A Defence of Emotivism
Leslie Allan
Published online: 16 September 2015
Copyright © 2015 Leslie Allan
As a non-cognitivist analysis of moral language, Charles Stevenson’s sophisticated
emotivism is widely regarded by moral philosophers as a substantial
improvement over its historical antecedent, radical emotivism. None the less, it
has come in for its share of criticism. In this essay, Leslie Allan responds to the key
philosophical objections to Stevenson’s thesis, arguing that the criticisms levelled
against his meta-ethical theory rest largely on a too hasty reading of his works.
To cite this essay:
Allan, Leslie 2015. A Defence of Emotivism, URL =
<http://www.RationalRealm.com/philosophy/ethics/defence-emotivism.html>
To link to this essay:
www.RationalRealm.com/philosophy/ethics/defence-emotivism.html
Follow this and additional essays at: www.RationalRealm.com
This article may be used for research, teaching, and private study purposes. Any substantial
or systematic reproduction, redistribution, reselling, loan, sublicensing, systematic supply,
or distribution in any form to anyone is expressly forbidden. Terms and conditions of access
and use can be found at www.RationalRealm.com/policies/tos.html
Leslie Allan
A Defence of Emotivism
1. Introduction
In this essay, my aim is to defend an emotive theory of ethics. The particular version
that I will support is based on C. L. Stevenson’s signal work in Ethics and Language [1976],
originally published in 1944, and his later revisions and refinements to this theory presented
in his book, Facts and Values [1963]. I shall not, therefore, concern myself with the earlier
and less sophisticated versions of the emotive theory, such as those presented by A. J. Ayer
in Language, Truth and Logic [1971] and Bertrand Russell in Religion and Science [1935].
It should be noted that although my meta-ethical views are based on Stevenson’s
analysis, there are points of difference. For example, I am not too sure that his dispositional
theory of meaning is correct and I find his account of personal deliberation as
self-persuasion inadequate. So, in the following exposition, where my comments are found
to diverge from Stevenson’s later views, it should be understood that on these points I have
found his analysis unconvincing, although this divergence will usually be explicitly
mentioned. Along with my debt to Stevenson, I am also particularly indebted to the many
worthwhile insights on semantics made by J. O. Urmson in his book, The Emotive Theory of
Ethics [1968], and to D. H. Monro for his clear and perceptive discussions on the psychology
of attitudes in his work, Empiricism and Ethics [1967].
In this essay, I do not propose to give a full exposition of an adequate emotive
theory, except to provide a brief summary of Stevenson’s position held at the time of his
compiling Facts and Values [1963]. It would be more worthwhile to assume a working
familiarity of his developed views, and in so doing leave room for a comprehensive rejoinder
to some of the more important objections made by his critics. This is what I shall proceed to
do. Significant details of a convincing emotive theory will be developed in the course of the
discussion.
Downloaded from http://www.RationalRealm.com
2
Leslie Allan
A Defence of Emotivism
2. Stevenson’s Emotivism
Stevenson’s views were the result of dissatisfaction with both restricted naturalist 1
and intuitionist accounts of ethical discourse. These theories, he claimed, which render the
meanings of ethical words purely in terms of natural and non-natural properties and
relations (hence they are termed ‘descriptivist’ theories), are unable to provide an adequate
motivating reason for promoting ‘good’, doing what is ‘right’, and so on. By drawing
attention solely to the cognitive function of ethical language, a bridge was placed between
ethical properties and relations and the immediacy and intimacy of personal decision and
action.
Stevenson’s emotive theory sought to correct this inadequacy of descriptivist
accounts by pointing to an even more important function of ethical language. Such words
not only serve to convey factual information; their primary function is to express the
attitude of the speaker and to invite-so-to-speak2 the hearer to share the speaker’s
attitude. These two aspects of ethical terms, expressing the speaker’s attitude and
inviting-so-to-speak a similar attitude in the hearer, constitute their emotive meaning.
Ethical disagreement, then, is often characterized by a disagreement in belief. But
more fundamentally, it is a disagreement in attitude. It is only with the concurrence of
attitudes that an ethical dispute is resolved, even though there may remain a divergence of
related beliefs.
The second major criticism that Stevenson had of descriptivist theories was their
disregard for the vagueness and ambiguities of ordinary language. The complexities, he
argued, are such that the cognitive meanings of ethical terms cannot be forced to fit a single
descriptivist mould and do justice to common language. Many such analyses of the ‘real’
meaning of ethical words are covert uses of their emotive power to persuade the hearer of
the speaker’s normative views. These ‘persuasive definitions’, as Stevenson called them, are
attempts to prove the speaker’s normative convictions to be true by definition.
To accommodate the wide variety of descriptive meanings of ethical terms,
Stevenson characterized their meaning in schematic form. To take ‘good’ as an example,
and including its emotive senses for the sake of completeness, Stevenson’s revised schema
is as follows: 3
1
It is important to note the difference between restricted naturalism and broad naturalism. Restricted
naturalism offers a logical reductionist account of ethical words, giving their meanings strictly in terms of
empirical phenomena. Broad naturalism explains ethical discourse in terms of empirical phenomena, such as
human desires. The former implies the latter, but not vice versa. So, emotivism is a form of broad naturalism,
but not a form of the restricted sort.
2
In his [1976], Stevenson wrote of ethical terms tending to ‘evoke’ the hearer’s attitude. He later came to
accept J. O. Urmson’s point that to talk of ‘inviting’ was semantically more correct and coined the term
‘invites-so-to-speak’ in order to allow for the variability of the strength of the speaker’s intention to influence
the hearer’s attitude (see Stevenson [1963: 208–10).
3
This is a revised version of the definition given in Stevenson [1976: 207]. It should be noted that the first
pattern of analysis offered in Stevenson [1976] was later rescinded in Stevenson [1963: 210–13].
Downloaded from http://www.RationalRealm.com
3
Leslie Allan
A Defence of Emotivism
‘This is good’ has the meaning of ‘This has qualities or relations X, Y, Z . . . ,’ except
that ‘good’ has as well a laudatory emotive meaning which permits it to express
the speaker’s approval and invite-so-to-speak the approval of the hearer.
The analyses of the other ethical terms, such as ‘right’, ‘ought’, ‘duty’ and
‘obligation’, are similar to that proposed for ‘good’. Stevenson simply noted their differing
ranges of applications and emotive emphases.
In contrast to earlier emotive theories, Stevenson recognized that reason has an
important role to play in ethical debate. He had noted the complex psychological
connections between our beliefs and our attitudes. Reason, he argued, is used to change
beliefs with the intention of causing a corresponding shift in attitude. But there is no strict
logical connection between facts and moral values. Two people can agree on the relevant
facts and yet neither be logically compelled to adopt the other person’s moral judgement. In
this respect, Stevenson’s thesis is sharply opposed to cognitive theories.
