The Normative Foundations of Defamatory Meaning

This paper assesses normative arguments regarding four views about defamatory meaning. The moralised view holds that a statement about a person is defamatory if and only if we ought to think less of that person if the statement is true. The nonmoralised view holds that a statement is defamatory if and only if people in fact think less of the subject on hearing the statement. A third - the dual view - can be split into two versions. The first version holds that a statement is defamatory if and only if it satisfies either the moralised or non-moralised views. The second version holds that statements satisfying either view can be defamatory, but they ought to be considered fundamentally different forms of personal defamation, with different remedies, defences and conditions of liability attached. Both the moralised and non-moralised views are rejected because they fail to acknowledge instrumental and intrinsic reputational value respectively. The first version of the dual view is rejected because it compromises the expressive value of defamation, implausibly suggests that truth should be a general defence and fails to recognise that different objections apply to the moralised and nonmoralised views. The upshot is that we ought to accept the second version of the dual view.


I. INTRODUCTION
Tort theory has a largely monogamous relationship with the study of negligence. This is understandable since the rise of negligence has dominated the story of contemporary tort law. 1 Perhaps because of this marriage, tort theorists have not investigated other elements of the subject in as much detail. As some writers have commented, one area receiving comparatively little attention is defamation. 2 Though discussions of freedom of speech abound, questions more specific to defamation law have been underexplored. More recently, however, there have been attempts to investigate the underlying values of defamation. 3 Following in this vein, I aim to consider how defamatory meaning should be defined. But I do not pose this as an interpretive, conceptual or historical question. I am not seeking to distil a test for defamatory meaning from the available legal sources, to formulate a test based on linguistic analysis of a folk concept of reputation, or to map the historical development of defamatory meaning. Instead, it is posed as a normative question. Since the definition identifies potentially actionable reputational harms and is a necessary condition of liability, it must be informed by the values and interests that defamation law seeks to protect. 4 To settle the matter by appeal to legal usage or linguistic intuition would be to elide the questions that any definition must address if it claims to protect something morally or socially valuable. Thus, the argument will not attempt an analysis of the common law approach to defamatory meaning, in all its complexity, but will rather identify and evaluate a series of theoretical arguments for and against the available views. I aim to evaluate four such views, which I will call The Moralised View (MV), The Non-Moralised View (NMV), The Dual View Version 1 (DV1) and The Dual View Version 2 (DV2). 5 Although all four views have something to recommend them, I will conclude that the DV2 is the most defensible view.
The MV states that: A statement, S, about a person, P, is defamatory if and only if one ought to think less of P if S is true.
On this view, if a statement made by Jennifer leads Melanie to believe that David is racist, David's reputation is damaged because Melanie ought to think less of David if he is racist. The view is openended in a number of ways. I do not specify what it means to 'think less of P', although obvious examples would include the belief that P has inferior moral or political status; that she has acted wrongfully or holds immoral attitudes; or that she is selfish, dishonest or cruel.
A second question is this: to which opinions do we refer when applying the MV? The MV as stated is incomplete: the criteria and procedure for moral judgment remain to be specified. Should it fall to judges to decide what morality entails in these situations? Should judges defer to public opinion? Or should there be a set of pre-designed principles? These are important questions, but I will not fully address them here, although I will return to legitimacy issues later. The topics I discuss can be raised regardless of what assumptions we make about how we ought to think of others (and in any case I hope that the assumptions I make on this point are uncontroversial). Although this formulation of the MV is therefore broad, it is importantly different from competing views in a few respects. It makes explicit that when this test is applied a judgement is made about what individuals ought to think on hearing certain statements. This also entails that the decision is not dependant, in any fundamental way, on actual or likely reactions. Despite this, actual reactions can have two subsidiary functions. First, they may play an evidentiary role in indicating how people ought to react to the statement, and second, they will help to determine the level of harm caused and thus the appropriate remedy.
The NMV states that: A statement, S, about a person, P, is defamatory if and only if people in fact think less of P on hearing S.
On this view, if Melanie thinks less of David for being racist then he is defamed. But he is defamed because Melanie in fact thinks less of him, not because she ought to. So conversely, if Jennifer leads Melanie to believe that David is not racist, and Melanie thinks less of him for being a naïve, bleeding-heart liberal, then David's reputation THE NORMATIVE FOUNDATIONS OF DEFAMATORY MEANING might also be damaged. The court need not make any normative judgement in applying this view, but rather reach a conclusion about actual or likely responses to the statement. Again, I will leave openended what it means to think less of P and will rely on uncontroversial cases that involve ostracism or open hostility. The DV1 states that: A statement, S, about a person, P, is defamatory if and only if either one ought to think less of P if S is true or people in fact think less of P on hearing it.
This view requires little initial explanation, since it is a combination of the first two views, except for two clarifications. First, the disjunctive term is inclusive, so a statement fulfilling both limbs of the test satisfies the view. Second, strictly speaking the court need not decide which limb of the test is satisfied. This seemingly minor detail has significant implications to which I will return later. The fourth approach to defamatory meaning, the DV2, holds that: Statements satisfying both the MV and the NMV are defamatory but a strict symbolic and substantive distinction should be drawn between them.
