Intentional and Performative Persuasion: The Linguistic Basis for Criminalizing the (Direct and Indirect) Encouragement of Terrorism

Article 5 of the Council of Europe Convention on the Prevention of Terrorism requires member states to criminalise ‘public provocation to commit a terrorist offence’. In the U.K., the realisation of this obligation is found in the ‘Encouragement of terrorism’ offence contained in section 1 of the Terrorism Act 2006. As well as fulfilling the U.K.’s treaty obligation, this offence was intended to stop the spread of violent extremist ideology. Although the compatibility of this offence with the right to freedom of expression enshrined in Article 10 of the European Convention on Human Rights has been queried, both the domestic courts and the European Court of Human Rights have held that it complies with Article 10’s demands. So, instead of taking Article 10 as its starting point, this article draws instead on work from the field of linguistics: namely, speech act theory (SAT). By using insights from SAT, and by examining some of the linguistic strategies that may be used to encourage acts of terrorism, the article seeks to advance the legal understanding of the concept of encouragement. In particular, the article draws out two features of encouragement that have important implications for the appropriate boundaries of the encouragement of terrorism offence -encouragement is intentional and it is performative - and argues that, as currently drafted, the offence does not reflect the nature of encouragement as an intentional activity. The article concludes by drawing out from its analysis a series of proposed amendments that together address the rights-based concerns about the offence whilst maintaining its effectiveness as a counterterrorism tool.


I INTRODUCTION
Article 5 of the Council of Europe Convention on the Prevention of Terrorism (CECPT) requires member states to criminalise Ôpublic provocation to commit a terrorist offence'. In the U.K., the realisation of this obligation is found in section 1 of the Terrorism Act 2006. Titled ÔEncouragement of terrorism', section 1 criminalises the publication of statements that are likely to be understood by a reasonable person as encouraging some or all of the members of the public to whom the statement is published to commit, prepare or instigate acts of terrorism. 1 In addition to fulfilling the U.K.'s treaty obligation, the offence was intended to stop the spread of violent extremist ideology by supplementing the then-existing common law offence of incitement. Whilst the common law offence of incitement applied to statements inciting a specific offence (e.g., incitement to murder, incitement to rob), it did not extend to statements that encourage terrorist activity in general without identifying a specific offence. The creation of the encouragement of terrorism offence ensured this broader reach, in particular by introducing the notion of indirect encouragement.
There have been a number of rights-based critiques of the encouragement of terrorism offence that argue that it constitutes an unjustifiable infringement of the right to freedom of expression. There is somewhat of a disjuncture, however, between the conclusions reached in this literature and those of the judiciary, both at the national and transnational level. The right to freedom of expression is enshrined in Article 10 of the European Convention on Human Rights (ECHR), but it is not an unqualified right. Derogations are permitted, inter alia, in the interests of national security and for the prevention of disorder or crime, provided that the derogation is prescribed by law and necessary in a democratic society. Indeed, the European Court of Human Rights has held that restrictions on indirect incitement to commit terrorist acts can be Article 10 compliant, 2 and it declared a challenge to the offence contained in section 58 of the U.K.'s Terrorism Act 2000 (collection of information or possession of a document likely to be useful to a terrorist) on the grounds of vagueness and freedom of expression to be Ômanifestly illfounded' and inadmissible. 3 Challenges brought before the domestic 1 As amended by s 5 of the Counter-Terrorism and Border Security Act 2019. 2 Hogefeld v Germany, App. no. 35402/97 (20 January 2000). 3 Jobe v United Kingdom, App. no. 48278/09 (14 June 2011). courts to this and other terrorism offences on the grounds of freedom of expression have also been unsuccessful. 4 These divergent perspectives should not surprise us, since rights Ôexpress the resolution within society of situations of conflict between the interests of different members of society' and so Ôany approach may find expression in terms of rights, and we do not advance any debate by diverting attention from the issue of what right-duty relationships are appropriate, to a confused discussion over whether one side has, or should have, rights or not'. 5 For this reason, we do not take as the starting point of our critique the right to freedom of expression. This is not to suggest that this right is not important in this context. On the contrary, in terms of counterterrorism policy there are cogent reasons to encourage ideological debate and discussion. Barendt states that ÔWe can only respond intelligently to undesirable extremist attitudes, and remove or reduce the reasons why they are held, if we allow them, to some extent, to be disseminated'. 6 Indeed, one of the core objectives of the Prevent strand of the U.K. Government's counterterrorism strategy is to counter terrorist ideologies. 7 One of the methods for achieving this is to debate extremist ideas and espouse counter-narratives. 8 It is important, therefore, to ensure that there is Ôcareful coordination' between this work and efforts to prosecute those who encourage acts of terrorism (which fall under the Pursue strand). 9 Counterterrorism offences that overreach blur the line between the work of Prevent and Pursue, can render individuals unwilling to participate in Countering Violent Extremism (CVE) programmes for fear of criminal prosecution, and can aggravate feelings of suspicion and resentment on the basis that such programmes are simply a pretext for spying and surveillance. 10 Instead of the right to freedom of expression, this article draws instead on work from the field of linguistics: namely, speech act theory (SAT). Using insights from SAT, the article seeks to advance the legal understanding of the concept of encouragement. When explaining the meaning of this and related terms, the general approach of the courts has been to resort to the use of synonyms. 11 This can result in circularity. For example, in R v Marlow 12 the defendant was convicted of the common law offence of incitement. When discussing the meaning of the word incitement, the Court of Appeal stated that the word encourages Ôrepresents as well as any modern word can the concept involved', adding that for the purposes of the offence of incitement encouragement involves Ôwords or actions amounting to a positive step or steps aimed at inciting another to commit a crime'. 13 The analysis in this article seeks to shed light on what encouragement is by examining some of the linguistic strategies that may be used to encourage acts of terrorism. In addition, it draws out two further features of encouragement that have important implications for the appropriate boundaries of the encouragement of terrorism offence: encouragement is intentional; and, it is performative. It will be shown that, by focusing on how others are likely to understand a published statement, the current version of the offence does not reflect the nature of encouragement as an intentional activity. This not only has important consequences in terms of the breadth of the offence, it also means that a person may be convicted of the encouragement of terrorism offence even when their publication of a statement was not an act of encouragement at all. The article applies these insights to the debate over whether the encouragement of terrorism offence is human rights compliant, drawing out a series of proposed amendments to the offence that together address the 10 Keiran Hardy, ÔHard and Soft Power Approaches to Countering Online Extremism' in Maura Conway, Lee Jarvis, Orla Lehane, Stuart Macdonald and Lella Nouri (eds.), Terrorists' Use of the Internet: Assessment and Response (IOS Press, 2017). 11 Joseph Jaconelli, ÔIncitement: A Study in Language Crime,' Criminal Law and Philosophy, online first access <https://link.springer.com/content/pdf/10.1007% 2Fs11572-017-9427-8.pdf> 19 January 2018. 12 [1997] EWCA Crim 1833. 13 [1997] Crim LR 897 (emphasis added).
rights-based concerns about the offence whilst maintaining its effectiveness as a counterterrorism tool. The article begins, in section II, by introducing SAT and explaining the performativity of language. Section III then outlines the framework for analysing speech acts. Sections IV, V and VI work through the three facets of speech acts within SAT in turn: locution, illocution and perlocution. Section IV focuses on locution, arguing that the encouragement of terrorism offence rightly encompasses indirect, as well as direct, encouragement and offering suggestions for how to elucidate the meaning of indirect encouragement further. Section V focuses on illocution, arguing that the mens rea of the encouragement of terrorism offence should reflect the nature of encouragement as an intentional activity. Section VI focuses on perlocution, arguing that this part of the offence should be framed in terms of language performativity. The article concludes by presenting proposals for the reformulation of the offence.
