Part 1

Theorists and others who reject so-called ‘rhetorical’ deployments in favour of scholarly or principled approaches construct a false and ultimately unproductive dichotomy. By identifying illicit or unprincipled usage with ‘rhetoric,’ they miss the important point that all uses of the word ‘terrorist’ are rhetorical in one sense or another. This is true in the general sense that all acts of communication can be evaluated in rhetorical terms.Footnote ⁸ It's also true in the narrower sense that, as a moral-normative word, many (perhaps most) uses of the word – including principled and careful ones – have an emotive dimension as well as a descriptive one, enjoining those to whom they are directed to take a particular moral view on cases of political violence.Footnote ⁹ Of course there will be some contexts in which one uses the word ‘terrorist’ with little or no emotion or emotive intent (in discriminating between ordinary murder and political murder, for example – here ‘terrorism’ is sometimes felt to be less bad than crimes motivated purely by private greed).Footnote ¹⁰ But even where the moral force of the word isn't invoked directly, a normative judgement is usually presupposed; and, as a political term, whether used in ordinary speech, in the media, or in the discourses of government policy, the word ‘terrorist’ generally has the illocutionary forceFootnote ¹¹ of moral censure, repudiation or outright condemnation.Footnote ¹² Political theorists as well as most security analysts and historians of terrorism also participate in the discourses within which ‘terrorist’ carries these normative-emotive connotations (though they sometimes try to challenge them).Footnote ¹³ I therefore think it helpful to see what can be learned by interpreting their usages in light of the rhetorical possibilities that they seek to deploy, open up, or close down.

My approach requires a reformulation of the questions usually asked about defining the word ‘terrorist’: first, the question of definition would be answered when we had indicated the circumstances in which the illocutionary force of ‘terrorist’ as a term partly of condemnation could defensibly be invoked; secondly, instead of seeking a definition free from rhetoric, that is, reinforcing a divide between conceptual and rhetorical deployments, we would seek a means of distinguishing between different kinds of rhetorical use and perhaps even between good and bad rhetorical deployments.

The different rhetorical possibilities I will consider are distinguished by the way speakers attempt to do things in relation to three different kinds of definitional criteria: first of all, basic moral or political values; secondly, descriptive criteria that are understood to be derived more or less directly from these basic norms; and thirdly, more conventional descriptive criteria, not arising directly from the basic values and not necessarily postulating any further evaluative-illocutionary force. There are three different ways in which we may see speakers as trying to do things with the word ‘terrorist’ in relation to the three definitional criteria. All of them typically involve ‘redescription’ (or paradiastole), a mainstay of ordinary moral language as well as forensic oratory.Footnote ¹⁴ Redescription occurs where the application of a term of opprobrium or of moral praise is debated in relation to a particular entity or class of entities.

I want to expand this idea of redescription and distinguish between three different kinds of redescriptive rhetoric, starting with the classical idea, as it occurs around ‘terrorist’ as a normative term. In the first and most basic approach, speakers use the term ‘terrorist’ to apply a form of censure based on widely recognised moral and political norms to particular acts or entities. They do so, first, according to descriptive criteria understood to arise as corollaries of the norms themselves. Thus, for example, the inviolability of civilian lives and bodies as a human right is central to approaches defining ‘terrorism’ by attacks upon the ‘innocent’ or ‘non-combatants.’ The value invoked provides a descriptive criterion allowing commentators to determine whether a given agent has engaged in terrorism, that is, based on whether deliberate targeting of the relevant class of persons has occurred (and whether it has occurred to a sufficient degree, a point to which I return below). Similarly, where the norm concerns abhorrence for the use of ‘terror’ as a political instrument, the descriptive criteria will distinguish between those seeking to cause fear in a wider ‘audience’ and those not.Footnote ¹⁵

