Cross-linguistic Delegitimisation of Women Leaders in Online Political Discourse

Cross-linguistic Delegitimisation of Women Leaders in Online Political Discourse | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Cross-linguistic Delegitimisation of Women Leaders in Online Political Discourse Sergei Sikorskii, Maria Luisa Carrió-Pastor, Giovanni Garofalo This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7820069/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 18 Dec, 2025 Read the published version in Corpus Pragmatics → Version 1 posted You are reading this latest preprint version Abstract This study examines evaluative configurations in hostile digital political discourse targeting women leaders across Spanish and Italian contexts. Through systematic analysis of 2,000 replies to Isabel Díaz Ayuso and Giorgia Meloni on X, we identify culturally distinct evaluative repertoires that function as delegitimization mechanisms. Spanish discourse emphasizes moral sanctioning through Propriety-Normality combinations (59.6% vs. 37.0% Propriety in Italian replies), while Italian discourse foregrounds competence-authenticity challenges through Capacity-Veracity configurations. Logistic regression reveals systematic relationships between evaluative configurations and identity-salient stance markers, with Tenacity (OR = 6.45) and Normality (OR = 3.83) judgements emerging as strongest predictors. The findings advance theoretical frameworks in stance analysis and Appraisal Theory by demonstrating how evaluative copatterns create compound delegitimizing effects beyond individual category contributions, while revealing persistent cultural influences on evaluative practice within shared digital environments. Figures Figure 1 Figure 2 Figure 3 1. Introduction Contemporary digital political communication reveals systematic patterns in how hostile discourse targets women leaders. When political actors engage on platforms like X, women face attacks that are not merely random incivility but patterned through recurring evaluative strategies that question competence, integrity, and legitimacy. Such communicative practices function as delegitimising mechanisms that may discourage participation in public debate (Esposito & Breeze, 2022 ; Southern & Harmer, 2021 ), while men politicians, though also criticised, tend to face abuse less systematically tied to identity markers (Erikson et al., 2023 ; Meriläinen, 2024 ). Cross-national evidence confirms these dynamics across contexts. In the UK, women MPs encounter both overt abuse and "everyday" microaggressions that frame them as unqualified or inappropriate for leadership roles (Southern & Harmer, 2021 ). Nordic studies demonstrate that while overall volumes of abuse may not differ by gender, its quality does: women politicians face more personalised, identity-focused attacks, particularly when occupying portfolios traditionally coded as "masculine" (Erikson et al., 2023 ; Håkansson, 2021 , 2024 ; Meriläinen, 2024 ). Recent analyses of Spanish political discourse show that women leaders like Isabel Díaz Ayuso tends toward moral condemnation/infantilisation, while Italian studies of Giorgia Meloni reveal patterns focused on competence and authenticity (Combei & Reggi, 2024 ; del Saz-Rubio, 2024 ). These patterns suggest that hostile evaluations embed gendered assumptions within apparently neutral assessments of political performance. Following (Du Bois, 2007 ), we treat evaluations as components of stance (a configuration of evaluation, positioning, and alignment) whose social meanings emerge through indexical processes (Agha, 2006 ; Ochs, 1992 ; Silverstein, 2003 ). Evaluative language thus functions not merely as opinion expression but as a resource for public positioning: critics who portray a woman leader as 'unable to cope with pressure' or recast policy disagreement as moral failure mobilise evaluations that appear to address performance while potentially invoking identity-based assumptions about legitimacy. We examine how such identity-salient linguistic markers (evaluative patterns that may index social positioning beyond policy critique) cluster with specific Judgement configurations, drawing on work on delegitimisation and legitimation in political discourse (Reisigl & Wodak, 2001 ; Van Leeuwen, 2007 ). This study examines hostile replies on X addressed to Giorgia Meloni (Italy) and Isabel Díaz Ayuso (Spain) over a 40-day window, using the Judgement subsystem of Appraisal Theory to identify systematic co-patterns of evaluation. Rather than treating categories in isolation, we analyse their co-occurrence (e.g., Capacity and Propriety) as compact configurations that index delegitimising stances (Bednarek, 2008 ; Hood, 2010 ; Martin & White, 2005 ). The comparative dimension addresses how evaluative strategies vary across linguistic and contexts. Spanish and Italian replies to these leaders reveal distinct emphases: Spanish discourse tends toward moral sanctioning and behavioural evaluation, while Italian discourse foregrounds competence and authenticity assessment. However, rather than proposing universal cultural mechanisms, we interpret these as contrastive repertoires emergent within our corpus (Thompson & Alba-Juez, 2014 ). Research questions How do Judgement categories distribute and co-pattern in hostile replies on X to two women leaders across Spanish and Italian? Which evaluative configurations cluster with identity-salient linguistic markers in this corpus? What Spanish-Italian contrasts appear in evaluative repertoires, and how do these patterns relate to identity-salient stance cues? Our contribution lies in specifying how evaluative co-patterns function as delegitimisation mechanisms in digital political discourse. We develop an operational protocol for identifying identity-salient stance cues through Judgement category analysis, with empirical claims confined to our dataset of 2,000 annotated replies. By examining evaluative configurations rather than isolated categories, this study advances understanding of how linguistic resources index social positioning in antagonistic digital interaction. 