Downloaded from http://www.RationalRealm.com
4
Leslie Allan
A Defence of Emotivism
3. The Role of Meta-Ethical Analysis
Before I go on to examine the major objections to Stevenson’s thesis, it will be
instructive to digress for a moment to reflect on the particular concerns of meta-ethical
inquiry and to see how Stevenson’s emotivism fits into this scheme. In this way, the
criticisms of Stevenson’s theory will be better understood. If meta-ethics is not to be
revisionary, then its first major task is to elucidate the meanings of moral terms as used in
ordinary discourse. Let us call this function, ‘linguistic analysis’.
Secondly, it is to give some account of the psychological and social functions of
ethics. Moral discourse may be seen in terms of social contracts, as with Hobbes and Rawls,
or the analysis may be brought down to a more personal level. Intuitionists, such as Ross
and Bradley, see ethics in terms of the individual directly apprehending certain synthetic
a priori truths. I shall call this aspect of meta-ethics, the ‘extra-linguistic analysis’.
The third function is to explore the epistemological status of normative utterances
and the logical relations between them and non-moral statements. It is this aspect that is
probably most peculiarly philosophical in its methods and I shall call this function,
‘epistemological analysis’.
It seems easy enough to define these three functions and see how they differ from
each other. The problem arises when we attempt to delimit them in practice, for in any
particular meta-ethical theory we care to examine, we shall find complex interconnections
between the three. An answer to one of these three problems invariably affects the answers
to the other two. Furthermore, to answer one of these questions adequately and
competently, we must first draw on the knowledge gained in answering the other two. All in
all, a competent meta-ethical theory is not built up by answering one problem at a time in
isolation from the others, but is a gradual refinement of a world-view that sees these three
aspects of ethical discourse, the linguistic, extra-linguistic and epistemological, as an
integrated whole.
Looking at Stevenson’s theory, we can readily see this if we consider, for example,
his contention that emotive meaning is a fundamental part of the meaning of ethical terms.
This, of course, is a linguistic analysis. But it also has repercussions for the other parts of the
theory. The extra-linguistic analysis could not neglect to point out that the social function of
ethical discourse is to influence attitudes and not only beliefs. Furthermore, this
dependence of extra-linguistic analysis on linguistic analysis also operates in reverse. As with
any ‘ordinary language’ analysis, Stevenson’s linguistic theory depends on a careful
observation of the social functions of ethical language.
Much work needs to be done in making clear these complex interconnections, not
only in Stevenson’s meta-ethic, but in all those that we wish to consider seriously.
Unfortunately, to do this with Stevenson’s thesis, or any other meta-ethic, is well beyond
the scope of this essay, but it is sorely needed to overcome some of the misconceptions of
his analysis. Some major objections to his theory have rested on an inattention to this
three-fold division of labour within his analysis. So, I shall rest content to roughly categorize
it according to its three major functions, stressing once again that there is much overlap
between the divisions.
Downloaded from http://www.RationalRealm.com
5
Leslie Allan
A Defence of Emotivism
In a nutshell, Stevenson’s linguistic analysis comprises his theory of emotive and
varying descriptive meanings of ethical terms. The extra-linguistic aspect is the stress on
attitudinal disagreement as being fundamental to ethical discourse, while the
epistemological analysis rests on the mostly causative psychological function of reasons in
ethics.
Downloaded from http://www.RationalRealm.com
6
Leslie Allan
A Defence of Emotivism
4. Objectivism in Moral Language
J. D. Mabbott’s critique of emotivism in his An Introduction to Ethics [1977] is a clear
example of the confusion that results from a lack of recognition of this multi-functional
nature of a meta-ethical theory. Unfortunately, he also seriously misconstrues the basic
tenets of emotivism, often equating it with ‘introspective report’ type subjectivism. 4 None
the less, his confusions are instructive of what to avoid.
He maintains that emotivism, along with all other ‘subjectivist’ theories, could not
possibly be true for the simple fact that ‘normal users’ of ethical language are objectivists.5
‘Objectivist’ is here used in the sense of a person who believes that moral language ascribes
properties to things and events and that these properties are real independently of people’s
desires, approvals, and so on.
What Mabbott is doing here is falsely construing Stevenson’s emotivism to be a
purely linguistic theory and carrying the ‘subjectivist’ element of the extra-linguistic analysis,
the emphasis on attitudinal disagreement, into Stevenson’s linguistic analysis of ordinary
language. Now, it seems that to define ethical words in terms of a descriptive component
and the semantically necessary illocutionary forces 6 of expressing and inviting-so-to-speak
attitudes is to be non-committal on the question of the objectivity or subjectivity of moral
values.
A philosopher is still quite free to develop an objectivist meta-ethic while accepting
this epistemologically harmless ‘ordinary language’ analysis. The varying descriptive
component of Stevenson’s schemata allows ample room for this. Our philosopher need only
argue that an attitude is objective if and only if it conforms to the particular descriptive
meaning of ethical terms that he considers satisfies the requirements of objectivity. ‘Value’
would be objective in the sense that it is independent of the attitudes of particular people.
The emotivist would certainly deny that such undertakings have been or will be successful,
but it is certainly not ruled out a priori by Stevenson’s ‘ordinary language’ analysis of the
meanings of ethical terms.
A parallel situation can be found in theories of truth. If we surveyed the general
population, we would find many descriptive definitions of truth, including pragmatic,
foundationalist, positivist, and so on, with most people holding some type or other
objectivist view. As linguists, it is clear that to say ‘X is true’ is to commit the illocutionary
act of expressing a belief that ‘X’, and even possibly to invite-so-to-speak the similar belief
of others. Suppose we decide that these illocutionary forces are a semantically necessary
component of the meaning of ‘true’. Now, no-one would seriously suggest that just because
beliefs are private, that this linguistic analysis would commit us to a subjectivist or relativist
theory of truth. A philosopher need only argue that a belief is objective if and only if it
conforms to the particular descriptive meaning of epistemological terms that he considers
4
See, for example, J. D. Mabbott [1977: 76f, 78].
J. D. Mabbott [1977: 95–9]
6
The illocutionary force of a speech act is what the speaker is doing when making that utterance; for example,
commanding, questioning, promising and so on.
5
Downloaded from http://www.RationalRealm.com
7
Leslie Allan
A Defence of Emotivism
satisfies the requirements of objectivity. ‘Truth’ would be objective in the sense that it is
independent of the beliefs of particular people.
So, Stevenson’s linguistic analysis is similarly non-committal. From the perspective of
his comprehensive theory, and not just from its linguistic component, it is clear why so many
ordinary people are objectivists. They are only aware of the descriptive aspect of ethical
terms and do not notice that it is differences of attitude that are fundamental in ethical
disputes.