Three further criteria are very common, all arising independently of the basic normative concern but strongly grounded in the conventions of ordinary and political speech: first, the agents in question are generally required to have committed the ‘terrorist’ act as a means intended to achieve ends that are political in one sense or another. Secondly, the term is commonly used with reference to non-state entities rather than states.Footnote ¹⁶ The latter provokes concern amongst many conscientious normative and critical theorists, anxious that states will be exempt from an important form of condemnation, but it has also been challenged by some conservatives for different purposes.Footnote ¹⁷ These last two criteria are independent and rather arbitrary from the moral-normative perspective that sees ‘terrorist’ as a term expressing moral outrage or rebuke since speakers commonly see the targeting (or terrorising) of civilians as morally no better or worse whether carried out by conventional armies, agents of the police, or members of non-state organisations. The status of the agent and its aims, therefore, usually has little or no normative significance. But it is nevertheless important in determining correct usage arising, as it were, on grounds of linguistic convention rather than political theory. A third descriptive criterion is commonly introduced to reflect the idea that non-state groups often use violence communicatively or theatrically to dramatise a grievance or coerce a wider population. Cases of violence that exhibit all the features commonly regarded as relevant may be seen as paradigmatic.

Where debates occur over whether or not a given case properly constitutes ‘terrorism’ based on these criteria, then the rhetorical technique that interlocutors deploy is one of ordinary ‘redescripton’. The debate will be over whether the behaviour or entity possesses the relevant characteristics, both those arising from basic norms and others.Footnote ¹⁸ In these cases, interlocutors aren't trying to contest the meaning or range of applications of the word; they merely argue about whether it properly applies in the present case.

It's useful to say something here about the difference that arises between defining acts as terrorist and applying the term to actors. Political status definitions generally define agents as ‘terrorist’ first and then regard the actions of such agents as ‘terrorist’ as a consequence.Footnote ¹⁹ But definitions focusing on non-combatant targeting or the use of terror begin with the acts themselves and then seek to persuade interlocutors to transpose the idea from the act to the group or individual performing it. In both cases, the move between act and entity can be characterised in rhetorical terms as synecdoche, that is, where the name of a part is used to invoke the essential features of the greater whole to which it belongs (or vice versa). The question of when a synecdoche of this kind can be used with justice – and when it will be accepted widely amongst interlocutors – is a matter as serious as any other in the theory and ethics of political violence. I'll limit discussion to generalising from act to entity. First, there's the question of defining an entire political movement or organisation sometimes by a relatively small number of its actions or the actions of its members. What proportion, for example, should there be between attacks on civilians and attacks on military personnel before an entity like the Provisional Irish Republican Army can legitimately be denounced as a ‘terrorist’ organisation on a definition emphasising non-combatant targeting? Would only one properly terrorist act in the course of an armed campaign stretching over decades be sufficient to justify the terrorist label or would there have to be many more?Footnote ²⁰ I don't imagine that political or international relations theory could give an answer that is at once determinate and authoritative. Secondly, debates often hinge on whether a wider movement should be defined by the use of terrorist tactics by one of its subdivisions, or, as its supporters might prefer, by the role its other divisions play in constitutional politics or social work. We can see this, for instance, when we think of the relationship between Hamas operatives who organise suicide bombing missions and those who work on charity schemes in Gaza, for instance.Footnote ²¹ Similarly, in Northern Ireland, Sinn Fein has generally tried to differentiate itself for the purposes of political negotiation from its military counterpart, the PIRA, while unionist and other critics insist on referring to the entity as a whole as ‘Sinn-Féin-IRA,’ seeking to undermine the distinction. The stakes in the rhetorical game centrally concern whether one part of a greater organisation (or some actions by some of its parts) can be used to characterise the organisation as a whole and, again, the question is unlikely to meet with a final, determinate answer from philosophy. Whether a speaker is successful will depend in part at least on the wider sensibilities of listeners, the perception of the organisation in terms of its cause or political legitimacy, and other contingent factors.Footnote ²²