2. Theoretical Background 2.1. Evaluative Language as Stance Performance Stance encompasses the simultaneous acts of evaluation, positioning, and alignment that speakers perform through linguistic choices (Du Bois, 2007 ). In the stance triangle, evaluation involves speakers' subjective assessments of objects or propositions, positioning establishes the speaker's relationship to the evaluation, and alignment creates intersubjective relationships between speakers. This framework moves beyond treating evaluation as isolated opinion expression toward understanding how evaluative choices position speakers socially and create relationships with audiences. The social meanings of evaluative language emerge through indexical processes whereby linguistic forms point to social categories, relationships, and identities (Silverstein, 2003 ). Indexicality operates across multiple orders: direct indexical relationships (form-meaning pairings) become embedded within higher-order indexical fields that connect linguistic features to social personae and cultural models (Agha, 2007). When speakers evaluate political figures, their linguistic choices do not merely convey opinions but index stances toward authority, competence, and legitimacy that carry broader social implications. Indexical processes are particularly relevant for understanding how gender emerges in discourse. Rather than treating gender as a stable variable, sociolinguistic research demonstrates that gender identity is constructed through iterative linguistic practices that index cultural models of femininity and masculinity (Ochs, 1992 ). Evaluative language can indirectly index gender through stances, affects, and social activities that are culturally associated with gendered identities, creating what Ochs terms "constitutive indexicality" where repeated linguistic practices contribute to identity construction. 2.2. Appraisal Theory and Evaluative Configurations Appraisal Theory provides a systematic framework for analysing evaluative meaning in discourse, with the Judgement subsystem specifically addressing evaluations of human behaviour and character (Martin and White, 2005 ). Judgement categories capture two broad evaluative domains: Social Esteem (assessments of normality, capacity, and tenacity) and Social Sanction (assessments of veracity and propriety). These categories function as resources for stance-taking rather than fixed semantic labels, allowing speakers to position themselves and their addressees relative to social norms and expectations. Appraisal research has increasingly emphasised evaluative configurations— patterns of co-occurring evaluative meanings that create coherent attitudinal positions (Hood, 2010 ). Rather than analysing isolated evaluative instances, this approach examines how multiple Appraisal resources combine to create "prosodies" of evaluation that extend across texts and social interactions. Such configurations allow for more nuanced stance performances where, for example, capacity assessments combine with propriety judgements to create compound delegitimising effects. Cross-linguistic applications of Appraisal Theory reveal both universal and language-specific patterns in evaluative resource deployment (Thompson & Alba-Juez, 2014 ). While the semantic domains of Judgement appear cross-linguistically stable, their realisation patterns, co-occurrence tendencies, and social implications vary across cultural and linguistic contexts. This variation reflects differences in cultural values, social hierarchies, and communicative norms that shape how evaluative meanings are interpreted and deployed. 2.3. Delegitimisation and Identity-Salient Discourse Delegitimisation operates through discourse strategies that undermine the authority, credibility, or appropriateness of political actors (Reisigl and Wodak, 2001 ). These strategies typically involve negative categorisation, attribution of negative qualities, and questioning of competence or moral standing. Van Leeuwen's (2007) analysis of legitimation strategies reveals how discourse constructs and challenges authority through appeals to different value systems, with particular attention to how evaluative language indexes relationships between power and social acceptance. Identity-salient delegitimisation occurs when attacks on political figures deploy evaluative strategies that invoke cultural stereotypes or expectations associated with social identities. This process differs from general political criticism by embedding identity-based assumptions within apparently neutral performance assessments. For instance, questioning a woman politician's emotional stability or decision-making capacity may invoke cultural models of gender and leadership while ostensibly addressing professional competence. The digital context intensifies these dynamics through platform affordances that prioritise engagement and emotional response (Papacharissi, 2014 ). Social media platforms create "affective publics" where emotional expressions and evaluative stances circulate rapidly and gain visibility through algorithmic amplification. Within these environments, delegitimising discourse can achieve broader reach and impact while maintaining the appearance of spontaneous user response rather than coordinated attack. 2.4. Cross-Linguistic Evaluation and Digital Discourse Comparative analysis of evaluative discourse reveals how cultural repertoires shape the deployment of linguistic resources across different contexts (Thompson and Alba-Juez, 2014 ). These repertoires encompass not only lexical choices but also structural patterns, pragmatic conventions, and indexical associations that vary across linguistic communities. Cross-linguistic evaluation research demonstrates that while semantic categories of evaluation may be universal, their social meanings and deployment patterns reflect cultural values and communicative norms. Digital platforms introduce additional complexity by creating spaces where multiple linguistic and cultural repertoires intersect while operating under shared technological constraints (Zappavigna, 2012 ). Platform features such as character limits, threading structures, and visibility algorithms shape how evaluative meanings can be expressed and circulated, potentially favouring certain types of evaluative strategies over others. The resulting discourse represents a hybrid of traditional evaluative repertoires and platform-specific adaptations. Research on Twitter discourse reveals how platform constraints influence evaluative expression through compression, hashtag use, and intertextual referencing (Zappavigna, 2018 ). These adaptations create new forms of evaluative practice while drawing on existing cultural resources for stance-taking and social positioning. Understanding digital evaluative discourse thus requires attention to both linguistic repertoires and technological mediation of social interaction. 