to naturalism.
Many philosophers find irresistible the thesis that meaning is normative (Baker and
Hacker, 1984; Bloor, 1997; Brandom, 1994; Boghossian, 1989; Glock, 1994,
1996; Kripke, 1982; Lance and O’Leary Hawthorne, 1997; McDowell, 1993,
1998; McGinn, 1984; Millar, 2004; Miller, 1998; Pettit, 1990; Wright, 1980,
1984). They are moved, quite often, by the following reasoning. Meaningful
words have correctness conditions. The English word ‘green’ applies correctly to
something if and only if it is green. Furthermore, the fact that a speaker means green
by ‘green’ determines not that she will apply ‘green’ to something if and only if it is
green, but that she ought to do so. If she fails to apply ‘green’ correctly, she will
have failed to speak as she should. What someone means by a term thus determines
how she ought to use it.
If this reasoning is cogent, anyone committed to the assumption that meaningful
terms have correctness conditions seems equally committed to the normativity of
meaning. This poses a problem especially, though perhaps not exclusively, for
naturalists about meaning. By ‘naturalist’, I refer to those who do not think
meaning, intentionality or normativity can be among the building blocks of the
universe (Fodor, 1990). Naturalists seek to explain what it is for someone to mean
something by a term without making further appeal to normative, semantic or
intentional facts. The trouble is that naturalists trade in ‘is’ statements, whereas
meaning is allegedly fraught with ‘ought’s; and on the face of it, there seems to be a
yawning gap between ‘is’ and ‘ought’ or between the two distinct spaces—the
normative space of reasons and the causal space of nature (McDowell, 1998,
p. 392; Sellars, [1956] 1997).
Given its potency, the thesis that meaning is normative has invited some
suspicion (Bilgrami, 1993; Dretske, 2000; Fodor, 1990; Gluer, 1999; Gluer,
2001; Gluer and Pagin, 1999; Papineau, 1999 and Wikforss, 2001). Yet, most
Address for correspondence: St. Hilda’s College, Cowley Place, Oxford OX4 10Y, UK.Email: [email protected]
Mind & Language, Vol. 21 No. 2 April 2006, pp. 220–240.# 2006 The AuthorJournal compilation # 2006 Blackwell Publishing Ltd
people assume, without sufficient argument, that meaning is normative. Even those
who aim to naturalise semantic normativity often assume, along with those who
oppose naturalism, that meaning is normative; though they maintain that the
normativity of meaning can be naturalised (Blackburn, 1993; Dretske, 1986;
Millikan, 1995).
Against this overwhelming consensus, I will argue that we have no good reason
to believe that meaning is normative in a sense that justifies a presumption against
naturalism. In the next section, I will present what I take to be the most intuitive
argument in favour of the normativity of meaning, and will argue that it equivo-
cates between one sense of ‘normative’ that is no trouble for naturalism and a sense
of ‘normative’ as ‘prescriptive’, that does yield a presumption against naturalism. In
section 2, I will present some general arguments against the assumption that
meaning is prescriptive. In subsequent sections I criticise what I take to be the
most compelling reasons that might be given in favour of the thought that meaning
is prescriptive.
1. Correctness and Prescriptivity
What is a normative statement? Here are a few un-contentious examples: ‘you
ought to give some of your income to charity’; ‘torturing innocent people is
wrong’; ‘keeping your promises is right’. In contrast, here are some un-contentious
examples of descriptive statements: ‘snow is white’; ‘the earth is flat’; ‘elephants
never forget’. What seems special about normative statements, what distinguishes
them from descriptive statements, is that normative statements tell us what to do,
whereas non-normative statements simply describe how things are.
G. E. Moore ([1903] 1993) and David Hume ([1739] 1978) have both argued,
in different ways, that this fundamental difference between normative and non-
normative concepts and statements makes normative concepts and statements
irreducible to natural concepts and statements. Moore said that any attempt to
reduce normative concepts to natural concepts is to commit the Naturalistic
Fallacy. Hume argued that you cannot validly derive a normative statement from
a consistent set of descriptive statements. Both of these arguments are controversial,
and I cannot go into the controversies here (cf. Darwall, Gibbard and Railton,
1992; Hudson, 1969; Miller, 2003). Rather, I will simply grant the anti-naturalist
the assumption that some such arguments can be made good and ask instead
whether meaning is normative in the relevant sense.
There are few explicit arguments for the normativity of meaning. Most ‘nor-
mativists’ assume the thesis as a platitude, part of our intuitive or pre-theoretic
picture of meaning and understanding (Brandom, 1994; Boghossian, 1989; Glock,
1996; McDowell, 1993; McGinn, 1984; Kripke, 1982; Pettit, 1990; Wright, 1980,
1984). Since I do not share this intuition, I would like to try to reconstruct an
argument for the normativity of meaning that starts with assumptions we can all
share. The argument does, I think, capture the line of reasoning that leads many
Is Meaning Normative? 221
# 2006 The AuthorJournal compilation # 2006 Blackwell Publishing Ltd
people to assume that meaning is normative. However, as it turns out, the
argument equivocates between two senses of the claim that meaning is normative,
one of which is innocuous, and the other of which is damaging to naturalism.
Intuitively, it is a necessary condition for a term to have a meaning that it has
conditions of correct application. More precisely, where t is a term, F it’s meaning,
and f the feature or collection of features in virtue of which F applies, what we
might call the Meaning Platitude says:
Meaning Platitude: t means F ! (x)(t applies correctly to x $ x is f )1
The expression ‘applies correctly’ is a placeholder for the various semantic relations
a term can have to the world: ‘t applies correctly to x’ stands for ‘t refers to x,’
‘t denotes x,’ or ‘t is true of x’. Thus, if t means green, t is true of all and only green
things; if t means Socrates, it denotes Socrates and only Socrates. In short, we might
say, according to MP, the meaning of a term can be expressed by its correctness
conditions. Many people take MP to be a platitude because it hardly seems open to
us to deny that in order to have meaning, terms must have correctness conditions.
This is what distinguishes the use of language from the making of mere noise. Of
course, MP has been denied. In philosophy nothing is uncontroversial. However,
those who deny MP typically do so on the force of sceptical arguments (Quine,
1960; Kripke, 1982). Moreover, most of those who seek to naturalise meaning
seem to subscribe to some form or other of MP (Dretske, 1986, 2000; Fodor, 1990;
Millikan, 1984, 1995).
As I have formulated it, MP does not mention a speaker. The thought that
meaning is normative, however, is the thought that meaning is normative for a
speaker. Since MP does not imply that meaning is normative, where does the
assumption that meaning is normative come from? The answer is that it comes
from another, allegedly intuitive assumption: that a speaker who means something
by an expression must be following a rule (norm) for its correct application.
Sometimes, to say that meaning is normative just is to say that for someone to
mean something by a word, she must be following a rule. One meaning of the
word ‘normative’ is ‘pertaining to a norm or a standard’, so if a speaker must follow
a rule in order to mean something by a term, then, the reasoning goes, meaning
must be normative. But the fact that a speaker follows a rule does not, in itself,
make meaning normative in the sense that is required for the Humean and
Moorean arguments to take hold.
