Unmasking Malicious Stance Indicators and Attitudinal Priming: An ‘Evaluative Textbite’ Approach to Identity Attacks in Violent Extremist Discourse

Abstract

The article explores the patterning and functioning of attitude semantics in the practice of identity attacks within terrorist communications. Positioned in facework and stance-taking research (e.g. Tracy & Tracy, 1998, 2008, 2017), it introduces the concepts of ‘evaluative textbites’ and ‘attitudinal priming’ to linguistic examinations, advocating a functional approach to unravelling identity attacks, drawing on corpus analysis methods (e.g. word frequency, and concordance-line qualitative analysis) and the Appraisal framework (Martin & White, 2005). Findings reveal that, linked with stance-taking activity, attitudinal priming offers insights into how specific ideational targets are primed for particular attitudinal, evaluative functions. Evaluative textbites provide linguistic evidence of an author’s encoded hostile attitude and the nuanced patterning and functioning of ‘ideation-attitude’ co-occurrences in these attacks. Identity attacks are a rhetorical tool, normative and valuation-based, targeting individuals’ or out-groups’ immoral behaviours and devaluing victims by reference to their personal traits, power-distance relationships, interactional roles, and master identities. This article offers implications for future study of identity attacks in hate crimes, genocidal rhetoric, and defamation texts, and strengthens counter-extremism efforts by illuminating the investigative value of identity work.

Introduction

In our interconnected yet unstable world, the examination of extremist discourse has emerged as a vital interdisciplinary inquiry that plays a crucial role in countering the menacing communication and ideological challenges confronting the global community. Violent extremists (interchangeably, terrorists) often justify their violent acts by claiming to respond to perceived attacks on their in-group’s identity and interests (e.g. Etaywe, 2022a, b) (see Example 1, underlined). However, a prevalent feature of extremist illicit or objectionable public messages is the targeting of people from specific identities and ideologies (Etaywe, 2023a); verbal identity attacks are commonplace in terrorist inciting and threatening communications (Etaywe & Zappavigna, 2021, 2023). This identity-defence justification and act of identity attacks draw parallels with religiously aggravated hate crimes (e.g. Culpeper et al., 2017), highlighting the significance of identity-attacks in illegal communications and the need for a fuller understanding of the nature of identity attacks in extremist communications.

(1)

‘…here is the disbelieving world today having mobilized, called out, formed coalitions and parties, and gathered all its strength, partners, allies, and awliya to wage war against Islam and its people and to plot against the believers and their religion with all means available…’ (Excerpted from al-Baghdadi1)

Despite identity attacks in terrorist discourse (e.g. ‘dog of the Jews’ from al-Baghdadi’s texts inciting against the King of Jordan), and lexis from religious ‘symbolic capital’ sources (Bourdieu, 1991) (e.g. ‘apostate…governments’ from al-Baghdadi’s texts inciting against Arab rulers), a notable gap persists in understanding how extremists strategically use language for identity attacks. This article addresses this gap, examining the nature of and investigative usefulness of such attacks (Etaywe, 2023a) taking these attacks as stance markers. Threatening stances convey “attitudes towards another conveyed by word and phrase choices,” semiotic resources serving as “stance indicators” (Tracy & Tracy, 1998, p. 233). The paper argues that these words also serve the triune act of stance-taking: alignment, positioning, and evaluation (Du Bois, 2007). In inciting the toppling of Arab rulers and governments in Gulf countries (e.g. United Arab Emirates, Saudi Arabia), al-Baghdadi’s labelling them “apostate” seeks to morally raise doubts about their legitimacy, positioning them as outsiders, and forging value disalignment while aligning with the in-group.

The study considers identity attacks a face-threatening practice resulting from personal malice, spite and ideological positions, and designed to challenge the face wants of others or other groups, endangering inter-group relationships and presentations to achieve in-group goals in contested matters (e.g. resorting to violent actions to protect the in-group and its resources) (e.g. Tracy, 2017). An extremist’s identity attack is a deliberately nasty “face attack” (Tracy & Tracy, 1998, p. 227; Tracy, 2008) for strategic communicative purposes. In these attacks, the paper argues, terrorist communications strategically “drive particular patterns in structure and meaning, and leverage key social values and bonds” to nurture disaffiliation from ‘others’ (e.g. Etaywe, 2022a, p. 279, 2023b; Etaywe & Zappavigna, 2023). Building on Goffman’s (1967) notion of purposeful face threats, this study takes identity attacks in violent extremism as being purposefully offensive, and a form of calculated aggressive facework or identity work. Additionally, we consider identity attacks to be strategic communicative acts that are motivated, inter alia, by (i) the goal of enacting social disaffiliation with ‘others’ and affiliation with in-group members (Etaywe, 2022a), and (ii) by the terrorist-text author’s personal aggressive identity becoming fused with the master identities of the group the individual claims to defend against a threat to the group’s essence—which has been referred to as ‘identity fusion’ (Buhrmester et al., 2018; Ebner et al., 2022). This purposefulness makes identity attacks acts for which individuals are held accountable and linguistic traces useful for predicting extreme pro-in-group orientations and malicious stance indicators.

Brown and Levinson’s (1987) theory of politeness is rooted in everyday interpersonal interactions, particularly those among friends and family. Their theory largely overlooks institutionalised norms and obligations within the framework of justifying verbal aggression (e.g. Tracy, 2017). Originally, their theory aimed “to disarm [aggression] and make possible communication between potentially aggressive parties” (Brown & Levinson, 1987, p. 1). However, this article contends that deliberate aggression and acts of ‘social disaffiliation’ by terrorists (e.g. Etaywe & Zappavigna, 2023; Etaywe, 2024a, b), as articulated by Tracy and Tracy (1998, p. 229), cannot be equated with politeness or its absence. Instead, the article argues for a complementary functional approach to criminal-text analysis, as proposed by May et al. (2021). This approach, grounded in social practice, examines how a text functions within its context or the purpose it serves. This paper focuses on the functions of attitudinal meanings in evaluative expressions within identity attacks, drawing on the Appraisal framework (Martin & White, 2005) that enables capturing institutionalised norms and attitudes as different ways of feelings targeted at different people, behaviours, things and phenomena. This functional approach adds to existing methods into facework and face-threatening practices (or strategic impoliteness) (e.g. Culpeper et al., 2017a, 2017b) where negative appraisal is designed to influence inter-group relations and treatment and allegedly serve the in-group by attacking the public face of out-groups and members thereof. The article introduces an ‘evaluative textbite’ approach and the concept of ‘attitudinal priming’ to explain the nature of identity-attack performance, arguing for how emerging attitudinal semantic patterns serve othering and binary opposition construction and function as hostile stance indicators. In terms of priming, the focus is on (i) the patterning of references to an out-group ‘priming’ negative attitudinal meaning, and (ii) the patterning and functioning of attitudinal meaning-making choices colouring the texts and encoded stances. The study contributes to the literature on threatening stances (e.g. Gales, 2011) and pragmatic functions of terrorist-threatening communications (Meloy & Gill, 2016; Meloy & Huffman, 2014; Shuy, 2021) by focusing on linguistic choices in the performance of identity attacks and how these patterns may indicate hostile and violent stances. To reiterate, this article aims to examine the nature of how extremists strategically use language for identity attacks, as well as the investigative usefulness of identity attacks, taking these attacks as stance markers. Specifically, the article explores the patterning and functioning of attitude semantics in the practice of identity attacks within terrorist communications.

A Functional Approach to Identity Attacks: ‘Evaluative Textbites’ and ‘Attitudinal Priming’

Interrogators analysing threatening communications rely on linguistic features to anticipate violence and assess threats (Smith, 2004, 2006). These features include affect words, emotional expressions (e.g. anger), mentions of weapons, violent language, death themes, swearing, pejorative and racist language, and references to moral outrage, grievances, and victim blaming (Meloy & Gill, 2016; Pennebaker & Chung, 2011; Etaywe et al., forthcoming). This study contributes to this line of research into threatening and inciting communications by systematically examining attitudinally loaded expressions in identity attacks. The semantics and collocations of these expressions, as well as the frequency thereof, provide evidence of encoded violent values and construed hostile stances, contributing to describing the terrorist-text author’s overall ‘attitudinal profile’ (White, 2011).