We can see the potential usefulness of characterising ascriptions of ‘terrorism’ to organisations in this way in connection with an argument by Virginia Held. She maintains that basing definition on the criterion of attacks on civilians must be wrong-headed because it means, counter-intuitively, that attacks on soldiers can never be seen as terrorist.Footnote ²³ The problem disappears in some cases, however, when the rhetorical nature of speech acts using the word is better understood. Speakers using the word as synecdoche can move from identifying particular acts as ‘terrorist’ on a definition emphasising the deliberate targeting of non-combatants to an ascription that refers to the agent carrying out the attack. Then a second synecdoche is also possible, by which actions even of a conventional military nature may be characterised as ‘terrorist’ where they have been carried out by an organisation known previously to have carried out a sufficient number of terrorist acts in the narrower sense. Successful attempts to ascribe ‘terrorism’ using this kind of double synecdoche – moving from act to agent and back again – reducing the perceived legitimacy of an agent and its actions as a whole.Footnote ²⁴

Revisionary redescription is the second rhetorical approach. Where the redescription of an entity not normally thought to belong in the ‘terrorist’ class based on conventional descriptive criteria aims not just at rhetorical emphasis but at a revision of the criteria themselves then it constitutes what is often called ‘persuasive definition,’ that is, an attempt to alter permanently the criteria by which ascriptions are judged appropriate.Footnote ²⁵ Through this kind of redescription speakers try to alter public attitudes and achieve wider ideological change. They can do so in two quite different ways, the most common being to challenge the descriptive criteria that determine the proper range of the word. However, it is also possible to challenge the normative force of the term too. Carlos Marighela and Virginia Held, for instance, have both adopted this approach in contrasting ways. Since revisionary redescription is a form of rhetoric central to much scholarly and polemical debate about the meanings of ‘terrorism,’ I'll discuss it at greater length in part 2.

The third way for speakers to approach ordinary definitional criteria involves emphatic redescription. This occurs where an entity manifests some but not all of the relevant descriptive features and speakers seek to emphasise its moral affinity with terrorism. This would typically occur where an entity fulfilled the more fundamental evaluative or descriptive criteria but not some of the non-normative, conventional criteria. Thus, even if people normally only used the word to refer to non-state groups, a speaker could still describe a particular state as ‘terrorist’ if it engaged in certain kinds of war crime. Using the word here would draw attention to features of the entity's behaviour that are morally identical to some of those associated with non-state terrorist organisations. As a metaphorical use of the word, it's effectively a truncated way of saying that ‘the actions of that state are little or no better than those of a terrorist organization.’

I now want to turn to some examples to illustrate different ways in which speakers and theorists have put these rhetorical techniques to use. First, I'll look at two more polemical approaches, the kinds that Samuel Scheffler is especially concerned with.Footnote ²⁶ Then I'll talk about some of the more influential theorists of terrorism and show how they too have employed rhetorical techniques of redescription.

Part 2a. Partisans

My first illustration of polemical deployment is Yasser Arafat's address to the United Nations General Assembly in November 1974. A key aim of the speech was to convince Arafat's audience that the Palestine Liberation Organisation and the entities it represented were not ‘terrorist.’ The reason he gave for this, simply, was that they were fighting for liberation: ‘The difference between the revolutionary and the terrorist,’ he declared,

Thus, the argument goes, if we accept that the PLO should be condemned as terrorist, it means that one would also have to condemn other liberation movements in similar terms, including those that are widely remembered as liberating forces in both Western and other historical traditions. Arafat makes the familiar move of arguing that the real terrorists are those who fight against the liberation movements and who use violence to support colonial oppression: ‘those who fight against the just causes, those who wage war to occupy, colonize and oppress other people – those are the terrorists, those are the people whose actions should be condemned, who should be called war criminals: for the justice of the cause determines the right to struggle.’Footnote ²⁸ For Arafat, therefore, the defining criterion is what just war theorists call ‘just cause.’ If the cause is just, then the movement which fights for it must be ‘revolutionary,’ not ‘terrorist.’ It's to other entities trying to prevent the cause from being realised that the ‘terrorist’ label properly applies. All non-state movements, therefore, that can reasonably claim to fight for liberation are exonerated from the charge of terrorism, irrespective of their methods.