2.5. Analytical Framework: From Categories to Configurations Our analytical approach treats Judgement categories as resources for stance performance rather than discrete semantic labels. We focus on evaluative configurations, systematic co-occurrence patterns such as Capacity-Propriety combinations, that create compound evaluative effects beyond individual category meanings. These configurations function as compact strategies for positioning social actors within evaluative frameworks that index broader cultural models of competence, authority, and legitimacy. The identity-salient dimension of our analysis examines how evaluative configurations may invoke cultural repertoires associated with social identities. Rather than assuming direct gender indexing, we analyse patterns that cluster with linguistic markers potentially associated with identity-based evaluation, such as infantilising language, appearance focus, or emotional stereotyping. This approach allows for empirical identification of identity-salient patterns without presuming their cultural significance. Cross-linguistic comparison focuses on contrastive repertoires within our corpus rather than universal mechanisms. Spanish and Italian evaluative patterns are analysed as emergent configurations specific to responses to these particular political figures rather than as representative of broader cultural systems. This approach acknowledges the limitations of our data while providing insight into how evaluative resources operate across different linguistic contexts within the constraints of digital political discourse. 3. Methods This study employs a mixed-methods approach combining corpus linguistics, discourse analysis, and statistical methods to examine evaluative configurations in hostile digital political discourse. The methodology integrates theoretical insights from stance analysis and Appraisal Theory to identify systematic patterns in how Judgement resources index delegitimising positions across Spanish and Italian contexts. 3.1. Research Design and Corpus Construction We used a two-stage design to balance coverage and analytic depth. Stage 1 built the sampling frame of candidate replies; Stage 2 drew stratified analytical subcorpora. Stage 1.We collected reply objects on X between 1 April and 9 May 2025 using leader-specific queries that combined name variants and verified handles (e.g., "Isabel Díaz Ayuso" OR @IdiazAyuso; "Giorgia Meloni" OR @GiorgiaMeloni), with language filters (lang:es, lang:it). We excluded retweets and retained posts with “is_reply”; quote tweets were included but flagged as a subtype. After de-duplication and light hygiene (removal of exact duplicates, promotional spam patterns, and off-language leakage identified via spot checks), the sampling frame comprised 88,872 unique replies (Ayuso 42,950; Meloni 45,922). We preserved conversation IDs and source-post IDs to maintain microcontexts and recorded engagement at collection time (likes/retweets) binned into low/medium/high for later modelling. Stage 2. From this sampling frame we sampled 2,000 replies (1,000 per leader). Following social-media corpus guidelines (Moreno-Ortiz & García-Gámez, 2023 ), we combined a lexicon-triggered stratum with a 10% random stratum to balance evaluative density with coverage. The lexicon targeted replies likely to contain Judgement resources or incivility, including competence (incapaz/incompetente; incapace/inetta) , morality (corrupta/mentirosa; corrotta/bugiarda) , and behavioural characterisation (histérica/débil; isterica/debole) . Sampling was stratified along three axes: (i) time (uniform across the window to cover distinct events), (i) microcontext (multiple source posts per leader; cap on replies per thread to prevent thread dominance), and (iii) post subtype (standard replies vs quote tweets, analysed with a subtype flag). With n = 1,000 per leader, proportion estimates carry a ≈ ± 3pp 95% margin of error at p = .5. The final analytical corpus comprised 2,000 replies (1,000 per politician) selected through stratified sampling across three dimensions: temporal distribution to capture responses to different political events, conversational diversity to prevent single-thread dominance, and post-type variation to include both original posts and replies. This sampling framework follows recommendations for maintaining discursive diversity while enabling systematic comparative analysis (KhosraviNik, 2017). 3.2. Operationalising Identity-Salient Evaluation Identity-salient evaluation was operationalised through a systematic protocol targeting linguistic patterns that potentially invoke cultural repertoires associated with social identity categories. Rather than assuming direct indexical relationships, we developed detection criteria based on convergent evidence from gender and political discourse research (Jane, 2014 ; Mantilla, 2015 ; Southern & Harmer, 2021 ). The protocol employed three detection criteria applied sequentially: (i) Explicit identity markers: lexical items directly referencing identity categories or diminutive constructions that may invoke patronising stances; (ii) Differential targeting: evaluative patterns documented in previous research as disproportionately directed at women politicians relative to male counterparts (emotional instability attributions, appearance focus, maternal role invocation); and (iii) Stereotypical bundling: co-occurrence of multiple evaluative themes that collectively invoke cultural models of gender and authority (competence questioning + emotional characterisation + infantilising language). Each reply meeting any criterion was flagged as potentially identity-salient, with criterion-specific coding (1 = explicit, 2 = differential, 3 = bundled) to enable sensitivity analysis. Inter-criterion overlap was tracked, with 23% of flagged replies meeting multiple criteria. This approach acknowledges the indirect nature of identity indexing while providing empirically grounded detection methods that can be replicated across contexts. 