That a significant minority of ordinary people are relativists or subjectivists in ethics,
and the number is probably increasing, is not sufficiently appreciated by objectivist ethical
philosophers. Mabbott’s 7 contention that all ‘normal users’ of ethical language are
objectivists is a case in point. That there are just as many differences amongst objectivists as
between objectivists and subjectivists seems to have been ignored. The problem for Mabbot
here is to pick out just which objectivist view is the paradigm of ‘normality’. It seems we
may, with equal ease, pick out some popular characteristic that is common to subjectivists
and many objectivists (say, for example, the belief that ethics is independent of religion) and
brand all other uses of ethical language as ‘abnormal’. It is all too easy to label difficult
people as ‘abnormal’ compared with taking serious account of them in one’s linguistic
theory.
This ends my preliminary remarks. I have tried to show, using Mabbott’s example,
that to take a too simple view of the function of Stevenson’s theory will inevitably lead to
misunderstanding it. We may now progress to a consideration of Stevenson’s more
sophisticated critics, bearing in mind the lessons we have learned.
7
J. D. Mabbott [1977: 96]
Downloaded from http://www.RationalRealm.com
8
Leslie Allan
A Defence of Emotivism
5. Objections to Stevenson’s Emotivism
5.1 Personal Moral Deliberation and Advice
In the following discussion, I shall concentrate on the objections of three critics of
Stevenson’s analysis; R. B. Brandt, H. J. McCloskey and J. O. Urmson. Brandt is a
multifunctionalist and his views on emotivism can be found in his Ethical Theory [1959:
ch. 9]. McCloskey is an intuitionist. His criticisms are published in Meta-Ethics and
Normative Ethics [1969: ch. 3], while Urmson, who holds a grading theory of ethical terms,
published a book devoted to emotivism, The Emotive Theory of Ethics [1968]. I shall discuss
their criticisms in order of importance, leaving aside their more peripheral objections for
reasons of space.
The most telling and obvious objection to emotivism, for many people, is that it
simply does not describe what we do when we are making up our mind on moral issues.
According to this objection, deciding what we ought to do in a particular situation is
different from asking ourselves what we want to do, sorting out our attitudes or engaging in
self-persuasion. As Brandt [1959: 220] says, ‘Very often we conclude, after moral reasoning,
that what we most insistently want to do must simply be set aside (or at least that it ought
to be).’ McCloskey [1969: 61] makes the same objection.
In his earlier book, Ethics and Language, Stevenson saw self-persuasion as the
psychological process we go through for overcoming the conflicts in our attitudes ‘in almost
all of our personal deliberations’ [1976: 148]. I doubt whether this is true, and in his later
analysis of personal deliberation, Stevenson [1963: 197–204] made no mention of
self-persuasion. I do not, therefore, care to defend this aspect of his earlier view. But I do
think that his general view that moral indecision reflects uncertainty over what attitude to
adopt or is a manifestation of conflict of attitudes is correct.8
A simple psychological point now needs to be made, and it is a point that has been
duly emphasized by Stevenson [1963: 200–2] in his later work and developed at length by
D. H. Monro, in his book Empiricism and Ethics [1967: ch. 17]. It is this. Most, if not all, of us
have ideals; ideals about what kind of person we would most like to be and about what kind
of world we would most like to see. These ideals are the result of reflective thinking and are
relatively permanent. They are readily contrasted with our more impermanent and less
reflective preferences. We have, then, differing levels of attitudes, ranging from the
highest-level to the lowest-level. Morality is concerned with these highest-level, long-lasting
and most reflective attitudes.
The second point that I wish to make is that, in ordinary language, different terms
are ascribed to different preferences, depending on their level. Our highest-level
preferences are labelled ‘values’, while our lowest-level preferences are labelled ‘wants’.
Some terms, such as ‘desire’, are somewhat ambiguous, sometimes being applied to
low-level preferences and sometimes to high-level preferences. Once this clarification has
8
See Stevenson [1976: 130–2; 1963: 197–204] for a statement of his position.
Downloaded from http://www.RationalRealm.com
9
Leslie Allan
A Defence of Emotivism
been made, it becomes clear why moral decision making is not especially concerned with
determining ‘wants’ or low-level ‘desires’.
McCloskey, in Meta-Ethics and Normative Ethics [1969: 60f], draws the example of
the person who discovers ‘on reflection that he really does think members of other races
are inferior, and that he has been treating them as such all his life’ and suggests that
Stevenson is committed to saying that this person has ‘the problem of harmonising this
newly discovered attitude with his other known attitudes’ and not of ‘suppressing it and
cultivating it out of existence’. This is inferred, McCloskey presumes, from Stevenson’s
[1976: 132] talk of ‘systematising one’s actual and latent attitudes in a way which gives
them definite direction’.
However, McCloskey has here seriously misconstrued Stevenson’s position.
McCloskey [1969: 60] has mistakenly taken Stevenson’s view to be that personal
deliberation is the ‘discovering and developing one’s latent attitudes’ [italics mine]. But on
the very same page from which McCloskey quotes Stevenson, Stevenson [1976: 132]
illustrates the example of the conscript who, upon discovering his latent attitude of ‘fear of
being killed’, decides to ‘inhibit it out of shame’. So, there is nothing to stop, in my view, the
latent racist in McCloskey’s example above deciding to suppress his racist attitude.
Stevenson writes in the same place of ‘modification of one of the conflicting attitudes’
[1976: 131] and the causing of ‘one set of attitudes to predominate over the other’ [1976:
132] in personal deliberation. He further elaborates this analysis in his later work [1963:
199–202].
McCloskey [1969: 63] further objects to Stevenson’s view that to adopt a moral
position is to adopt an attitude on the grounds that we can cultivate new non-moral
attitudes but we cannot cultivate new moral beliefs. He uses the example of cultivating a
liking to English beer when living in England and accustomed to German beer.
But are moral attitudes on par with attitudes to beer? According to Stevenson's
emotivism, they are not. They are concerned with our most permanent and reflective
attitudes, and these we do not choose any more than we choose our beliefs. We no more
choose to have a pro-attitude towards sexual equality than we choose to believe that
Ottawa is the capital city of Canada. In fact, for any non-cultivatable moral belief that
McCloskey cares to point out, the emotivist will point out a non-cultivatable moral attitude.
If McCloskey wants to say that all attitudes are cultivatable, then he must be in the position
of saying that there are no such things as moral attitudes. But it is not unusual to ask of
someone, ‘What is your attitude to war? or euthanasia? or in vitro fertilisation?’
So, not all attitudes are cultivatable, and moral attitudes are included in this
sub-class. McCloskey may be granting this point, but arguing that moral positions cannot be
identified with particular attitudes because some attitudes are cultivatable, and moral
positions are non-cultivatable. However, to argue that two classes A and B are not identical
because class A has a characteristic that some members of the larger class to which class B
belongs lacks is logically invalid. McCloskey’s argument, if this latter interpretation is
correct, has the same formal structure as this invalid argument: Laughing bipeds cannot be
identified with humans because laughing bipeds have teeth and humans are animals, some
of which do not have teeth.
Downloaded from http://www.RationalRealm.com
10
Leslie Allan
A Defence of Emotivism
McCloskey further supports his criticism of Stevenson’s view that personal
deliberation concerns uncertainty or conflict in attitudes by appealing to what we do when
we seek moral advice. According to McCloskey [1969: 48, 61f], when we seek such advice
we are neither sorting out our attitudes nor engaging in self-persuasion, but are
endeavouring to find objective moral truths.