Before continuing, three further points should be noted. First, the encouragement of terrorism offence encompasses the encouragement of Ôacts of terrorism' and ÔConvention offences'. 14 Whilst in this article we focus solely on the encouragement of acts of terrorism, the arguments that we advance may equally be applied to the encouragement of Convention offences. Second, in order to remain faithful to the original context, throughout the article we use the word encouragement when referring to the text of the Terrorism Act 2006 and the word incitement when referring to the scholarship on SAT. We do not address the question whether the words encouragement and incitement have subtly different meanings; any such differences that do exist have no bearing on the arguments that we advance. Third, in the article we use a number of illustrative examples. These are taken from a wider research project on online jihadist propaganda. The exclusive use of examples from jihadist propaganda reflects only the materials that were available to use and should not be regarded as downplaying other forms of violent extremism.
14 According to section 20(2) of the Terrorism Act 2006, the term ÔConvention offence' means Ôan offence listed in Schedule 1 or an equivalent offence under the law of a country or territory outside the United Kingdom'. The offences listed in Schedule 1 include offences involving explosives, hostage-taking, hijacking of aircraft and ships, and biological, chemical and nuclear weapons

II THE PERFORMATIVITY OF LANGUAGE
The case for an action-based approach to language -that is, one that recognises that language not only reflects reality but also shapes ithas been argued within a number of disciplines, particularly from the second half of the 20th century onwards. For example, in 1960 one of the most influential linguists of the 20th century, Roman Jakobson, proposed a theory of communicative functions of language that comprised six functions: referential, emotive, conative, phatic, metalingual and poetic. 15 Only the first of these concerns language as an exchange of information. The remaining five all involve some action dimension: the Ôconative' function refers to relationships between speakers and what language achieves in this social realm; the Ôpoetic' function attends to the aesthetic dimension of linguistic form and how it links to specific effects, and so forth. 16 Jakobson's theory of communicative functions was critical in challenging the view that language simply reflects reality. A similar concern with the functions of language was held by another highly influential linguist of the 20th century: the ordinary language philosopher J.L. Austin. His work -which originally developed as a series of lectures delivered at Harvard University in 1955 and was posthumously collected and published in the 1960s -includes the famous How to Do Things with Words. 17 This book represented a step forward in terms of articulating how language shapes reality. In it, Austin argued that language is performative, specifically that it consists of a series of speech acts by means of which speakers and hearers achieve things communicatively. 18 Examples of speech acts include apologising (e.g. for wrong doing), naming (e.g. a building or 15 Roman Jakobson, ÔClosing Statements: Linguistics and Poetics' in Thomas A. Sebeok (ed.), Style in Language (MIT Press, 1960). 16 The Ôemotive' function attends to the kind of effects that the expressive and subjective aspects of talk can have; the Ôphatic' function refers to the role played by ritualised communication; and, the Ômetalingual' or Ômetalinguistic' function refers to how language can reflect on itself, including the role it plays in social interaction. 17 J. L. Austin, How to Do Things with Words (OUP, 1962). 18 According to SAT, as originally conceived by Austin and especially Searle, even statements are performative. Searle explained that making a statement (a constative) carries an illocutionary force and is therefore also a performative -a declaration. When a speaker utters an indicative sentence (e.g. ÔI like fish') the speaker takes on a commitment to the truth of the expressed propositional content and, hence, a change is effected (John Searle, ÔHow performatives work,' 12 Linguistics and Philosophy 535 (1989)). a ship), making a declaration (e.g. in court), complimenting, promising, etc.
In order for speech acts to be successful -that is, for their functions to be fulfilled -several Ôfelicity conditions' need to be satisfied. 19 For instance, the felicity conditions for the speech act of warning (i.e. the conditions that need to be satisfied for a warning utterance to be considered to have been successfully issued 20 ) include: -The Propositional Content condition: a warning refers to a future event. -The Preparatory conditions: • The speaker (S) believes that the event will occur and that it will be detrimental to the hearer (H) • S believes that it is not obvious to H that the event will occur -The Sincerity condition: S genuinely believes that the event will be detrimental to H -The Essential condition: the warning counts as an attempt by S to have H recognise that a future event will be detrimental. 21 An example would be a public information campaign informing citizens of the offence created by section 1 of the Terrorism Act 2006. The campaign warns citizens that the publication of statements that encourage acts of terrorism may attract criminal liability and result in a sentence of (up to 15 years') imprisonment. It refers to a future event (conviction and imprisonment) (Propositional content condition). The speaker (the campaign co-ordinator) believes that: conviction and imprisonment will follow publication of a proscribed statement; conviction and imprisonment will be detrimental to the offender; and, without awareness of the offence being raised, citizens would not know the consequences of publishing such a statement (Preparatory conditions). The campaign co-ordinator genuinely believes that criminal conviction and imprisonment would be detri-19 See Austin supra n. 17. 20 It is worth noting that a speech act can be successful in providing a warning even though the warning is not heeded. Similarly, a speech act can be successful in offering encouragement even if the hearer does not respond to the encouragement. 21 John Searle, ÔIndirect Speech Acts' in Peter Cole and Jerry Morgan (eds), Syntax and Semantics: Speech Acts (Academic Press, 1975). Reprinted in John Searle, Expression and Meaning (CUP, 1979). Note that, in those circumstances in which H can take effective action against the event referenced in the warning, the illocutionary force of the speech act (i.e., the warning) is that of seeking a change of behaviour in H so as to prevent the event. mental (Sincerity condition), and the campaign constitutes an attempt to inform citizens that being convicted of the encouragement of terrorism offence will be detrimental (Essential condition).
SAT also helpfully informs the rationale for including certain uses of language within the encouragement of terrorism offence. For example, issue three of (what was then 22 ) Islamic State's flagship English-language magazine, Dabiq, contained an article titled ÔHijrah: From Hypocrisy to Sincerity'. This used the speech act of warninglinguistically realised via different means, from affirmative to conditional statements -to warn readers about faith that does not manifest itself in deeds. Hypocrisy, it said, Ôis discrepancy between what the inner self encloses and what the outer self discloses', adding that Ôminor hypocrisy can beget major hypocrisy' and Ôan unkept promise of hijrah to Allah could result in a devastating ending for the slave'. Applying this to Muslim professionals living in the West (in particular medics and engineers), the article urged them to Ôanswer the call to hijrah' and move to the Caliphate. Failure to do so Ôwill become a greater proof against him on Judgment Day'.
The above two examples illustrate the performativity of language. By warning citizens that it is a criminal offence to publish a statement that encourages acts of terrorism, the public information campaign seeks to dissuade them from publishing such statements. In fact, one of the two key criticisms of the encouragement of terrorism offence is that its dissuasive effect is even greater than this, and that it inhibits wider ideological discussion and debate. The article from Dabiq, meanwhile, used warnings of eternal judgment to try and persuade Muslims living in the West to uproot and move to the Caliphate. Thousands responded to the call, some of whom may have read this article. 23 So whether it was refraining from publishing proscribed statements and participating in ideological discussion, or travelling to the Caliphate, action was achieved via a series of speech acts performed under appropriate (Ôfelicity') conditions. It is this ability of language to change the world into which words are spoken that earmarks speech acts as a potential candidate for criminalisation. Importantly, these two examples also illustrate a further point. In linguistics, it is now generally accepted that the meaning of any utterance depends on the interplay between the actual words being relayed and a range of extra-linguistic circumstances (context). For instance, the text contained in section 1 of the Terrorism Act 2006 derives its authority from the fact that it was debated in both chambers of the Houses of Parliament and received Royal Assent. Similarly, the article from Dabiq was published at a time when Islamic State's territory was expanding: a fact that the magazine used to try and establish the authority of its warning, by claiming evidence of Allah's blessing. 24 More generally, readership of Dabiq magazine must be viewed alongside a range of other factors that contribute to the process of radicalisation (on which more below). In short, all language is situated. 25 This poses a considerable challenge for any legislator wishing to criminalise some form of speech act. To give citizens fair warning of the boundaries of the offence created by section 1, the definition of the offence must specify a priori the statements (and hence the speech acts) whose publication will attract criminal liability as clearly as possible. 26 This requirement is enshrined in Article 7 of the European Convention on Human Rights and its importance has been underlined by the House of Lords. 27 The second of the two key criticisms of the encouragement of terrorism offence is that the offence fails to meet the demands of fair warning. Yet, as we will see below, statements may encourage via a number of different linguistic devices, the import of which may differ markedly depending on their context. It follows that seeking to construct an abstract formula that definitively distinguishes encouraging from non-encouraging statements in advance of their publication is a near impossible task. 24

III UNDERSTANDING SPEECH ACTS
Any speech act involves three facets, namely: -The locution, which is the act of saying something. In Austin's words, this Ôincludes the utterance of certain noises [the phonetic act], the utterance of certain words in a certain construction [the phatic act], and the utterance of them with certain Ômeaning' in the favourite philosophical sense of that word, i.e., with a certain sense and with a certain reference [the rhetic act]'. 28 -The illocution, which explains the reason for which S is using the locution, i.e., Ôfor asking or answering a question, giving some information or an assurance or a warning', etc. 29 -The perlocution, which is the effect of what was said: ''the perlocutionary act always includes some consequences, as when we say ÔBy doing x I was doing y'''. 30 To illustrate these three facets, consider the following example. It was presented by the jihadist propaganda organisations Global Islamic Media Front and Sawt al-Jih ad in Nusantara as being an official statement from Abu Bakr Bashir (on behalf of the group Ansar al-Tawhid in Indonesia).