Though I will elaborate further on this view later, it effectively means that statements meeting the moralised and non-moralised definitions should be considered fundamentally different forms of personal defamation and should have different remedies, defences and conditions of liability attached to them. The DV2, therefore, is not merely a view about what words defame but a view about the nature of such defamations: the values that they violate and thus the ways they should be treated, at least presumptively, by courts. There are two last preliminaries. I will remain silent on the relation between reputational damage and harm. I will speak of reputational damage without committing to the idea that such damage constitutes harm. Whether it is harm in a strict sense depends on which theory of harm we adopt. On a preference-based theory, according to which a person is harmed if her preferences are frustrated, a defamatory statement can be harmful only if the victim's preferences are affected. Alternatively, on an objective theory, such statements may be harmful even if the victim is indifferent to them. It is not clear what turns on the question of whether reputational damage is harm on the best metaphysical ac-count of harm. It is natural to think that compensation aims to respond to harm, but perhaps not exclusively. It may be that ostracism or lost career opportunities are harmful, and are the proper subject of compensatory damages, whilst imputation of wrongful conduct is not, and should be the subject of some other remedy (such as vindicatory damages). Alternatively, perhaps compensation is a proper response to a wider variety of claims than those arising from harms. Either way, fully exploring the relationship between reputational damage and harm should not be necessary to adjudicate between the four views I have stated.
Finally, on the meaning of reputation, I take a person's reputation to be the beliefs and attitudes that others hold about her. Although this distinguishes one's reputation from one's self-conception, it is otherwise broadly conceived. This definition is more neutral than others proposed in the literature. Lawrence McNamara takes reputation to be 'a social judgement… based upon facts which are considered relevant by a community'. 6 This definition conceptually links reputation to a moral community thus tilting it in favour of his version of the MV.
McNamara does not deny that other definitions are possible and aims to defend his proposal through normative arguments, but I prefer to adhere to a more neutral conceptual framework so that adjudicating between different views is settled directly by their normative appeal rather than conceptual claims about the nature of reputation.

II. THE MORALISED AND NON-MORALISED VIEWS
The MV has been defended by several scholars. Lyrissa Barnett Lidsky argues that the courts' role in screening out undesirable value judgments about reputation is both inescapable and beneficial. 7 Similarly, McNamara states that, 'A determination of defamatory capacity requires the courts to constitute a jurisdiction as a moral community'. 8 And Leslie Treiger-Bar-Am asserts that defamation law should be 'an instrument for promoting moral virtue. Hence, the standard for defamatory meaning… ought to be interpreted so as to promote a reasoned, principled morality rather than reflect societal prejudices'. Treiger-Bar-Am's argument has an instrumentalist flavour: defamation law should be used to promote virtues and discourage vices. By contrast, I suggest that the primary motivation for the MV is that it captures the sense in which a person's reputation can be seriously affected independently of any consequential harms. In other words, it captures the non-instrumental value of reputation.

A. Non-instrumental and Instrumental Reputational Value
The non-instrumental value of reputation is the value of the mere fact that others have particular attitudes or beliefs towards one. Its instrumental value, on the other hand, is the usefulness of those attitudes and beliefs in securing further goods. 10 For example, the belief that David is trustworthy may bring him career or business opportunities; expressions of respect or admiration; opportunities to enter new relationships; and the conditions necessary to maintain close social and familial ties. Reversal of this belief might destroy his career or business opportunities; cause him to be ostracised from his community; prompt expressions of disapproval or spite; and erode his close social or familial ties. These consequences set back the instrumental value of David's reputation.
Defamatory statements typically diminish both instrumental and non-instrumental reputational value, but we can imagine cases in which only the former is affected. Consider the following: Pub: David goes to his local pub to watch a football game. After many drinks, he strikes up a conversation with another fan, Melanie, which drifts into politics. Only later, as he pieces together his memory of the conversation, does David realise that Melanie formed the mistaken belief that he is a member of a far right organisation and harbours racist beliefs about minorities.
In Pub, David's reputation is seriously compromised even if Melanie's beliefs and attitudes about him have no adverse consequences. Melanie believes that David holds attitudes that are 10 Some take non-instrumental value and intrinsic value to be interchangeable. Others hold that a thing has intrinsic value by virtue of its internal properties, rather than its extrinsic properties, i.e. its relation to other things. In contrast, a thing has extrinsic value by virtue of its relation to other things with intrinsic value. Thus, the distinctions between intrinsic and extrinsic value and instrumental and non-instrumental value are orthogonal. In the present case, I use the terminology of non-instrumental value since this best captures the relevant phenomenon. deeply in conflict with his convictions and values, and those he ought to hold. 11 The mere fact that Melanie possesses these beliefs is sufficient to damage his reputation.