3.3. Appraisal Annotation and Configuration Analysis Evaluative analysis employed Martin and White's (2005) Judgement framework, focusing on systematic identification of category co-patterns rather than isolated instances. The annotation protocol operationalised five Judgement categories: Capacity (ability, competence, skill), Propriety (ethics, morality, legality), Veracity (honesty, authenticity, credibility), Tenacity (resolve, dependability, determination), and Normality (conventionality, expectedness, typicality). Each reply was coded for all present categories with polarity marking (positive/negative) and salience weighting (primary/secondary) to capture evaluative prominence hierarchies. Configuration analysis identified systematic co-occurrence patterns within individual replies, with particular attention to cross-category combinations (e.g., Capacity + Propriety, Veracity + Tenacity) that create compound evaluative effects. Configurations appearing in < 3% of replies were grouped as "miscellaneous" to focus analysis on substantive patterns. Language-specific annotation incorporated pragmatic markers documented in Spanish and Italian evaluative discourse research. Spanish coding captured ¡qué..! constructions for ironic intensification, diminutive-evaluative combinations (politiquilla, ministra), and moral-register lexicon (sinvergüenza, criminal). Italian annotation identified subjunctive doubt constructions (non sia onesta), metaphorical competence frames (maschera, teatrino), and authenticated/performance contrasts (finge, recita). This ensures crosslinguistic comparison reflects genuine differences in evaluative deployment rather than analytical artefacts. 3.4. Annotation Procedures and Reliability Assessment Two annotators, trained on a 200-item pilot, independently coded the corpus in 500-reply batches with periodic calibration. The protocol targeted Judgement at two layers: (i) category selection (presence/absence and polarity, with salience weighting of primary vs. secondary instances), and (ii) configuration identification (co-occurring Judgement bundles) and identity-salience flags (three detection criteria). Disagreements were adjudicated after each batch. We report reliability across multiple indices tailored to task granularity and class imbalance. For polarity, Fleiss' k = 0.52. For fine-grained Judgement type selection, Fleiss' k = 0.32, converging with imbalance-robust metrics (Krippendorff's a = 0.31, Gwet's AC1 = 0.34) and consistent with evaluative annotation facing implicit cues (Bednarek & Caple, 2017 ; Fuoli, 2018 ; Martin & White, 2005 ). Raw agreement was 63%. For the category presence/absence + salience layer, overall Krippendorff's a = 0.52; per-category a values were Capacity = 0.57; Propriety = 0.62; Veracity = 0.55; Tenacity = 0.52; Normality = 0.6 (Hayes & Krippendorff, 2007 ). Configuration identification yielded a = 0.58. Identity-salience flagging reached a = 0.63, with criterion-specific variation (explicit markers a = 0.68; differential targeting a = 0.52; stereotypical bundling a = 0.58). Confusion analysis showed non-random proximity between Propriety and Normality, where moral critique blurred with characterological characterisation. To reflect construct adjacency rather than coder error, we adopted compound labels (permitting > 1 Judgement per unit); 26.3% of units received co-assignments. Final datasets comprise adjudicated consensus with compound labels preserved for transparency. 3.5. Ethical Considerations and Data Governance This research analysed publicly available social media content directed at public political figures, following established protocols for digital discourse research (Townsend & Wallace, 2016). Data collection utilised X's public API within rate limits and terms of service, with no attempts to access private or protected content. Anonymisation protocols removed individual user identifiers while preserving sufficient context for discourse analysis. Quoted examples are selected to illustrate systematic patterns rather than individual attacks, with preference for analytically relevant rather than sensationalist content. Research outputs focus on linguistic mechanisms and cultural repertoires rather than individual users or personal characteristics. Hostile content handling follows established practices in political incivility research (KhosraviNik, 2023 ), prioritising scholarly analysis over reproduction of harmful discourse. The research contributes to understanding of digital political communication dynamics without amplifying or endorsing the delegitimising strategies that form its analytical foundation. All procedures received institutional research ethics approval prior to data collection. 4. Results This section reports findings from the stratified corpus of 2,000 replies (1,000 directed at Isabel Díaz Ayuso in Spanish; 1,000 directed at Giorgia Meloni in Italian). Results are organized by research question. 