(1) With a deep heart, I ask you to forgive me as I place this great wish on your strong shoulders, and to continue to be steadfast in your blessed jihad, and to be blessed with either of the honours: victory or martyrdom. 31 Strictly speaking, this is not a phonetic but a textual (written) act, namely a post on the website of a jihadist propaganda organisation. However, whilst SAT was originally developed to account for spoken language, it has subsequently been applied to any communication, including written language. In (1) the speech act being performed is that of requesting: S [the speaker; in this case, the message's author] requests that H [the hearer; in this case, those reading the statement] continues in their Ôblessed jihad' so as to receive the honour of either victory or martyrdom, and also that H forgives him for making the 28  request. 32 This request is explicitly stated by S: ÔI ask you to forgive me … and to continue to be steadfast …, and to be blessed with…' The speech act's locution follows a particular linguistic structure, the core element of which may be represented as follows: First person singular subject pronoun [I] + present simple verb [ask] + second person plural direct object pronoun [you] + to-introduced subordinate clause [to forgive me; to continue to be steadfast …] As explained above, the context of any speech act is critical in shaping its meaning. In (1), for example, the official statement's context tells us something about the meaning S attaches to the word Ôjihad'. In addition, it is important to examine the specifics of the locution, because the particular choice of wording of any speech act also shapes its meaning. 33 This is especially the case when persuasive purposes are at stake. The order in which information is presented is known to be crucial when it comes to persuasion. 34 So too is framing, such that the negative consequences of non-persuasion may be emphasised. 35 In (1), framing and ordering are evident in the choice of words that precede and follow the core aspect of the locution: the request that H continues to pursue (S's understanding of) jihad. Opening the message with the prepositional phrase ÔWith a deep heart' seeks to appeal to H on the grounds of sincerity and emotion, as does S's asking for forgiveness for making the request. Wording the request as Ôa great wish', too, may seek to minimise the perceived degree of imposition on H. Similarly, the final part of the statement refers to beneficial consequences (being blessed, by either victory or martyrdom). Together, these features may increase the likelihood of S's intention being favourably acted upon by H.
Turning next to illocution, in (1) S expressly states that he is performing a request. The request is communicated via the use of 32 The terms speaker and hearer are used throughout this article, so as to preserve the connection to the theory's origins. 33  what in SAT is known as an explicit performative: ÔI ask you'. Alongside other explicit speech act formulae (e.g. ÔI want to express my gratitude' for the speech act of thanking), performative verbal structures such as ÔI ask you' are what characterise direct speech acts -that is, acts of communication in which S's intention is explicated to H, rather than in any way needing to be implied or inferred by H. Clearly, not all communication is realised via direct speech acts. This is especially true of attempts at persuasion, where there are multiple reasons why a persuader might prefer to employ indirect speech acts, as we explain further below.
Turning lastly to perlocution, determining the perlocutionary effect of (1) is far from straightforward, for two reasons. First, any attempt to establish whether or not the intended perlocutionary effect (that H continues in his Ôblessed jihad') has been achieved faces significant evidential difficulties. Since the locution of (1) occurred in the context of a public, online environment, and S did not specify any particular H to which the request was addressed, (1) can be said to have been addressed to anyone reading the locution act that understood its illocutionary force. Establishing the identity of these Hs is difficult, particularly in the case of those that have taken steps to shield their identity, as is investigating their response to S's request. And even if it could be proved that one particular H (out of maybe thousands) read S's request and continued in his Ôblessed jihad'perhaps because this H posted a message to the same website using some form of words that communicated ÔYes, I will continue' -this merely raises a further set of difficulties. Other speech acts and extralinguistic factors may also have played a part in (this particular) H's decision to continue. This is particularly likely to be the case in message-saturated digital environments, in which Hs may read messages such as (1) within a long string of related messages, and may also be exposed to differently mediated (face-to-face; telephone; social media, etc.) messages about the same or similar propositional content. Hs therefore can do nothing but weigh all this information, plus any additional factors relating to other contexts in which they may be involved, before deciding whether or not to accede to S's request. After all, Hs are independent agents, not automatons whoupon recognising a given illocutionary force -rush to realise its intended effects. 36 36 Dennis Kurzon, ÔThe speech act status of incitement: Perlocutionary acts revisited,' 29 Journal of Pragmatics 571 (1998).

IV ENCOURAGING ACTS OF TERRORISM: THE LOCU-TIONARY ACT
To establish liability for the encouragement of terrorism offence, the first requirement that must be proved is that the defendant published a statement or caused another to publish a statement. 37 A Ôstatement' is defined as a Ôcommunication of any description', and includes communications Ôwithout words consisting of sounds or images or both'. 38 This definition is deliberately expansive, and mirrors the application of SAT to any communication. It also recognises the important role that images play in persuasion in general and in terrorist propaganda specifically. 39 ÔPublishing' is defined in a similarly expansive manner, as Ôpublishing [the statement] in any manner to the public' and expressly includes providing an electronic service Ôby means of which the public have access to the statement' and Ôusing such a service … to enable or to facilitate access by the public to [it]'. 40 The legislation's accompanying explanatory notes explain that Internet Service Providers and website administrators may therefore be regarded as publishing statements on their platforms/websites. There is one restriction, however: the statement must have been published to the public. The offence does not apply to private communications. 41 The other actus reus requirement focuses on the content of the statement and its likely interpretation. The prosecution must show that the statement was Ôlikely to be understood by a reasonable person as a direct or indirect encouragement or other inducement to some or all of the members of the public to whom it is published to the commission, preparation or instigation of acts of terrorism or Convention offences'. 42 The Ôpublic' is defined as the public (or any section thereof) of any part of the UK or of another country, and 37 Terrorism Act 2006, s 1(2)(a). 38  expressly includes public meetings or gatherings (regardless of whether payment is required to attend). 43 Our focus in the remainder of this section is the means of persuasion employed in the published statement, specifically, first, the express inclusion of indirect encouragement and then, second, the inclusion of glorification as an illustrative example of indirect encouragement.