The reason is that there are different kinds of rules that a speaker might be said to
be following, in order to mean something by an expression (Gluer and Pagin,
1999). But it is only if the rules speakers follow are prescriptive, it’s only if they tell us
1 I am assuming a truth-theoretic interpretation of MP throughout, but it could just as wellbe given an assertion-theoretic interpretation. See Gluer, 1999 for further discussion of theassertion-theoretic version and normativity. I will stick to the truth-theoreticinterpretation for simplicity and because it is one that many naturalists hope to capture.
222 A. Hattiangadi
# 2006 The AuthorJournal compilation # 2006 Blackwell Publishing Ltd
what to do, that the Humean and Moorean arguments could be made to apply.
However, very often, the thought that meaning is normative is presented as the
thought that a speaker must be following a rule that simply specifies the conditions
of correct and incorrect use of an expression. For instance, Boghossian says that
‘the normativity of meaning turns out to be, in other words, simply a new name
for the familiar fact that . . . meaningful expressions possess conditions of correct use’
(1989, p. 513). A rule that specifies the correct use of a term (assuming MP) is the
following:
R1: (x)(t applies correctly to x $ x is f ).
Though R1 is indeed a norm, the fact that I must follow it does not make meaning
normative in the appropriate sense. R1 simply states the correctness conditions of
an expression; it does not tell me what to do (Gluer, 1999; Gluer and Pagin, 1999).
Rather, if I am following R1, then R1 supplies a description of my uses of t as those
that are ‘correct’ and those that are ‘incorrect’. Furthermore, recall that the
correctness conditions of a term are just the conditions under which the term refers
to or is true of an object to which it is applied. R1 just tells the speaker what these
conditions are. For example, in the case of ‘square’, a rule of the form of R1 would
state that for all x, ‘square’ refers to x if and only if x is square. It does not follow
from the fact that I am following such a rule that I ought to apply ‘square’ to
something if and only if it is square.2 The mere fact that I am following such a rule
implies only that if I apply ‘square’ to something square, my application can be
described as ‘correct’, i.e., in accordance with the rule that I am following, and if I
apply ‘square’ to something non-square, my application can be described as
‘incorrect’, i.e. failing to accord with the rule that I am following.
On this interpretation of ‘normative’, to say that meaning is normative is to say
that when I mean something by an expression, I must be following a rule that
distinguishes between those uses that accord with the meaning and those which do
not. This can be captured by the principle I will call Correctness.
2 One might argue that for me to mean something by an expression, a rule of the form ofR1 must be ‘in force’ (Gluer and Pagin, 1999). If R1 is ‘in force’ for a speaker, then it is‘binding’; the speaker ought to aim to comply with R1 (thanks to an anonymous refereefor this suggestion). That is, R1 will tell me which uses of t are correct, and an additionalrule will tell me that I ought to aim to use t correctly, in accordance with R1. Thisadditional rule will, indeed, be normative—it tells me what I ought to do—even thoughR1 remains descriptive. However, if meaning something by a term implies that I ought toaim to comply with a rule like R1, then this view will suffer precisely the same objectionsas Prescriptivity, which I discuss in the following. Moreover, there is a more usual sense inwhich we say that, for example, a legal rule is in force—i.e., in the sense that it isaccepted and enforced by sanctions. So, to say that a rule like R1 is ‘in force’ need not beto say that one ought to comply with it, but merely that it is a rule that is accepted, in alinguistic community, for example, and non-compliance with it will result in sanctions.These are all descriptive statements, as are the statements that are made true by the factthat R1 is ‘in force’ (i.e. regarding which actions accord with R1 and which do not).
Is Meaning Normative? 223
# 2006 The AuthorJournal compilation # 2006 Blackwell Publishing Ltd
Correctness: S means F by t ! (x)(S applies t ‘correctly’ to x $ x is f ).
Where S is a speaker, and the other terms are as defined above. If the slogan
‘meaning is normative’ just is Correctness, then meaning is not normative in the
sense that would engage Hume’s and Moore’s arguments—for those arguments to
take hold, meaning-determining rules must tell the speaker what to do. Assuming
Correctness, however, the fact that I mean green by ‘green’ is merely the fact that
when I apply ‘green’ to something green, my use can be described as ‘correct’ and
when I apply ‘green’ to something non-green, my use can be described as
‘incorrect’.
It might be argued that ‘correct’ is a normative term, so Correctness itself
constitutes a presumption against naturalism. Indeed, many philosophers take the
statement that some use of an expression is correct to imply that it is the use that a
speaker who grasps the meaning of an expression ought to make that application of
the expression (e.g. Boghossian, 1989, p. 509; Brandom, 1994, p. 27). However,
deontic terms, such as ‘correct’ and ‘right’, are not always normative. Sometimes,
to say that something is right does not imply a prescription; rather, it is to say that it
meets a certain standard. For example, think of theme parks where there is a
minimum height requirement for some of the more dangerous rides. This is a
standard children must meet if they are to go on the ride. But however happy
Niblet may be to meet the standard, whether or not she does is a straightforwardly
non-normative, natural fact—it is the fact that she is four feet tall. We might say
that Niblet is the ‘right’ height or the ‘correct’ height, but this is clearly not to say
that it is a height she ought to pursue quite independently of any of her desires.3
Rather, it is simply to say that she is four feet tall. In contrast, if I say something like
‘the right action is the one that maximises overall happiness’, I mean that such
actions ought to be taken even in the absence of any relevant desire.
It is clear, however, that we should not assume that the ‘correct’ in Correctness
has prescriptive force (Gluer, 1999; Gluer, 2001; Soames, 1998; Wikforss, 2001
and Wilson, 1994). Given MP, it seems more plausible to take ‘correct’ to have
descriptive force. Remember that in the initial formulation of MP, ‘correct’ is just a
catch all phrase for the various semantic relations terms can have to the world. To
say that some use of a term is ‘correct’ is thus to say that it accords with an
application rule that specifies the conditions under which it refers—it is to say that
the term refers to or is true of the thing to which it has been applied. If we keep this
firmly in view, it no longer makes sense to treat ‘correct’ prima facie as a
prescriptive term. Consider, for instance, the statement that the word ‘square’
applies correctly to a particular table top. This is not a normative statement—it is
not equivalent to the statement that you ought to say that the table top is square
3 It may be true that if Niblet wants to go on the ride, she ought to aim to be four feet tall,but this is a hypothetical means/end imperative, where being at least four feet tall is ameans to satisfying her desire to go on the ride. However, hypothetical means/endimperatives are not troubling for the naturalist. I discuss this further in Section 3.
224 A. Hattiangadi
# 2006 The AuthorJournal compilation # 2006 Blackwell Publishing Ltd
quite independently of any desire (such as the desire to tell the truth, for instance);
rather, it is equivalent to the straightforwardly non-normative judgement that the
table top is square and that, in being square, it meets the standard for the correct
application of the word ‘square’. To say that some use of a term is ‘correct’ is thus
merely to describe it in a certain way—in light of the norm or standard set by the
meaning of the term.
For meaning to be normative in the required sense, the slogan ‘meaning is
normative’ must be glossed as ‘meaning is prescriptive’. It is this sense of ‘norma-
tive’ that is crucial in distinguishing normative from non-normative statements in
Hume’s and Moore’s arguments, so anyone who claims that meaning is in principle
irreducibly normative needs to assume that meaning is normative in this sense.