I shall not attempt to defend the view that we seek advice for self-persuasion,
although this may be true for a minority of cases. It seems to me that we seek advice in
order to settle tensions between our attitudes or to alleviate uncertainty in what attitude to
take to a particular issue. For the sake of clarity, I shall divide people that seek advice into
two broad categories. In the first category are those that think they know in general terms
what is good or what they ought to do, but do not know whether a particular thing is good
or a particular act is what they ought to do. A clear example of this is the newly-wed woman
who believes that she ought to act in accordance with God’s will but is not sure if taking the
contraceptive pill is against his fiat.
In the second category are those persons who are not sure what is good or
obligatory in general terms and find themselves forced into making a moral decision. An
example of this is the young father who must decide between placing his aged mother into a
nursing home and caring for her himself. This is a vexed question for him as the latter option
involves disruption to his own family. Here, he is unsure of the criteria for making the right
choice.
That those people in the first category believe they are seeking facts is easily
explained on Stevenson’s second pattern of analysis. The newly-wed woman in the example
above has descriptively defined ‘ought’ as referring to a constraint imposed by God’s will.
Her objective in seeking advice is to discover whether taking the contraceptive pill satisfies
the requirement of the definition, thereby, in effect, alleviating her uncertainty in her
attitude to this contraceptive method. Opponents of emotivism may object that in cases
such as these, in which only purportedly factual information is being sought, it is not strictly
correct to say that moral advice is being sought. However, to say of the woman’s question
to her pastor, ‘Ought I refrain from taking the Pill?’, that it is not a request for moral advice
is surely to stretch the meaning of ordinary language.
This example also makes it clear why the woman’s wishes do not seem relevant in
the solution to her moral dilemma. She is settled in her highest-level attitude; that is, she
most desires to do God’s will. The advice she seeks is the answer to the question: What, in
fact, is God’s will? Her lower-level wants and desires are not relevant to the answering of
this question, although, it should be noted, they may have sparked her dilemma. In this
case, her desire for a safe and efficient contraceptive method to avoid pregnancy most
probably necessitated a visit to her pastor.
Let us imagine that the woman’s pastor now tells her that non-natural contraceptive
methods are absolutely forbidden by God and that, for a while, she accepts this restraint. In
time, however, she comes to realize that this ban causes needless suffering on a large scale.
She may begin to doubt that God really forbids non-natural contraceptive methods and,
after further inquiry, decides that her pastor was wrong. Alternatively, she may question
whether God really does command what is right and, after further reflection, give up her
belief that God exists on this count. If she now seeks advice for this reason, that she does
not know what the proper criteria are for determining the morality of contraception, she
has fallen into my second category.
Downloaded from http://www.RationalRealm.com
11
Leslie Allan
A Defence of Emotivism
She may, as with most people who seek advice on these grounds, believe the criteria
in question to be objective; that is, that they are substantially or completely independent of
human attitudes and desires. The sophisticated emotivist does not deny that most people
believe this. But whatever criteria our newly-wed decides upon after seeking advice,
whether it be a system of respect for innate rights, a system of prima facie obligations or a
single criterion, such as maximising happiness, the emotivist is committed to saying that she
has, fundamentally, settled her highest-level attitudes, even though she may not have
realized this herself. This is part of Stevenson’s extra-linguistic and epistemological analysis
of ethics I referred to in §3 above. It is his attempt to explain what is happening in ethics
from an empiricist viewpoint. At this level of analysis, McCloskey’s objection that people
seeking moral advice do not implicitly accept Stevenson’s analysis is beside the point.
Let us, for a moment, consider a parallel case. Suppose an atheistic philosopher and
sociologist provided us with a psycho-social explanation of theistic belief systems, showing
their psychological origin in some basic human need, the means of social reinforcement of
such beliefs and their dependence on the adoption of irrational criteria for theory choice.
And suppose his general theory, with the addition of further details, explained the
phenomenon of church attendance in these terms. Now, for a theist to object to this
analysis of church attendance on the grounds that church attendees really do believe they
are worshipping God at church, and so do not implicitly accept his theory, is to engage in
irrelevance. What people think they are doing and what they are in actual fact doing are not
necessarily identical.
McCloskey has here confused an account of the meaning of terms (in this case,
ethical terms) with an analysis of their psycho-social function and epistemological import.
He then proceeded to criticize the former on the grounds that ordinary people do not agree
with the latter analysis. As we have seen in §3 above, there are complex interconnections
between these three aspects of meta-ethical analysis. However, if McCloskey had wished to
criticize Stevenson’s linguistic analysis on the grounds that it is dependent on a faulty
extra-linguistic analysis, this requires more work to be done than simply pointing out that
ordinary people do not believe it.
My second point on this second category of advice is that not all the people that
seek such advice are objectivists. That a significant minority of these people are subjectivists
and relativists is little appreciated by objectivist moral philosophers. These people are
usually defined out of existence. When such people seek advice, they explicitly aim to settle
the tensions between their highest-level attitudes. By drawing to their attention new facts
or re-emphasising familiar ones, they hope that they will see the situation in question in a
new light. They anticipate that this will consequently bring to bear a psychological influence
in reorientating their highest-level attitudes.
I think that requests for moral advice are more complex than the schemata here
suggests. There will be borderline cases and cases that fall into both categories. None the
less, I believe that I have satisfactorily answered McCloskey’s objection. Stevenson’s
‘descriptive meaning’ component in his second pattern of analysis accounts for the apparent
fact-finding nature and lack of relevance of personal attitudes in the first category of moral
advice. The second category of moral advice is accounted for by his extra-linguistic and
epistemological theory of what moral uncertainty really is.
Downloaded from http://www.RationalRealm.com
12
Leslie Allan
A Defence of Emotivism
5.2 Expressing and Directing Attitudes
The next objection that I wish to consider is that ethical statements do not always
express the attitude of the speaker. In his book, Ethical Theory [1959], Brandt illustrates this
criticism with two examples. The first is this.
Suppose someone, whom you meet by chance on a train, asks your advice about
the selection of a college for his daughter. You may give careful thought to his
description of the problem, and conclude by saying, ‘I think she ought to apply at
Pembroke College, at least among other places.’ But do you really care where she
goes? Or does your hearer suppose you care?
[Brandt 1959: 229f]
I think that a more instructive question to ask is, Is this an ethical statement? This
example seems on par with the case of the person who asks me how to get to the nearest
phone box and I answer, ‘You ought to turn left at the next corner’. These uses of ‘ought’ do
not seem to be ethical, but instead indicate a causal relationship between a course of action
and an intended outcome. Their meaning appears to be: ‘For your daughter to succeed,
send her to Pembroke College or a similar college.’ and ‘To get to the nearest phone box,
take the next turn left.’