Ironically, Arafat's approach bears close comparison with that of Alan Dershowitz whose polemics point in the opposite direction.Footnote ²⁹ Both adopt a rhetorical strategy which effaces the distinction between legitimate and illegitimate uses of violence that their audiences would intuitively make and would wish to defend. ‘Terrorism,’ according to Dershowitz's usage, includes both those groups for which the targeting of civilians is a central tactic and those who use violence more discriminately (like the French, Polish and Jewish resistance to Nazism during the Second World War who ‘targeted primarily Nazi military and police personnel.’)Footnote ³⁰ He argues that all attempts at a stricter definition of ‘terrorist’ have ultimately proven unsuccessful, reinforcing the seemingly relativistic cliché that ‘One man's terrorist is another man's freedom fighter.’Footnote ³¹ He responds to the problem, not by trying to achieve a more secure distinction between one and the other, but by assimilating all non-state groups using violence for political purposes under the same ‘terrorist’ heading.

By including all such groups in the same category, Dershowitz – like Arafat – prohibits any possibility of making a distinction between terrorist and non-terrorist uses of armed force beneath the level of the state. He then further reinforces the point by suggesting that, while there are various criteria by which people might still wish to distinguish a class of acceptable ‘freedom fighting’ within the broader terrorist category, none can be applied authoritatively. They might try, for instance, to place them on a sliding scale from the least bad (those that restrict targeting to active military personnel) to the worst (those that target innocent civilians) or to distinguish those with a justifying cause from those without.Footnote ³² But there will always be disagreement about the right criteria, both from interested parties and ‘reasonable people’,Footnote ³³ so it's impossible to make principled, authoritative differentiations, condemning some non-state armed groups while endorsing others. Since Dershowitz believes that in order to reduce the threat of terrorism, democracies must denounce it and ensure that it is never rewarded with political success, this means that all armed, non-state movements have to be condemned uniformly.Footnote ³⁴

For both Dershowitz and Arafat, therefore, all cats are black at night, as the saying goes. Both aim at a similar result for obvious – but different – polemical purposes and each exemplifies one of the dangerous and unhelpful tendencies cited by Scheffler. One seeks to denounce all enemy states and the other to condemn all non-state armed groups as ‘terrorist.’Footnote ³⁵ So what exactly are they doing with the word ‘terrorist’ here in terms of the different rhetorical approaches I outlined?

On the face of it, each presents himself as employing the first type of redescription: neither indicates any revisionary intention, and both address themselves to their audience in a manner suggesting that they are using the word ‘terrorist’ in the same way anyone would. But putting on this appearance is itself a manipulative ploy (and hence a case of ‘bad’ rhetoric, as I'll suggest). Both are in fact really performing a radical, revisionary redescription of the second kind. Each innovates quite deliberately in two respects: first, they choose criteria that are relevant to questions of political violence, but that aren't usually seen as sufficient to uphold the judgement they hope others will make: non-sovereign status isn't sufficient to define an armed group as terrorist; and the justice of a political cause isn't sufficient to guarantee that a group fighting for it will qualify as non-terrorist.Footnote ³⁶ Second, by employing the wrong criterion, both seek to eliminate the effectiveness of the word ‘terrorist’ in making distinctions within the relevant category, that is, between different kinds of sub-state armed groups, all claiming to fight for a just political cause.

Hence, the problem that Scheffler and others refer to is one of political, polemical rhetoric. But I want to show how non-partisan theorists too engage in forms of rhetorical redescription. This will, on the one hand, support my argument that it's necessary to make distinctions within and between rhetorical approaches. It will also form the basis, on the other hand, for suggestions about how to distinguish between ‘good’ and ‘bad’ rhetorical techniques in section 4.

Part 3

There is therefore some truth behind the fears expressed by the theorists I cited at the beginning: the term ‘terrorist’ is used in a range of different ways, some more innovative than others; and the differences in usage do seem to result sometimes from ideologically motivated rhetorical goals pursued by partisans either of states engaged in fighting against terrorism so-called or of the so-called terrorists themselves. How should theorists deal with these problems? I've suggested that a rhetorical approach to the question could offer some constructive answers to this question and I'll show how it could, first, by comparing a series of different possible philosophical responses before looking, secondly, at some criteria in part 4.