To facilitate counting stance indicators and analysing semantic patterning in attitudinal expressions, combining corpus analytic tools with discourse analytic approaches proves useful for examining forensic texts, as in this study. This combination provides strong semantic evidence and a functional perspective on threatening stances and commitment to violence, focusing on lexicogrammatical choices and their repeated functions (Gales, 2010). Understanding this interaction often requires a close discourse analysis within the functional perspective of the Appraisal framework (Martin & White, 2005), as demonstrated in this study of identity attacks and the attitudes they reveal. The approach to attacked aspects of identity factors in dimensions such as speakers’ or writers’ stances towards phenomena.

To systematically describe attacked aspects of identity, this study adopts Tracy and Robles’s (2002) taxonomy. This taxonomy includes master identities (sex, ethnicity, and religion), interactional identities (roles in communicative contexts like mentor, manager, and leader), and personal identities (individuals’ personality traits, stances, and interpersonal relationships) (Tracy & Robles, 2002, p. 23). According to this taxonomy, personal identities are intertwined with master and interactional identities, influenced by cultural beliefs and in-group expectations (Tracy & Robles, 2002, p. 23). Personal identity encompasses traits like timidity, seriousness, and aggressiveness, as well as individuals’ relationships (e.g. superior, subordinate, distant, close) and their attitudes towards people, events, and societal matters. In the context of relationships, personal identity is also referred to as “relational” identity (Tracy & Robles, 2002, pp. 22–23). For instance, individuals in an employee-supervisor relationship navigate unequal dynamics at times (Tracy & Robles, 2002). This article endorses Tracy and Robles’s assertion that personal relationship-linked identities are inseparable from master and interactional identities, allowing for explaining identity fusion as a psycholinguistic phenomenon. This study’s use of Tracy and Robles’s (2002) framework emphasises analysing repeated, attitudinally loaded expressions to uncover the aspect of identity these expressions indicate as personal, relational, interactional, or master identities within the ‘evaluative textbites’ (Etaywe, 2022a, 2023b). For example, a repeatedly used lexical item in Shekau’s texts is ‘infidel’. Examining an evaluative textbite of ‘infidel’ (i.e. a concordance line that contains ‘infidel’ + who is evaluated as being ‘infidel’, as revisited in the “Macro Description of the Authors’ Attitudinal Profile: Patterning in Choice of Attitudinal Meaning Types and Features” section, Fig. 2), we have the following: “Whoever follows the Qadiryya is an infidel! Shiites are infidels! Whoever follows democracy is an infidel!”. By verbally attacking the ideational targets as infidels, Shekau portrays others (who are victimising Sunni Muslims adherent to Boko Hamam) as having differential master religious identities. He simultaneously portrays himself as relationally distant from infidels and as an insider/a Muslim supporter of the victimised, who is assuming an interactional role or identity inciting violence against the targeted out-groups, as in: “…take your weapons and slaughter the infidels.” More elaboration about the concept of ‘evaluative textbites’ is provided next.

Evaluative Textbites for a Forensic Examination: Towards a ‘Functional’ Linguistic Fingerprint

Previous linguistic research for forensic purposes utilised collocational patterns, or ‘n-gram textbites,’ to describe distinctive patterns in an author’s writing (Johnson & Wright, 2014). This research focuses on repeated structures (i.e. syntagmatic features), where word n-grams are taken as linguistic “evidence of distinctive text-encoding by authors” (Johnson & Wright, 2014, p. 39). The present article extends beyond syntagmatic features, proposing that ‘evaluative textbites’ (Etaywe, 2022a, p. 17, 2023b, p. 15) describe an extremist author’s writing by operating at both syntagmatic and paradigmatic features, specifically the functions of repeated, attitudinally loaded lexical choices. While the ‘n-gram textbites’ approach emphasises the distinctiveness of features in terms of structure frequency, the ‘evaluative textbites’ approach focuses on distinctiveness in terms of both frequency and function. Thus, distinctive ‘evaluative textbites’ can provide evidence of encoded attitude and extremist threatening stance and personae.

Given the importance of quantification and qualitative analysis of texts in forensic linguistics research (Coulthard et al., 2017), the ‘evaluative textbites’ approach achieves both analytical dimensions. Firstly, it focuses on “[t]he [repeated] lexical selection [which] forms the basis of [identifying]…the salient encoded attitude” and coupling disposition (i.e. coupling of the attitudinal lexical item, plus who/what is being evaluated) (Etaywe & Zappavigna, 2021, p. 4). This focus on repeated, attitudinally loaded lemmas aligns with Kniffka’s (2007) argument that forensic analysis must identify not only lexicogrammatical features in offenders’ repertoires but also how offenders use these features in combination and how they interact in forensic text. Secondly, the approach offers “a way of reading between and beyond [lexicogrammatical] formal properties” (Canning, 2014, p. 63). Through the close study of ‘evaluative textbites,’ particularly collocating ideational targets with negative attitudinal meanings appearing in proximity (e.g. concordance lines), this article argues that empirical evidence of an author’s “functional linguistic fingerprint” can be obtained (Etaywe & Zappavigna, 2023, p. 201; Etaywe, 2022a). That is, the way attitudinal, evaluative meanings are construed to enact identity attacks are realisable through attitudinally charged lexical choices that encode an author’s hostile identity. There are various useful linguistic approaches to examine identity in political (violent) discourse (e.g. Chilton, 2004; Hurt & Krzyzanowski, 2010; Wodak et al., 2009). The approach to identity in this study factors in dimensions such as the author’s stances towards people, entities, opinions, behaviours, happenings, or states of affairs being construed by the text (White, 2011). This ‘functional fingerprint’ sheds light on the overall tendency of ideation-attitude ‘coupling disposition’ (Etaywe & Zappavigna, 2021) in these textbites, revealing semantic evidence of how these textual segments function in the context of terrorist inciting or threatening violence, that is, what job they are doing in context. Focus on function moves us beyond the concept of ‘linguistic fingerprint’: “the linguistic ‘impressions’ created by a given speaker/writer [which] should be usable […] to identify them” (Coulthard, 2004, p. 432) to focus not only on language patterning but also on language functioning. This approach to enacted identity and identity attacks accounts for dimensions of terrorist personae, such as (in)tolerance and aggression (e.g. Etaywe & Zappavigna, 2021).

Repetition Priming, and Attitudinal Semantic Priming for Threatening Stance Indication

The ‘evaluative textbite’ approach, focusing on quantification of lexical items and qualitative semantic analysis of their attitudinal load and co-occurrences (evaluative couplings) (Knight, 2010), allows exploration of the binary opposition structure of negatively evaluated groups. Given the in-group versus out-group concept’s primacy in analysing violent extremism communications and related practices like dehumanisation of out-groups (Comerford et al., 2021), this article adopts the theoretical orientation that ‘We’ versus ‘They’ ideational meanings facilitate the establishment of "category-bound features" (Antaki & Widdicombe, 1998, 3–4) and "positioning [others unfavourably] by acting as loci of moral valuations" (Tann, 2010, p. 175). In other words, broader categories of ideational meanings facilitate what this article terms aggressive ‘attitudinal priming,’ where every ideational category is primed for co-occurrence with specific evaluative functions. The attitude-ideational target co-occurrence can be explained by assuming that every ideational category is primed for co-occurrence with other pragmatic/semantic functions. Specifically, out-group ideational targets are primed for negative evaluative functions, while in-group ideational targets are primed for positive evaluative functions.

The article analogises with Johnson and Wright (2014) and Hoey (2005, pp. 8–15), arguing that collocation, like in ‘n-gram textbites’ (Johnson & Wright, 2014), is accounted for by assuming that every word is primed for co-occurrence with other words or pragmatic/semantic functions. Attitudinal priming is likened to lexical priming, a subjective process associated with stance-taking activity, activated based on the writer’s linguistic, sociocultural, and ideological background. Similar to Zappavigna et al., (2008, p. 171) exploration of patterned couplings, this article contends that evaluative textbites enable analysts to simultaneously identify and describe patterns in (i) paradigmatic options across interpersonal and ideational meanings and (ii) syntagmatic options, allowing the study of two psycholinguistic processes at the same time—“repetition priming” and “semantic priming” (Hoey, 2013, p. 3342).

Akin to Hoye’s lexical priming, attitudinal priming accounts for co-occurrences of evaluative expressions and evaluated objects within the terrorism context, drawing upon the writer’s ideological schemas and linguistic repertoire stored in the language user’s cognition or mind. Capturing repeated, attitudinally loaded lexical items (e.g. ‘infidel,’ ‘dog,’ ‘enemy’) and references to ideational targets or primary participants in their linguistic context can be considered a form of repetition priming at the syntagmatic axis, serving the purposes of attitudinal profiling. Furthermore, attitudinal priming works when expressions of specific polarity and type (e.g. negative judgment, negative appreciation) are associated at the paradigmatic axis of language with particular categories of ideational targets in the writer’s mind—semantic priming par excellence. This study focuses on this semantic, evaluative association to explain the nature of identity attacks within the context of terrorism, representing a pragmatic association.