Brandt’s next example is more troublesome. It is this.
Suppose a fundamentalist is reading the Biblical account of the—apparently
pointless—execution of someone (say, all men, women, and children among the
Amalekites) at the behest, allegedly, of God. He says to himself, ‘That’s strange.
But it must have been right; otherwise God would not have ordered it.’ This use
of ‘right’ appears to be a normal one.
[Brandt 1959: 230]
In this case, Brandt claims, the function of the statement is not to express the
person’s attitude but to direct it. Brandt may be correct in naming the directive function of
the statement in this case. However, this does not automatically preclude it from also
having an expressive function. Consider, for a moment, a parallel case. An amateur admirer
of science is reading a popular account on relativity and comes across the view that the
mass and length of a body are dependent on its velocity. He says to himself, ‘That’s strange.
But it must be true; otherwise physicists would not accept it.’ Now, it may be the case that
the statement serves to direct his belief, but I see no reason to deny that when he says, ‘But
it must be true’, he is also expressing his belief.
Similarly, when I tell my friend, ‘There’s ants in your pants!!’, my statement serves to
direct his attention and to command (to take off his pants). However, it also serves to
express my belief. So, the one statement may have more than one illocutionary force.
Brandt does not say why he thinks the fundamentalist in his example cannot also be
expressing an attitude at the same time. I conclude that neither of Brandt’s examples
succeed in showing that ethical statements sometimes do not express attitudes.
Downloaded from http://www.RationalRealm.com
13
Leslie Allan
A Defence of Emotivism
Brandt [1959: 227] also makes the claim that ethical statements do not always
function to direct the attitudes of the hearer. (I shall take it that ‘inviting-so-to-speak’ is a
form of directing.) Two of the examples that he gives are particularly instructive, so I shall
examine these first. The examples are; ‘If the policy is morally wrong, then that settles the
matter.’ and ‘It must be right or wrong—so let’s think it through and find out which.’ It is
certainly true that neither of these statements have the illocutionary force of commending
their relative ethical subject matters. But neither do they serve to express the speaker’s
attitudes to them. In fact, these two examples would have better served the point that
Brandt had wished to make in his previous criticism.
Does the fact that some ethical statements do not have the illocutionary forces of
expressing attitudes and commending9 debilitate Stevenson’s theory of emotive meaning? I
do not think so. Urmson’s work, in his The Emotive Theory of Ethics [1968: ch.11], on the
relevance of illocutionary forces to the meanings of words is particularly instructive. His
analysis, briefly, is this. (Note that the examples given here are mine and not Urmson’s.)
Although the sentences in which a particular word appears vary widely in their
illocutionary forces, knowledge of certain illocutionary forces in some paradigm uses of the
word is essential to understanding that word’s meaning. Those illocutionary forces that are
essential to understanding a word’s meaning are termed ‘central illocutionary forces’. In
many cases of the use of a word, these central illocutionary forces will not be present.
Let us take an example. Young Robert is at primary school and overhears his peers
calling his friend Samuel a ‘nigger’. Robert, believing the term to be simply a reference to
Samuel’s African origin, adopts the term himself in referring to his friend. He has failed to
understand the meaning of ‘nigger’ because he does not realize that to say, ‘You’re a
nigger’, is to have the illocutionary force of derogation. This illocutionary force is a central
one.
Robert begins to have doubts concerning his understanding after hearing his peers
mock Samuel. He asks his father, ‘Is Samuel a nigger?’ Here, the central illocutionary force of
derogation is not present at all; the function of the sentence is to ask a question.
One more example should suffice. Suppose a space traveller from a more technically
advanced planet arrived on earth and is in the process of mastering our language. He
informs us that our theory of relativity is seriously deficient and, after speaking to a
physicist, proclaims, ‘Einstein’s theory is true, but I do not believe it’. Let us further suppose
that we have good independent reasons for believing that the first part of his statement is
merely mimicking what he had heard from the physicist, and so we duly inform our
extra-terrestrial friend that he has committed a semantic error. We explain to him that to
say sincerely, ‘X is true’, is, by linguistic convention, to express a belief.
Until he has recognized this central illocutionary force, he has not fully understood
the meaning of the word ‘true’. The knowledge of this central illocutionary force is
semantically essential, even though all sentences of the form, ‘If A is true, then B’, do not
9
I shall henceforth use the term ‘commending’ in place of Stevenson’s ‘inviting-so-to-speak similar attitudes in
the hearer’ for the sake of readability. For those readers that do not consider the two terms equivalent,
please substitute the latter for the former wherever it appears.
Downloaded from http://www.RationalRealm.com
14
Leslie Allan
A Defence of Emotivism
express a belief that ‘A’. What I think these examples show is that the fact that not all uses
of ethical terms have the illocutionary forces of commending and expressing is not
detrimental to the theory that knowledge of these forces is essential to the linguistic
analysis of these terms.
Brandt [1959: 227f] follows his examples with a criticism using, as an illustration, a
mother reproving her child. His remarks here are more appropriately directed to Hare’s
prescriptivist theory, for Stevenson disdained ‘imperativist’ analyses and did not believe that
the function of ethical language was to directly influence behaviour. In fact, Brandt [1959:
228] appears to be conflating the two theories. He necessarily connects the ‘intent to
influence the attitudes of hearers’ with the use of imperatives. In another place [1959: 226],
he claims that to have ‘special directive influence on the feelings and attitudes of others’ is
to tend ‘to make a person behave much as he would if he were obeying the order “Don’t do
x!”’. It is certainly true that Stevenson’s view in Ethics and Language was that an attitude is
a disposition to behave in certain ways [1976: 60], but it was not his view that in ethics we
act as if we were obeying imperatives [1976: 32f].
Brandt [1959: 218f] has another argument against the view that ethical statements
serve to direct the attitudes of the hearer. He puts it thus. We must distinguish, he claims,
between what a person thinks he ought to do and what he wants to do. The two are not
always congruous. Similarly, he says, we must also distinguish between two types of
argument: that a person ought to do something and that he do something. Ethical argument
is the former but not the latter type. We may persuade a person that he ought to do
something, and thus succeed in our ethical argument, but he may not want to do it. For
Brandt, our subsequent attempt to redirect his attitude so that he wants to do what he
ought to do is no longer within the bounds of ethical argument.
The distinctions that Brandt draws here are correct. However, his argument does not
prove what he intended it to prove; that ethical argument does not function to redirect
attitudes. This is because he uses the term ‘attitude’ to refer exclusively to ‘wants’. We have
already seen in §5.1 above that ‘wants’ are not the particular concern of ethical argument,
and here Brandt is right. However, it is simply not the case that all attitudes are ‘wants’ in
this ordinary low-level sense. Ethical language concerns our highest-level attitudes—our
ideals—and it is these that come into play in our ethical disputes. 10 We have persuaded a
person that he ought to do X when he agrees with us that doing X is consonant with his
highest-level pro attitudes, either by changing his beliefs about X or by changing his
highest-level attitudes, or a combination of both.