Kripke certainly suggests that meaning is fraught with ‘ought’s. He says, for
instance: ‘Ordinarily, I suppose that, in computing ‘‘68 þ 57’’ as I do . . . I follow
directions I previously gave myself that uniquely determine that in this new
instance I should say ‘‘125’’’ (Kripke, 1982, p. 10, emphasis added). And again, he
says, ‘I feel confident that there is something in my mind—the meaning I attach to
the ‘‘plus’’ sign—that instructs me what I ought to do in all future cases. I do not
predict what I will do . . . but instruct myself what I ought to do to conform to the
meaning’ (Kripke, 1982, pp. 22–21). Later, he says, ‘The point is not that, if I
meant addition by ‘‘þ’’, I will answer ‘‘125’’, but that, if I intend to accord with my
past meaning of ‘‘þ’’, I should answer ‘‘125’’’ (Kripke, 1982, p. 37). On this view,
the rule that I must be following in order to mean addition by ‘plus’ is a prescriptive
rule, such as R2:
R2: (x)(apply t to x $ x is f !).
Unlike R1, R2 prescribes a course of action; it tells me what I ought to do. If
meaning something by t requires that I follow a rule like R2, and if following a rule
implies that one ought to do what it tells one to do, then in order to mean
something by t, I ought to to apply t to x $ x is f. On this construal of ‘meaning
is normative’, if I mean green by ‘green’, I ought to say that something is green if
and only if it is. This thought is captured by the following principle:
Prescriptivity: S means F by t ! (x)(S ought (to apply t to x $ x is f )).4
4 Prescriptivity should not be expressed so as to have narrow scope, that is, as follows:S means F by x ! (a)(S ought (to apply x to a) $ a is f ). So formulated, the principlestates that if S means F by x then, under certain conditions (namely, if and only if a is f ), Shas an obligation (to apply x to a). In contrast, as it is formulated in the main text, the‘ought’ has wide scope. This is to say that if S means F by x, then S has an obligationwith a conditional content; that is, she has an obligation (to apply x to a $ a is f ). Theformulation with wide scope seems to better capture the suggestion that if a speakermeans something by an expression, she incurs an obligation—to use the expressioncorrectly, given its meaning. Krister Bykvist pointed out this scope ambiguity to me.
Is Meaning Normative? 225
# 2006 The AuthorJournal compilation # 2006 Blackwell Publishing Ltd
If meaning is normative in the sense expressed by Prescriptivity, the normativity of
meaning will pose a serious problem for naturalism. Crucially, Prescriptivity engages
Hume’s Law—that you cannot derive an ‘ought’ from an ‘is’—because according
to Prescriptivity it is a necessary condition for someone to mean, say, square by
‘square’ that she ought to apply ‘square’ to all and only squares. If Prescriptivity is
true, there may turn out to be a gap between the explanatory resources of
naturalism, appropriate to the ‘space of nature’, and meaning, which remains
inaccessible in the ‘space of reasons’. Naturalists will face the additional and
potentially intractable difficulty of bridging that gap.
In contrast, even if Correctness is true, meaning still remains firmly on the ‘is’ side
of the ‘is’/‘ought’ gap because what someone means only makes true a special kind
of description of her behaviour. Hence, Correctness does not yield a potential objec-
tion to naturalism a priori. Of course, it may turn out to be difficult for naturalists
to accommodate Correctness—because they will need to find a natural relation that
determines the correctness conditions of an agent’s terms (Dennett, 1998; Fodor,
1990; Stich and Warfield, 1994). Yet, difficult though this may prove, it is just the
familiar problem of explaining intentionality naturalistically (Fodor, 1990;
Wikforss, 2001).
As I will argue in the following, naturalists have no need to worry about the
normativity of meaning. Prescriptivity, though potentially troubling for naturalism,
is false. In the next section, I will consider some general objections to Prescriptivity.
In the following sections, I address some arguments that might be made in defence
of Prescriptivity. Though I cannot consider all possible arguments for Prescriptivity,
I will consider those that are either the most prominent or the most compelling,
and will argue that they fail to show that meaning is fraught with ‘ought’s.
2. Prescriptivity
The Prescriptivity Principle, as I have presented it, says that if I mean horse by ‘horse’,
I ought to apply ‘horse’ to something if and only if it is a horse. My semantic
obligation has this biconditional content because Prescriptivity was initially presented
as an attempt to capture what it is for a speaker to mean something by an
expression, assuming MP. Since MP says that t applies correctly to x if and only
if x is f, Prescriptivity says that if someone means something by an expression she
incurs an obligation with biconditional content. However, on this construal,
Prescriptivity seems to be too strong. As a prerequisite for carrying out her obligation
to apply her words correctly, a speaker will have to ensure that:
(1) (x) (S applies t to x $ x is f )
(2) (x) (x is f ! S applies t to x)
Unpacking the biconditional highlights a difficulty. Though an agent may be able
to ensure that (1) is true, she surely cannot ensure that (2) is true. In many cases, it
226 A. Hattiangadi
# 2006 The AuthorJournal compilation # 2006 Blackwell Publishing Ltd
is not up to her whether a is f; so, in order to make (2) true, she would have to
apply t to every a that is f. That is, if she meant horse by ‘horse’, in order to carry
out her semantic obligation, she would have to apply ‘horse’ to all the horses in the
universe. Because ‘ought’ implies ‘can’, (2) is not just too demanding, it is false.
This difficulty should already arouse our suspicions. Unlike Prescriptivity,
Correctness can be formulated so as to capture MP: That is, according to
Correctness, if I understand ‘horse’ to mean horse, ‘horse’ will apply correctly (i.e.
refer) to all and only horses. There is no analogous difficulty in saying that the
correct application of ‘horse’ outstrips what I can do—‘correct’ or ‘refers’ does not
imply ‘can’.5 Thus, it seems as though Correctness can capture what it is for a
speaker to mean something by a word, whereas Prescriptivity cannot.
Since Prescriptivity is clearly too strong, perhaps we should assume a weaker
principle, such as the following:
Prescriptivity*: S means F by t ! S ought: (x)(S applies t to x $ x is f ).
According to Prescriptivity*, a speaker who means something by an expression
ought to ensure that she uses it only when it is correct to do so. But
Prescriptivity* is also false: it is not the case that in order for someone to mean
something by a term, it is necessary that she ought only to use it correctly. Under
some circumstances, I might be obligated to tell a lie, which does not imply that I
mean something non-standard by my expressions. For example, I want to teach my
son never to touch the oven, so I say ‘the oven is hot’ even when it is cool. Indeed,
given the advantages of my son’s avoiding the oven, I ought to say to him ‘the oven
is hot’ even when it is not. In this case, I ought to apply ‘hot’ to something that is
cool, which does not imply that I do not mean hot by ‘hot’.
The general point against Prescriptivity is that the correct use of a term is not
always the use that we ought to make. Sometimes we ought to lie for the sake of a
greater good—such as to protect someone from danger. On other occasions, we
ought to use words incorrectly simply because it will have a more powerful effect
or because it will make people laugh. Yet, according to Prescriptivity, a speaker who
means something by a term ought to apply it correctly, where ‘applies correctly’
stands in for ‘refers to’, ‘is true of’, ‘denotes’, and so forth. Thus, Prescriptivity makes
it a necessary condition of meaning something by a term that a speaker ought to
speak the truth. But this requirement is too strong to be a purely semantic require-
ment. It implies that on occasions when someone ought not to tell the truth, she
does not mean what she ordinarily means by her terms.