What explains the reaction, for those who share it, that David's reputation is damaged in Pub? I suggest we can find an explanation by developing Thomas Scanlon's theory of blame. Scanlon argues that 'to claim that a person is blameworthy for an action is to claim that the action shows something about the agent's attitudes towards others that impairs the relations that others can have with him or her'. 12 I am sceptical about Scanlon's ideas as an account of blame, 13 but blame frequently does have the effect of impairing moral relationships in the way Scanlon describes, and it is easy to see how statements imputing immorality might do the same. When we come to believe that someone has acted wrongfully, for example, we might reconsider the basis of our relationship with her, cease to value spending time together or raise a complaint. 14 We must be careful here to separate those modifications of our moral relationships that affect the instrumental value of a reputation and those that affect its non-instrumental value. Some of these reactions, such as the possibility of raising a complaint, affect the former as well. If a complaint is raised against us, this might have other adverse effects on our relationships with people, such as spending less enjoyable time together. But one of Scanlon's insights is that our relationships are not just constituted by the quality of our experiences or the goods that those relationships make possible. They are also constituted by the reasons we take ourselves as having to treat and to view others in certain ways, and those reasons can be affected merely by belief in another's wrongdoing. What if a person commits a wrong against another and neither the victim nor anyone 11 There is another view that I do not have space to consider in this paper, which we can call The Self-Conception View. This holds that a statement, S, about a person, P, is defamatory if and only if it lowers P's estimation according to P's sincere self-conception. This view is non-moralised, in the sense that it defers to the claimant's judgments rather than imposing independent principles, but it can also recognise the non-instrumental value of reputation, although the explanation for that value must be different to the one I offer here. else detects it? Although I do not rule out the possibility that the wrongdoer and the victim's relationship is compromised, the wrongdoer's reputation is not damaged unless a third party forms some belief about the wrongdoing. This means that interference with a moral relationship is a broader category than reputational damage. My claim is rather that the former explains one facetsetback to non-instrumental value -of the latter.
Scanlon suggests that what matters are the reasons we take ourselves as having to treat others in certain ways. 15 This implies that there is no reputational harm in cases where others do not acknowledge reasons to think less of us. Consider the following variation on Pub: Pub 2: The same as Pub, except Melanie is a member of a far right organisation and approves of David on the basis of her mistaken belief that he is a fellow member. She disseminates this belief to other members of the organisation.
Does the fact that this belief is held only by organisation members who approve of David based on this mistaken assumption vitiate the sense that his reputation is compromised? I do not think so. We would sympathise with David's desire to correct the mistake, even if it had only been spread amongst members who perversely approve.
At this point, some might claim that any value interfered with here is purely instrumental. What is damaging, it might be suggested, is that David experiences unpleasant emotions -such as embarrassment and shame -when he contemplates Melanie's beliefs about him. Even if these emotions depend primarily on the content of the belief rather than the evaluative attitude of the believer, it is the unpleasantness of the emotional response that is bad for him. But this claim seems mistaken. It is intuitive that David's reputation is damaged even if he does not experience any negative emotions. He may recognise that it is bad for him that others have false beliefs about his moral character even if he does not experience any emotional discomfort. Moreover, David's embarrassment may be counterbalanced by other instrumental benefits. Perhaps Melanie sends him an annual hamper, motivated by her appreciation for their supposedly shared values. This benefit might render the instrumental value of Melanie's belief positive overall, but this does not mask a 15 Ibid., pp. 128-129. setback to non-instrumental value. Finally, the claim that the damage to David's reputation consists only in the negative emotions he experiences mischaracterises David's reactions in Pub and Pub 2. When David contemplates Melanie's belief about him, the affective content of his reaction merely points towards what David is concerned about. The fact that David ought not to be racist, and that others ought to think less of him if he is racist, is explanatorily prior to David's experiencing the sensation of embarrassment.
This example casts doubt on Scanlon's suggestions that it is the reasons we think we have that affect our moral relationships. Although he agrees that it can be wrong to withhold blame when blame is called for (and thus wrong even though the withholder does not take herself as having any reason to blame), he explains this by appeal to the relationship between the party who withholds blame and the victim of the wrong for which the blame would be appropriate: we owe it to victims to blame wrongdoers and failure to do this can impair our moral relations with them. This suggests that the moral relationship between the person who wrongly withholds blame and the wrongdoer is left untouched. A better explanation is that our moral relationships with others are constituted, not only by the reasons we take ourselves as having to treat them in certain ways, but also by the reasons we ought to take ourselves as having. If I wrongly praise a wrongdoer, my relationship with her is impaired because I ought to blame her for her actions. As in Pub 2, my positive reaction reflects a reason, which I fail to recognise, for me to think less of her. It should not be surprising that our relationships can be affected by facts such as this, about which we may be mistaken. For example, a person's relationship with her spouse is impaired if she is blithely unaware that he has betrayed her; the relevant notion of relationship is broader than a purely experiential one. 16 She may mistakenly praise him for his loyalty, but though her erroneous belief might yield instrumental benefits for them, the fact that she has a reason to denounce him undermines the foundation of trust on which (some of) the value of their relationship depends.