The Express Inclusion of Indirect Encouragement
During the Parliamentary debates on the legislation, concern was expressed that the notion of indirect encouragement would encompass statements that express understanding and which, as a result, have the effect of providing encouragement. 44 Cherie Blair's comment, referring to Palestinian suicide bombers, that ÔAs long as young people feel they have got no hope but to blow themselves up you are never going to make progress' 45 was repeatedly offered as an example, as was the following statement from Liberal Democrat MP Jenny Tonge: This particular brand of terrorism, the suicide bomber, is truly born out of desperation … Many, many people criticise, many, many people say it is just another form of terrorism, but I can understand and I am a fairly emotional person and I am a mother and a grandmother, I think if I had to live in that situation, and I say this advisedly, I might just consider becoming one myself. And that is a terrible thing to say 46 These concerns about the breadth of the offence were exacerbated by its stipulation that it is sufficient that the published statement is Ôlikely' to be understood as indirectly encouraging Ôsome' of the members of the public to whom the statement is published to terrorism. 47 Exactly how many members of the public amounts to 43  Ôsome' is unclear, particularly given that the statement may have a global audience of millions. Marchand accordingly commented: It is unreasonable to expect an individual to foresee or ascertain how a statement will be perceived due to the diverse potential audience and subjective, personal interpretation of a statement. By basing the universe of criminalized statements on such vague and unforeseeable concepts, the definition is also potentially too broad 48 Aside from the ambiguity of the wording Ôlikely to be understood' and Ôsome … of the members of the public to whom it is published', and the uncertainty involved in its application, it is also unnecessary and places undue weight on a minority's potential understanding of the statement. Section 1(4) instructs that, when assessing a statement, its contents and the circumstances and manner of its publication must be taken into account. This assessment would include the target audience of the statement, the diversity of this audience and their likely interpretation(s) of the statement. Section 1(1) could therefore simply ask whether the statement (directly or indirectly) encouraged acts of terrorism. In deciding this, the fact-finder would take into account all relevant circumstances, including likely minority understandings of the statement, as per the instruction contained in section 1(4). At present, by contrast, the fact that the statement is likely to be understood by a reasonable person as encouraging a small minority of the target audience to terrorism is not merely a relevant consideration to be considered, but is sufficient in and of itself to establish the actus reus of the offence. When considering statements with a potentially global reach and an audience of millions, elevating the potential views of a minority in this way creates uncertainty and risks a chilling effect on freedom of expression.
In an effort to address concerns about the notion of indirect encouragement, Forcese and Roach suggest a schema that presents ''speech Ôspace''' as five concentric circles of increasing size and argue that offences like the one contained in section 1 of the Terrorism Act 2006 should not encompass more indirect forms of incitement and encouragement. 49 The outer ring of their schema -labelled ÔIncitement propaganda and operations' -encompasses Ôspeech intentionally targeted at furthering the objectives of terrorist groups, whether in terms of recruiting, counselling, threatening, inciting and the communication of operational tools, and techniques that further terrorist purposes'. 50 The inner circle -labelled Ôfree speech core' -covers speech Ôthat raises no concern from the optic of terrorist radicalization'. 51 In between these two are three other categories: ideological speech; apologia; and, radicalized boasting. Forcese and Roach argue that the first two of these should not be proscribed, since ideological speech (exchanges that Ôinclude debates on religious and political doctrine') is not, on its face, linked to violence and apologia (speech that Ôinvolves celebrations and justifications of past acts of violence') Ôis not linked to violence except to the extent that such statements communicate approval of conduct that might then be emulated (but which is not itself called for in the statement)'. 52 Radicalized boasting (statements that Ôfavour and endorse future acts of violence, but may be (and presumably usually are) a form of chest-thumping, far removed from operational intent or ability') is, they state, Ômore difficult', but Ôfalls short of the incitement to hate associated with a hate crime … or outright counselling or instructing a terrorism offence'. 53 Ultimately, Forcese and Roach's schema fails to distinguish adequately between the different categories of speech they identify. First, they appear to place reliance on the speaker's intention, as shown by their definition of incitement propaganda and operations quoted in the previous paragraph. As we will explain below, we agree that the encouragement of terrorism offence should only apply to those with a terrorist intention. However, instances of incitement propaganda cannot be distinguished from ones of radicalized boasting, apologia or even ideological speech by the presence or absence of a terrorist intention. A person might engage in any of these forms of speech with the intention of persuading others to materially support or commit acts of terrorism. In fact, these more indirect forms of encouragement are often more persuasive. Marc Antony's ironic funeral oration is an oft-cited example. 54  and, in so doing, may also address our own face needs as speakers. 55 Indirectness can also serve to underline common ground between S and H, and/or construct S's identity. 56 These are benefits that are well understood by marketers, advertisers and politicians, amongst others.
Forcese and Roach's description of radicalized boasting also contains self-contradictions. Consider the following statement: Such boasting may favour future acts of violence, but it is not directly tied to operational intent or ability. It is speech that falls short of the incitement to hate associated with a hate crime, and does not directly intend to incite or threaten an offence. Moreover, to the extent that it amounts to instruction, the risk posed by this colourful speech does not cross a de minimis harm threshold. Statements like Ôall real Muslims should engage in military jihad' would rarely cross the threshold from Ôradicalized boasting' to Ôincitement propaganda or operations' 57 In this extract, it is said that radicalized boasting Ôdoes not directly intend to incite or threaten an offence'. Two sentences later, the statement Ôall real Muslims should engage in military jihad' is offered as an example of radicalized boasting. Yet, depending on the context, such a statement could well have been made with the intention of inciting a terrorist act -in which case it would not satisfy Forcese and Roach's definition of radicalized boasting but would satisfy their definition of incitement propaganda and operations. This uncertainty is compounded by the fact that, a few pages earlier, Ôassertions of a personal duty to take up arms' was offered as an example, not of incitement propaganda or radicalized boasting, but of apologia. 58 It is hard to know, therefore, which category of Forcese and Roach's schema the statement Ôall real Muslims should engage in military jihad' falls within. This should not come as a surprise since, as we explained in section III, the meaning of any utterance cannot be ascertained solely from the words used, shorn of their context. An utterance's meaning depends on the interplay between the actual words relayed and a range of extra-linguistic circumstances. Indeed, the description of radicalized boasting from Forcese and Roach above points not only to the locutionary act (endorsement of future violence) and its illocution (supposed absence of a terrorist intention), but also the perlocutionary effect (does not cross a de minimis harm threshold). We will return to perlocution below. Turning next to ideological speech and apologia, Forcese and Roach suggest that these are marked by the lack of an overt link to terrorist violence. Ideological speech, they state, Ôis not, on its face, linked to violence', whereas apologia is only linked to violence to the extent that it suggests that certain conduct should be emulated. 59 Choudhury attempts to distinguish between direct encouragement and indirect encouragement in a similar way. Drawing on anti-discrimination law, he says that Ôin the case of direct encouragement, a suggestion of encouragement can be found on the face of the statement', whereas indirect encouragement Ôincludes statements which have the effect of encouragement even though this cannot be gleaned from the face of the statement'. 60 From a linguistic perspective, however, neither of these suggestions is sustainable.