I've already pointed towards the first two responses in the introduction. Debate tends to oscillate between these two views: on the one side, theorists pursue a philosophically compelling and unifying definition. They do this either by arguing that there is really only one criterion by which entities or actions ought necessarily to be defined as ‘terrorist’; or by seeking a ‘consensus definition’ that identifies a cluster of features associated with the things commonly called ‘terrorist.’Footnote ⁵⁸ As I mentioned, however, theorists who think of definition in this way sometimes decide to abandon questions of definition altogether when faced with the plethora of ‘rhetorical’ (re)definitions seen above. The Platonic pursuit of a single essence with a singular conceptual identity, then, collapses into its opposite, effectively a Schmittian approach that sees all definitions of moral terms as assertions of power and as attempts to achieve the hegemony of particular interests or partisan commitments.

Schmitt's view on the inescapability of partisan perspective and polemic in the definition of political concepts which I cited at the beginning of this article reflects Friedrich Nietzsche's insights into the radical contingency of meaning attaching to moral terms.Footnote ⁵⁹ These insights have more recently been brought to bear in historical inquiry into ideological change by Quentin Skinner in his re-appropriation of classical theories of rhetorical redescription.Footnote ⁶⁰ This suggests an important perspective in which to think about the way terms like ‘terrorist’ are subject to change meaning on the long run. It is clear, for instance, that the term ‘terrorist’ is open to a range of more or less radical shifts in meaning – both denotative and connotative – according to the relative rhetorical success of various parties with an interest in ideological change. Or, as the theorists I cited in the introduction warn, its usefulness could be negated altogether if consensus on its meaning breaks down completely, that is, if no innovators succeed in establishing new, hegemonic usages. This gives some idea of the stakes in the game of rhetorical redescriptions of the more radical kind, that is, concerning the directions that moral and political discourse are likely to take and the extent to which theorists and activists could influence the outcome. But taking up Nietzsche and Skinner in this way points toward a third approach through which we can think about the way the word is used and may be subject to contention at this particular moment in historical time.

While the first approach may seem natural to someone advocating a particular moral or ideological view on definition and the second might allow us to say something about conceptual change over the longue durée,Footnote ⁶¹ a third perspective is needed to mediate between them and provide a critical framework for adjudicating between different definitional positions at the present. The third perspective provides a route between the Scylla of Platonic essentialism and the Charybdis of radical contingency. Ironically, perhaps, its general form is suggested by the sceptical remarks that Walter Laqueur makes about defining terrorism: he writes that ‘people reasonably familiar with the terrorist phenomenon will agree 90 per cent of the time about what terrorism is, just as they will agree on democracy or nationalism or other concepts. In fact, it is an unmistakable phenomenon, even if the search for a scientific, all-comprehensive definition is a futile exercise.’Footnote ⁶² Instead of postulating a definition explicitly, Laqueur defers to wider, common sensibilities, assuming in effect that people generally know terrorism when they see it. My view is similar to this, though I resist the essentialising tendency of his text. I assume that people generally know what they intend when they use the word ‘terrorist’ and that they are able to use it because they have internalised a set of rules for correct usage (rules that may apply differently across different contexts) and because they feel the force of certain moral and political values which they need their linguistic usages to reflect. Philosophically, the possibility I'm interested in can be described using terms suggested, on the one hand, by Jürgen Habermas and, on the other, by Richard Rorty. Both provide ways of conceptualising aspects of the interactions of speakers engaged in redescriptions both of the ordinary and the revisionary kinds. A brief look at their suggestions about public discourse in general will allow me to indicate tentatively the possibilities that exist for mediation between different speakers and views on terrorism.