4.1. RQ1. How do Judgement categories configure in hostile replies to these leaders, and what co-patterns emerge? Table 1 summarizes the distribution of Judgement categories across the two corpora with 95% confidence intervals. Propriety judgements were most frequent in Spanish replies to Ayuso (59.6%, CI: 56.5-62.6%), while Capacity judgements dominated Italian replies to Meloni (35.0%, CI: 30.0-40.3%). Normality evaluations showed the largest cross-linguistic contrast, appearing in 28.5% of Spanish replies (CI: 25.8-31.4%) versus only 2.2% of Italian replies (CI: 1.2-3.6%). Tenacity judgements were infrequent in both corpora (Ayuso: 1.9%, Meloni: 1.2%). Table 1 Distribution of Judgement categories across Spanish and Italian replies with 95% confidence intervals Leader Category n k prop ci_low ci_high Ayuso Capacity 1000 289 0.289 0.2617539675950579 0.3178609836083989 Ayuso Propriety 1000 596 0.596 0.5652749017650743 0.6259903337063386 Ayuso Veracity 1000 207 0.207 0.1830325088651358 0.23321005370648906 Ayuso Tenacity 1000 19 0.019 0.012196787737404145 0.029484688702703543 Ayuso Normality 1000 285 0.285 0.25788529526982207 0.3137602711223259 Meloni Capacity 1000 350 0.35 0.3210621906513813 0.3800858789245359 Meloni Propriety 1000 370 0.37 0.34062614660286095 0.4003688470296006 Meloni Veracity 1000 326 0.326 0.2976606245950845 0.3556711361129795 Meloni Tenacity 1000 12 0.012 0.006877579092159562 0.020857473928157594 Meloni Normality 1000 22 0.022 0.014572598673771597 0.03308591637481774 Co-occurrence analysis reveals systematic category clustering within replies. Table 2 presents conditional probabilities for key category combinations. The most striking pattern emerges in competence-to-morality transitions: Spanish replies show (Propriety|Capacity) = 0.408, indicating that 40.8% of competence critiques co-occur with moral evaluation. In contrast, Italian replies show P(Propriety|Capacity) = 0.209, representing a near 2:1 difference in moralization rates. Table 2 Conditional probabilities for major category co-patterns leader A B_given_A num den P(B|A) Ayuso Capacity Propriety 118 289 0.4083044982698962 Ayuso Capacity Tenacity 2 289 0.006920415224913495 Ayuso Propriety Tenacity 16 596 0.026845637583892617 Ayuso Veracity Propriety 89 207 0.42995169082125606 Ayuso Veracity Tenacity 2 207 0.00966183574879227 Ayuso Tenacity Propriety 16 19 0.8421052631578947 Ayuso Normality Propriety 124 285 0.43508771929824563 Ayuso Normality Tenacity 4 285 0.014035087719298246 Meloni Capacity Propriety 73 350 0.20857142857142857 Meloni Capacity Tenacity 3 350 0.008571428571428572 Meloni Propriety Tenacity 4 370 0.010810810810810811 Meloni Veracity Propriety 65 326 0.19938650306748465 Meloni Veracity Tenacity 6 326 0.018404907975460124 Meloni Tenacity Propriety 4 12 0.3333333333333333 Meloni Normality Propriety 7 22 0.3181818181818182 Meloni Normality Tenacity 1 22 0.045454545454545456 Additional co-patterns include Propriety-Tenacity combinations (P(Tenacity|Propriety) = 0.027 for Ayuso, 0.011 for Meloni) and Capacity-Normality linkages, though these occur at lower frequencies. Chi-square tests confirm significant associations between categories, with phi coefficients indicating moderate to strong effect sizes for key combinations. We observe asymmetric co-occurrence (e.g., P(Propriety|Capacity) > P(Capacity|Propriety)), consistent with an alignment pathway from competence assessment to moral sanction; our design is associational and does not establish temporal or causal ordering. The distributions show that Spanish replies lean toward Propriety-, while Italian replies foreground Capacity-. These tendencies are visible in the corpus excerpts. In Spanish, Ayuso is attacked through moralised disgust (Ex. 4.1a), while in Italian Meloni is targeted via explicit reference to failure (Ex. 4.1b). Ex. 4.1a (Ayuso, ES; Propriety−) (tweet_id: AYS_16_2025_0057) Original (ES) “@IdiazAyuso Qué ascazo de mujer !!!!” Translation (EN) “What a disgusting woman !!!!” Reading: Propriety− (moralised disgust; sanctioning register). Ex. 4.1b (Meloni, IT; Capacity−) (tweet_id: MEL_11_2025_0055) Original (IT) “@user Aspettiamo un post che commenti anche i tuoi fallimenti .” Translation (EN) “We’re waiting for a post that also comments on your failures .” Reading: Capacity− (incompetence via repeated failure). 4.2. RQ2. How do these evaluative configurations function as resources for delegitimising stance performance? Logistic regression analysis identifies the strongest predictors of identity-salient stance cues (Table 3 ). Tenacity evaluations emerge as the most powerful predictor (OR = 6.45, 95% CI: 2.84–14.66, p < 0.001), indicating that vulnerability-based attacks are central to delegitimization strategies. Normality judgements show the second-strongest effect (OR = 3.83, 95% CI: 2.85–5.15, p < 0.001), reflecting the role of behavioral conformity demands in identity-salient positioning. Table 3 Logistic regression results: Predictors of identity-salient stance (ORs, 95% CIs, p-values) Predictor OR CI_low CI_high p_value Intercept 0.1456602376181486 0.10966486103553034 0.19347040266709906 2.2445312240528477e-40 Capacity 0.8751150260502529 0.62133743604804 1.2325449335386953 0.4452090941524517 Propriety 2.1096318561888068 1.5908175824884663 2.7976473340738286 2.1769011435196684e-07 Veracity 1.2099033323092494 0.9305163823786222 1.5731760356448905 0.15491822270940414 Tenacity 6.4496850040381535 2.837034150416943 14.662649247701562 8.646946902848037e-06 Normality 3.8284395825972344 2.8478334966990855 5.1467017487455315 5.995285189933957e-19 CapxProp 1.8670563554474753 1.159597995865985 3.0061275087092123 0.01019053885020874 ES 1.6532440488181457 1.313399687350815 2.081023706093331 1.8531048351199657e-05 Figure 2 presents these results as a forest plot, visualizing the relative strength of different evaluative configurations. Model diagnostics indicate adequate fit (McFadden pseudo-R² = 0.191, classification accuracy = 73.2%). Figure 3 illustrates the Capacity×Propriety interaction by leader, demonstrating how competence-morality configurations vary across linguistic contexts. The interaction shows divergent patterns: Spanish replies exhibit steep increases in identity-salient probability when both categories co-occur, while Italian replies show more moderate effects. The co-pattern analysis shows that Spanish replies frequently bundle Capacity- with other Judgements, often escalating into moral sanction, while Italian replies combine incompetence and dishonesty in compound attacks. This is exemplified by a Spanish tweet that links dishonesty and incompetence (Ex. 4.2a) and an Italian tweet that pairs incapacity with dishonesty (Ex. 4.2b). Ex. 4.2a (Ayuso, ES; Capacity− + Veracity−) (tweet_id: AYS_16_2025_0004) Original (ES) “@Morcego_Man @IdiazAyuso Ahora dilo pero sin las mentiras , que eres una inútil .” Translation (EN) “Now say it but without the lies , because you’re useless .” Reading: Veracity− ( mentiras = dishonesty) + Capacity− ( inútil = incompetence). Ex. 4.2b (Meloni, IT; Capacity− + Propriety−) (tweet_id: MEL_14_2025_0002) Original (IT) “@user ‘Onoriamo’? Ma che devi onorare se pensi solo a te stessa, incapace e disonesta .” Translation (EN) “‘We honor’? What do you have to honor if you only think of yourself, incompetent and dishonest .” Reading: Capacity− ( incapace ) + Propriety− ( disonesta ); compound configuration. 