In section III above we explained that direct speech acts are acts of communication in which S's intention is explicated to H. The request is explicated via an explicit performative in which S expressly states that he is performing a request. Statement (1) was an example of a direct speech act, the explicit communicative being the words ÔI ask you'. As we indicated above, there are other ways in which speech acts may be performed that do not rely on directness. The reality of language use is that Ss often do things with words without employing specific verbs or verb phrases that explicitly state that they are performing such activities. For example, the requestive illocutionary force of the statement ÔI ask you to forgive me' could be realised via many other locutions, including different degrees of indirectness, 61 such as: 59 Ibid, 48-49. (1-a) ÔWould you be able to forgive me?' (1-b) ÔTo err is human, to forgive is divine, as the old saying goes' Whilst (1-a) and (1-b) are requests, neither employs an explicit performative. (1-a) makes use of a Ôquery preparatory' strategy, that is, an utterance (worded as an interrogative here) Ôcontaining a reference to preparatory conditions (e.g. willingness, ability) as conventionalized in any specific language'. 62 For its part, (1-b) uses the requestive strategy known as Ôstrong hint', that is, an utterance Ôcontaining reference to object or element needed for the implementation of the act'. 63 The strong hint in (1-b) is worded as a culturally-rooted statement (a proverb or saying). In other words, both (1-a) and (1-b) use indirectness to realise the act of requesting forgiveness. Which realisation of the request will be favoured in a particular situation will depend on a range of factors, some of which (e.g. cultural style) may not even be linked to the specific context of the communication. 64 Indirect speech acts are of considerable practical importance. As noted above, they can carry more persuasive force than direct acts. Moreover, some communicative behaviours may hardly ever be realised directly. Incitement itself is an example. The statement ÔI incite you to make an explosive' is syntactically possible, but in practice S is very unlikely to incite by making such a statement. 65 Mey has accordingly developed a theory that accounts both for the performativity of language and the Ôindirect speech act dilemma' in linguistics. 66 His theory works Ôfrom the outside in': Ôthe focus is on the environment in which both speaker and hearer find their affordances, such that the entire situation is brought to bear on what can be said in the situation, as well as on what is actually being said'. 67 Within this theory, two concepts are key. First, pragmeme. This refers to a Ôgeneral situational prototype, capable of being executed in the situation', such as denying, bribing, co-opting, warning, inciting and so forth. 68 Second, practs. These are the instantiated pragmatic acts via which a pragmeme may be realised. Two points follow from this. First, attempting to distinguish between direct and indirect encouragement by asking whether a suggestion of encouragement and/or violence can be found on the face of the statement has no basis in action-based theories of language. In fact, from a speech act perspective such a distinction is incoherent since it would create a prioristic, arbitrary distinctions between practs belonging to the same pragmeme, when such distinctions can only be made in situ (i.e., in relation to a specific context). For example, only one of the statements Ôdon't you think it is time you inflicted harm and suffering on the West?' and Ôhaven't you had enough of being humiliated by the West?' mentions violence on its face; yet, both are examples of the same pragmeme and use the same linguistic realisation, namely a strong hint syntactically expressed as a rhetorical question.
Second, an understanding of the distinction between direct speech acts and indirect speech acts reveals the limited scope of the former. To limit the encouragement of terrorism offence to statements that encourage via a direct speech act -that is, statements that employ an explicit performative such as ÔI encourage you to …' -would give the offence such a narrow scope as to render it practically worthless. Both the utility and the integrity of the offence require that it also encompasses indirect speech acts. In this regard, it is important to 69 Ibid, 221. note that the inclusion within section 1 of indirect encouragement did not, in fact, represent a significant extension of the common law of incitement. 70 As Hunt observes, ÔAs far as Common Law incitement is concerned, what is said need not expressly exhort the commission of the offence in order to constitute Ôencouragement' and the inciter will be guilty of incitement if they imply that others should commit the offence'. 71 Concerns about the reach of the section 1 offence, which critics often attributed to its express inclusion of indirect encouragement, were in fact the product of other features of the offence -particularly its mens rea requirement, which we discuss in section V below. 72

The Inclusion of Glorification as an Example of Indirect Encouragement
As we saw earlier, section 1 of the Terrorism Act 2006 is itself a speech act. It specifies the conduct for which citizens may be convicted of, and punished for, the crime of encouragement of terrorism. We have also examined the distinction between direct speech acts and indirect speech acts and seen that, within the category of indirect speech acts, pragmemes have multiple practs. Moreover, in the case of persuasion-related pragmemes (including incitement), which are rarely realised via explicit performatives, it is impossible to determine their illocutionary force without considering factors outside of the locution itself. The meaning of a given statement of encouragement depends not only on the specific locution act but also the circumstances in which it is performed, including the broader extra-linguistic context. This is recognised in section 1(4), which instructs fact-finders to consider the contents of a statement and the circumstances and manner of its publication. At the same time, the need to assess statements on an individualized basis poses a serious challenge to attempts to give advance warning of the statements that, in general terms, will and will not be held to amount to encouragement.
One response to this dilemma is to seek other methods of communicating the boundaries of the offence, besides a generalized dictionary-style definition. One such method is the use of illustrative examples. 73 Collectively, a set of illustrative examples can communicate the key features of the type of situation the legislator is seeking to prohibit. 74 Importantly, such an approach is capable of engendering as much certainty as one based exclusively on rules (if not more). 75 Indeed, during the Parliamentary passage of the Terrorism Act 2006, the then Home Secretary stated that Ôglorification features in the Bill as an example of what is encompassed by the concept of indirect encouragement'. 76 Section 1(3) explains that a statement indirectly encourages terrorism if it satisfies two conditions. The first is that it glorifies the commission or preparation of acts of terrorism. Here, glorification means Ôany form of praise or celebration', 77 and the glorification can relate to a past or future terrorist act or to acts of terrorism in general. 78 Importantly, the second condition is that the statement is one from which members of the public Ôcould reasonably be expected to infer that what is being glorified is being glorified as conduct that should be emulated by them in existing circumstances'. 79 So glorification alone is insufficient to amount to indirect encouragement; there must be glorification with the inference of emulation.
During the Parliamentary passage of the Act the Joint Committee on Human Rights expressed concern at section 1's use of the term Ôglorification', suggesting that it is Ôtoo vague to form part of a If the celebration of acts of terrorist violence is understood by listeners as encouraging them to emulate them, in circumstances where it is also likely that they will soon commit further terrorist acts, then there is no reason of principle why a speaker glorifying the acts should not be held as guilty as a speaker using more direct language 83 We would add that, in our opinion, it is ironic that those who criticised the encouragement of terrorism offence for being too vague should also have called for the removal of an illustrative example that (2) It is mandatory for every The practs in (2) -(5) are linguistically realised via different levels of indirectness. Examples (2) and (3) are statements. In (2), reference is made to an obligation (ÔIt is mandatory…') and a prohibition (Ôit is forbidden…'). In the example, S (the individual who posted the message) does not present himself/herself as its Ôauthor', that is, as Ôthe agent who puts together, composes or scripts the lines that are uttered'. 88  authorship is to be attributed to an impersonal Ôit'. For his/ her part, S appears as the message's Ôanimator', that is, Ôthe sounding box from which the utterances come'. 89 That being the case, and there being nothing in (2) that indicates that the roles of Ôanimator' and Ôauthor' overlap, one cannot infer that the obligation is owed to S. It has been suggested that appealing to Ôoutside forces' in this way Ômay be understood as a means of avoiding reactance and preserving -even nurturing -the goodwill of sympathisers'. 90 Identifying authorship in (3) may be easier given that the example is part of an official statement from the Islamic Emirate of Afghanistan. Therefore, one may attribute authorship to officials from that Emirate, assigning the individual/s working for them the production role of Ôanimator/s'. In addition, in (3), the content of the statement makes use of a common argumentation strategy for the construction of polarised identity groups: Ôthe ideological square.' 91 Developed by Teun van Dijk to theorise social identity formation, this concept relies on four discursive strategies (the four corners of the square) via which language users seek to construct the identity of the groups/s with which they wish to affiliate (the Ôin-group/s') as diametrically opposed to those they seek to disaffiliate from (the Ôout-groups'), namely: jured and detained'. Such out-grouping or Ôthey-ing' practices do not explicitly call for retaliation in kind, i.e., an invasion of America by Afghan people. Instead, they state facts that may be linked to the illocutionary force of incitement.
Examples (4) and (5) share the fact that they directly address the intended Hs. In (4), this is done by explicitly stating: ÔIn this section, the OSJ, we give our readers suggestions on how to wage their individual jihad. Here is one idea of how an individual Muslim may do so'. What follows is a series of statements that seek to encourage following the suggestions through repeated reference to their simplicity. In (5) the S prefaces the strong hint strategy with a deontic modality utterance: ÔI have to say this'. The two rhetorical questions that follow are the locutions via which two specific requests for action against America and the American Government are respectively hinted at: not to co-exist peacefully with America and not to be loyal to the American Government. The rhetorical essence of the questions comes from their containing out-grouping or they-ing discourse, specifically referring to America as Ôa nation that is responsible for the tyranny and crimes committed against our own brothers and sisters', and to the American Government as Ôleading the war against Islam and Muslims'. Given such emphasising of bad properties (tyranny) and actions (crimes and war leadership) against the ingroup (Ôour own brothers and sisters' and ÔIslam and Muslims'), the only, obvious answer -and hence the encouragement force -is Ôno, I cannot'.