More accurately, we should distinguish between belief-relative, evidence-relative and fact-relative reasons. 17 And we should note that any of these reasons can be ones that an agent does not recognise, even though she ought to. So in Pub 2, Melanie has belief and evidence-relative reasons to condemn David (given her beliefs about his racist attitudes and the evidence available to her, she ought to condemn him), but she does not have a fact-relative reason (given the fact that David is not racist, she ought not to condemn him). These can be distinguished from the reason she takes herself as having to praise David for his racist attitudes, which is a reason she ought not to recognise.
That Melanie has these belief and evidence-relative reasons even though she mistakenly believes she does not is a controversial claim that I cannot fully defend here, 18 but one of its virtues is that it explains what goes wrong in cases where blame is absent. In Pub 2, although Melanie does not blame or exhibit any other negative reactive attitudes 19 towards David, she has belief-and evidence-relative reasons to do so, and this is sufficient to impair their moral relationship. Just as justified negative reactions can impair our moral relations with others, so too can unjustified positive reactions.
Scanlon also argues that we have a relationship, in a general sense, with everyone who can understand and respond to reasons. 20 According to him, it is possible to alter our reasons to view others in certain ways even if we've never met them and in this sense our relationship with them can be impaired. This is not to deny that personal relationships have greater significance to the people involved than the 'moral relationship' we have with all other reason-following beings, but it explains both why we still care about our reputation amongst strangers and why we generally care more about personal relationships.
Two objections might be raised at this point. First, it might be argued that personal relationships are those with a minimal experiential dimension, consisting of a sustained pattern of interaction. no personal relationships with other reason-following beings, so in what sense do we have any relationship at all? It is misleading to call our connection to strangers a relationship since its value cannot be analysed in terms of any of the typical features of personal relationships, particularly those that depend on positive interactions between members such as affection or mutual support, and it is difficult to see what normative reasons we have to care about our reputation amongst those with whom we have no relationship. The absence of meaningful relationships with strangers cannot be papered over by metaphorical terms such as the 'moral relationship', unless we have a clear account of what this relationship consists in, which Scanlon does not provide. The second objection, closely related to the first, is that the concept of the moral relationship does no independent work in the argument. If the main reason that we value our reputation amongst strangers is that this will prevent us entering relationships with them in future, then the explanatory work is done by personal relationships and the concept of the moral relationship is redundant.
Rather than attempt to plug the gap in Scanlon's theory of blame by providing an account of the moral relationship, I will take a different approach. I reject the notion of the moral relationship, but argue that a key feature of personal relationships identified by Scanlon -the reasons we have to view others positively or negatively -also obtains between strangers. By arguing in this way, the troublesome apparatus of moral relationships becomes entirely dispensable. To rely on this claim, it is important to clarify which features of Scanlon's account of blame I must adopt for present purposes. We can identify two distinct claims implied by Scanlon's account: (1) Statements can alter our reasons to view others in certain ways even if we have never met them.
(2) Altering our reasons to view others in certain ways can impair our relationship with them, even if we have never met them.
The most important claim for our purposes is (1). Scanlon is right, I argue, at least that personal relationships are partly constituted by our reasons to view or treat people in certain ways, rather than merely their experiential or affective dimensions. However, the kind of value on which my argument rests is a form of non-instrumental value attached to beliefs and attitudes that others hold towards us, THE NORMATIVE FOUNDATIONS OF DEFAMATORY MEANING which need not be limited to such relationships. This form of disvalue, although it is an important constitutive feature of personal relationships, can exist both inside and outside of them. In other words, it can obtain between strangers who have no relationship with each other. Therefore, contrary to Scanlon's view, we need only rely on (1). In Pub, the intuitive worry regarding David's reputation stems from the reasons that Melanie and others have to think differently of him. If we alter the facts of Pub so that David and Melanie do not meet, it is doubtful that David's legitimate concerns would evaporate merely because the false beliefs no longer impair a personal relationship. Equally, this does not mean that reference to Scanlon's view is redundant, since it helps us to see that our relation to strangers shares some features of personal relationships that we have reason to care about. The fact that they are important constitutive features of personal relationships explains why we ought to care about the same value elsewhere.
It is worth noting that, if this response is not convincing, the objections limit rather than refute the argument. Even if the noninstrumental value in question is limited to personal relationships, it explains how some statements can damage those relationships. This does not leave our overall account of defamatory meaning implausibly limited, since it is still possible for statements made to strangers to damage the instrumental value of reputation, which is an issue to which I will return.
One damages the non-instrumental value of a person's reputation, then, if one undermines the reasons others have for viewing that person in a morally favourable light. 21 But although its acknowledgment of this value counts in favour of the MV, its exclusive focus on it means that it has implausible implications in a range of cases. Consider the following: Victim Blaming: As a result of a statement made by David, Melanie's community believe that she has been the victim of a rape. The community has a culture of victim-blaming and Melanie is ostracised.
Homophobia: As a result of a statement made by David, Melanie's community believe that she is homosexual. Owing to a culture of homophobia, Melanie is ostracised.
Heresy: As a result of a statement made by David, Melanie's community believe that she is an atheist. The community has a culture of intolerance for contrary beliefs and Melanie is ostracised.
Informant: As a result of a statement made by David, Melanie's community believe that she has informed the police about an act of domestic violence. The community despises police informants and Melanie is ostracised.