4.3. RQ3. What Spanish-Italian contrasts appear in evaluative repertoires, and how do these patterns relate to identity-salient stance cues? Cross-linguistic analysis reveals substantial effect sizes for key categories (Table 4 ). Normality judgements show the largest contrast (Cramér's V = 0.365, p < 0.001), confirming Spanish discourse's emphasis on behavioral policing. Propriety evaluations demonstrate a large effect size (V = 0.226, p < 0.001), while Veracity shows medium-sized differences (V = 0.135, p < 0.001), with Italian discourse featuring more credibility-focused attacks. Table 4 Cross-linguistic effect sizes (Cramér's V) for category contrasts Category Cramer´s_V p_value Ayuso Meloni Capacity 0.06541093292775553 0.0034416206734931088 289 350 Propriety 0.2261307413636883 4.84418793725353e-24 596 370 Veracity 0.13457621205415032 1.7611557411151298e-09 207 326 Tenacity 0.028333115696007258 0.20512116079234521 19 12 Normality 0.36480273166894667 7.787392189432449e-60 285 22 Leader-specific analysis reveals distinct cultural pathways for identity-salient stance performance (Table 5 ). In Spanish contexts, Propriety judgements strongly predict identity-salient stance (OR = 3.62, 95% CI: 2.41–5.41), while the same category shows nonsignificant effects in Italian contexts (OR = 1.08, 95% CI: 0.71–1.65). Conversely, Normality evaluations predict identity-salient stance in both contexts but with different magnitudes (Ayuso: OR = 4.54; Meloni: OR = 5.68). Table 5 Leader-specific odds ratios for identity-salient stance predictors Leader Predictor OR CI_low CI_high p_value Ayuso const 0.15423637510495014 0.10173182398208255 0.23383891563474454 1.321685001577727e-18 Ayuso Capacity 1.001460051903562 0.6011986725178501 1.6682043414340828 0.995528829094319 Ayuso Propriety 3.6149401264576655 2.4147827684633394 5.411580821487112 4.3018810100922885e-10 Ayuso Veracity 1.2615017957069514 0.8661385727546785 1.8373350761998686 0.2259430498484959 Ayuso Tenacity 29335490064.573086 0.0 inf 0.9988590384664778 Ayuso Normality 4.542449305475346 3.2174902807763477 6.413025026398749 7.870942260837633e-18 Ayuso CapxProp 1.633461336374365 0.8452535247433561 3.1566812315157984 0.1443402590231145 Meloni const 0.20717622682060402 0.14561666985633484 0.29476013290215414 2.1246657345483944e-18 Meloni Capacity 0.7688707820468026 0.4885754237442349 1.209971379556599 0.25591361179501493 Meloni Propriety 1.079652768195933 0.7068680298957951 1.6490349691511414 0.7228547555635039 Meloni Veracity 1.1322662290345151 0.7801486732335287 1.643310893676648 0.5133552827202998 Meloni Tenacity 1.1995658248902334 0.30613689944510414 4.700374802425989 0.7939845064298733 Meloni Normality 5.683590576594332 2.3903640098087653 13.513925791133468 8.426322155890154e-05 Meloni CapxProp 2.441562066432025 1.170692585882572 5.0920501215489615 0.017303868255399178 These patterns indicate culturally specific evaluative repertoires: Spanish replies systematically moralize political critique through Propriety-Normality combinations, while Italian replies foreground competence-authenticity challenges through Capacity-Veracity configurations. The interaction model confirms that category effects vary significantly by linguistic context (Category × Leader interactions, p < 0.05 for Propriety and Normality), supporting the hypothesis of culturally embedded evaluative pathways. Cross-linguistic contrasts are clear in the excerpts: Spanish users often frame Ayuso as abnormal or "out of line" (Ex. 4.3a), while Italian users depict Meloni as politically mad (Ex. 4.3b). These examples reflect the broader repertoires identified in the statistical analysis. Ex. 4.3a (Ayuso, ES; Normality−) (tweet_id: chemay59_2025-04-11_11:30:43) Original (ES) “@user Dudo que seas Ayuso oficial… Te has ido de la olla .” Translation (EN) “I doubt you’re the official Ayuso… You’ve lost it .” Reading: Normality− (abnormal behaviour/being “out of line”). Ex. 4.3b (Meloni, IT; Normality−) (tweet_id: MEL_18_2025_0047) Original (IT) “@GiorgiaMeloni Prospero? Siamo ad una soglia di follia politica.” Translation (EN) “Prospero? We’re at a threshold of political madness .” Reading: Normality− (abnormalisation via follia ). The Spanish pattern emphasizes moral sanctioning and behavioral conformity as delegitimization strategies, while the Italian pattern prioritizes competence questioning and authenticity challenges. These contrasts persist after controlling for temporal effects, consistent within this corpus and window; broader stability is a hypothesis for future work. 5. Discussion 5.1 RQ1. How do Judgement categories configure in hostile replies to women leaders, and what co-patterns emerge? The distributional results show that hostile replies are not scattered individual attacks but are structured by recurring evaluative resources. This confirms the premise that digital hostility functions communicationally, as patterned stance performance rather than idiosyncratic abuse (cf. Du Bois, 2007 ; Silverstein, 2003 ). The dominance of Propriety- judgements in Spanish replies and Capacity-/ Veracity- judgements in Italian replies-visible in the aggregate (Table 1 ; Fig. 1 ) —is also instantiated in typical excerpts: Spanish moral sanction with abnormalisation (corrupta + emphatic capitals) in Ex. 5.1a, and Italian competence/ authenticity framing (hai sempre fallito) in Ex. 5.1b. Taken together with the cooccurrence counts (Table 2 ), these examples indicate that categories of moral sanction and competence assessment are mobilised as collective repertoires of delegitimisation. These findings extend prior research on gendered political aggression, which identified a shift