The Appraisal Framework as a Theoretical Foundation

The theoretical foundation of this study primarily lies within the Appraisal framework (Martin & White, 2005), a framework rooted in the social semiotic paradigm of Systemic Functional Linguistics. The framework provides us with a discourse semantic approach to evaluative language and a systematic way to describe the evaluative choices and their associated attitudinal meanings as used in the identity attacks. Concerned with the attitude system of meaning, the study focuses on values for expressing positive or negative stances and attitudes towards phenomena. Within the attitude system, three interdependent systems (i.e. choices) operate simultaneously: attitude types, explicitness (whether it is inscribed or invoked, i.e. directly explicitly states or implied), and polarity (whether the attitude is positive or negative) (see Fig. 1).

Fig. 1

The attitude system used to analyse the attitude semantics (adapted from Martin & White, 2005)

The attitude type system comprises three semantic regions or sub-systems:

• affect: this sub-system encompasses resources used to express emotional reactions and states, which may be subcategorised as (un)happiness, (in)security, (dis)inclination and (dis)satisfaction.

• judgement: this sub-system involves resources for construing assessments of behaviour with respect to values of social esteem (i.e. normality, capacity and tenacity) and values of social sanction (i.e. veracity and propriety) in terms of the rules and regulations set by society.

• appreciation: this sub-system provides resources for estimating the value of things, events, processes and products in terms of aesthetic dimensions, and includes reaction, composition and valuation parameters.

This study is concerned with the function of repeatedly used attitudinal lexical items and their use in decoding the most frequently salient type of attitude type of attack in the dataset (described in the next Section). The ‘attitude’ system permits us to approach the linguistic resources offered in texts as part of identity attacks as systematic constructions of interpersonal meanings.

Method

Data

This study analysed eleven online public statements made by two jihadists affiliated with distinct terrorist organisations: former Boko Haram leader Abubaker Shekau (hereafter Shekau) and former ISIS leader Abubaker al-Baghdadi (hereafter al-Baghdadi). The choice’s aim is to give a representation of threatening and inciting texts by adherents to violent extremist ideology (Salafist jihadism), considered among the most lethal worldwide (Global Terrorism Index, 2019, 2020). Table 1 provides an overview of the dataset. The final corpus size was determined after a data cleaning process, normalising the frequency to 10,000 words per author.

Table 1 Overview of the dataset

The selection of authors aims to offer a representation of transnational terrorist groups operating in diverse geopolitical contexts, shedding light on any commonalities and differences in the style of identity attacks across different text types used in the terrorism context. Al-Baghdadi incites Sunni Muslims in the Arabian Peninsula, Syria, Iraq, and Jordan to violence against a range of targets, primarily pro-U.S. regimes and states, Arab regimes such as the Saudi monarchy, UAE rulers, King of Jordan, apostates, polytheists, al-Asad regime of Syria, and pro-Iranian militias (Table 1). Shekau’s discourse generally recruits violent actors and threatens the stability of Nigeria’s federal government and neighbouring countries. Shekau threat texts in this study are delivered against various victims, including General Rogers of the Nigerian Army, vigilant youth troops fighting against Boko Haram, Nigerians not following Boko Haram’s creed (in Shekau1); President Buhari (in Shekau2); Nigerian soldiers, defence minister, General Buratai, Civilian Joint Task Force, and vigilant troops (in Shekau3); President Buhari, the Nigerian government, and people of Chibok, etc. (in Shekau4); Emir of Kaun (in Shekau 5); Nigerians and Nigerian government (in Shekau 6); Nigerian security and military forces (in Shekau 7); President Goodluck and the president of the Christian Association of Nigeria (in Shekau8); The people of Kanu (in Shekau9).

The provided texts are public statements. Shekau’s texts, produced between 2012 and 2018, were publicly released on YouTube, and then in legitimate English translations by Nigerian media outlets. Al-Baghdadi’s texts (2016-2018) were publicly online available in English through ISIS’s al-Hayat Media Centre and Rumiya magazine, addressing ISIS members and the discontented youth in Iraq, Syria, Jordan, and Arabian Peninsula and inciting violence against those non-adherents to ISIS creed.

Procedure of Analysis: An Evaluative-Textbite Approach

The texts were analysed by focusing on evaluative textbites patterns, examining the repeated, attitudinally loaded lexical items and how specific attitudinal meanings co-occur with out-group’s ideational targets in concordance lines. Akin to Etaywe and Zappavigna (2021), AntConc, a corpus analysis tool (Anthony, 2019), was employed to generate a word frequency list and to qualitatively explore evaluative meanings in concordance lines. A list of repeated, attitudinally loaded lexical items was generated and their concordance lines were examined to disambiguate those items that were not used attitudinally, the evaluative load and polarity of the words and their specific meanings. This procedure enables a systematic depiction of how interpersonal meanings are “strung throughout the clause [or concordance line] as a continuous motif or colouring” and are, in essence, “distributed like prosody throughout the continuous stretch of discourse” (Halliday, 1979, p. 66). The attitudinal lexical items of the same stem, for example, destr* (destroy, destroying, destruction), were grouped together under one lexical entry (lemma). A list of attitudinal lemmas of each author was manually sorted and tabulated. Then, the examination involved:

• Categorising attitudinal lexical items: understanding how these words colour the attitude towards the victims’ identity aspects. This includes classifying attitudinal lexical items into judgement-, appreciation-, or affect-based lexis, and identifying sub-categories, explicitness (inscribed vs. invoked), and polarity (positive vs. negative) (as per Martin & White, 2005). The codes of major types, subtypes, polarity and explicitness described previously in "The Appraisal Framework as a Theoretical Foundation" section are used for manual labelling of the identified attitudinal features. In evaluative textbites that contain attitudinally charged terms (e.g. ‘apostate’ and ‘atheist’ from the religious register, which are frequent in jihadist discourse) and ideas (e.g. ‘tyranny’ and ‘freedom) which convey implicit evaluative language known as affording attitude (Doran, 2019), the following coding is used:

  • o The terms are coded as instances of judgement when describing people and their actions.

  • o The ideas are coded as instances of appreciation when describing things, such as a regime or entity.

• Identifying aspects of targeted identity: determining which aspects of identity were targeted, namely personal (e.g. rats), relational (e.g. fearful), interactional (e.g. tyrant), or master (e.g. atheist) (following Tracy & Robles, 2002).

• Referencing ideational targets: Assessing whether the general ideational targets are referenced by ‘functionalisation’ (i.e. in terms of an activity social actors do, e.g. government, immigrants) or ‘identification’ (in terms of what social actors are, e.g. White, men, my brothers) (following van Leeuwen, 2008).

The resulting patterns of attitudinal meanings were used to describe the ‘attitudinal profile’ (White, 2011) of the authors, and taken as evidence on how linguistics can get at the performative nature of identity attacks by studying repeated discursive patterns which is what corpus analytical methods support (e.g. Bednarek, 2010).

For example, in:

(2)	‘Rise, O lions of war in Hijaz, Najd, and Bahrain, and make preparations. Prepare detachments and mobilize battalions, and seek Allah’s help in order to root out these treacherous murtadeen [these (rulers in Arabian peninsula) / - judgement].’ (al-Baghdadi 2)

the ideational meaning ‘these’ is coupled with negative judgement, annotated between brackets as follows: [these (rulers in Arabian Peninsula)/-judgement]. In this textbite, ‘treacherous’ and ‘murtadeen’ (i.e. apostate) are attitudinal lexis characterised by the function ‘-judgement’; that is, they explicitly (hence, ‘inscribed’) negatively judge the Arab rulers in the Arabian Peninsula for being considered disloyal to Muslims (hence non-Muslims or ‘apostate’) and for their collaboration with ‘enemies’ of Islam the U.S.-led coalition against ISIS and Sunni Muslims in Syria and Iraq (hence ‘treacherous’). This text employs ideational targets that can be categorised in terms of ‘functionalisation’, that is, in terms of what rulers do. This identity attack is manipulated by being directed at the ‘personal identity’ (‘treacherous’) of rulers, for example, in Saudi Arabia, the United Arab Emirates, and Bahrain, for not doing the right thing. This personal identity attack is simultaneously linked to the rulers’ disloyalty to their master religious identity of Islam, arguably invoking referencing to the rulers by ‘identification’ in terms of who they are (‘apostate’). Repetition of this coupling of rulers (and similarly non-Muslims and the U.S led coalition against ISIS, as a general ideational target) with only negative attitudinal meanings gives rise to semantic evidence of the ‘attitudinal priming’ phenomenon. That is, the broad category of out-group’s ideational targets acts as loci for negative attitude. Additionally, the identity attack, in this case arguably serves the function of inciting terrorist acts to ‘root out’ these rulers to win for Islam and to signify loyalty to the in-group. Given that an evaluative act also has positioning and alignment functions (Du Bois, 2007), this example of identity attack also operates to position the incited-against as a threat to Muslim nations, position al-Baghdadi as a ‘caring’ insider, and simultaneously foster communal disalignment between people and their governments or ruling regimes.