In each of these cases, Melanie is not defamed on the MV because we ought not to think less of her, even if the statements have devastating effects on the instrumental value of her reputation. One might even point out that, in Informant, the non-instrumental value of Melanie's reputation is possibly enhanced as we ought to respect her courage. But it is difficult to believe that the harm she suffers cannot be actionable in tort. This seems to ignore the very real instrumental value of Melanie's reputation.
One argument in favour of this conclusion is that a person's reputation is useful in securing goods that are themselves uncontroversially valuable. For example, the belief that someone is trustworthy may be necessary to secure an income and meaningful employment, or to maintain social and familial relationships, and few would deny that these are valuable. Cases such as the four cited above, in which only the instrumental value of reputation is affected, are slightly more difficult. It might be argued that the value of these social ties depends on basic conditions of trust and respect that are absent in these intolerant communities. These social ties lack value because they are sensitive to bigotry: a social group that ostracises its members for prejudicial reasons is not a social group worth belonging too.
We must be careful not to throw the baby out with the bathwater here. These examples are fit for purpose as long that the social relationships therein have some value, even if the value of these relationships is substantially mitigated by the prejudicial attitudes of the community members. It would be difficult to maintain that relationships that may possess many other positive dimensions are rendered valueless for this reason alone. The NMV thus has the advantage of grounding reputation in actual social attitudes -with all their flaws -rather than notional ones. Further support for this approach can be found in Thomas Gibbons' observation that reputation is not always dependent on moral evaluations, and a person's social standing is 'as likely to be assessed according to the audience's preferences and tastes '. 22 It must also be remembered that the issue is whether Melanie suffers a setback to the value of her reputation in the above cases. More generally, our aim is to determine whether others' beliefs and attitudes towards us have a form of instrumental value that partially explains why our reputations matter. This can be distinguished from the question of whether, in any individual case, a person should be liable in tort for causing this reputation damage. Although I claim that Melanie suffers reputational damage in these cases, it does not follow from this that David should be liable to compensate her for it. Nevertheless, identifying and distinguishing the relevant forms of reputational value is a necessary step towards a theory of liability in defamation, since a clearer understanding of reputational value will help us to determine which liability standards are most appropriate. I return to this issue in Section III.C.
A further objection is that, if popular attitudes are to blame for the harm Melanie suffers in these cases, the problem should be addressed through social and political change. If Melanie loses employment opportunities, the solution is to introduce more robust anti-discrimination laws rather than pursue an action in defamation. Compensation is not only an individualistic solution, but fails to prevent reputational harms by addressing their underlying causes. This is a crucial point, but not a damaging one, since I do not advocate the NMV as an alternative to root-and-branch social reform, but rather a supplement to it. It may also be impossible to eradicate such prejudices entirely, or it may be achievable only after a protracted struggle, so these cases may be unavoidable even if we accept that a right of action in defamation is not an optimal solution. Moreover, although the NMV is a test for defamatory meaning and does not impute wrongdoing to the defendant, the mediation of reputational harm through the wrongful attitudes of others does not preclude wrongdoing on behalf of the defendant either. It is an open question, for example, whether those who negligently make statements that satisfy the NMV ought to compensate victims. This position is arguable even if others are also responsible for the harm.
Finally, it might be suggested that the NMV impedes rather than supplements social reform by giving credence to prejudicial views. This concern may be due to the failure to distinguish between the MV and the NMV. The distinction between these views captures the idea that damaging statements can be of two fundamentally different types: those that, if true, ought to lower a person's estimation and those that do lower a person's estimation. Once this distinction is properly clarified, statements satisfying the NMV should not be perceived as giving any credence to the attitudes of those who think less of the claimant. In fact, as we will see in the next section, recognising the expressive value of defamation law supports the DV2.
Our conclusion, then, is that a person's reputation can have instrumental value. However, the problem with the NMV is the corollary of the objection to the MV, namely that it fails to consider the non-instrumental value of reputation. It is hard to believe that imputations of seriously wrongful conduct can be rendered nonactionable simply because the victim's associates, or the wider community, perversely approve of them. As Pub shows, David can legitimately claim that his reputation is damaged if others come to believe that he is racist, even if they are more positively disposed to him as a result. To dismiss this setback to the victim's reputation would stifle her attempt to vindicate her character and conduct. If, for example, a published retraction of such a statement would disabuse people of these beliefs, it would be wrong to deny the victim this remedy because she has benefitted from this change in her reputation. The NMV, like the MV, assumes an inaccurate and fragmented picture of the value of reputation.
For similar reasons, it should be clear that a definition requiring that a statement satisfy both the MV and NMV is also untenable. Such statements will represent the best claims in defamation and may attract bigger remedies, but the conceptual structure of defamation should not rule out claims that only satisfy one of the tests. Such a view would be vulnerable to the same criticism levelled at the tests individually: they assume a fragmented picture of the value of the reputation by failing to accept that impairment of the non-instrumental (or instrumental) value of reputation is significant on its own.