from overtly sexist invective to subtler, performance-framed delegitimation (Esposito & Breeze, 2022 ; Southern & Harmer, 2021 ). Crucially, these recurrent configurations are recognisable to participants themselves: the pairing of incompetence with corruption or of fakeness with failure functions as a discursive shorthand for stance alignment, inviting uptake and circulation within hostile publics. Ex. 5.1a — Ayuso (ES; Propriety− + Normality−) ( tweet_id : PepBausset_2025-04-11_08:04:06) Original (ES) “@user Ayuso súper corrupta , a los tribunales POR FIN .” Translation (EN) “Ayuso super corrupt —to the courts AT LAST .” Reading: Propriety− ( corrupta ); Normality− (social deviance frame); Graduation (force) via capitals. Ex. 5.1b — Meloni (IT; Capacity− + Veracity−) ( tweet_id : MEL_18_2025_0069) Original (IT) “@GiorgiaMeloni Basta cazzate… hai sempre fallito .” Translation (EN) “Enough nonsense… you’ve always failed .” Reading: Capacity− (repeated failure), Veracity− (misrepresentation implied). The co-occurrence analysis highlights that evaluative meanings rarely appear in isolation. Instead, they are bundled into compact configurations-such as Capacity- → Propriety- in Spanish and Capacity-»Veracity- in Italian-that intensify delegitimisation. In the Spanish set, this escalation is instantiated in Ex. 5.1c (cobarde recruiting sinvergüenza), while the Italian route is instantiated in Ex. 5.1d (falsa, bugiarda alongside a fitness attack). These pairings correspond to the significant asymmetries detected by McNemar tests and the conditional probabilities reported in Table 2 (e.g., higher P(Propriety-|Capacity-) in Spanish; higher P(Veracity-|Capacity-) in Italian). From an Appraisal perspective, such pairings operate as evaluative prosodies (Hood, 2010 ), whereby one Judgement resource recruits another to consolidate stance and shift footing—from an apparently technical competence assessment to moral or authenticity condemnation. In interactional terms, this escalation provides a portable template for attack sequences, enabling users to move from local critique to categorical exclusion in one turn or across turns. Ex. 5.1c (Ayuso, ES) (tweet_id: Fran_Toro_2025-04-10_12:12:55) Original (ES) “@user Cobarde , sinvergüenza , te largas a Ecuador…” Translation (EN) “ Coward , shameless , you’re off to Ecuador…” Reading: Capacity− ( cobarde ) recruiting Propriety− ( sinvergüenza ). Ex. 5.1d (Meloni, IT) (tweet_id: MEL_18_2025_0049) Original (IT) “@GiorgiaMeloni Cara fascistona falsa e bugiarda .” Translation (EN) “Dear big fascist, fake and a liar .” Reading: Veracity− ( falsa, bugiarda ) recruited alongside Capacity− (leadership fitness). These results also refine theories of delegitimisation (Reisigl & Wodak, 2001 ; Van Leeuwen, 2007 ) by demonstrating that the undermining of authority in digital environments is not a matter of single evaluative acts but of recurrent configurational patterns. What is delegitimised is not just a leader's individual action but their broader eligibility to hold authority, achieved through the patterned layering of Judgement categories. Such findings underscore the communicational function of hostility: it organises alignment among users by drawing on shared evaluative repertoires, making the discourse legible and resonant in wider publics. Finally, the cross-linguistic contrasts in distribution (Propriety-/Normalitydominance in Spanish vs. Capacity-Neracity- in Italian) demonstrate that evaluative repertoires are culturally inflected. This aligns with Thompson and AlbaJuez’s (2014) claim that while the semantic system of Judgement is stable across languages, its deployment patterns vary according to cultural norms of political critique. In our data, Spanish users more readily mobilise abnormalisation and infantilisation as delegitimising resources, as in Ex. 5.1e, while Italian users foreground authenticity and competence challenges, often coupling abnormalisation with veracity attacks, as in Ex. 5.1f. These contrasts suggest that digital political communication not only reflects but amplifies culturally specific models of what counts as legitimate leadership. These repertoires also differ in how they naturalise gendered delegitimisation. Spanish abnormalisation/infantilisation frames women as intrinsically unsuited to politics, whereas Italian competence/authenticity critiques cast exclusion as ostensibly rational performance evaluation. Ex. 5.1e (Ayuso, ES) (tweet_id: FerminYugueros_2025-04-11_07:12:34) Original (ES) “@user Eres ridícula a límites insospechados.” Translation (EN) “You are ridiculous beyond belief.” Reading: Normality− (abnormalisation/infantilisation). Ex. 5.1f (Meloni, IT) (tweet_id: MEL_15_2025_0002) Original (IT) “@GiorgiaMeloni Ma smettila cessa carciofara bugiarda …” Translation (EN) “Oh stop it, ugly artichoke-peddler, liar …” Reading: Normality− (demeaning label) coupled with Veracity− ( bugiarda ). 5.2 RQ2. How do evaluative configurations function as resources for delegitimising stance performance? The modelling shows that Propriety- and Normality- are the most reliable predictors of identity-salient stance cues across the corpus, with an additional Capacity×Propriety compound effect (significant in Italian). In communicational terms, delegitimising force is intensified when technical incompetence is coupled with moral sanction, and when abnormalisation/infantilisation frames are invoked. These patterns are visible in the corpus excerpts: in Spanish, a Capacity-Propriety- bundle escalates from cowardice to shamelessness (Ex. 5.2a) and a stand-alone Normality- evaluation targets behavioural deviance (Ex. 5.2b). In Italian, Capacity×Propriety compounds crystallise incompetence together with corruption (Ex. 5.2c), and Normality- invocations (often cooccurring with sanction) act as direct identity-salient triggers (Ex. 5.2d). These examples instantiate the configurational operation of evaluative meaning documented in the statistical results. Operationally, these configurations function as reusable attack scripts: once competence and sanction are paired, subsequent turns can repeat or vary the pairing with minimal linguistic work, sustaining