Habermas's synthesis of speech act theory with truth-condition semantics affords a means of linking the use of the term ‘terrorist’ to background norms. If Habermas and his forerunners are correct in assuming that the meaning of a statement is available only to those who understand the conditions in which it could be judged valid, then it points towards the way in which statements involving the word ‘terrorist’ relate to criteria that are known both to speakers and to their intended audiences.Footnote ⁶³ Even disagreements about meaning, on this view, are possible only on the basis of some prior background in shared normative and conceptual frameworks. The validity-condition idea gives us a way of conceptualising how we can regard all deployments of the term ‘terrorist’ as simultaneously grounded in common understandings, concerning both moral norms and rules of linguistic practice, and rhetorical, insofar as speakers try to persuade their audiences to apply emotive moral norms by endorsing and participating in speech acts of reproach. The validity conditions in Habermas's theory correspond to the moral, descriptive, and conventional-linguistic norms analysed above in part 1, norms amounting to the ‘rule-knowledge’ internalised by participants in a particular life-world as part of its social reproduction.Footnote ⁶⁴ Particular speech acts can take more or less radical approaches towards these norms. On the one hand, they can simply apply them in a given case, seeking the agreement of listeners that the application is correct according to normal conditions of validity. Ascriptions can then be contested and interlocutors can challenge purely arbitrary, tendentious or downright propagandist deployments of the term. Cynical deployments of this kind include those speech acts that ‘misfire’, as J. L. Austin put it, because speakers fail to act within conventional rules or in the relevant institutional contexts.Footnote ⁶⁵ Propagandist ‘abuses’ (Austin's word again) correspond to ‘strategic’ actions aimed at perlocutionary effects that are concealed rather than being made fully transparent through the their enunciations. Habermas distinguishes these from ‘communicative’ action aiming at common, fully transparent understanding.Footnote ⁶⁶

On the other hand, speech acts can be directed towards engagement in the radically redescriptive project of trying to challenge the norms of usage themselves, arguing that the validity criteria accepted by the linguistic-moral community are inadequate to its experience or moral sensibilities, or that they are incoherent. The rhetorical idea of revisionary ‘redescription’ which I have developed out of Skinner's reappropriation of classical thought describes a technique for engaging in higher-level debates of this kind and, as the discussion suggests above in part 2, interlocutors may or may not be open and transparent about doing so. In this respect, the word ‘terrorist’ has become an important piece in the ideological language games through which interlocutors in contemporary political debate contest the acceptable limits of armed force in relation to widely recognised moral and political values.

Approaching in this way the definitional controversies surrounding the concept of terrorism highlights the concrete, diachronic forms of verbal and practical interaction through which communities construct and debate meanings and generate phronesis but it also points toward the possibility of broadening the inclusiveness of discourse by admitting new participants.Footnote ⁶⁷ Perhaps the most urgent question concerning plural meanings of the word ‘terrorist’ arises in considering the relationship between democratic discourses – or conversations, to use Richard Rorty's metaphor – orientated by a commitment to human rights and other discourses with different cultural orientations, in a global context. In Rorty's view, the validity of propositions is determined and judged in the first place in and through each particular conversation but the inclusiveness of a particular conversation can then extend beyond and between discursive or cultural communities. Conversations beyond the borders of particular communities that lead to an interweaving of the norms belonging to each can then give rise to wider, broadening conversations and communities.Footnote ⁶⁸ Within communities determined in this way by the scope of a conversation, the achievement of ‘objective’ understandings is always reducible to the desire to establish agreement on some things that are ‘so well justified that, for the moment, further justification is not needed.’Footnote ⁶⁹ Public discourse and debate aims, on this view, not at an objective correspondence between social knowledge and the nature of things, but at ‘as much intersubjective agreement as is possible.’ The wider the agreement, the greater ‘the reference of “us”’ will be; the deeper the consensus, the stronger the solidarity within the community.Footnote ⁷⁰