III. EVALUATING THE DUAL VIEWS
The arguments considered so far suggest that statements made by others can set back both the non-instrumental and instrumental value of a person's reputation. The advantage of the DV1 over both the MV and the NMV is that it has plausible implications in a wider range of cases. It implies that both imputations of immoral conduct that have no consequential harms and consequential harms that do not follow from imputations of immoral conduct can be defamatory. Unfortunately, this advantage raises its own problems.

A. Expressive Value
On the DV1, both the moralised and non-moralised elements are internal to the definition of defamation. However, the expressive value of a judgement that a statement has satisfied either the moralised or non-moralised element of the definition should be sharply distinguished. (Here, following Anthony Duff, I refer to the expressive value of judgements rather than communicative value. This distinguishes legal pronouncements from the interactive communicative process of, say, a criminal trial). 23 Although judgements on defamatory meaning lack the condemnatory force of criminal punishment, they do embody a form of what Joel Feinberg calls 'symbolic non-acquiescence'. 24 On the MV, by finding that a person ought to think less of another on hearing some statement pertaining to her, the judgement gives content to a legal conception of the attitudes and beliefs of the moral person; it conveys a message about which statements ought and ought not to lead us to think less of others. Putting the point another way, McNamara claims that the refusal to recognise a judgement of a person as defamatory 'has an exclusionary function with respect to the constitution of the juris-diction as a moral community' because it rejects the values that underpin that judgement. 25 This reveals a powerful objection to the DV1, namely that it compromises the expressive value of judgments about defamatory meaning. Consider the contrast between a statement that a person has committed a rape and a statement that a person is the victim of a rape. On the DV1, both might be defamatory: the former satisfies the moralised element and the latter, in certain circumstances, satisfies the non-moralised element. But unless an explicit decision is made about which element of the definition is satisfied -and on the DV1 this decision need not be made if it is clear that at least one element is satisfied -the expressive value of the judgement is ambiguous. This is a serious problem given the radical difference in the expressive content of the respective judgements. If an imputation of victimhood is seen to satisfy the moralised element, the law endorses victim-blaming, and if an imputation of rape is seen to satisfy only the non-moralised element, the law fails to express the appropriate condemnation for a serious wrong.
The problem is especially worrying if definitions of defamation fail explicitly to distinguish the two elements. But even where judgements attempt explicitly to distinguish between moralised and non-moralised elements, troubling ambiguity can still creep into a decision. In Youssopoff v Metro-Goldwyn-Mayer Pictures Ltd, 26 although the judges denied that a film suggesting that Princess Natasha of Russia had been raped imputed moral discredit, they may have implicitly applied a subtler version of the MV based on the moral value of female chastity, particularly with respect to members of the royal family. Leslie Treiger-Bar-Am argues, in her analysis of the case, that a rape victim is 'tainted with sexual impurity, and hence, as purity entails moral worth or value, with immorality', 27 demonstrated in part by the judges' slippage between the terms 'raped' and 'ravished'. The DV1 may seem like a practical compromise, but in forging this compromise it creates an ambiguous space in which courts can rely on suppressed moral judgments.

B. The Truth Defence
There is another equally significant problem: it is not clear whether the defence of 'truth' ought to be available. 28 A successful application of the truth defence denies the unlawfulness of the injury to the claimant. 29 A plausible basis for the defence might be stated as follows: if A has committed some wrong, she is morally liable 30 to suffer the appropriate consequences, such as legal censure and a lower estimation in the minds of others, and therefore lacks a right to be compensated for them. 31 But if this is the basis of the truth defence, or even part of its basis, it should only be available with regard to statements that satisfy the moralised definition, for only in this instance has A done anything that could make her liable for damage resulting from public knowledge of her actions. If truth can be a defence to an action based on a statement that satisfies only the NMV, this has unacceptable implications in many cases involving damage purely to the instrumental value of reputation. For example, in Victim Blaming, the applicability of the truth defence has the disastrous implication that Melanie is morally liable to suffer ostracism because she is a victim of rape; in Homophobia, it implies that Melanie is morally liable to suffer ostracism because she is homosexual, and so on. These implications obviously provide a further reason not to combine both views within a unified test for defamatory meaning.