Once again, Brandt appears to have reduced all emotivist theories to one form or
another of Hare’s imperativist theory, in which Hare postulated a direct relationship
between the acceptance of an ethical position and voluntary behaviour. For Hare, to have
convinced a person that he ought to do X logically entails that the person is convinced to
want to do X. Stevenson agreed with Brandt that this is not the case; that to say, ‘I ought to
10
In his argument, which has been simplified here, Brandt has also treated the term ‘motivation’ as referring
solely to short-term motivations to do a particular act. The fact that ‘motivation’ may also refer to long-term,
reflective intentions is ignored by him. This lack of distinction leads Brandt to equate all emotivist theories
with the view that a direct function of ethical language is to guide low-level, short-term, unreflective motives.
Downloaded from http://www.RationalRealm.com
15
Leslie Allan
A Defence of Emotivism
do X but I do not want to’ is not necessarily to use ‘ought’ in a non-ethical sense. Stevenson
would have accepted Brandt’s argument here as a point against Hare’s prescriptivism.
In places, emotivists such as Stevenson [1963: 199–202] and Monro [1967: ch. 17]
explain the fact that a person does not always want to act in a way that he thinks he ought
to act by referring to the person’s different levels of attitudes. If, for example, he thinks he
ought to mow his invalid neighbour’s lawn, but decides not to from laziness, he has a
low-level con attitude to mowing his neighbour’s lawn. But he has also a high-level pro
attitude to helping invalid people. So, we may say, he has a high-level con attitude to his
low-level con attitude to mowing his neighbour’s lawn. This psychological incongruity
between the two levels of attitudes will manifest itself as feelings of guilt. We may conclude,
then, that Brandt’s argument is not applicable to emotivist views that provide a logical gap
between ethical language and voluntary behaviour, such as that of Stevenson’s.
We are now in a position to distinguish between moral and non-moral attitudes. It is
important to note that I am not here distinguishing between moral and immoral attitudes,
which is a normative judgement, but between those attitudes that are in the domain of
morality and those that are not. The latter separation of attitudes is a formal meta-ethical
distinction.
A moral attitude, then, is either a highest-level attitude concerned with character or
‘state of the world’ ideals, or is a lower-level attitude that is the object of one or more of
these highest-level attitudes, or is the object of an intermediate-level attitude that is itself
the object of one or more highest-level attitudes. A lower-level moral attitude will either be
in conformity or in conflict with highest-level attitudes, either directly or via
intermediate-level attitudes. A lower-level attitude, on this view, may be called ‘bad’,
‘wrong’ or ‘immoral’ by its holder, and yet be a pro attitude, because it is incongruous with
his highest-level attitudes, as our lawn-mowing example above illustrates. Attitudes that are
considered to be neither in conformity nor in conflict with highest-level attitudes are
non-moral. For example, for many people in our society, a person’s preference for either
white wine or red wine, or for wine at all, is considered to be outside the domain of
morality.
Whether a particular attitude expressed is a moral attitude or a non-moral attitude
will depend on the person who expressed it. For example, an Australian abattoir owner who
says, ‘Slaughter sheep facing Mecca’ (in order to obtain export contracts from Islamic
countries), is expressing a non-moral attitude if he considers such means of increasing his
profits to be morally irrelevant. On the other hand, for an Arabic Moslem to say, ‘Slaughter
sheep facing Mecca’ (in order to please Allah), is to express a moral attitude. Whether an
attitude to something, such as facing sheep towards Mecca before slaughtering or a taste
for alcoholic beverages, ‘ought’ to be a moral attitude is sometimes itself the subject of
ethical debate. The solution to such arguments, of course, requires normative judgement,
which is outside the purposes of this essay.
Downloaded from http://www.RationalRealm.com
16
Leslie Allan
A Defence of Emotivism
5.3 Reasons in Ethical Argument
The next objection that I wish to address is that Stevenson’s analysis of reasons in
ethical argument is inadequate. McCloskey and Urmson have devoted significant attention
to developing this criticism. Unfortunately, much of it is based on a misunderstanding of
Stevenson’s views. I shall here concentrate on McCloskey’s criticisms.
For Stevenson [1976: 115–29], a ‘reason’ in moral argument is a statement that
expresses a belief intended to influence the moral attitudes of an opponent by means of
altering his beliefs. He lists four categories of reasons and gives illustrations of each. 11 In his
Meta-Ethics and Normative Ethics [1969: 57f], McCloskey claims that because the reasons
cited in the third and fourth categories are ‘reasons which influence behaviour’ and ‘are
directed at terminating the conflict but not the disagreement nor the divergence of
attitude’, then, for Stevenson’s list to have been complete, it should have included such
‘reasons’ as ‘threats and torture (where this takes the form of statements, e.g. that one’s
wife and children have been or are being cruelly killed’. Clearly, McCloskey concludes,
threats and torture are not moral reasons.
Is it true that Stevenson had confused influencing attitudes with influencing
behaviour? I see no reason to believe that the types of reasons given in Stevenson’s [1976:
121–6] third category (examples 8–12) are not primarily directed to influencing attitudes.
And McCloskey neglects to say why he thinks otherwise. Stevenson [1976: 127–9] prefaces
his fourth category (examples 15–18) with the statement that the motives behind the
reasons in this category are ‘less concerned with resolving disagreement in attitude than
with temporarily evading the force of a disconcerting influence, or altering the means by
which it is exerted’. I think that Stevenson had underestimated the importance of
redirecting attitudes in the examples that he gave. Never the less, he would have
maintained that the reasons in this category are ethical reasons only in so far as they serve
to redirect moral attitudes. Once again, McCloskey has not given us his reasons for believing
that they do not serve to redirect attitudes at all. To demonstrate here that the reasons in
these examples do in fact serve this function would unnecessarily extend the scope of this
essay. I shall leave it to the reader to examine these examples for themself.
McCloskey [1969: 57f] makes a second point and it is this. Stevenson’s list of reasons
that influence attitudes is ‘seriously incomplete, as it should include as group (vi) “reasons”
such as brainwashing, subliminal influencing (if through the use of statements) and the like’.
Clearly, McCloskey concludes again, these are not moral reasons. Strictly speaking,
‘brainwashing’ and ‘subliminal influencing’ are not ‘reasons’, even if they do make use of
‘statements’. These means are more properly regarded as methods of influence. They are
not themselves ‘statements of belief’, which is what, for Stevenson, reasons are.
More importantly, Stevenson would have agreed whole-heartedly with McCloskey
that brainwashing and subliminal influencing are not rational; that is, that they are not
‘reason-using’ 12 methods. Stevenson [1976: 139f] lists some of these non-rational methods:
11
McCloskey [1969: 57] attributes a fifth category of reason to Stevenson. In fact, this category is not a
category of reason, but a category of non-rational persuasion (see Stevenson [1976: 139–47]).
12
The theoretical distinction between reason-using and non-reason-using methods is given in Stevenson [1976:
139f].