So the pragmemes of incitement and glorification may be realised via a number of different practs, all of which could be used in the legislative definition of the offence as illustrative examples of indirect encouragement. It should be emphasised that this list of practs is not intended to be exhaustive. An exhaustive list is impossible because, as Kurzon states, Ôany statement may constitute incitement given the appropriate situation'. 94 Given this adaptability of the term Ôindirect encouragement', we do not believe that the offence definition should add the gloss Ôor other inducement' to the words Ôdirect or indirect encouragement'. Not only is it unnecessary surplusage, its implication that there are other types of statement that are not (directly or indirectly) encouraging but which are an inducement is also potentially misleading. 94 See Kurzon supra n. 36, 588.

V ILLOCUTIONARY FORCE
The mens rea of the encouragement of terrorism offence is either an intention to (directly or indirectly) encourage members of the public to commit, prepare or instigate acts of terrorism, or recklessness as to whether the statement will have this effect. 95 A defendant may therefore be convicted of this offence absent proof of a terrorist intention. 96 There is, however, a defence of non-endorsement. This applies where: (a) the statement neither expressed the defendant's views nor had his endorsement; and, (b) in the circumstances it was clear that the statement neither expressed his views nor had his endorsement. 97 It should be noted that the non-endorsement defence is only available in cases of reckless encouragement. 98 It is therefore unavailable, not only to those whose purpose is to encourage acts of terrorism, but also those who foresaw such encouragement as a virtually certain consequence of the statement's publication (i.e. those who had an oblique intention to encourage acts of terrorism). 99 (It should also be noted that, in the case of an electronically published statement, the non-endorsement defence is unavailable if, in the opinion of a police constable, the statement is unlawfully terrorism-related and the defendant fails to comply with a take-down notice from the constable within two days. 100 Concern has been expressed here about the Ôlack of appropriate checks and balances'. 101 ) As Chan and Simester explain, mens rea has an important role in establishing culpability, but Ôit serves other purposes too'. 102 These other purposes include: first, Ôsometimes mens rea constitutes the morally wrongful character of D's behaviour in a more fundamental way, by identifying what kind of action D is performing'; and, sec- ond, Ômens rea plays a key mediating role in criminalisation, being part of the trade-off between the protection of potential victims and the preservation of liberties for potential defendants'. 103 Each of these functions is relevant to the encouragement of terrorism offence. We will consider them in turn.
First, the mens rea of an offence may assist us in determining the character of an action. This is certainly true of speech acts. As we saw in the previous section, according to SAT it is the illocution -S's intention behind the locutionary act -that is the key aspect to consider in determining the intrinsic meaning of a speech act. If S lacked any intention to encourage, it follows that S's speech act lacks the necessary illocutionary force to be regarded as an act of encouragement. This insight is valuable in considering the question whether reckless encouragement ought to be criminalised. In his work on the law on complicity, Cowley discusses this question and concludes, tentatively, that Ôreckless enabling by encouragement' should be. 104 As an example, he offers the declaration of English King Henry II Ôwill no one rid me of this turbulent priest?' -following which, four of the King's knights travelled to Canterbury to kill the Archbishop Thomas Becket. Cowley observes, ÔI think it is possible that Henry was genuinely reckless rather than calculating on this occasion, but that such recklessness would be enough to charge him as an accessory to Becket's murder'. 105 From a linguistic perspective, the King's utterance could have been issued as either a conventionally indirect request or an indirect complaint. In SAT, requests have various components: a head act (the request proper; a mandatory component) and a series of grounding moves (e.g. apologies, statements of need, etc; all optional). Blum-Kulka et al identify a total of nine head acts. 106 These vary in terms of directness, with mood derivables (imperatives) being the most direct and mild hints the most indirect. In between these two extremes are suggestory formulae (suggestions to do something) and query preparatories, that is, utterances that contain a conventionalised reference to a preparatory condition, such as ability or willingness (e.g. ÔWould you mind walking the dog?' or ÔCould you pass 103 Ibid, 384. the salt?'). King Henry II's utterance uses a query preparatory formula. Given this indirectness, it is particularly important to consider the context in which the utterance was issued in order to assess its illocutionary force and, therefore, whether it was indeed intended as a request. The fact that the utterance was issued by a monarch, whose role regularly entails asking others to do things on his behalf, would support an interpretation of the utterance as a conventionally indirect request. Viewed in this way, the King would be deemed to have intentionally requested the killing, albeit indirectly. No question of recklessness arises.
The competing interpretation is that the utterance lacked the illocutionary force of a request and was instead an indirect complaint. Various contextual factors might support such an interpretation. Perhaps the King normally expressed his requests directly, e.g. using mood derivables, or using certain formulae, e.g. ÔI hereby instruct you …'. Or perhaps the King only issued requests in certain settings, such as meetings of a particular type, and the utterance was not issued in one of these. If the utterance is interpreted in this way, it follows that the King was not requesting that the priest be killed. Yes, the utterance may have been reckless, as Cowley opines. But the nature of the speech act is determined by its illocutionary force and, on this interpretation of the King's utterance, he did not issue a request and he did not utter words of encouragement.
To be clear, our argument is not that reckless statements are never blameworthy. Whether there are some (limited) circumstances in which such statements may justifiably be criminalised as a distinct wrong is a separate question that is beyond the scope of this article. 107 Our point is that statements that were made without an intention to encourage terrorism should fall outside the section 1 offence, since absent such an intention such statements lack the necessary illocutionary force to be regarded as the speech act of encouragement in the first place.
At this point, one might object that we are placing too much weight on how the term encouragement is understood in SAT. Legislators can extend terms and give them a technical legal meaning if they wish -particularly if there are strong policy reasons for doing so -and it might be argued that there are sound policy reasons for ensuring that reckless encouragement falls within the section 1 offence. 108 We disagree, for two reasons. First, there are equally strong (if not stronger) policy reasons for excluding reckless encouragement from the section 1 offence, as we explain in the next paragraph. Second, a recklessness mens rea standard would be incongruous with our analysis of the concept of encouragement in the previous section and the proposed clarification of the term Ôindirect encouragement' that we offer in the conclusion. As we have explained, it is the illocutionary force of a statement -the reason for which S is using the locution -that is the key aspect to consider in determining the meaning of a speech act. On the SAT-based approach we have proposed, a recklessness mens rea standard would therefore be ineffectual. Even if a defendant were found to have published the statement in question recklessly, so that the mens rea of the offence was established, without proof of an intention to encourage the inevitable conclusion would be that the actus reus of encouragement was absent. King Henry II's statement illustrates this. If the statement is interpreted as an intentional (indirect) request, no question of recklessness arises. Or, if the statement is not interpreted as an intentional request, it does not (on a SAT perspective) amount to encouragement even if its utterance was reckless. The only way in which a recklessness mens rea standard would broaden the scope of the encouragement of terrorism offence would be if the defendant's recklessness was also relevant to the assessment of whether there was the actus reus of encouragement in the first place. But to accept this would be to depart from the SAT-based approach for which we have argued and, by undercutting the relevance of SAT in this context, leave us with the current ambiguity and confusion that surrounds the notion of indirect encouragement.