5 This problem holds even if MP is understood in the assertion-theoretic sense. I cannot beobligated to assert all of the terms I understand every time I am warranted in doing so,because there are just too many statements, at any given time, that I have warrant toassert.
Is Meaning Normative? 227
# 2006 The AuthorJournal compilation # 2006 Blackwell Publishing Ltd
3. Hypothetical and Categorical Obligations
It might seem intuitive to suppose that meaning is normative because if a speaker
wants to tell the truth, or if a speaker wants to communicate, then she ought only to
use her words correctly. For instance, Kripke says ‘if I intend to accord with my past
meaning of ‘‘þ’’, I should answer ‘‘125’’ (to 68 þ 57¼?)’ (Kripke, 1982, p. 37,
emphasis added). However, on the face of it, this is an example of a hypothetical
means/end prescription, and if Prescriptivity is to undermine naturalism, the semantic
‘ought’s it introduces cannot be of this kind. Hypothetical means/end prescriptions
are conditional prescriptions in which the consequent and antecedent bear a rela-
tionship of means to end. An example of an ordinary hypothetical means/end
prescription is this: ‘If you want to get from Oxford to Cambridge by noon, you
ought to take an early morning train’. Taking an early morning train is a means of
getting from Oxford to Cambridge by noon. A categorical prescription, in contrast, is
not contingent on an end. This distinction is not just about form: categorical
prescriptions can have a conditional structure so long as the antecedent and con-
sequent are not related as end to means. Hence, ‘If you are a moral agent, you ought
to aim to maximise happiness’ is a categorical prescription: aiming to maximise
happiness is not a means to being a moral agent; it is a condition of your being one.
The distinction between categorical and hypothetical ‘ought’s is difficult to
draw.6 Nevertheless, the distinction is decisive, since many hypothetical ‘ought’s
pose no difficulty for naturalism. The reason is that on our usual, normative
interpretation of ‘ought’, hypothetical ‘ought’-statements seem to be plainly false.
How can it be that someone’s desire or intention to do something makes it the case
that she ought to do it? It is clearly not the case that just because George W. Bush
wants or intends to invade Iraq, he ought to do so. Hence, as R. M. Hare
suggested, the force of many hypothetical ‘ought’-statements must be descriptive,
rather than prescriptive (Hare, 1952, Ch. 3). To say ‘if you want to get from Oxford
to Cambridge by noon, you ought to take an early morning train’ is merely to
describe a way of getting from Oxford to Cambridge by noon. Similarly, to use
Hare’s somewhat outdated example, to say ‘if you want to go to the largest grocer
in Oxford, you ought to go to Grimbly Hughes’ is to say that Grimbly Hughes is
the largest grocer in Oxford. Since these statements seem to have descriptive,
rather than prescriptive force, they do not tell us what we ought to do, despite
appearances. Thus, it is important that the ‘oughts’ foisted on the naturalist are not
hypothetical oughts, and thus descriptive. If meaning is normative, if meaning is
genuinely ‘oughty’, then the fact that an agent means something by an expression
must amount to the fact that she ought to use the expression in accordance with its
meaning quite independently of what she wants to do (cf. Gluer, 1999, 2001;
Wikforss, 2001).
6 I have now and will in the following drop the qualification ‘means/end’ and simply use‘hypothetical’ as short for ‘hypothetical means/end’ prescriptions.
228 A. Hattiangadi
# 2006 The AuthorJournal compilation # 2006 Blackwell Publishing Ltd
4. Mistakes and Lies
One reason that is often cited in defence of semantic normativity is that a theory of
content must allow for the possibility of error. For instance, Kripke (1982) presents
the normativity thesis in the course of his argument against the dispositionalist,
whose failure to accommodate error is diagnosed as a failure to capture semantic
prescriptions. Kripke’s dispositionalist maintains that the correctness conditions of a
speaker’s terms, and thus what she means by them, can be read off of her
dispositions. That is, the dispositionalist says that a speaker who means horse by
‘horse’ is disposed to apply ‘horse’ to some creature if and only if it is a horse.
Kripke points out that the dispositionalist will have difficulties dealing with people
who are disposed to make mistakes. Imagine, for example, someone who system-
atically forgets to carry digits when doing complex sums. We would normally want
to say that such a person means addition by ‘plus’ despite systematically making
mistakes (Kripke, 1982, p. 29). The dispositionalist, however, cannot take this into
account. If someone is disposed to give the right answer sometimes and the wrong
answer at other times, which disposition determines what the speaker means?
Somehow, the dispositionalist must specify, without circularity, which of the
speaker’s dispositions are meaning constituting and which are error producing.
But this, Kripke argues, the dispositionalist cannot do (Kripke, 1982, pp. 26–37).
Kripke diagnoses the dispositionalist’s failure as the failure to capture semantic
prescriptions. He says that a full specification of my dispositions will tell you what I
will do, never what I ought to do (Kripke, 1982, p. 36). The fact that I mean addition
by ‘plus’ implies that I ought to apply ‘plus’ in accordance with its meaning; the
fact that I am disposed to use ‘plus’ in a certain way implies no such thing. Hence,
no matter how much we know about a speaker’s dispositions, we will not thereby
know that she means addition by ‘plus’. The dispositionalist fails, that is, because he
fails to capture semantic ‘ought’s. Indeed, the ‘problem of error’ is a variant of the
familiar problem, made prominent by Fred Dretske (1986): that any theory of
representation must allow for the possibility of misrepresentation.7 It is tempting to
gloss this, as Kripke does, in prescriptive terms. But, in the end, the gloss is just
gloss and nothing prescriptive lies beneath (Wikforss, 2001).
Kripke is right to point out that dispositionalism, at least as he formulates it, faces
a difficulty. The dispositionalist seems to say that the correctness conditions of a
term can be read off of a speaker’s dispositions to use that term; that we can find
out what ‘horse’ refers to by looking at the conditions under which a speaker is
disposed to use the term ‘horse’. The problem is that under some conditions, the
speaker will be disposed to apply ‘horse’ to non-horses. Sometimes, as Kripke
points out, when she does so, she will have made a mistake. But this is just a special
case of a more general problem that ultimately has nothing to do with semantic
7 However, Dretske (2000) argues, against his former self, that misrepresentation hasnothing to do with normativity.
Is Meaning Normative? 229
# 2006 The AuthorJournal compilation # 2006 Blackwell Publishing Ltd
‘ought’s. Sometimes, when the speaker applies ‘horse’ to a non-horse, she will be
lying; at other times, she will be joking, speaking ironically, or sarcastically. In
these cases, we hardly want to say that she has made a mistake, that she has failed
to speak as she ought. Indeed, because the dispositionalist tries to derive truth
conditions from dispositions, she will have difficulty accommodating cases in
which the speaker is disposed to say nothing whatsoever in the presence of a
horse. It may be that under some circumstances—such as when the speaker has
better things to think about—she will not say, even to herself, ‘lo, a horse’ in the
presence of a horse. Since the dispositionalist generates the extensions of terms
from the stimulus conditions of dispositions, the dispositionalist seems constrained
to say that ‘horse’ neither refers only to horses, nor to all horses. But once again, we
do not always think that the speaker has made a mistake when she is disposed either
to apply ‘horse’ to a non-horse, or disposed not to apply ‘horse’ to a present horse.