Downloaded from http://www.RationalRealm.com
17
Leslie Allan
A Defence of Emotivism
‘rhetorical cadence, apt metaphor, stentorian, stimulating, or pleading tones of voice,
dramatic gestures, care in establishing rapport with the hearer or audience’ . . . ‘use of
material rewards and punishments, and also (for instance) the various forms of public
demonstration and display.’ In fact, any method that goes beyond a dispassionate
statement of beliefs is to that extent considered, by Stevenson, to be non-rational. He goes
on to devote a whole chapter to the consideration of such methods in Ethics and Language
[ch. IV]. Brainwashing and subliminal influencing indisputably fall within this category of
non-rational methods. Emotivists such as Stevenson, therefore, are in agreement here with
McCloskey. 13
McCloskey devotes considerable space to his next point:
Stevenson is committed to identifying a valid, adequate reason, i.e. a good
reason, with a successful, influencing factor which is in the form of a statement.
The test for him is whether it leads to agreement in attitude . . . Whether a
consideration successfully influences belief or attitude is not the factor which
determines whether it is a relevant or a good reason.
[McCloskey 1969: 58]
We may agree with McCloskey’s view that the relevance and goodness of a reason is
not determined by whether it is in fact successful. However, to attribute a divergent view to
Stevenson, as McCloskey does, is to seriously misconstrue his position. In support of his
argument, McCloskey [1969: 57] quoted Stevenson’s [1976: 114f] statement that ‘Whether
this reason will in fact support or oppose the judgement will depend on whether the hearer
believes it, and upon whether, if he does, it will actually make a difference to his attitudes
. . .’ This quote may appear to coincide with the view that McCloskey attributes to him.
However, for Stevenson to say that a reason ‘supports’ a judgement is no more to imply that
the reason is ‘good’ or ‘relevant’ than it is for McCloskey to say that a reason is ‘successful’
in altering a judgement. In fact, this is all that Stevenson meant by saying that a reason
‘supports’ a judgement. Whether a reason is ‘good’, ‘adequate’ or ‘relevant’ is an entirely
different question for Stevenson, requiring a normative judgement.
That this is Stevenson’s view is clear from his forceful and extensive treatment of this
issue. He had devoted a complete chapter in Ethics and Language [ch. VII] to the task of
demonstrating that meta-ethical inquiry could not distinguish ‘valid’ from ‘invalid’ reasons,
because any attempt to do so was a normative judgement. 14 His analysis was continued
with respect to the term ‘justified’ in a later essay:
A methodological inquiry, when it attempts to find the R’s [factual reasons] that
will justify a given E [evaluative conclusion], does not stand apart from an
evaluative inquiry but simply continues it, yielding ordinary value judgements that
are expressed in a different terminology.
[Stevenson 1963: 89]
13
The comments offered here are also relevant to Brandt’s similar objection he makes in his [1959: 219].
See especially Stevenson [1976: 170f].
14
Downloaded from http://www.RationalRealm.com
18
Leslie Allan
A Defence of Emotivism
Stevenson believed that to judge reasons as being ‘good’, ‘relevant’, ‘adequate’, and
so on, was to morally evaluate them, and not to describe them as being successful or
effective. In his meta-ethical inquiry, Stevenson [1976: 113, 159f] had continually sought to
distance himself from such evaluations. He not only did not hold the view that evaluations
of reasons followed from his descriptive methodology, and which McCloskey erroneously
attributed to him, he also explicitly warned against it. As Stevenson [1976: 165] had said, ‘It
has been explained that “validity” introduces nothing novel into ethics, and that no chaotic
implications (such as “One method is as good as another, so long as it impresses people”)
need be feared from this conclusion.’ 15
It is curious that following more than one full page of criticism on this misguided
point, McCloskey gives an accurate representation of Stevenson’s view on reasons in ethics.
McCloskey [1969: 59] writes, ‘Equally serious is the suggestion implicit here that ‘relevant’,
‘irrelevant’, ‘adequate’, ‘inadequate’, ‘genuine’, ‘pseudo’, ‘good’ and ‘bad’ are emotive
words which admit of arbitrary persuasive definition in the moral sphere . . .’ Leaving aside
the point that McCloskey’s use of the word ‘arbitrary’ is an unnecessarily harsh
representation of Stevenson’s view, I make the point again that this view of Stevenson’s
that McCloskey now recognizes is most explicit in Stevenson’s writings.
What is equally perplexing is that McCloskey can find this view ‘implicit’ in his
incorrect interpretation of Stevenson’s position when the two views appear to be logically
contradictory. The view that McCloskey initially falsely attributes to Stevenson is a
descriptivist, relativist theory; that ‘X is a good reason’ means ‘X is considered a good
reason by person Y’ or ‘X is effective in influencing person Y’s attitudes’. The view that
McCloskey now correctly attributes to Stevenson is that ‘X is a good reason’ means ‘X has
qualities or relations A, B, C . . .’ along with its emotive components of expressing and
inviting-so-to-speak certain attitudes. I know of no way of reconciling these two views.
McCloskey’s [1969: 60f] next points to Stevenson’s view that ‘disagreements about
reasons and methods of argument are simply second order disputes of the same general
character as first order normative disputes, with the same sorts of reasons available to
resolve them’. McCloskey contends that Stevenson is wrong here and offers the following
argument in support:
‘The person who refuses to accept as a good reason for judging A to be good, that
it is identical with B, which he does judge to be good, is fully as absurd as the
person who denies that he has been offered a good reason for not crossing the
road now that a truck is speeding around the corner at him. For Stevenson, the
former is a good reason only if we have a pro attitude towards consistency; if we
have a con attitude, it is not a good reason. It is entirely up to us!’
15
The above clarification of Stevenson’s views is also relevant to Brandt’s similar objection in his [1959: 219].
McCloskey [1969: 58] claims that Stevenson’s discussion in his [1963: 194–7] of the case of a father
considering lying to his son, to influence his choice of a career, was explicitly designed to deal with the
difficulty that Stevenson’s analysis faced in implying that successful lies constitute ‘good reasons’. Contra
McCloskey, it is not the case that Stevenson felt compelled to deal with this difficulty because, as I think I
have demonstrated, it was not a problem for his theory. Stevenson, in fact, used his ‘father and son’ example,
along with another ‘swinging voter’ example, to deal with a quite unrelated problem for his theory: that it
stressed uncertainty and disagreement in ethical discussions while neglecting other motivating factors.
Stevenson had stated this purpose clearly and unambiguously.
Downloaded from http://www.RationalRealm.com
19
Leslie Allan
A Defence of Emotivism
McCloskey here again misunderstood Stevenson. Stevenson had allowed exceptions
to his rule that reasons in first order and second order disputes are psychologically and not
logically related to ethical judgements. On first order disputes, Stevenson [1976: 116]
writes, ‘In general, ethical statements, like all others that have some descriptive meaning,
are amenable to the usual applications of formal logic.’ Those cases in which formal logic is
applied ‘present exceptions to the rough but useful rule mentioned previously—the rule
that ethical judgements are supported or attacked by reasons related to them
psychologically, rather than logically’ [1976: 115].