This leads us to the next of the purposes served by mens rea requirements, namely, that they may preserve the liberties of potential defendants. As Chan and Simester explain, ÔSetting a more stringent mens rea requirement means that the prospect of liability will have less impact on the day-to-day lives of law-abiding individuals'. 109 This is important in the current context, given the range of statements that may foreseeably have the effect of indirectly 108 Thank you to one of the reviewers for encouraging us to address this point. 109 Chan and Simester supra n. 102, 395. encouraging individuals to engage in terrorism-related activity. As explained above, jihadist propaganda frequently encourages acts of terrorism through the construction of in-group and out-group identities. 110 Ironically, these identities can be reinforced by statements that were intended to denounce terrorism. An example is the statement from former US President George W. Bush, in the aftermath of the 9/11 attacks, that ÔEither you are with us or you are with the terrorists', which is used in Islamic State's Dabiq magazine to evidence the claim that there are Ôtwo camps before the world for mankind to choose between, a camp of Islam -without the body of Khilafah to represent it at the time -and a camp of kufr -the crusader coalition'. 111 Similarly, statements emphasising the importance of tolerance are used to reinforce in-group and out-group identities by highlighting the supposed incompatibility of Islam with democracy, 112 as are discussions of U.S. foreign policy. 113 Given the broad range of statements that may foreseeably have the effect of indirectly encouraging terrorism, the key question within the current version of the section 1 offence is whether the publication of the statement was reckless -which in turn depends on whether or not the statement's publication was justified. Whilst many might feel that this test of justifiability provides them with a sufficient safeguard against prosecution, this may not be the case for others, such as those from ethnic minorities and/or those curious about or contemplating extremist worldviews. For these individuals, this uncertainty contributes to the reluctance to engage in ideological debate and discussion, the feelings of suspicion and resentment and the unwillingness to participate in CVE programmes that we identified in the introduction. Limiting the mens rea of the offence to intentional encouragement, as we have suggested, would reduce this chilling effect on freedom of expression and more carefully demarcate the Prevent and Pursue strands of the U.K.'s counterterrorism strategy.

VI PERLOCUTION
The definition of Ôpublic provocation to commit a terrorist offence' found in Article 5(1) of CECPT includes the requirement that the communication in question Ôcauses a danger that one or more [terrorist] offences may be committed'. Section 1 of the Terrorism Act 2006 does not contain an equivalent provision. In fact, according to section 1(5)(b) it is irrelevant Ôwhether any person is in fact encouraged or induced by the statement to commit, prepare or instigate any such act or offence'. This lack of a danger requirement was criticised by a number of commentators, including Barendt, who pointed out that someone could be convicted of the encouragement of terrorism offence even if the leaflet he published was only circulated to a handful of people and was unlikely to persuade any of them to terrorist offences. 114 Performativity refers to the ability of language to shape reality -to our doing things with words, as per Austin's famous dictum (see above). Perlocutionary effect is one of the three facets of performative speech acts (or pragmemes, in Mey's terms) -alongside locution and illocution. In fact, it is arguably the most controversial aspect of SAT, with some suggesting that it falls outside the remit of SAT altogether. 115 Some of the difficulties in establishing perlocutionary effect were outlined above. These are illustrated by R v Faraz. 116 At trial Faraz, the manager of a bookshop in Birmingham, was convicted on eleven counts, including seven counts of disseminating a terrorist publication. 117 As part of its case, the prosecution had sought to adduce evidence that a number of known terrorists -including the 7/7 bombers and members of the airline liquid bomb plot conspiracy -had been in possession of publications that had been available at Faraz's bookshop. The Court of Appeal held that such evidence may only be adduced if: (1) there is admissible evidence that the known terrorists 114 Barendt, ÔIncitement to, and Glorification of, Terrorism,' supra n. 6. 115 Over two decades ago, one of the most influential scholars within Pragmatics (the field in which SAT is placed) argued that Ôperlocutionary effects do not form part of the study of pragmatics, since pragmatic [illocutionary] force has to do with goals rather than with results' (G. N. Leech, Principles of Pragmatics (Longman, 1983), 203). Kurzon's detailed study of the incitement pragmeme fully supports Leech's view, taking the further step of arguing that perlocutionary acts are Ôoutside the domain of speech act theory' and calling for the replacement of the term perlocution with Ôuptake' (supra n. 36, 595). Marcu's theoretical take on SAT is tellingly titled ÔPerlocutions: the Achilles' heel of speech act theory' (supra n. 34). 116  had in fact been encouraged by the publications to commit terrorist offences (since this would be relevant to the jury's consideration of the likely effect of the publication upon its readership) ([43]); or, (2) it is admitted for the narrow purpose of determining whether the readership of the publication included people who were likely to regard the contents of the publication as encouragement to commit acts of terrorism, in which case it should be accompanied by strict guidance as to its relevance and Ôa firm health warning as to its limitations' ([48]). Importantly, the Court stated: [I]t was not in doubt that many of the convicted terrorists had other publications in their possession. It was entirely possible that if they had been influenced at all they were influenced by that other material rather than that which was indicted in the present case. It was entirely possible that those Ôof a jihadi disposition' acquire all sorts of material as a form of declaration or selfjustification. It did not follow that they were converted to jihadism or were encouraged by the terms of the publications themselves to commit terrorist acts In other words, perlocutionary effect could not be assumed from the fact that the known terrorists were in possession of the relevant publications.
The difficulties in establishing the perlocutionary effect of any public, let-alone mass-mediated, message are as -if not more -acute in the online realm. The leading empirical study of online radicalisation found that, whilst the Internet has created more opportunities for individuals to become radicalised and appeared to facilitate the radicalisation process, there was little evidence of self-radicalisation (that is, where radicalisation occurred without virtual and/or physical interaction with others). 118  radicalisation is a social process. 120 And even if a person does selfradicalise, they are likely to have consumed a large quantity of online material, so identifying the specific role played by any individual item will be extremely difficult. To require proof of perlocutionary effect would, therefore, be to impose an insurmountable obstacle in the vast majority of cases involving public statements that encourage acts of terrorism.
Turning to performativity, a distinction may helpfully be drawn between a defendant's forum internum (that is, his or her inner convictions and beliefs) and the forum externum (the external realm of action). This distinction is used in German criminal law to justify the criminalisation of conspiracy. 121 If a lone individual merely thinks about committing a crime, this remains part of his forum internum. But when an individual agrees with others to commit a crime, his thoughts join with those of the others and move from the forum internum to the forum externum. The resultant group dynamic has a number of important effects, including: the removal of internal constraints to becoming involved in criminal activity; a sense of group identity that solidifies loyalty, encourages risky behaviour, and leads to individuals acting against their self-interest; and, specialisation of labour and economies of scale. 122 It was for this reason that the Law Commission for England and Wales opted to retain the (statutory) offence of conspiracy, commenting that agreements to commit crime create Ôa special kind of threat'. 123 In other words, such agreementsverbal or otherwise -are performative; they create a danger to the public that did not previously exist. Similarly, when an individual publishes a statement to the public that intentionally encourages acts of terrorism, his thoughts move from the forum internum to the forum externum. Others may now read his thoughts and beliefs and be influenced by them. The act of publication thus serves to materialise the performativity of language; it creates a risk that did not previ-ously exist that others will read the statement and be encouraged to act upon it.
The requirement in Article 5(1) of the CECPT that the communication in question Ôcauses a danger that one or more [terrorist] offences may be committed' should be understood as stipulating that the communication was performative, not that it is necessary to prove perlocutionary effect. This interpretation is consistent with the use of the word Ôdanger', which points to the potential of the communication to shape real-world events but does not require this potential to have been realised. By contrast, the statement in section 1(5)(b) that it is irrelevant whether the statement in fact encouraged anyone to commit, prepare or instigate an act of terrorism should be understood as saying that perlocutionary effect is not a prerequisite of liability for the encouragement of terrorism offence. Understood in this way, there is no inconsistency between Article 5(1) and section 1(5)(b). The problem is rather that each only presents part of the picture. Article 5(1) focuses on performativity, whilst section 1(5)(b) focuses on perlocutionary effect. Whilst it might perhaps be suggested that the section 1 offence's requirement that the statement was published to the public obliquely limits the offence to performative statements, this must be rejected. In some cases, a statement that encourages terrorism could be published publicly without creating such a danger -such as where someone shares excerpts of violent extremist propaganda on a closed, password-protected forum that is open only to individuals with official security clearance -and so it would be wrong to effectively render the danger requirement redundant in this way. Our proposed reformulation of the offence accordingly amalgamates the wording of Article 5(1) and section 1(5)(b) in order to require performativity -in the sense we have described -but not perlocutionary effect.