The point is that if the dispositionalist solution is a failure, it is a failure because our
dispositions do not generate correctness conditions. Its failure does not indicate that
there are semantic obligations.
5. Usually, Mostly, Normally
One might argue that lying is parasitic on truth telling, and that what someone
means by an expression is constituted by obligations she incurs on occasions of
sincere use.8 So, what someone means by a term is constituted by the obligations she
incurs when she aims to tell the truth. Though contingent on her aim to tell the
truth, these obligations might nevertheless be essential to what she means—in
order for her words to have meaning, she must aim to use them correctly at least
most of the time. On occasions when she does not strive to tell the truth, the
meaning of the term—as constituted on occasions of sincere use—simply carries
over. Thus, a speaker can meaningfully lie, but at the same time, meaning is
constituted by semantic obligations.
There are several difficulties with this suggestion. The first is that the obligations
incurred on occasions of sincere use will not yield the right pattern of semantic
obligations. Suppose, for example, that Matilda (Who Told Such Dreadful Lies)
wants, on some occasion, to tell the truth—that her house is on fire. In that case,
given what she wants, given what she means, and given that her house is on fire,
she ought to say ‘my house is on fire’. Matilda does not acquire, on that occasion of
sincere use, the obligation to use ‘is on fire’ of something only if it is on fire.
Rather, she acquires the obligation to say that that house is on fire, at that time.
This obligation concerns only a singular occasion of use. Now suppose that Matilda
8 Martin Kusch and Simon Blackburn have suggested this to me in conversation. It couldbe interpreted as a Davidsonian move, but for further reasons why this does not defendPrescriptivity see Bilgrami, 1993; Gluer, 1999, 2001 and Wikforss, 2001.
230 A. Hattiangadi
# 2006 The AuthorJournal compilation # 2006 Blackwell Publishing Ltd
has incurred obligations, on separate occasions, to say ‘this house is on fire’, ‘this
log is on fire’, ‘this oven mitt is on fire’ and so forth. All these taken together do
not add up to the obligation to say that something is on fire only if it is. Matilda just
has a list of obligations to say one thing in one circumstance, another thing in
another. Since it is permissible for Matilda to lie, on the suggestion under con-
sideration, there will be some occasions on which it is not the case that she ought
to apply her terms though it is correct to do so. There is thus no way that the
conjunction of Matilda’s singular obligations will add up the obligation to use her
expressions only if correct. If the suggestion does not even give us Prescriptivity*,
however, it is difficult to see how this could provide an adequate defence of
Prescriptivity.
The second problem is that these ‘ought’s seem to be hypothetical, and thereby
not troubling for the naturalist. If we apply Hare’s reasoning, to say ‘if Matilda
wants to tell the truth, and if she means fire by ‘‘fire’’, she ought to say that the
house is on fire’, is simply to say ‘‘the house is on fire’’ is true in Matilda’s
language’. The ‘ought’ simply drops out. Thus, even if it is true that a speaker
must aspire to use her words correctly most of the time, this only gives rise to
hypothetical ‘ought’s of a kind that pose no difficulties for the naturalist.
6. Prima Facie Obligations
Perhaps Matilda does have a semantic obligation to tell the truth, which is not
contravened when she wants to lie, but overridden. Obligations that can be
overridden by other obligations are called prima facie obligations (Ross, 1987).
For example, if I promise to meet Cathy tomorrow for tea, some would say that
I undertake a prima facie obligation to meet her tomorrow for tea. This is only a
prima facie obligation because I might be justified in breaking my promise under
extenuating circumstances—for instance, if at the time of our tryst, Krister requires
me more urgently. My obligation to Krister overrides my obligation to Cathy, but
this does not mean that I had no obligation to Cathy in the first place, nor does it
mean that the obligation went away; it was just ‘trumped’, as it were, by my
obligation to Krister. The fact that I have a prima facie obligation to Cathy is the
fact that, all things being equal, I ought to drink tea with Cathy. Hence, one might
argue that my semantic obligations are prima facie obligations. If I mean something
by an expression, I have an obligation to tell the truth, all things being equal; my
obligation will be overridden if I ought to tell a lie.
I suspect that the surface plausibility of this suggestion trades on our smuggling
in desires on which the obligation to tell the truth is contingent. In order to test
whether I have a prima facie obligation to use an expression correctly, we need to
consider a case in which all is genuinely equal. So, not only should we assume that
I have no desire to tell a lie, but that I have no desire to tell the truth—otherwise,
my obligation to apply a term correctly might just be contingent on that desire.
Moreover, we have to rule out the possibility that the obligation is really moral or
Is Meaning Normative? 231
# 2006 The AuthorJournal compilation # 2006 Blackwell Publishing Ltd
prudential. So, we have to assume that I have no desire to communicate; that if I
have an audience, my audience is utterly indifferent to whether or not I tell the
truth; and that nothing whatsoever hangs on what I say—it will not lead me to act
imprudently or irrationally. The question is, given that my audience and I are
indifferent to whether I tell the truth, and given that nothing hangs on what I say,
am I still obligated to tell the truth? I see no reason why. If what I say affects
nothing and no one, it hardly seems to matter whether I apply the term ‘horse’
only to horses.
One might suppose that it is necessary for an account of a speaker’s meaning that
we posit a prima facie obligation for a speaker to use her terms correctly. However,
this does not seem to be true. Given that I mean horse by ‘horse’, MP tells us that
‘horse’ refers to all and only horses. Given Correctness, however I use the term
‘horse’, it will be correct in application to something if and only if it is a horse. This
seems to be sufficient to account for the meaning of ‘horse’; nothing is added by
saying that, ceteris paribus I ought to apply ‘horse’ only if it is correct to do so.
Furthermore, there is a difficulty in supposing that a speaker has a prima facie
obligation to use her words correctly—because the obligation to speak correctly
would be overridden not only by an obligation to lie, but also, it seems, simply by
the desire to lie. In contrast, my obligation to meet Cathy for tea cannot be
overridden by a desire not to do so, but only by another obligation. If the
obligation to speak the truth can be overridden by a mere desire not to do so,
this seems to undermine the idea that an obligation, even a prima facie obligation,
places a categorical constraint, whether I like it or not. Thus, instead of saying that
I have a prima facie obligation to speak correctly that can be overridden by a desire
to lie, it seems more plausible to suppose that what I have is not a prima facie
categorical obligation, but a hypothetical one: if I want to tell the truth, I ought to
use my words correctly; if I want to lie, I ought to use them incorrectly. What
might have looked like a prima facie obligation to tell the truth is more plausibly
construed as a hypothetical obligation contingent on a desire to tell the truth. Once
again, this poses no serious problem for the naturalist.