It may be pointed out that truth is no defence to breach of privacy rights and legal regimes that protect such rights provide sufficient protection. McNamara argues that the imputations just described should be actionable as invasions of privacy but have no place in defamation law. This follows from his claim that a statement about a person that imputes no moral discredit cannot harm her reputation, since reputation is exclusively concerned with the moral judgements 28 I will not consider whether truth should be an absolute or partial defence, but there is a strong argument for the latter interpretation. See R. R. Ray, 'Truth: A Defense to Libel' Minn. L. Rev. 16 (1931) 43. 29 See Eric Descheemaeker, 'Mapping Defamation Defences' The Modern Law Review 78(4) (2015) 641-671, 650-652. 30 Here I use the concept of moral liability rather than legal liability. On one view, to say that a person who has committed a wrong is morally liable to suffer damage to her reputation is to say, roughly, that she is not wronged by others thinking less of her. For more detailed discussion of this concept of moral liability, see Victor Tadros, 'Orwell's Battle with Brittain: Vicarious Liability for Unjust Aggression', Philosophy and Public Affairs 42(1) (2014) 42, 47-51. 31 For discussion of the possible bases of the truth defence, see Ray, 'Truth: A Defense to Libel'. of a community, and statements lacking any moral implications about their subjects are irrelevant for these judgements. 32 There are several problems with this reply, however. First, it rules out the possibility that reputation can have only instrumental value. As I have been arguing, this seems wrong: the beliefs that others hold about us can be useful in securing a range of goods and opportunities, even if these beliefs are morally irrelevant or abhorrent, and this is most plausibly categorised as a form of reputational value. Although what is of primary importance is that valid claims are recognised and invalid claims rejected, the legal categorisation of claims is still of secondary importance since separate (though overlapping) sets of values underpin reputation and privacy, and claims in these respective domains of law have different justificatory bases. 33 Violations of privacy, for instance, might be grounded in the value of autonomous control of personal information rather than material damage. 34 Since the purpose of defamation law is primarily to remedy wrongful damage to reputation, it would be odd if these cases were dealt with by privacy law rather than defamation.
Secondly, legal protection of privacy may not apply to all relevant cases. On any legal regime for the protection of privacy there might be victims who have suffered reputational harm and who cannot justify an action for invasion of privacy who will be left without a remedy. This may be because the information is entirely false, in which case it is questionable whether this constitutes an invasion of privacy, 35 or because the information is not sufficiently personal to fall within the protected sphere of privacy, even though it causes reputational harm.

C. Opting for the Dual View Version 2
The arguments regarding expressive value and the truth defence already suggest that a symbolic and substantive distinction should be drawn between claims based on the moralised and non-moralised  Affairs, 12 (1983) 269-288. 35 Although note the US category of invasion of privacy consisting in publicity that unreasonably places the claimant in a 'false light'. However, it seems more arbitrary to include false information causing damage to the instrumental value of others' attitudes towards one under privacy law than defamation. views. A further reason is that the different types of claim are vulnerable to different objections. Depending on the strength of these objections, it may be mistaken to apply identical conditions of liability. Of course, what is offered here falls short of a complete argument that statements satisfying the moralised and non-moralised views should be considered fundamentally different forms of defamation, but it may be enough to raise concerns about the tendency to combine the two very different tests in a unified analysis of defamatory meaning.

Liberal Legitimacy
The MV faces a problem of legitimacy, which can be derived from the liberal critique of the use of state power to promote values that citizens can reasonably reject. This important strand of anti-perfectionist liberal thought suggests that vesting the authority to determine how we ought to react to a given statement with judges and lawmakers, whose decisions will affect the legally enforceable duties imposed by defamation law, raises a problem of legitimacy. 36 The claimant or defendant, who is subject to the coercive authority of the civil court, may be able reasonably to reject an opinion about whether a statement imputes immorality. With what justification does the court coerce an individual based on a controversial judgment that the individual rejects?
The view that liberal ideas of political justice in pluralistic societies should be brought to bear on the study of private law has recently gained some support. Writing from a Rawlsian perspective, Samuel Scheffler has rejected the idea that private law is excluded from the basic structure of society. 37 McNamara argues that the assessment of defamatory capacity should be based on a Rawlsian overlapping consensus. On his view, 'a vast range of imputations would be actionable under a liberal framework for the presumptive content of ''the right-thinking person''', 38 because an imputation ought to be actionable if a subsection of society with a reasonable, comprehensive religious, philosophical or moral doctrine would consider it relevant for moral judgement. Courts should therefore refer to a hypothetical person who represents a member of an inclusive liberal community, committed to 'equal moral worth, freedom of moral choice, protection from harm and moral diversity'. 39 This seems to misapply Rawls' thesis in Political Liberalism, however, which defends a constraint against the exercise of certain forms of political power 40 without reasons that can be accepted by any reasonable, comprehensive doctrine. McNamara loses sight of the fact that the civil court's authority is exercised over both the claimant and defendant, and imputations that are defamatory on the claimant's comprehensive view might be reasonably rejected by the defendant, thus raising again the problem of legitimacy. What the liberal view recommends, therefore, is not a more expansive approach to defamatory capacity in which any reasonable belief can be invoked, but a more restrictive one in which only those statements that would damage a person's reputation on all reasonable comprehensive conceptions are admissible. It is only statements satisfying the latter test that can in principle be justified from the perspective of both the claimant's and the defendant's comprehensive doctrines.
This liberal challenge does not destroy the MV, but if correct it does restrict its scope. One way to temper the objection is to appeal to a threshold of moral seriousness that any defamatory statement should meet in order to be actionable. Imputations of minor improprieties are too trivial to merit a legal claim. Even more serious allegations such as abusive language may not, in many contexts, sufficiently damage a person's reputation. Graver imputations of immoral conduct, however, such as conduct that seriously violates the rights of others, are likely to meet the relevant threshold even if the statements have no further negative effects. But, crucially, as imputations of immoral conduct become more serious, reasonable 39 Ibid. disagreement about whether the conduct in question is genuinely immoral correspondingly decreases. The most serious imputations undermine a person's standing as a free and equal citizen, which might justify legal recourse, at least within a liberal framework.