VII CONCLUSION
This article has shown that encouragement is, by its very nature, an intentional and a performative activity. As a pragmeme, its illocutionary force is that of getting another person or group to do something. This, as we have seen, can be realised via numerous practs that predominantly share indirectness as a linguistic strategy, from rhetorical questions and statements of obligation and prohibition to statements of Ôfact' and suggestory formulae. Across all these realisations, the intention (or illocutionary force) of encouragement points to the commission or preparation of an act of terrorism -be it the construction of a bomb or weaponisation of a vehicle, to name but two examples discussed in the article. It is the combination of uttering such practs in spoken/written/visual form (i.e., locution) with the intention in their given context of use (i.e., illocution) that make the speech act of encouragement a performative activity.
Recent U.K. legislation has recognised the importance of tightening overly broad counterterrorism powers. 124 In this spirit, using the insights gleaned from SAT, this article has evaluated the current formulation of the encouragement of terrorism offence and proposed a number of amendments. The following is a revised version of the first five subsections of section 1 of the Terrorism Act 2006 that incorporates all of these changes:

Encouragement of terrorism
(1) A person (D) commits an offence if: (a) D publishes a statement to the public, or causes another to do so; (b) D publishes the statement (or causes the other to do so) with the intention of encouraging the commission or preparation of acts of terrorism; and, (c) the statement encourages the commission or preparation of acts of terrorism.
(2) Encouragement may be direct (by statements such as ÔI encourage you to …' or ÔI urge you to …') or indirect (as detailed in subsection (3)).
(3) Statements that indirectly encourage acts of terrorism include those that use statements of fact, obligation or prohibition, suggestory formulae and hints (for example in the form of rhetorical questions) to: (a) glorify the commission or preparation of acts of terrorism (whether in the past, in the future or generally); or (b) denigrate a section of the public where the statement may reasonably be understood as seeking to persuade those to whom it is published to commit or prepare acts of terrorism.
(4) To assess whether a statement encourages the commission or preparation of acts of terrorism, regard must be had to the content of the statement, its intention, and the circumstances and manner of its publication. (5) It is immaterial whether anything mentioned in subsections (1) - (3) relates to the commission or preparation of one or more particular acts of terrorism, of acts of terrorism of a particular description or of acts of terrorism generally. (6) An offence is only committed under this section if the statement created a danger that other people would be encouraged by it to commit or prepare acts of terrorism, though there is no requirement that other people were in fact so encouraged.
Subsection (1) states the actus reus and mens rea requirements of our revised version of the offence. It removes the existing reference to how the statement is likely to be understood, as well as the reference to the instigation of acts of terrorism. 125 It also requires proof that the statement was published with the intention of encouraging the commission or preparation of acts of terrorism. This stricter mens rea requirement means that there is no longer any need for a defence of non-endorsement. The three requirements for liability are also deliberately listed in an order that requires the illocutionary force of the statement to be considered before an assessment is made of whether or not the statement encourages terrorism.
The next two subsections provide some clarification of the meaning of encouragement. Subsection (2) specifies that encouragement may be direct or indirect, and offers examples that illustrate the limited scope of direct encouragement. Subsection (3) provides a more comprehensive set of illustrative examples of indirect encouragement than the current version of the offence. As well as including four specific examples of the practs that might be used, it highlights the fact that a statement may seek to encourage by using these practs to denigrate an out-group as well as to glorify past or planned acts of the in-group.
The next three subsections deal with wider aspects of the offence. Subsection (4) largely mirrors the existing section 1(4), but importantly adds the intention of the statement to the list of things to consider when determining whether the statement encourages the 125 Our reason for removing the reference to encouraging another person to instigate an act of terrorism is that it is unnecessary; such a person can already be charged under the Serious Crime Act 2007 with intentionally encouraging the encouragement of terrorism offence. commission or preparation of acts of terrorism. Meanwhile subsection (5) retains what is currently section 1(5)(a), making it clear that there is no requirement that the statement in question encourage a specific type of terrorist act. Subsection (6) merges the existing section 1(5)(b) with Article 5(1) of CECPT, clarifying that whilst the offence only encompasses statements that are performative it does not require proof of perlocutionary effect.
Our reformulation is consistent with the offence's original rationale. By including indirect forms of encouragement and retaining the lack of a requirement that the defendant be shown to have encouraged a specific criminal offence, it encompasses those who, in the words of the then Home Secretary, Ôincite terrorism more obliquely'. 126 At the same time, it addresses the key criticisms of the current version of the offence. Concerns about vagueness are addressed by providing further clarification of the term indirect encouragement and by providing that the actus reus may no longer be established solely on the basis of whether a fact-finder believes a reasonable person would be likely to understand the statement as encouraging some or all of its audience to terrorism, whilst concerns about over-breadth are addressed by limiting the reach of the offence to those with a terrorist intention. Our proposed reform is thus more protective of those who engage in ideological discussion and debate without any terrorist intention since it excludes them from the scope of the offence, as opposed to requiring them to rely on prosecutorial discretion. 127 Reducing the overreach of the offence in this way would more clearly demarcate the Pursue and Prevent strands of the Government's counterterrorism strategy, enabling greater coordination between these two strands of work and reducing the suspicion that sometimes surrounds CVE programmes.
Whilst our analysis of the U.K.'s encouragement of terrorism offence did not take as its starting point the Article 10 ECHR right to 126 Charles Clarke (Hansard HC Vol. 438 Col. 334 (26 October 2005)). 127 A prosecution for the encouragement of terrorism offence may only be brought with the consent of the Director of Public Prosecutions (Terrorism Act 2006, s 19). This was described by the Government as an important safeguard against the possibility of the offence being used inappropriately (Joint Committee on Human Rights, Government Response to the Committee's Third Report of this Session: Counter-Terrorism Policy and Human Rights: Terrorism Bill and related matters (10 th Report of Session 2005-06, HL Paper 114, 2006)). In R v Gul [2013] UKSC 64 the Supreme Court criticised this practice of combining overly-broad offence definitions with reliance on prosecutorial discretion, commenting that it amounts to Parliament abdicating its legislative responsibility to an unelected official. freedom of expression, it is worth noting, by way of conclusion, that our reformulated version of the offence would be deemed to comply with the requirements of the Human Rights Act 1998 and the ECHR. 128 Perhaps more tellingly, given the relatively low threshold imposed by the domestic courts and the European Court of Human Rights, it would also comply with the higher standards set by the U.S. First Amendment. 129 Our proposal may therefore serve as a blueprint, not just for the UK legislature but also for those in other states seeking to fulfil their obligations under the CECPT whilst maintaining respect for human rights. 128 As noted in the introduction, the jurisprudence to date suggests that even the current version of the offence complies with the Human Rights Act 1998 and ECHR. See supra n. 2, n. 3, n. 4 and accompanying text. 129 Arguably, requiring proof that it was the defendant's intention to encourage terrorism would, on its own, be sufficient to satisfy the First Amendment (see Elfbrandt v Russell 384 U.S. 11 (1966) and Noto v United States 367 U.S. 290 (1961)). Should Brandenburg v Ohio 395 U.S. 444 (1969) apply, the position is less straightforward. In this case the U.S. Supreme Court drew a distinction between Ôadvocacy of the use of force' and advocacy that is Ôdirected to inciting or producing imminent lawless action and is likely to incite or produce such action' (447). In terms of advocacy that is directed to inciting or producing imminent lawless action, our reformulation of the offence requires proof that it was the defendant's intention to encourage acts of terrorism. In terms of advocacy that is likely to incite or produce such action, for the reasons already explained we do not believe this should be interpreted as requiring proof of perlocutionary effect. It should instead be understood as requiring that the advocacy was performative, in which case it is satisfied by the requirement in our revised version of the section 1 offence that the published statement created a danger that others would be encouraged to commit or prepare acts of terrorism. In any event, in terms of the Brandenburg distinction it seems hard to us to regard statements that encourage acts of terrorism and that are published publicly with the intention of encouraging others to commit or prepare acts of terrorism as mere Ôabstract advocacy' that lacks a close connection to any likely or intended act (see Barendt supra n. 6, 458).
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creativecommons.org/licenses/by/4.0/.

Funding
Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.