As with first order disputes, Stevenson made the same exception with second order
disputes: ‘To evaluate or recommend an ethical method (whenever validity can have no
bearing on the case) is to moralize about the ways of moralists.’ 16 So, to criticize or support
an ethical method on the grounds of logical or inductive validity is not to engage in moral
evaluation. Stevenson [1976: 158] had expressly stated his view that logical principles are
not normative: ‘Some may wish to contend that “validity” itself, even in the conventional
sense that applies to logic and science, is a normative term; but the writer suspects that any
such contention would involve a misleading use of either the term “validity” or the term
“normative. ”’
So, to consider the person in McCloskey’s example, if A and B are strictly empirically
or logically identical, 17 Stevenson would have agreed that that person had a ‘good reason’
to judge A to be good irrespective of whether he or we (whoever ‘we’ are, McCloskey does
not tell us) have a pro or con attitude to logical consistency. And this is for the simple reason
that the ‘good reason’ given to him is a logical reason. Similarly, if that person refused this
reason, he would be ‘absurd’ irrespective of our attitudes, simply because it is a logical
absurdity that is being described here. As we have seen, for Stevenson, logical reasons are
not dependent on normative judgements and so are independent of our attitudes.
In his book, The Emotive Theory of Ethics [1968], Urmson had also misconstrued
Stevenson’s view on the role of logic in ethical argument. It is Stevenson’s contention,
Urmson [1968: 71] urged, that ‘no fact is logically more relevant to a disagreement in
attitude than any other fact’. 18 Much of Urmson’s critique of Stevenson’s emotivism
depends on this false understanding of his position. Without examining Urmson’s views in
detail here, I will simply note that my earlier clarification of Stevenson’s position is equally
relevant here.
16
Stevenson is here using the term ‘validity’ in its logical and inductive sense. See his [1976: 158] and f. n. 4 on
same page.
17
If it was simply a matter of class identity, then there would be no logical contradiction involved in the person
saying, ‘A is not good and B is good, and A and B belong to the same class’.
18
See also Urmson [1968: 48, 64].
Downloaded from http://www.RationalRealm.com
20
Leslie Allan
A Defence of Emotivism
5.4 Persuasive Definitions
The final criticism that I wish to consider is McCloskey’s [1969: 53] objection that
Stevenson’s second pattern of analysis elucidating the function of ‘persuasive definitions’ in
moral discourse is inadequate because moral philosophers are not moral propagandists and
do not act as such. This shows, McCloskey claims, that Stevenson’s ‘persuasive definition’
theory ‘bears on only one type of use—or abuse—of moral expressions’.
Behind this criticism, I think, lie two false assumptions. The first is that all definitions
of moral terms are given by professional philosophers. If we grant, for the sake of argument,
McCloskey’s point that no moral philosopher defines moral terms persuasively, this does
not show, ipso facto, that in those cases in ordinary discourse where moral terms are
defined, their function is not frequently persuasive.
The second false assumption is that Stevenson had argued that all philosophical
definitions are persuasive. McCloskey labours this point. As McCloskey [1969: 53] wrote, ‘If
all definitions of “good” really were the arbitrary definitions this [Stevenson’s] theory
represents them as being, the reactions to Moore’s contentions are astonishing . . .’ and so
on. However, this is a misunderstanding of Stevenson’s position. Here is what Stevenson
had to say on the matter:
It must not be thought, of course, that all definitions of emotive terms are
persuasive, or that they need always have an inadvertent persuasive effect. A
speaker whose purposes are mainly descriptive can neutralize the effects of
emotive meaning by intonation or explicit admonition. At times even this is
unnecessary, for if the general situation is one which militates against persuasion,
or if the speaker and hearer concur on the relevant evaluative matters, the actual
emotive effects may be without practical bearing. The definition can then
proceed as though the term were emotively neutral, and may serve a primarily
descriptive purpose. It remains the case, however, that a great many definitions
of emotive terms are persuasive, in intent and in effect.
[Stevenson 1976: 212]
From this, we can safely conclude that Stevenson regarded the dispassionate
analyses of moral philosophers as lying outside the ambit of persuasive definitions.
Downloaded from http://www.RationalRealm.com
21
Leslie Allan
A Defence of Emotivism
6. Conclusion
This concludes my defence of Stevenson’s emotivism. In this essay, I tried to show
how the three aspects of his thesis relate to the broader realm of meta-ethical inquiry. I also
considered a number of significant objections to it: that most people are objectivists, that it
provides an inadequate account of personal indecision, moral advice and reasons, that the
‘expressive’ and ‘directive’ theses are mistaken, and that moral philosophers do not use
persuasive definitions. These criticisms, I argued, result from misunderstandings regarding
the nature of a meta-ethical theory and an inattentive reading of Stevenson’s views. In
attending to these objections, I further elaborated on the psychological nature of moral
attitudes and attempted to present a clear account of the distinction between these and
non-moral attitudes. Many criticisms of Stevenson’s emotivism have been left unanswered
and a number of points have not been elaborated fully. Notwithstanding these
shortcomings, my purposes here will have been fulfilled if I have gone some way to
dispelling some of the more troublesome confusions.
Downloaded from http://www.RationalRealm.com
22
Leslie Allan
A Defence of Emotivism
References
Ayer, A. J. 1971. Language, Truth and Logic, Harmondsworth: Penguin.
Brandt, R. M. 1959. Ethical Theory, Englewood Cliffs: Prentice Hall.
Feldman, F. 1978. Introductory Ethics, Englewood Cliffs: Prentice Hall.
Foot, P., ed., 1977. Theories of Ethics, Oxford: Oxford University Press.
Hare, R. M. 1972. The Language of Morals, Oxford: Oxford University Press.
Mabbott, J. D. 1977. An Introduction to Ethics, London: Hutchinson.
McCloskey, H. J. 1969. Meta-Ethics and Normative Ethics, The Hague: Martinus Nijhoff.
Monro, D. H. 1967. Empiricism and Ethics, London: Cambridge University Press.
Nagel, T. 1980. The Limits of Objectivity, in The Tanner Lectures on Human Values, Volume II,
ed. S. McMurrin Utah: University of Utah Press.
Singer, P. 1981. Practical Ethics, Cambridge: Cambridge University Press.
Stevenson, C. L. 1963. Facts and Values, London: Yale University Press.
Stevenson, C. L. 1976. Ethics and Language, London: Yale University Press.
Russell, B. 1935. Religion and Science, London: Thornton Butterworth.
Urmson, J. 0. 1968. The Emotive Theory of Ethics, London: Hutchinson.
Warnock, M. 1978. Ethics since 1900, Oxford: Oxford University Press.
Wellman, C. 1961. The Language of Ethics, Cambridge: Harvard University Press.
Downloaded from http://www.RationalRealm.com
23