In cases where a textbite could arguably imply a fit into more than one category of the attacked aspect of identity (as in the case of ‘treacherous’ above, which some may argue it could be both ‘personal’ or ‘interactional’ identity given the role of a traitor), the intuitive category is selected following these questions to differentiate the attacked aspect of identity:

• Does the aspect encompass a trait (e.g. loyalty, seriousness, treachery)?

• Does the aspect encompass relationships of distance or power?

• Does it encompass a role (e.g. spy)?

A ‘yes’ answer to the first question indicates a personal identity, the second a relational identity, and the third an interactional one. That said, adopting Tracy and Robles’s (2002) argument that personal and interactional identities are often informed by or bound up by master identities, the paper argues that an attack on one identity aspect such as personal identity (e.g. treacherous) in the context of terrorism simultaneously activates or implies that this identity aspect is intertwined with master identities. It also argues that this intertwining is influenced by the text author’s cultural beliefs and in-group expectations regarding attributes and behaviours serving the in-group; that is, the authors’ meta-values of loyalty and care for perceived harm caused to the in-group (e.g. Etaywe, 2024b; Etaywe & Zappavigna, 2023).

To add an additional lens for the discussion of how attitudinal lexical items map onto broader sociocultural values, the lexical choices with the salient attitude type (which is here judgement) are discussed in relation to the sociocultural values underpinning these choices in terrorist discourse—using the following social value categories proposed by Smith (2004): dominance and aggression; morality and truthfulness; justice and tolerance; and culture and religion. The identity attacks were discussed in relation to their interpersonal function within the context of acts of threatening and acts of inciting hostility and violence, particularly in terms of their rhetorical functions of ‘identity work’ (e.g. Coupland, 2001; Tracy et al., 2006), taking forms of identities that extremists ‘work on’ in discourse through contesting and/or affirming their membership of social categories (e.g. religion, race, nation, organisation) as a form of ‘legitimacy work’ (e.g. Brown & Toyoki, 2013). That is, identity attack patterns, as a form of identity work, serve to construct an extremist group’s membership and actions as legitimate by affirming and/or contesting inter-group identities, hence influencing inter-group relationships, specifically via evil-othering and legitimating violence against immoral others (e.g. Graham et al., 2004).

Results and Discussion

This section analyses and interprets the overarching semantic patterns in the selection of attitudinally loaded lexis used in identity attacks, revealing the authors’ macro attitudinal profile. The profile is constructed based on discernible patterns of attitudinal meanings realised through repeated lexis (“Macro Description of the Authors’ Attitudinal Profile: Patterning in Choice of Attitudinal Meaning Types and Features” section). Transitioning from the broader picture, the section then illuminates the scrutiny of identity attacks within evaluative textbites containing these repeated, attitudinally loaded lexis. It explores insights gained in terms of two primary rhetorical functions achieved through (i) the utilisation of affect and appreciation-based identity attacks (the least utilised attitudinal meanings in the dataset) (“AFFECT-Based and APPRECIATION-Based Identity Attacks: Justifying Violence Against Sources of Insecurity and Less-Human Others” section), and (ii) predominantly through negative judgement-based lexis, explaining the prevalence of this attitudinal meaning (i.e. judgement) in the execution of identity attacks (“JUDGEMENT-Based Identity Attacks: Out-Groups’ Behaviour-Based, Normative Licencing of Violence” section). The discernible patterns in attitudinal meanings serve two rhetorical functions, as discussed in subsequent subsections. These findings accord with Graham et al. (2004) observations on the rhetorical functions of identity-related othering resources employed by politicians and extreme religious leaders:

• Construction of evil out-groups: Identity attacks are employed to construct out-groups as a source of insecurity and by depreciating and passing negative judgments on their identities. This negative portrayal positions these groups as the malevolent ‘Other,’ establishing and/or fostering a dichotomy and disalignment in values within society.

• Appeal to and alignment with in-group audiences while face-attacking and disaligning with threatening others: Identity attacks also serve to appeal to in-group audiences, fostering affiliation under the banner of shared identities. Simultaneously, they inflict harm on perceived threatening others.

The analysis shows that identified repetitions in meaning and co-occurrences of ideational categories with particular attitudinal meanings play a crucial role in ‘work on’ contested identities and inter-group relationships, situating extremist texts within their social contexts and divisive functions, positioning the threatened and/or the incited-against as evil, source of insecurity and wrongdoers, and ultimately serving to align the authors with the wider in-group’s members.

Macro Description of the Authors’ Attitudinal Profile: Patterning in Choice of Attitudinal Meaning Types and Features

The analysis has revealed four aspects of identity as targets of negative evaluations—personal, relational, interactional and master aspects of identity (unpacked and exemplified in the subsequent subsections). The evaluative meanings used in these identity attacks (chiefly, personal identity attacks) also provide clues to the authors’ identity as aggressive and antagonistic towards their victims, while the authors’ ‘work on’ relational-identity attacks operates to construe the victims’ identity as distant from/or inferior to the authors’. In other words, the patterning of encoded attitudes, ‘aggressive’ and ‘distant’, are the two salient features that colour the texts and characterise the personae constructed in discourse. The personal and relational identity attacks tend to appear together with the interactional and master-identity attacks. The latter, in identity work terms, serves to affirm the authors and victims as belonging to distinct traditional social categories (e.g. religion, nationality), facilitating cultural/ideological positioning based on contested traditional social structures and hard boundaries of membership the extremists claim to defend. This finding accords with Pape’s (2005, p. 87) argument that “[t]he harder the boundary between groups—the more exclusive are membership rules—the more extreme is the “us” versus “them” dichotomy”. In addition, while the two authors demonstrate variation in the extent to which they repeat attitudinal lemmas, they reveal commonalities in the salience (of counts) of attitude types. Al-Baghdadi employs a higher number of repeated, attitudinally charged lemmas compared to Shekau, utilising 171 lemmas repeatedly (ranging from 2 to 46 times), totalling 1012 occurrences, which constitutes approximately 10% of his texts. Similar lemmas constitute over 7% of Shekau’s texts (see Table 2). These repetitions colour texts with strategic verbal aggression choices motivated by the functions of the texts as explored in the subsequent subsections.

Table 2 Overview of the repeated, attitudinally-loaded lemmas and occurrences

The concordance line analysis of all of these repeated occurrences of attitudinal lemmas (as key words in context, KWIC) has covered all these occurrences. For example, examining the repeated use of the lemma ‘infidel*’ in Shekau’s texts reveals an investigative value in terms of providing information about those considered infidel, thus othered and verbally and/or (would-be) physically targeted (as in ‘…take your weapons and slaughter the infidels’—as in Fig. 2). These others who are verbally attacked include, as in Fig. 2, President Jonathan, Ndume, soldiers, members of the Tijani sect, the Izala sect, the Qadiriyya sect, Shiites, and whoever follows democracy), contributing to tools for identifying potential targets of violence and effective risk management (e.g. Etaywe, 2023c).

Fig. 2

Example of concordance lines of the lemma infidel* in Shekau’s texts

Exploring the attitude semantic features has revealed a profile snapshot. Below is a summary of the attitudinal profile of both authors, in terms of repetition of choices of attitude type categories (see Table 3), polarity and explicitness (see Table 4) (Note: most used feature indicated in bold). The analysis of the repeated attitudinal meanings has revealed a pattern of attitudinal features that are characteristic of the extremist evaluative style—in terms of ‘Our’ ideational categories primed for positive attitudinal meanings versus ‘Their’ ideational categories are primed for negative attitudinal meanings—particularly judgement expressions. This attitudinal priming pattern serves, inter alia, to constantly emphasise contested inter-group identities, affirm the extremist-in-group relationship, and ultimately legitimise ‘our’ violence against ‘them’. The heightened use of judgement-based expressions in terrorist threatening communications accords with Gales’ (2010) results of analysis of threatening communications in workplaces as well as Hurt and Grant’s (2019) findings on violent fantasies that judgement is a salient attitude type in threatening texts and can serve as evidence of pledging to harm. Additionally, the increased judgement versus decreased affect provides clues to texts with a less affective mode of violence (e.g. Meloy, 2001, p. 1212) and more institutionalised or ideology-based antipathy towards the out-group's social actors and their behaviours.