7. Truth as a Conceptual Norm
Many philosophers are inclined to argue that though meaning may not be norma-
tive, conceptual content is. Though insincere or otherwise incorrect uses of
language may violate no semantic norms, misapplications of concepts do. When
Matilda tells another whopper, she chooses to use her words incorrectly. Yet,
Matilda can hardly choose to apply her concepts incorrectly. She can hardly help
thinking that her house is on fire when she feels the heat of the flames, whatever
she might say about it. However, Matilda’s being compelled to apply the concept
fire only to fire is not tantamount to her being obligated to apply the concept fire
only to fire. Matilda’s compulsion to believe that the house is on fire can be explained
naturalistically—in terms of her dispositions. The fact that she cannot help but
232 A. Hattiangadi
# 2006 The AuthorJournal compilation # 2006 Blackwell Publishing Ltd
apply the concept fire as she does is simply the fact that she is disposed to think fire!
when confronted with fire. There is a disposition here, not an obligation.
Nevertheless, it could be argued that though we may not be obligated to speak
the truth, we ought to believe only what is true (Horwich, 1998, Ch. 8). If this
supposition is correct, it will turn out to be a prerequisite for believing only what is
true that we ought to apply our concepts correctly. To defend this line of reason-
ing, one might employ Michael Dummett’s argument for the value of truth
(Dummett, 1978). Dummett makes an analogy between the concept of truth and
that of winning at a game such as chess. He observes that you could specify all the
rules of chess, specify how all the pieces move, what constitutes winning or loosing
the game, and still something would be left out: that the whole point of the game is
to win. Similarly, Dummett argues, specifying the conditions under which a belief
is true leaves out the important fact that truth is valuable, that truth is the aim of
belief. If truth is valuable then we ought to believe only what is true. And, if we
ought to believe only what is true, then we ought to apply concepts only when
correct.9
Dummett’s argument is compelling because most of us do value truth—we
clearly want to believe only what is true. But our wanting to believe what is
true and our being obligated to believe only what is true are two different things. If
we ought to believe only truths only as a means to satisfying a desire to believe only
truths, then this is merely a bypothetical ‘ought’ and so not troubling for the
naturalist. Moreover, it does seem as though this ‘ought’ is hypothetical. I want to
believe only what is true because having true beliefs will make me more successful
in getting what I want. This is obviously true of particular beliefs. Consider Pam,
who wants some ice cream, and who believes that there is ice cream in the freezer.
Clearly, Pam wants her belief to be true because if it is true, then she can
successfully act on it—she can go to the freezer and find ice cream. This is not
just to say that Pam wants there to be ice cream in the freezer; rather, she wants to
believe that there is ice cream in the freezer only if there is. If there is no ice cream
in the freezer, she would rather believe that there is no ice cream in the freezer, so
she will be prevented from making an unnecessary trip. In general, Pam wants her
beliefs to be true because they form part of an interconnected system that helps her
to get about in the world. Thus, Pam wants all of her beliefs to be true (Whyte,
1990; Mellor, 1990; Horwich, 2000). The obligation to believe only what is true is
thus a hypothetical obligation contingent on the desire—which most of us have—
to believe only what is true.
If the argument from the value of truth is to support Prescriptivity, it must be the
case that we ought to believe only what is true quite independently of any desire to
do so. However, while it may be plausible to suppose that we want to believe only
truths, it is difficult to support the claim that we categorically ought to believe only
what is true. This is because, in some circumstances, the weight of evidence will
9 For another Dummettian argument for Prescriptivity, and criticism, see Gluer, 2001.
Is Meaning Normative? 233
# 2006 The AuthorJournal compilation # 2006 Blackwell Publishing Ltd
favour a falsehood, in which case we ought to believe the falsehood. If we have
overwhelming evidence that the butler committed the crime, it would be irrational
to believe that he is innocent, even if it is true that he is innocent. Hence, even if
truth is intrinsically valuable, we can at best be obligated to seek the truth, not to
believe only what is true.10 But an obligation to seek the truth will not imply that
we ought to apply our terms or concepts only correctly.11
8. Communication and Community
Some readers will be impatient to point out that the community supplies the
elusive normativity; that it is because I am a member of a linguistic community that
I ought to use my words correctly. In one sense, this is true: if I want to be
understood by others, then I ought to use my words as they do, or in such a way
that they would understand. Most of the time, communication is best served by the
correct use of words. But this obligation is of course contingent on my desire to
communicate. If I have no such desire, then the obligation goes away. If I wish to
lie, or make you laugh, or even if I wish to tell the truth but know that you think I
will lie, then I ought to use my words incorrectly.
Some people maintain, however, that meaning is essentially social; that correct-
ness conditions are constituted by communal agreement. Moreover, communal
agreements, like promises, create obligations. When I promise to return your book
tomorrow, I undertake an obligation to return your book tomorrow. I create an
obligation by making the promise. Similarly, when I am initiated into a commu-
nity, I must undertake an obligation to abide by the conventions of that commu-
nity. Where this is a linguistic community, I undertake an obligation to use my
expressions correctly. This, in rough outline, is a view defended at length by
Brandom:
The particular norms of concern . . . are discursive normative statuses, the sort
of commitment and entitlement that the use of concepts involves. These
norms, it will be claimed, are instituted by social practices . . . Elaborating an
account along these lines is pursuing three of Wittgenstein’s grand themes: the
insistence on the normative character of language and intentionality, the
pragmatist commitment to understanding these norms in terms of practices . . .
and the recognition of the essentially social character of such norms (Brandom,
1994, p. 55).
10 For an argument as to why it is not the case that we ought to seek the truth, seePapineau, 1999.
11 For further discussion of semantic normativity and the value of truth, see Wikforss, 2001.
234 A. Hattiangadi
# 2006 The AuthorJournal compilation # 2006 Blackwell Publishing Ltd
To paint it in broad strokes, Brandom’s picture is one in which communities
‘institute’ semantic or conceptual conventions by taking attitudes of approval or
disapproval towards bits of behaviour. Any behaviour that tends to meet with
disapproval is thereby wrong, by the community’s lights, whereas behaviour that
tends to meet with approval is thereby right. What a given speaker ought to say,
then, is what tends to meet with approval rather than disapproval. Thus, the social
practice of treating types of behaviour as correct or incorrect creates semantic
obligations.
Whatever the merits of this picture of meaning, it cannot be marshalled in defense
of the thesis that meaning is normative. Some argument must be given for saying
that meaning is essentially social, and in Brandom’s case, the argument that is given
depends on the assumption that meaning is normative. Like Kripke, Brandom
assumes that the normativity of meaning is basic to our intuitive view. He says, for
instance, that ‘our ordinary understanding of states and acts of meaning, under-
standing, intending, or believing something is an understanding of them as states
and acts that commit or oblige us to act and think in various ways (Brandom, 1994,
p. 13).’ He then proceeds to argue that no naturalistic theory of meaning would
do—and concludes that normativity must be assumed as primitive (Brandom,
1994, pp. 42–5). It is not clear why this would make meaning essentially social;
since whatever normativity is brought in at the social level might equally be
brought in at the individual level (Blackburn, 1993). Nevertheless, by introducing
communal practices, Brandom purports to explain how we create obligations—
obligations he takes to be there, standing in need of explanation. This just fails to
speak to my concerns; my worry is that there are no semantic obligations to be
explained in the first place.