Another possibility is that the court should defer to the comprehensive doctrine of the party who had a substantially better opportunity to avoid the outcome. An example will help to illustrate the point. Suppose that Melanie makes a statement implying that David is a Scientologist. David reasonably believes that this statement damages his reputation whilst Melanie does not. If Melanie made the statement maliciously to ruin David's reputation, then the statement ought to be actionable on the MV because Melanie could easily have avoided making the statement by not forming the malicious intention. Conversely, if Melanie conveyed the implication accidentally and non-negligently, it would be unfair to allow David to bring an action against her when she can reasonably reject David's conviction that his reputation is damaged.
I lack space here to investigate further the impact of liberal concerns on the legitimacy of defamation judgements. My point at this stage is that, depending on how the objection is resolved, it is prudent to leave open the possibility that claims based on the MV of defamatory capacity must meet different substantive requirements.

The Wrongs of Others
Let us now turn to an objection that applies to the NMV. If a victim can claim against a publisher of a statement, and the statement satisfies only the NMV, then the defendant might be held responsible for the wrongful attitudes of others. The publisher of the statement may not endorse the prejudicial attitudes of those who think less of the claimant, and this raises two questions: first, whether causation for the damage should be attributed to the defendant, and secondly, whether she should owe a corrective duty to the victim when the harm for which the victim complains is primarily the responsibility of others. Allowing such claims in defamation appears to conflict with other areas of tort law such as negligence, in which defendants are rarely liable for property damage or physical injury inflicted by third parties.
Two points are worth making about this apparent inconsistency. The first is that third parties who intentionally inflict injury or property damage are likely to be civilly and criminally responsible. Those who hold wrongful attitudes, however, generally enjoy the protection of rights to freedom of speech, conscience and association, even if their attitudes cause great harm. 41 Moreover, harm to the victim is often caused by these attitudes as a collective, and consequently no single individual will have primary responsibility. Victims do not have the same legal recourse to hold those with morally wrongful attitudes to account, which is one reason to afford them extra protection from defamatory statements.
Might this objection justify the introduction of a requirement for carelessness, if not actual malice, for statements satisfying the NMV? As before, it is not my purpose to answer this question here. The point is that, based on the discussion thus far, attaching different thresholds of liability to the moralised and non-moralised views is an open possibility.

IV. CONCLUSION
The upshot of this discussion is that we should adopt the DV2. Because it endorses a sharper symbolic divide between different kinds of statements, it is less likely to compromise the expressive content of defamatory judgements. It also forces the court to label its normative judgements more clearly, making them more amenable to scrutiny. We have also explored reasons why claims in respect of the moralised and non-moralised views might be substantively different. My interest in this paper is in the theoretical foundations of the law, but our discussion also has practical implications for the legal framework on defamatory meaning. I will conclude by briefly noting two of them.
First, the distinction between non-instrumental and instrumental reputational value has implications for any requirement regarding the harm inflicted by a defamatory statement. The upshot of the present argument is that such requirements should not be purely instrumental. If an imputation is morally serious, such as an accu-41 These rights are not absolute, of course. Even the right to choose one's company ought to be limited in certain ways. See Kimberly Brownlee, 'Freedom of Association: It's Not What You think' Oxford Journal of Legal Studies 35(2) (2015) 267-282. sation of a violent offence, it would be inappropriate to require a claimant to prove any other kind of damage, as this would fail to protect the non-instrumental value of reputation. Arguments in favour of a general requirement for claimants to prove actual harm fail to appreciate this value, or to consider the possibility of limiting the scope of the requirement rather than imposing it uniformly. 42 Equally, if the claim is in respect of instrumental harm to reputation, proof of other damage -such as financial loss, hostility or ostracism -should be required, and proof that people in fact think that the imputation is morally serious is irrelevant. For example, if someone is accused of assisting the police in a domestic violence enquiry, it is irrelevant that a section of the community thinks that 'grassing' is morally objectionable, but if the victim is ostracised by the community, proof of this fact alone may justify an action for damages.
Moreover, the distinction between non-instrumental and instrumental value maps onto a familiar distinction in tort between wrongs that are actionable per se -without proof of damage -and those that are actionable only when the claimant shows that she has suffered a recognised form of harm. One reason in favour of actionability per se is to prevent the burden of proof falling on the claimant. But the wrong of damaging the non-instrumental value of one's reputation offers another explanation why libel is actionable per se.
Second, distinguishing between different forms of reputational value helps to properly itemise reputational harm. A person who is ostracised because he is accused of serious wrongdoing is harmed more than a person who is ostracised for unjustified reasons, even if the material harm is identical. This may in turn affect the appropriate selection of remedies. Vindicatory damages might be appropriate for the immaterial damage, perhaps combined with compensatory damages to reflect any harm to instrumental value. Although the availability of vindicatory damages is controversial and the arguments against them must be addressed, their appropriateness for this purpose can be added to the case in their favour. 42 See, e.g. David Anderson, 'Reputation, Compensation and Proof' William and Mary Law Review 25 (1984) 747, 749-751.