Table 3 Summary of attitudinal features encoded in repeated, attitudinal lexical items

Table 4 Summary of attitudinal features encoded in repeated, attitudinal lexical items

Analysis of repeated attitudinal lemmas shows that both authors are very interpersonally involved in their communication. The general tendency of repeated attitudinal lexicons is of negative polarity (exemplified in the lexical choices in Table 5 above which can give an impression of the content of the analysed texts and their violent nature), inscribed attitude, and judgement type of attitudinal meaning. Judgement type is, thus, identified as the most salient type (which hence receives more attention, “JUDGEMENT-Based Identity Attacks: Out-Groups’ Behaviour-Based, Normative Licencing of Violence” section), implying that identity attacks are behaviour and ideologically oriented and are centred around the moral distance between ‘Us’ and ‘Them’. By contrast, the category of affect type is identified as the least repeatedly used in the dataset, implying emotional disengagement and more ideological determination (as unpacked in the subsequent subsections). Noticeably, both authors favour judgement with Shekau using Appreciation the most. Given the salience of the choice of judgement type and negative polarity, this warrants further investigation as a salient encoded attitude used in identity attacks (see the “JUDGEMENT-Based Identity Attacks: Out-Groups’ Behaviour-Based, Normative Licencing of Violence” section).

Table 5 Examples of repeated, attitudinally-loaded lexical items primed for out-groups’ ideational targets (in Shekau’s threat texts and al-Baghdadi’s incitement texts)

Noteworthy, although identity attacks typically use negatively inscribed (i.e. explicit) words as in Table 5 above—hence the focus on the emerging patterning on negative attitudinal priming in the subsequent subsections—the use of positive attitudinal meanings instances tended to co-occur with the in-group ideational targets. Even when an extremist like Shekau positively evaluates, for example, the violence enacted by members of the in-group as an act of defence (Example 3), it implicates a negative evaluation of others as transgressors. Aligned with Chandler’s (2017) argument that a mention of ‘good’ evokes ‘bad’ and ‘black’ evokes ‘white’ in the context of the construction of binary opposition of the world, the positive values co-occurring with ‘Our’ ideational targets also operate (as a form of identity work) to constantly imply that others are ‘bad’. This assumed persistent axiological binary opposition (whether implicit or explicit) reflects in-group expectations of loyalty and opposition to out-groups. This binary opposition also functions to construct a textual “code for” ideological positions (van Dijk, 1998, p. 207), positioning the self and victims as morally, affectually, and valuation-wise distant, as discussed in the following subsections.

(3)	We decided to defend ourselves.

affect-Based and appreciation-Based Identity Attacks: Justifying Violence Against Sources of Insecurity and Less-Human Others

The authors employ lemmas encoding affect- and appreciation-based identity attacks, as a form of identity work, to justify aggression. affect lexis portrays out-groups as in-group threats, while appreciation lexis dehumanises adversaries. Al-Baghdadi uses in/valuation lemmas to depict out-groups (‘the enemy of Allah’) as unworthy, contrasting them with the in-group’s activities in Syria as ‘blessed.’ Adversaries are linked with terms like ‘falsehood,’ Arab rulers (e.g. King of Jordan) allied with ‘the Romans’ and ‘the Jews’ are labelled ‘dog’ or ‘puppet’ (see Example 4), and the Rafida (Iran and its followers) are associated with ‘filthy,’ dehumanising ‘others’ to morally disengage from them, as per Bandura’s (2016) terms. appreciation serves as a rhetorical tool, justifying violence, portraying adversaries as less than human, hence belonging to distinct identity categories, and inciting the audience to harm others while reconciling with it. Additionally, from Goffman’s (1955, p. 216, 1967) lens on aggressive facework and how it compares and or contrasts with “the traffic rules of social interaction”, these negative appreciation-based acts of identity attacks provide evidence of authors going through social red lights. Example 4 attacks identity aspects related to traits (personal identity), power dynamics (relational identity), contested roles (interactional identity), and religious (master) identity within a political or ideological struggle. The negative appreciation of the ‘King of Jordan’ involves referencing the King’s role portrayed as being manipulated (functionalisation), and he is associated with a particular religious out-group (Christians and Jews) (identification) using the terms ‘puppet’ and ‘dog,’ respectively. This referencing reflects that identity attacks can involve ‘work on’ framing the target in terms of their actions and associating them with certain identities or groups to legitimise disloyalty or violence against them.

(4)	O ahl al-[S]unna in Jordan […] Where are the men of power? so that they can make the puppet of the Romans and dog of the Jews understand who the sons of Islam are, and who the mujahideen are...

In addition to appreciation, al-Baghdadi uses more repeated affectual lemmas than Shekau, predominantly falling into the in/happiness category (e.g. ‘sorrow’, ‘distress’). This is followed by lemmas encoding the in/security type (e.g. ‘horrors’) and the inclination type (e.g. ‘desire’). Such instances serve to present opponents as a threat, a source of negative emotions, and with an ‘ambition’ against the in-group’s physical and ideological territories. ‘Distress’ and ‘condole’ couple with the in-group’s ideational targets to present them as influenced by others’ illegitimate ambitions. Through this in-group versus out-group identity construction, and in identity work terms as a form of violent relationship legitimacy work, the in-group’s addressees are encouraged to prove themselves as ‘jealous’ (i.e. angry with righteous indignation) or caring for their religion by fighting to defend it. Elaborating on identity attacks, Example 5 references religious identity by mentioning ‘Crusader Europe,’ arguably constructed as a trait of Europe as well. In opposition to the crusader others, the example also references the ‘khilafa’ in the ‘cradle of the Khilafa and fortress of Islam,’ highlighting a connection to Islamic identity. The author portrays the role of ‘Crusader Europe’ as having an ambition to attack, suggesting a specific role in the context of conflict and potential aggression. The term ‘Crusader’ indicates a negative trait assigned to ‘Europe,’ power dynamics, religious identity within the context of a perceived threat to the cradle of the Khilafa and fortress of Islam in Iraq and Sham, and the historical role of Crusade armies in aggression. In the Muslim world, the term ‘Crusader’ is associated with cultural, physical, and economic invasion aiming to subjugate Muslims and desecrate Islam (Ray, 2017). Regarding referencing the ideational target, the example primarily involves ‘functionalisation’ by portraying the actions and ambitions of ‘Crusader Europe.’ Simultaneously, identification is present through the association of ‘Crusader Europe’ with a particular religious and geographical group, combining functional and identity-based references to convey a negative perception of the actions and intentions of ‘Crusader Europe’ towards the Muslim regions.

(5)	Crusader Europe has not ceased in their ambition to attack the cradle of the Khilafa and fortress of Islam in Iraq and Sham…

As regards Shekau, he primarily uses the in/security subcategory (e.g. fear) to describe social groups’ security circumstances. While coupling positive affect with ‘Allah’ and Boko Haram’s supporters, he constructs himself as a person with no ‘mercy’ and no ‘love’ towards his victims, deeming them not love-worthy (see Example 6). This binary construction of actors on an affectual basis presents Shekau as someone with more antipathy, emotional distance, and ideological determination to cause harm to out-group members because they do not deserve ‘love.’ The author suggests a specific trait related to love, implying that there is a person whom God has not indicated should be loved. The focus here is on the falsely claimed divine command regarding love, which may influence relationships of distance and disalignment with threatened Christians. While the sentence does not reference a religious term of evaluative load, it references God, suggesting a connection to religious identity and moral guidance in relationships with others.

(6)	God has not said we should love such a person.