Moreover, there is a distressing circularity in the claim that semantic obligations
are constituted by communal attitudes. Brandom seems to think that we can
catapult ourselves into language via our attitudes. But this is like pulling ourselves
up by our own socks, or turning ourselves into promise-keepers by promising to
keep our promises. If semantic obligations are supposed to be essential to content,
then there can be no content without the relevant obligations. But if the obliga-
tions are created by our attitudes, then they are themselves the products of
contentful states. If I approve of your saying ‘red’ in the presence of this apple,
my endorsement has content—I endorse that you say ‘red’ in the presence of this apple.
If I can have an attitude with this content, then for one thing, I must grasp the
concept apple. Assuming Prescriptivity*, this means that I ought to apply the concept
apple only if it is correct to do so. But if obligations such as these are uniformly
created by attitudes, then how is my obligation to apply the concept apple created?
If by another attitude, we embark on a regress. If by my dispositions, the view will
ultimately bottom out into a dispositional theory, which Brandom rejects on the
grounds that it fails to accommodate normativity (Hattiangadi, 2003). So, Brandom
faces a dilemma. Either way, we have no reason to think that we must assume that
there are semantic obligations, standing in need of an explanation.
Is Meaning Normative? 235
# 2006 The AuthorJournal compilation # 2006 Blackwell Publishing Ltd
9. Contractual Commitments
Kripke, Wright, and McDowell (among others) suggest that we are committed to
Prescriptivity* because we intuitively subscribe to what Wright calls the ‘contractual
theory’ of understanding (Brandom, 1984; Kripke, 1982; McDowell, 1993; Pettit,
1990 and Wright, 1980).12 According to the contractual theory, to understand an
expression is to be committed to a particular pattern of application of that expres-
sion. This commitment then confers an obligation on the speaker to carry out the
pattern of application in question. McDowell presents this assumption as follows:
We find it natural to think of meaning and understanding in, as it were,
contractual terms. Our idea is that to learn the meaning of a term is to
acquire an understanding that obliges us subsequently—if we have occasion
to deploy the concept in question—to judge and speak in certain determin-
ate ways, on pain of failure to obey the dictates of the meaning we have
grasped; that we are ‘committed to certain patterns of linguistic usage by
the meanings we attach to expressions’ (McDowell, 1993, p. 257 quoting
Wright, 1980, p. 21).
The reasoning behind this goes roughly as follows: to accept MP is to accept that a
term has content in virtue of its correctness conditions. In order for me to mean
something by an expression, I must have adopted some standard of correctness for
that expression. In adopting a standard of correctness, I must have formed an
intention towards that standard. Otherwise, what is it for me to have adopted the
standard in the first place?
The trouble is that the contractualist needs to show that when I form an
intention towards a standard, I incur the right set of obligations. That is, the
contractualist needs to show that if I grasp the concept horse, I ought to apply it
only to horses. The contractual theorist thus holds that in order for me to mean
something by an expression, I must be committed not just to using the standard in
any old way, but to meeting it; I am committed to carrying out a particular pattern of
use for the expression I understand. Wright suggests this when he says that we
intuitively suppose that ‘[w]hen I assent to the rule: F is to be applied only to
individuals which are f, I commit myself to a quite determinate way of using F’
(Wright, 1980, p. 36). Or, to put it more schematically, the Contractual Theory
(CT ) can be expressed as follows:
Contractual Theory: S means F by t $ S intends that: (a)(S applies t to a $ a is f ).
12 Kripke (1982) claims that intentions yield obligations, whereas Wright (1980) andMcDowell (1993) both suggest that commitments yield obligations. I discuss both versions.It should be noted that the contractual theory is presented as part of our intuitive view,one that is rejected both by Wright and Kripke, and modified by McDowell.
236 A. Hattiangadi
# 2006 The AuthorJournal compilation # 2006 Blackwell Publishing Ltd
If CT is indeed part of our intuitive view—which I doubt—we ought to revise our
intuitive view. To begin with, if an agent were to form the intention to use an
expression correctly, then given a referential standard of correctness, this would be
tantamount to her forming the intention never to lie. Think again of Matilda, who
claims that her house is on fire when she knows full well that it is not. In order for
her lie to be genuine—and not just a case of mistaken belief—she has to form the
intention to say ‘the house is on fire’ despite her awareness that the statement is
false. Assuming CT, though, if Matilda means fire by ‘fire’, she must have the
intention to apply the expression only if it is true. The trouble is that when she lies,
of course, she forms the intention to utter a statement that she believes to be false.
This obviously results in a difficulty: Matilda cannot simultaneously intend to apply
t to a only if a is f, and intend to apply t to a when it is not the case that a is f.
Intentions (as opposed to desires or beliefs) cannot stand in conflict, since inten-
tions are the result of firm decisions to carry out actions. Hence, at best, the
contractual theory makes it impossible for anyone to consistently form the inten-
tion to lie.
One might think that this objection can be met if the contractualist sticks to the
idea of someone being committed rather than merely intending to act in accordance
with a rule.13 However, if Matilda is committed to apply t to a only if a is f, then
she still cannot intend to apply t to a in full knowledge that it is not the case that a is
f without affecting her commitment. A one-off intention to lie might not destroy
her commitment, but if Matilda intended to lie too many times, presumably we
would no longer wish to say that she was committed to speaking the truth. But it is
odd to think that what Matilda means is hostage to her intentions in this manner.
Compulsive liars, like Matilda, do not gradually begin to attach non-standard
meanings to their expressions. When Matilda lies, jokes, or misleads, she presupposes
the ordinary or literal meanings of her terms, so she cannot change them just by
lying.
Perhaps the contractualist could meet this objection by restricting himself to
conceptual, as opposed to linguistic, content. In Matilda’s case, she lies because she
believes that the house is not on fire, when she says that it is. In her mind she
carries out her semantic obligations, while in her speech she uses the meanings of
her words to convey false information. Given that she grasps the concept fire, she is
committed at least to think fire! only when there are flames about. If this is the
contractualist’s argument, however, it will succumb to an objection that was made
previously: we have here a prudential obligation masquerading as a semantic one,
and the prudential obligation can be given a naturalistic explanation. Though there
might be value in only believing what is true, this seems to follow from a general
prudential obligation contingent on a desire to get what we want. The obligation
in question is therefore not semantic.
13 I am grateful to Bob Hale for making this suggestion.
Is Meaning Normative? 237
# 2006 The AuthorJournal compilation # 2006 Blackwell Publishing Ltd
10. Conclusion
The normativity thesis arises in the context of Kripke’s discussion of rule following.
Perhaps the most natural reason for supposing that meaning is normative is the
thought that to understand a concept is to follow the rule for its correct application.
And rules tell us what we ought to do. The Highway Code tells us how we ought
to drive. The Ten Commandments tell us how we ought to act. So, if there are
semantic rules, surely they will tell us what we ought to say.