The rhetorics within which the normal inviolability of civilian lives and life forms the normative centre for principled use of the word ‘terrorist’ can be seen as part of a wider conversation of this kind, one that is international in scope – if not actually universal – and which has the capacity to expand, embracing new interlocutors.Footnote ⁷¹ The degree of separation and the extent of misunderstanding between this conversation and others is easily – and, I suspect, commonly – exaggerated. Even Arafat's focus on the concept of terrorism in 1974 seems to reflect a concern to intervene successfully as a participant in conversation with the west under the auspices of the UN and his speech reflects an awareness of the norms of the conversation even in its attempt to subvert them. More recent interventions in the conversation by Arafat and others show an acceptance of the pragmatic necessity of accepting the rules in a more conventional form. Arafat, for instance, later condemned terrorism using the terms of the orthodox definition,Footnote ⁷² and Hamas has at times acknowledged the force of civilian immunity as a moral and legal principle, even if on a perverse reading of international law.Footnote ⁷³

Adapting norms could cut both ways, of course, and new participants in the rhetorics of ‘terrorism’ will sometimes seek to change the norms of dialogue. But thinking about the ways in which definitional criteria are addressed through different redescriptive strategies gives a clearer sense of how to differentiate between cynical attempts to subvert key norms and more principled attempts to open the conversation to genuine human needs and emancipatory concerns. Valid philosophical reasons for contemplating change in the terms of the conversation may also arise even without the encounter with newcomers. I now turn, therefore, to the criteria that the present approach towards defining and using the word ‘terrorist’ suggests for adjudicating between good and bad rhetorics.

Part 4

The various rhetorical possibilities available for using or debating the meaning of the word can thus be seen as occurring within a wider discursive ‘conversation’ about moral, political, and linguistic norms and their applications to particular actions and entities in contemporary politics.Footnote ⁷⁴ This approach to the question of meaning shows how the ‘rhetoric’ of terrorism may be conceptualised in a way that avoids the Schmittian subordination of meaning to the strategic imperatives of political adversaries while holding open the possibility of contesting meaning and the proper applications of the term ‘terrorist’ and the values it invokes. As an alternative conceptualisation, the approach I have outlined so far gives rise to a series of pragmatic considerations that will affect the way particular arguments about definition ought to be judged. These will indicate ways of distinguishing between ‘good’ and ‘bad’ rhetoric.

First and most crucially, good definitions and rules for use should reflect the commonly recognised need to distinguish between acceptable and unacceptable deployments of violence and enable principled, normative discrimination in speech. I would suggest that in ordinary speech the word usually differentiates between two possibilities: on the one hand, legitimate attempts to achieve political and social emancipation through resistance that may sometimes require the discriminate and proportionate use of force at a sub-state level; and, on the other, illicit violence that purports to be the same but that vitiates or brings into doubt any claim it has to emancipatory potential through human rights violations (‘terrorism’ qua targeting civilian life or lives). There may, however, also be circumstances which I'll discuss where the term applies to the actions of states or their agents.

As we've seen, the ability to discriminate morally using the word ‘terrorist’ may be – indeed probably has been – endangered by speakers who have attempted for strategic, polemical purposes to displace the criteria most directly relevant to evaluations of this kind. But it will also be endangered by attempts, like Virginia Held's, to neutralise the term's normative force.Footnote ⁷⁵ Two consequences could arise from this sort of strategy, neither of which is likely to help clarify the principles by which moral judgements should be made in public debate. First of all, if the word ‘terrorist’ no longer serves in making distinctions and expressing moral disapproval, then either the ability to make moral discriminations within the wider conversation will be blunted or lost altogether, or the problem will eventually be displaced onto other vocabulary. The same debates that now occur in relation to the word ‘terrorist’ would then simply recur in relation to some other word or words and the problem would then merely be replicated, not solved. Problematic as the term ‘terrorist’ may seem, this is an unpromising strategy as introducing a new vocabulary through which to conduct political debate would face even greater obstacles than using established terms and trying to shape them according to norms that are already widely understood.Footnote ⁷⁶

Secondly, and much more harmfully, if ordinary speakers continue to associate ‘terrorist’ with the targeting of innocents, as they are likely to do, but hear theorists following the inclusive definitional approach of Held or Dershowitz and then distinguishing between ‘good’ and ‘bad’ ‘terrorisms’, then this could cause needless moral doubt about clearly unethical acts. It could lead to ambivalence and increased permissiveness with respect to organisations that violate human rights as the false inference is made that these can be involved in ‘good’ targeting of innocents. Of course, sometimes ambivalence may be the only honest and adequate response to highly complex situations.Footnote ⁷⁷ But we should avoid creating unnecessary uncertainty through misguided or cynical attempts to change the accepted meanings of words.