Besides affect lexis, Shekau repeatedly uses appreciation-loaded lemmas in the form of insults or swear words, organised around certain lexical fields such as animals. He couples adversaries with negative valuation by likening them, for example, to ‘sheep,’ ‘spider,’ ‘dogs,’ ‘pigs,’ and ‘rats’ (see Examples 7–8). These identity attacks regarding aspects of personal identity reveal a dehumanising identity work in the form of an ideological reworking of feelings towards others into propositions about the value of others (cf. Martin & White, 2005, p. 45). They exemplify dehumanising groups to provide a warrant for violence, a practice prevalent in genocidal language (Wilson, 2017). This also suggests that some acts of identity attacks may be a “context-spanning” strategy (Tracy & Tracy, 1998, p. 231). This appreciation-based coupling not only indicates “doses of aggressiveness” (Mateo & Yus, 2013, 91) and aggression bursts but also evidences the author’s valuation of others which is a key part of terrorist’s mechanism of moral disengagement and rationalising the killing of others (e.g. Bandura, 2016). From Goffman’s (1955, p. 216) lens on aggressive facework, this verbal violation of “the traffic rules of [communication]” is a red light indicating hatred, licensing of violence, threats colouring a person’s intentional pursuit of a violent course of action, and threats that are a result of interpersonal and inter-group maliciousness and spite as well as the malicious cognitive perception of members of out-groups.

(7)	Any infidel is a sheep to be sold.
(8)	The information I have for you today concerns the attack [on] the […] barracks of the pigs […] dogs […] or the barracks of the rats!

Elaborating on the attacked aspects of identity, as part of ‘work on’ contested personal and master identities to license adversarial relationships, Examples 7 and 8 present attacks on aspects of identity related to negative traits (being ‘sheep’ like in personal value) associated with a master identity, being labelled an ‘infidel,’ a power dynamic suggesting exploitation, and a connection to religious identity with the dehumanising metaphor of being a sheep to be sold. Considering any infidel a ‘sheep’ suggests that targeted/threatened individuals are regarded as commodities to be sold. This evaluative construction also establishes a contested power dynamic indicating a relationship based on control or exploitation. Collocating ‘any infidel,’ a religious identity marker, with a dehumanising term suggests threatened violence connected to religious identity and dehumanisation of those outside a particular belief system. Example 8 uses derogatory terms such as ‘pigs,’ ‘dogs,’ and ‘rats’ to describe a group of people (the Nigerian military), implying negative individual traits (personal identity) and dehumanising language. This language establishes a power dynamic over those viewed as being less human, suggesting a relationship based on hostility and contempt. Regarding referencing the ideational targets, Example 7 above not only involves identification based on religious beliefs but also incorporates functionalisation where the metaphor of ‘sheep to be sold’ refers to the targeted people as a commodity, implying an identity work that takes the form of a reduction of the perceived worth of those labelled as ‘infidels’ to mere objects for transaction. Example 8 above also involves references to ideational targets by both ‘functionalisation’ related to what the military often does (and the action taken against specific military targets) and hostile and contemptuous ‘identification’ through the use of dehumanising metaphors associating these military targets with animals.

judgement-Based Identity Attacks: Out-Groups’ Behaviour-Based, Normative Licencing of Violence

Taking forms of contested identities that extremists ‘work on’ and attack as a form of ‘legitimacy work’ (Brown & Toyoki, 2013), the analysis shows that the heightened pattern of judgement-based identity attacks operates to position out-groups as wrongdoers and immoral actors, legitimising threatened and/or incited violence and contributing to constructing the “aloof, cool verbal style” of the authors, linked to their controlling behaviour (c.f. Weintraub, 2003, p. 145). These findings accord with Etaywe and Zappavigna's (2021) analysis of the discourse of the former al-Qaeda leader, Osama bin Laden. This distances authors from emotional or sympathetic connections, strengthening their overall appearance of ideological determination or pledge to violence (e.g. Gales, 2010). The repeated use of violent judgmental words can be indicative of “predatory violence,” ideologically motivated to gratify the desire for power, territorial dominance/control, and revenge (e.g. Meloy, 2001, p. 1212; Meloy et al., 2004, p. 1088). The repeated judgement lemmas in the dataset also align with four social value-categories proposed by Smith (2004) as characteristic of a terrorist group’s rhetoric.

Table 6 illustrates the value-based categorisation of repeated judgement lemmas, indicating a shared tendency among the two authors in basing their choice of judgement-lexicons along distinct value parameters. This value-based lexical selection serves as an indicator of rival identities/values and involvement in the activity of fuelling ideologically opposing poles in discourse, signalling authors’ ideological perceptions and positions. This finding accords with Etaywe and Zappavigna's (2021) and Thetela’s (2001) contention that patterns in value-based lexical selection can rhetorically characterise the ideological perception-semantics of conflict and violent extremism. The repeated judgement lemmas also offer support to earlier research findings (e.g. Smith, 2004) that lexical choices in terrorist groups’ statements are centred around socio-cultural values of aggression, morality, justice, culture, and religion. This representation of self and others constructs and uncovers authors’ relational and personal identities. Thus, this value-based representation gives valid grounds for investigators to suspect violent extremist communications where negative judgmental values are assigned to out-groups and positive judgments to members of the in-group. Given that the analysed datasets are predominantly of judgmental attitude, typical of threatening communications and a threatener’s/inciter’s identity, marking a pledge to harm, we may describe the judgement-based identity attacks as normative, as well as central to extremists’ work on identities as a violence-legitimacy work.

Table 6 Social value categories characterising judgement lemmas in the dataset

The dominating inscribed negative judgement of ‘Their’ ATs sets the threatening tone of the datasets as explicit, ethics-based confrontational texts affirming inter-group dis-alignment and contested values and ethicality. The increased use of the impropriety type colours the picture of the texts with a high degree of improper behaviours and violent attitudes/intentions. The patterning of personal and relational identities towards victims is construed in terms of (un)ethicality—in terms of what complies with or defies an author’s view of the social system and social necessity. Negative positioning of out-group participants, as well as construing related identities, is achieved by endowing social actors with semantic roles (e.g. van Leeuwen, 2008) and basing evaluative language and these roles on an author’s value system. The patterning of the representation of ‘Their’ actions (versus ‘Ours’) serves the goal of presenting menacing ‘others’ victimisers (and ‘Us’ as victims). This pattern serves in evil-othering the victimisers, providing moral reasoning for violent responses underpinned by the meta-value of ‘care’ for ‘harm’ imposed on ‘Our’ group. It also presents the author as close with and caring towards the in-group’s victims and morally distant from victimisers. For example, in Examples 9–10, outgroup ideational targets (underlined), as agents of negative processes fight, kill, and invade (in bold), are judged for their impropriety as killers, Islam-fighters, and invaders.

(9)	…the Government that is fighting Islam with the Christian Association of Nigeria who are killing Muslims… (Shekau)
(10)	They [US-led coalition against ISIS] raid your lands under the pretext of waging war against the Islamic State, then they do not depart until they have either killed your men and taken your women and children prisoner or until they have driven them out. (al-Baghdadi)

The emerging patterning of semantic roles indicates the authors’ feelings of victimisation and their hostile intentions to victimise others, a ‘work on’ victimiser-victimised identities in discourse that serves ‘our’ violence-legitimacy work. Others here are attacked for their roles as foreign fighters, killers, and invaders. In Example 9, Shekau attacks the Nigerian government’s identity based on its relationship with Christians and performance against Muslims. The identity attack intersects interactional and master identity; it is also bound up with the master religious identity and loyalty of government officials as well as their interactional identity as governors. The identity attack serves to license the motive for the communicated threat against the Nigerian government. The example suggests a negative trait associated with the Nigerian ‘Government’ portrayed as fighting Islam, and the ‘Christian Association of Nigeria,’ linked to the act of killing Muslims. The implication is that these entities are engaged in actions and relationships perceived as harmful or hostile, involved in roles of fighting against Islam and killing Muslims in the context of conflict. These discursively affirmed roles and agency are associated with discursively contested master identities relationships where the author explicitly mentions religious identities, with Islam and Christianity being central to the conflict described. Regarding reference to the ideational target (‘the Nigerian government’), it contributes to a hostile and accusatory tone. Referencing combines ‘functionalisation’ by ascribing not only what a government often does but also specific actions and alliances it creates and ‘identification’ by broadly categorising the government as actively involved in actions against Islam.