The reason why this argument fails should now be obvious from the foregoing
discussion. For this argument from rule following to go forward, semantic rules
clearly need to be both meaning constituting and prescriptive. Semantic rules will
have to be meaning constituting because otherwise they will not be essential to
meaning; and of course they will have to be prescriptive or they will not introduce
semantic ‘ought’s. The trouble is that these two constraints pull in opposite
directions (Gluer and Pagin, 1999). For a rule to be prescriptive, it must tell me
what I ought to do. According to MP, the meaning constituting rule for ‘horse’
must imply that ‘horse’ applies correctly to all and only horses. However, it is not
the case that I ought to apply ‘horse’ to all and only horses—I am not obligated to
apply ‘horse’ to all horses because I cannot do so, and ‘ought’ implies ‘can’. The
weaker rule, stating that I should apply ‘horse’ only to horses cannot constitute the
meaning of ‘horse’. The rule that tells me to apply ‘horse’ only to horses does not
distinguish between my meaning horse by ‘horse’ and something else, such as brown
horse or black horse. Moreover, I sometimes ought to tell lies, or use my words
incorrectly to prove a point or make people laugh, and the fact that I ought to use
my words incorrectly does not imply that I do not mean what I ordinarily mean by
my words. Since the weaker rule cannot distinguish my meaning any number of
things by ‘horse’, it simply cannot constitute meaning. Thus, Prescriptivity must be
false, and meaning is not normative in the sense that is required to generate a
presumption against naturalism.
St. Hilda’s College,
Oxford
References
Baker, G. P. and Hacker, P. M. S. 1984: Scepticism, Rules and Language. Oxford: Basil
Blackwell.
Bilgrami, A. 1993: Norms and meaning. In R. Stoecker (ed.), Reflecting Davidson.
Berlin: W. de Gruyter.
Blackburn, S. 1993: Essays in Quasi-Realism. Oxford: Oxford University Press.
Bloor, D. 1997: Wittgenstein, Rules and Institutions. London: Routledge.
Boghossian, P. A. 1989: The rule-following considerations. Mind, 98, 507–549.
238 A. Hattiangadi
# 2006 The AuthorJournal compilation # 2006 Blackwell Publishing Ltd
Brandom, R. 1994: Making it Explicit. Cambridge, MA: Harvard University Press.
Darwall, S., Gibbard, A. and Railton, P. 1992: Toward fin de siecle ethics: some trends.
The Philosophical Review, 101, 115–189.
Dretske, F. 1986: Misrepresentation. In R. J. Bogdan (ed.), Belief: Form, Content, and
Function Oxford: Clarendon Press, 17–36.
Dretske, F. 2000: Perception, Knowledge and Belief: Selected Essays. Cambridge:
Cambridge University Press.
Dummett, M. 1978: Truth In Truth and Other Enigmas. Cambridge, MA: Harvard
University Press, 1–25.
Fodor, J. 1990: A Theory of Content and Other Essays. Cambridge, MA & London:
Bradford Books/The MIT Press.
Glock, H-J. 1994: Wittgenstein Vs. Quine on logical necessity. In S. Teghrarian (ed.),
Wittgenstein and Contemporary Philosophy. Bristol: Thoemmes, 185–222.
Glock, H.-J. 1996: Necessity and normativity. In H. Sluga, and D. G. Stern (eds.), The
Cambridge Companion to Wittgenstein. Cambridge: Cambridge University Press,
198–225.
Gluer, K. 1999: Sense and prescriptivity. Acta Analytica, 14, 111–128.
Gluer, K. 2001: Dreams and nightmares. Conventions, norms and meaning in
Davdison’s philosophy of language. In P. Kotatko, P. Pagin and G. Segal (eds).,
Interpreting Davidson. Stanford CA: CSLI Publications.
Gluer, K and Pagin, P. 1998: Rules of meaning and practical reasoning. Synthese, 117,
207–227.
Hare, R. M. 1952: The Language of Morals. Oxford: Oxford University Press.
Hattiangadi, A. 2003: Making it implicit: Brandom on rule following. Philosophy and
Phenomenological Research, 66, 419–431.
Horwich, P. 1998: Meaning. Oxford: Clarendon Press.
Horwich, P. 2000: Norms of truth and meaning. In A. O’Hear (ed.), Philosophy: the
Good, the True and the Beautiful. (R.I.P. Supplementary Volume) Cambridge:
Cambridge University Press, 19–34.
Hudson, W. D. (ed.), 1969: The Is-Ought Question. London: Macmillan.
Hume, D. [1739] 1978: A Treatise of Human Nature. L. A. Selby-Bigge ed. Oxford:
Clarendon Press.
Kripke, S. 1982: Wittgenstein, Rules and Private Language. Cambridge, MA: Harvard
University Press.
Lance, M. N. and O’Leary Hawthorne, J. 1998: The Grammar of Meaning: Normativity
and Semantic Discourse. Cambridge: Cambridge University Press.
McDowell, J. 1993: Wittgenstein on following a rule. In A. W. Moore (ed.), Meaning
and Reference. Oxford: Oxford University Press. Originally published 1984,
Synthese, 58, 325–63.
McDowell, J. 1988: Mind, Value and Reality. Cambridge, MA: Harvard University
Press.
McGinn, C. 1984: Wittgenstein on Meaning. Oxford: Basil Blackwell.
Is Meaning Normative? 239
# 2006 The AuthorJournal compilation # 2006 Blackwell Publishing Ltd
Mellor, D. H. 1990: Telling the truth. In Mellor (ed.), Ways of Communicating.
Cambridge: Cambridge University Press.
Millar, A. 2004: Understanding People. Oxford: Oxford University Press.
Miller, A. 1998: Philosophy of Language. Montreal and Kingston: McGill-Queens
University Press.
Miller, A. 2003: An Introduction to Contemporary Metaethics. Cambridge: Polity Press.
Millikan, R. 1984: Language, Thought and Other Biological Categories. Cambridge, MA:
Bradford Books/The MIT Press.
Millikan, R. 1995: White Queen Psychology and Other Essays for Alice. Cambridge, MA:
The MIT Press.
Moore, G. E. [1903] 1993: Principia Ethica. T. Baldwin, ed. Cambridge: Cambridge
University Press.
Papineau, D. 1999: Normativity and judgement. Aristotelian Society Supplementary
Volume, 73, 17–43.
Pettit, P. 1990: The reality of rule-following. Mind, 99, 1–21.
Quine, W. V. O. 1960: Term & Object. Cambridge, MA: MIT Press.
Ross, D. 1987: Prima facie duties. Reprinted in C. Gowans (ed.), Moral Dilemmas.
Oxford: Oxford University Press.
Sellars, W. [1956] 1997: Empiricism and the Philosophy of Mind. Cambridge, MA:
Harvard University Press.
Soames, S. 1998: Scepticism about meaning: indeterminacy, normativity, and the rule-
following paradox. In Ali A. Kazmi (ed.), Canadian Journal of Philosophy: Meaning
and Reference. Supplementary Volume 23, Calgary: University of Calgary Press,
211–249.
Stich, S. and Warfield, T. (eds.) 1994: Mental Representation. Oxford: Blackwell.
Whyte, J. 1990: Success semantics. Analysis, 50, 149–157.
Wikforss, A. 2001: Semantic normativity. Philosophical Studies, 102, 203–206.
Wilson, G. 1994: Kripke on Wittgenstein and normativity. Midwest Studies in
Philosophy, Vol. 19: Philosophical Naturalism, 366–390.
Wright, C. 1980: Wittgenstein on the Foundations of Mathematics. London: Duckworth.
Wright, C. 1984: Kripke’s account of the argument against private language. The
Journal of Philosophy, 81, 759–777.
240 A. Hattiangadi
# 2006 The AuthorJournal compilation # 2006 Blackwell Publishing Ltd
Top Related