I have argued that in ordinary moral and political language, use of the word typically involves the redescription of an action or entity in a way that uses widely recognised descriptive criteria to invoke the force of common moral norms. From a strictly normative point of view it is essential that definitions appeal to the right kinds of values: on the one hand, those reflected in the rights ascribed to innocents; and on the other, those reflected in recognising a need for remedial actions to address social and political injustice where it occurs. Clearly, in other contexts where these values are not accepted, or where the term ‘terrorist’ is normally used in a purely ad hoc way, the task of theorists will be to challenge prevalent norms as robustly as possible. But within conversations built around the norms discussed above in part 1, definitional strategies should reflect a second kind of concern, a properly rhetorical one that's essential to the effective development and application of normative principles. This is the concern with persuasiveness, that is, the ability of principled interlocutors using the word ‘terrorist’ to persuade others of the validity of applying the term to particular cases. In thinking about usage, it's not sufficient to work with a definition whose lines of discrimination are only defensible in the abstract. For ordinary redescriptions of the kind the definition seeks to underpin to be meaningful and effective in public discourse, the definitional criteria by which they are proposed must be widely understood and accepted. The more deeply embedded the moral and descriptive norms upon which definitions depend, the greater the facility with which speakers can contest the applicability of the word ‘terrorist’ in particular cases. Also, the greater the facility with which such contestations can occur, the greater the chance for a principled, critical consensus in relation to concrete political situations. Theorists should thus try to harness where possible and shape where necessary the critical potential already immanent within the moral and political discourses of terrorism and human rights.

Thirdly, if effectiveness comes from being able to express and apply widely shared moral and political commitments and a widely felt need to differentiate in which the term ‘terrorist’ already has currency, then definitions should probably try to reflect established rules unless there are very good reasons for doing otherwise. I take it, then, that there will be some presumption in favour of semantic conservatism. In this respect, I follow Alison Jaggar's recommendation: she writes that, ‘a good account disturbs existing usage as little as possible. Proposals to modify common interpretations of a term must be given a plausible and appropriate rationale’.Footnote ⁷⁸ Rhetorical considerations regarding the utility of the term terrorist in principled critical arguments help clarify why this is important. Innovations risk reducing the cohesion of discursive norms and thus threaten to diminish the effectiveness of the term – that is, the ability of those using the word in a principled way to speak persuasively – at least in the short term. But revisionary approaches would be defensible if one discovered a troublesome inconsistency in ordinary usage, for example, between basic norms and descriptive criteria. Theorists would then try to shape usage not by introducing unfamiliar rules but by choosing between incompatible rules that are already established in practice. Following Coady and Rodin, for example, there may be valid possibilities for proposing definitional innovation based on revising the descriptive criteria where there is a perceived incompatibility with core normative commitments. This is especially true of the more peripheral, non-normative criteria, for example, whether the word only refers to non-state entities or encompasses all agents capable of violence. Revisionary redescription could also be used as a counter-measure against partisan attempts to subvert discursive norms.Footnote ⁷⁹

Finally, where speakers or theorists believe redefinition is necessary, their attempts to persuade others of this fact should be transparent, appealing to shared norms in an open way. The reasons for innovating should be stated as clearly as possible and hence made available for debate within the language community. This is the main thing that distinguishes the approach of the theorists discussed above in 2.b from the covert rhetorical manoeuvres of Arafat and Dershowitz and yields the most important dividing line between good and bad rhetorics.