Assigning semantic roles to adversaries, al-Baghdadi in Example 10 also assigns the value of ‘fighting’ to the US-led coalition targeting ISIS and Sunni communities in Syria to generate fear in the Sunni communities and incite jihad against the coalition. This identity attack is bound up with the master U.S. national identity and master religious identity. Example 10 suggests negative traits of killing Muslims associated with the ‘US-led coalition against ISIS,’ portraying them as aggressors who raid lands under the pretext of fighting the Islamic State and engaging in harmful actions, including killing, taking prisoners, and displacing women and children. These acts of killing and violence indicate an inter-group relationship marked by conflict and hostility. This identity attack on the U.S.-led coalition is associated with a religious master identity, where the sentence mentions the ‘Islamic State’ and frames the conflict in the context of a war against Islam. The choices in referencing the ideational target tend to condemn and oppose the U.S.-led coalition’s presence and actions, involving both ‘identification’ by attributing actions to the ‘U.S.’ and ‘functionalisation’ by describing the coalition’s activities, including killing Muslim men, taking their women and children as prisoners, or driving them out of their towns.

A main emerging pattern is also the use of lemmas of positive attitudinal inscription to describe the targeted others as loyal to and helping the religious adversaries. The mentioned text highlights a pattern where positive attitudinal terms are used to describe targeted individuals as loyal to and supportive of religious adversaries. This pattern links identity attacks to master identities, labelling ideational targets through identification (e.g. apostates, Muslims, Christians, Nigerians). Rather than focusing solely on lexical items, analysing concordance lines reveals semantic nuances. This pattern of coupling ideational targets serves to position them as loci of negative moral valuations, influencing interpretations of associated social actions and inducing further negative evaluations. For example, Shekau constructs Nigerians as ‘loyal’ to the national idea instead of religion, threatening and inciting violence against politicians not loyal to Boko Haram (Example 11). Al-Baghdadi portrays support for the in-group through ‘support’ concordance lines (Example 12). Non-ISIS-affiliated Muslims and ‘factions of apostasy’ are depicted as supporting ‘taghut’ (idols) and nationalism, while ISIS is presented as a protector or supporter of the oppressed and ‘the religion’ (Example 13). These constructed versions of people and selves (belonging to distinct social affiliations and identities) offer clues to how extremists discursively ‘work on’ reproducing existing social and power relations to serve extremists’ violence legitimacy work—a key part of identity work as a form of legitimacy work as put by Brown and Toyoki (2013).

(11)	Our second message is to attack anyone who pledge[s] to Nigeria [as] his country, to be faithful, loyal and honest, and serve Nigeria. (Shekau)
(12)	…the factions of apostasy surrendered the territories of ahl al-sunna and became a means of support for the Crusaders and the Nusayriyya in uniting their efforts to wage war against the Khilafa State in Iraq and Syria. (al-Baghdadi)
(13)	O Ahl as-Sunna! After Allah, you have nothing but the Islamic State to protect your religion… (al-Baghdadi)

Elaborating on identity work as a discursive activity (i.e. talk about ‘others’ and ‘us’) that serves to construct ‘our’ legitimacy, Example 11 implies negative traits associated with individuals pledging loyalty and service to Nigeria, suggesting an attack on those who express allegiance to a country with a non-religious constitution, regime, and educational system. This example, while communicating a threat against Nigerians, incites Boko Haram followers by instructing an attack on individuals loyal to Nigeria, indicating a relationship marked by hostility and opposition to those not loyal to Boko Haram's beliefs. It assigns a citizen role to those pledging loyalty, framing them as targets for attack, implying a specific role within the threatening message. Example 12 implies negative traits associated with ‘actions of apostasy,’ suggesting betrayal and surrendering territories held by ahl al-Sunna (people of the Sunni tradition). The example establishes a power dynamic by describing the surrender of territories and alignment with ‘Crusaders’ and ‘Nusayriyya,’ indicating a relationship marked by collaboration against the Khilafa State. This identity attack is religiously informed, referencing ‘ahl al-Sunna’ and the ‘Khilafa State,’ framing the conflict in the context of religious identity and political allegiance. Example 13 positions the Islamic State as the sole protector of the religion of Ahl as-Sunna, establishing a power dynamic and a relationship based on dependency for religious protection.

The analysis reveals a functionalisation-based pattern of referencing ideational targets (rulers), depicting rulers and officials of Muslim countries as morally aligned with outsiders (e.g. America) and morally distant. This is exemplified through the use of the same lemmas for both identification-based categories (e.g. America) and functionalisation-based categories of ideational targets (e.g. Arab rulers). The lines of lemmas such as hypocrite, infidel, oppressor, tyrant, unjust, enslaving, lying, and apostate illustrate this pattern and highlight “legitimacy contesting” (Brown & Toyoki, 2013, p. 876) as a kind of identity work. In Example 14 by al-Baghdadi, the term ‘tyrant’ is employed to criticise the Saudi monarchy, implying negative traits associated with the ‘state of Al-Salul’ and expressing strong disapproval. The sentence establishes a power dynamic by condemning the Saudi monarchy’s actions, indicating a relationship marked by opposition and condemnation, targeting both the political and religious identity of the Saudi monarchy.

(14)	…the tyranny and disbelief of the state of Al-Salul [Saudi monarchy] – may Allah disfigure them – is no longer hidden from anyone, even your children.

This coupling of negative judgement with officials’ and rulers’ actions functions, as per Etaywe and Zappavigna (2021), as a rhetorical mechanism delegitimising and defaming the general ideational targets. Social actors’ personal aspect identities, associated with their interactional identities (as rulers), become targets, with labels such as ‘tyrants’ attacking their personal and interactional values of justice. Terms like ‘infidel’ or ‘disbelief’ constitute an attack on the master religious identity. These impropriety -based “face attacks” implicate “an absence of a desirable identity [and] the presence of a disvalued one” (Tracy, 2008, p. 176), legitimising violence and mobilising followers against the incited-against rulers. Shekau justifies aggression by repeatedly using ‘infidel’ to couple impropriety with non-Muslims globally, Muslims under the Nigerian constitution or accepting democracy and ‘western’ education, and the military (e.g. ‘Giwa barracks’), thus demonising others based on their master, religious identity.

In the discourse of terrorism, particularly evident in al-Baghdadi’s texts, a deliberate pattern emerges wherein master identities, notably religious identity, are systematically attacked, fostering a dichotomy of ‘Us’ versus ‘Them’ (Brookes & McEnery, 2020; Wignell et al., 2017). This approach, rooted in a distinct worldview and (mis)reading of religious register terms, serves as a means to legitimise aggression and establish ideological opposition (Brookes & McEnery, 2020; Wignell et al., 2017). The attitudinally-loaded religious terms deployed in these attacks not only delegitimise and victimise others but also reinforce the positive/negative positioning of social actors based on inclusionary/exclusionary categorisation (van Leeuwen, 2008). The intensified use of such terms amplifies the manufacturing of fear texts (Cap, 2017, p. xi) and oppositional stances, compelling the audience to defend their in-group ideology against perceived threats, as indicated by the coercive function of these inciting. This strategic utilisation of language underscores the symbolic power of religious discourse in shaping identities, justifying violence, and influencing attitudes within the context of extremist communication.

(15)	Indeed, the enemies of Allah — the Jews, Christians, atheists, Rafida, apostates, and all the nations of disbelief — dedicated their media, wealth, armies, and vehicles to wage war against the Muslims.

(16)

Here is Halab, facing the most tyrannical and vicious Nusayri [pro–al-Assad] campaign with kafir, Magian [i.e. Majusi/Iran and its backed militias], and Russian support, through which they aim to establish an alternative, Nusayri entity, all in the midst of the treachery of the apostate factions, which […] strive to remove Allah’s rule from the earth for the sake of the interests of their masters and backers from among the nations of kufr.

The rhetorical function of repeated, attitudinally loaded religious terms, exemplified in Examples 15–16, in close proximity—referred to elsewhere as ‘attitudinal burstiness’ (Etaywe, 2022a, p. 10, 2023a, p. 233)—aims at delegitimising and coercive symbolic power. These terms attack others’ master religious identity, deriving power from authoritative sources and texts, particularly potent in the terrorist context, justifying violence (e.g. Brookes & McEnery, 2020). Terrorists leverage religious language for its ‘symbolic capital’ (Bourdieu, 1991), rooted in the power of Islam. Example 15 assigns negative traits like ‘enemies of Allah,’ ‘atheists,’ ‘Rafida’ (derogatory term for certain Shia Muslims), and ‘apostates’ to labelled groups, establishing a power dynamic by framing them as enemies waging war against Muslims. References to religious identities like ‘Jews,’ ‘Christians,’ and ‘Rafida’ frame the conflict within the context of religious master identities linked to perceived actions of groups targeting Muslims, a discursive construction that serves the identity work of ‘our’ violence “legitimacy affirming” and ‘others’ “legitimacy contesting” (Brown & Toyoki, 2013, p. 876). Example 16 similarly constructs a relationship and framing of religious confrontation.

A pattern emerging is that terrorists tend to rely on specific attitudinal terms to label others, deploying an ‘intertextual im/propriety criterion’ coded in divine sources, often in their original language (Arabic). These lexicons shape a social group’s discourse and inter-group relationships that transcend borders and construct transnational identities by indexing cross border religious identities (e.g. Zappettini, 2016). For example, ‘shirk’/ ‘mushrik’ (polytheism/polytheist) serves as an attack and symbolic logic for religiously categorising and labelling, establishing categories of good and bad, inside and outside, leading to aggression (Example 17 from al-Baghdadi’s texts). ‘Shirk’ and ‘nifaq’ (polytheism and hypocrisy) associated with ‘they’ are coupled with impropriety, contrasting propriety values of ‘tawhid’ (monotheism) and ‘iman’ (faith). Terrorists view technical terms as consecration or revelation, deriving their influence from powerful sources like Allah. This intertextual im/propriety value shapes a desired ‘reality’ or subverts targeted social orders and ideational targets (cf. Bourdieu, 1989, p. 23; Bourdieu, 1991, p. 170). The same exclusionary, legitimising, and polarising function applies to lemmas like ‘taghut’ (tyrant) and ‘al-Salul’ (a name for a historical figure used by al-Baghdadi to label the Saudi regime). Example 17 contrasts negative traits of ‘shirk’ and ‘nifaq’ with positive traits of ‘tawhid’ and ‘iman,’ implying a power dynamic and a relationship marked by ideological and religious conflict.

(17)	Their shirk and nifaq will never overcome your tawhid and iman.

To conclude this section, delving into identity attacks has uncovered three distinct types of attitudes characterising these linguistic phenomena: affect-, appreciation-, and judgement-based identity attacks. The analysis has shed light on how evaluative expressions play a crucial role in deciphering authors’ suspicious stances. The identification of attacked aspects of identity, categorised into master, interactional, personal, and relational identities, has provided depth to our understanding of the multifaceted nature of identity attacks, illustrating how linguistic choices contribute to establishing inclusion and exclusion statuses, power relations, and ideological oppositions. Scrutinising identity attacks can also contribute to suspicions of extremist texts and ideological schema. Considering identity attacks from the identity work lens provides insights into two broad kinds of identity work (i.e. discursive activity of constructing extremists and their in-group, as well as pro-in-group relationships and actions, as legitimate, while constructing out-groups and their actions as illegitimate): (i) discursive acts that serve to affirm the legitimacy of ‘our’ violence, and (ii) discursive acts that serve to contest ‘others’ violence, referred to respectively as ‘legitimating identity work’ and ‘delegitimating identity work’ in the terms of Brown and Toyoki (2013, p. 880). Additionally, building on Goffman (1955, p. 216) that characterises studies of facework as “study[ing] the traffic rules of social interaction”, the analysis revealed that the practice of identity attacks includes not only instances of aggressive attitudes but also context-spanning acts of, for example, dehumanisation. These acts serve as markers of “people going through red lights,” in the terms of Tracy and Tracy (1998, p. 226), and have forensic investigative value in the study of extremist discourse.

Conclusion

This study has introduced a novel approach to understanding identity attacks in the context of terrorism and threatening terrorist communications, contributing valuable insights to the linguistic literature on identity- and face-work in the context of language aggression and conflict. Grounded in the Appraisal framework and key corpus analysis methods, it has presented the concepts of ‘evaluative textbites’ and ‘attitudinal priming’ for forensic text examinations. By focusing on extreme negative stances encoded in textbites, the research advances our comprehension of extremist discourse, highlighting the prevalence of identity attacks in various terrorist groups’ texts, namely Boko Haram, and ISIS. This commonality is attributed to shared ideology and rhetorical functions, such as constructing the ‘evil others’ and appealing to in-group members to support violent causes.

The recurrent use of identity attacks has revealed that identity not only emerges as aggressive facework bursts that violate rules of social interaction but is also wielded ‘as a tool’ (e.g. Antaki & Widdicombe, 1998), specifically, a rhetorical tool that functions to heighten the positioning of evil-others, and illocutionary force of a terrorist’s purposeful call for violence, whether in incitement or a declaration of intent to carry out threatened acts, indicating a determination to carry out these acts. Repeated identity attacks serve as valuable indicators of suspicious stances for threat assessment practitioners, reflecting a biased and aggressive perspective that fosters an environment of animosity. Additionally, the study has illustrated how an assault on a specific identity aspect (such as personal or interactional identity) is intricately linked to the master identity of an out-group, offering insights into the mechanisms of identity fusion. The social values characterising the repeated lemmas—particularly cultural/religious and aggressive values (refer to Table 6 in the previous section)—provide clues about authors who value and prioritise certain beliefs and traditions, as well as violent methods for in-group defence. These insights suggest that it is not the religion itself that motivates violence, but rather the individual’s concern about in-group identity and perceived threats to their essence (e.g. religious identity and Muslim lands), leading to ‘identity fusion’ (Buhrmester et al., 2018). The threatened or incited-against individuals are portrayed as belonging or being loyal (in actions and roles) to an out-group, involving functionalisation and identification in references to the targets of verbal and threatened aggression.

Supported by corpus analysis methods, the article has provided a systematic approach to building an attitudinal profile of an author and to understanding how linguistics can get at the performative nature of identity attacks by studying repeated discursive patterns that cross social red lights. Identity attacks primarily emerge as judgement- and appreciation-based, with a relatively lower frequency of being affect-based. This pattern reveals a form of identity work driven by strong ideological motivations that highlight inter-group moral disparities and a perceived sense of inferiority of values attributed to out-groups.

The introduction of ‘attitudinal priming’ has shed light on how specific ideational targets are primed for co-occurrence with particular negative evaluative functions within the textual fabric. Emphasising ideational and interpersonal functions, the study illuminates how extremists construct a binary oppositional framework (e.g. Chandler, 2017) to interpret the world axiologically. The study enhances threat assessment practices by introducing violent behaviour indicators tied to ideology and linguistic choices. Additionally, presenting evaluative textbites as a unit of analysis allows future analysts to focus on meaningful segments and get closer to identifying the extremist’s functional linguistic fingerprint which offers useful investigative impressions created by an author’s encoded stances and attitudinal priming. The study has demonstrated the usefulness of the metafunctional analysis for forensic investigations (Canning, 2014), particularly for describing attitudinal priming. It shows the potential for attitudinal priming to enable linguists supporting threat assessment protocols to discern the semantic, pragmatic, and textual/discoursal contexts linked to attitudinal expressions in evaluative textbites. In Tracy and Tracy’s (1998) words, the examination of stance indicators and face attacks can provide linguistic clues that enable connecting context to linguistic strategy use (e.g. identity attacks). The analysis has shown the potential to use attitudinal priming, the ‘ideation-attitude’ patterning, by investigators to glean insights into verbal aggression and indications of violent behaviour as well as terrorist threatening or inciting genre/style associated with identity attacks. Using Hoey’s (2013, p. 3344) terminology, violent extremists are inherently predisposed or “primed to associate [an attitudinally loaded] word” or phrase “with social/stylistic context” in which it appears (Hoey, 2013, p. 3344). They tend to associate negative attitudinal meanings with the ideational targets of the ‘Them’ group in a context where language choices are strategically made to meet specific functional and ideological needs of the text producers and the norms of the genre.

The study suggests the use of the evaluative textbite approach and attitudinal priming concept in future research into identity work and identity fusion in phenomena such as hate speech, defamation, genocidal rhetoric, and political discourse to better understand how language users leverage identity and encode attitudes to lay the socio-psychological premise for the perception and fear of others’ threat, legitimising political violence against opponents. The study shows that the attacked aspects of identities offer a lens to interrogators on the motivations of violence and themes frequent in violent extremist texts, such as concern about shared religion, ethnicity, people’s behaviours (perceived as immoral), feelings of insecurity, and the worthiness of members of ‘out-groups.’

While this study makes significant contributions to the understanding of identity attacks, stance indicators, and attitudinal priming in extremist discourse, the results were based on a limited sample of eleven terrorist public messages by authors sharing the same ideology. Extending the study to include a more extensive and diverse corpus of extremist texts could enhance the robustness of the conclusions drawn. Despite these limitations, this research lays a foundation for future investigations and underscores the need for a comprehensive understanding of language’s role in extremist communication. Continued research in this direction can contribute to the development of proactive strategies for prediction, attitudinal profiling, and countering terrorism through linguistic analysis.