a stable delegitimising frame across replies and threads. Ex. 5.2a (Ayuso, ES; Capacity− → Propriety−) (tweet_id: Fran_Toro_2025-04-10_12:12:55) Original (ES) “@user Cobarde , sinvergüenza , te largas a Ecuador…” Translation (EN) “ Coward , shameless , you’re off to Ecuador…” Reading: Capacity− ( cobarde , lack of resolve) recruiting Propriety− ( sinvergüenza ). Ex. 5.2b (Ayuso, ES; Normality−) (tweet_id: FerminYugueros_2025-04-11_07:12:34) Original (ES) “@user Eres ridícula a límites insospechados.” Translation (EN) “You are ridiculous beyond belief.” Reading: Normality− (abnormalisation/infantilisation). Ex. 5.2c (Meloni, IT; Capacity×Propriety) (tweet_id: MEL_15_2025_0058) Original (IT) “@GiorgiaMeloni … patriota? Incapace e corrotta .” Translation (EN) “Patriot? Incompetent and corrupt .” Reading: Capacity− × Propriety − compound (configuration significant in IT models). Ex. 5.2d (Meloni, IT; Normality− ± Propriety−) (tweet_id: MEL_15_2025_0041) Original (IT) “@GiorgiaMeloni VERGOGNATI FASCIO…!” Translation (EN) “ Shame on you , fascist…!” Reading: Normality− (abnormalisation via vergognati ) with Propriety− (moral sanctioning epithet). In interactional terms, such scripts exhibit high uptake value: users can cite, echo, or minimally edit prior formulations (e.g., swapping "incapace" for "fallito") while preserving the compound stance, which facilitates cross-turn propagation in hostile publics. The significance of Normality- is particularly notable. While often under-analysed in Appraisal Theory, this category captures accusations of being abnormal, inappropriate, or lacking conventional comportment. Its predictive weight in both corpora suggests that abnormalisation is a salient pathway for identity-salient delegitimation: women leaders are constructed as "out of place," thereby undermining their legitimacy (see Ex. 5.2b, 5.2d). Critically, Normality- provides a shortcut to identity-salience: charges of being "out of place" render performance failings constitutive, not contingent. This converts what might appear as episodic critique into status-based exclusion, aligning with indexical accounts of stance and personhood. This finding supports recent observations that gendered aggression frequently operates through the language of behavioural deviance and infantilisation (Esposito & Breeze, 2022 ; Southern & Harmer, 2021 ). The Capacity×Propriety interaction confirms that delegitimisation is most powerful when technical incompetence and moral failure are combined. The odds ratios show that such combinations are over twice as likely to trigger identity-salient cues than either category alone. The interaction remains directionally stable under week fixed effects and in a 200reply non-keyword subsample, indicating it is not an artefact of temporal spikes or trigger words. Marginal effects show the largest step change occurs when both Capacity- and Propriety- are present, consistent with a threshold (stacking) dynamic rather than linear additivity. This supports the theoretical claim that evaluative meaning operates configurationally, with bundled resources producing emergent stances (Hood, 2010 ; Bednarek, 2008 ). From a communicational perspective, these compound configurations facilitate escalation: a competence critique ("unable to govern") readily recruits moral sanction ("corrupt"), thereby shifting the discourse from conditional critique to categorical exclusion (cf. Ex. 5.2a, 5.2c). Importantly, these results bridge appraisal theory with broader discourse studies of delegitimisation. They show that gendered hostility is not reducible to lexical items or overt insults but emerges from systematic co-patterning of evaluative resources. Delegitimisation thus functions as a discursive strategy that is recognisable to participants, portable across contexts, and resonant in digital publics. By grounding delegitimising force in configurations rather than isolated categories, this study demonstrates how evaluation operates as a communicational mechanism for stance alignment and authority contestation in online interaction. Taken together, this points to a template-based communicational mechanism shaped by platform affordances (brevity, reply threading): compact evaluative bundles travel efficiently, organise stance alignment, and scale delegitimisation beyond single turns. 5.3 RQ3. What Spanish-Italian contrasts appear in evaluative repertoires, and how do these patterns relate to identity-salient stance cues? The comparative analysis demonstrates that hostile replies to women leaders are structured by culturally differentiated evaluative repertoires. Spanish discourse was characterised by higher frequencies of Propriety- and Normality-, whereas Italian discourse foregrounded Capacity- and Veracity-. These distributional contrasts were reinforced by effect sizes (Cramér´s V ≥ .27 for Propriety- and Normality-), indicating substantive differences in the evaluative resources recruited across languages. In keeping with these distributional tendencies, Spanish examples commonly mobilise moral sanction and abnormalisation (see Ex. 5.Зa-b), whereas Italian examples foreground competence/authenticity challenges, often with abnormalising invective (see Ex. 5.3c-d). These contrasts indicate not only different inventories of evaluative resources but different mechanisms of portability: Spanish moral-sanction/abnormalisation pairings circulate as ready-made formulas, whereas Italian competence/authenticity pairings circulate as performance-assessment scripts. Ex. 5.3a (Ayuso, ES; Propriety− + Normality−) (tweet_id: PepBausset_2025-04-11_08:04:06) Original (ES) “@user Ayuso súper corrupta , a los tribunales POR FIN .” Translation (EN) “Ayuso super corrupt —to the courts AT LAST .” Reading: Propriety− ( corrupta ); Normality− (social deviance frame; graduation via capitals). Ex. 5.3b (Ayuso, ES; Normality − with infantilisation) (tweet_id: FerminYugueros_2025-04-11_07:12:34) Original (ES) “@user Eres ridícula a límites insospechados.” Translation (EN) “You are ridiculous beyond belief.” Reading: Normality− (abnormalisation/infantilisation). Ex. 5.3c (Meloni, IT; Capacity− + Veracity−, with sanctioning register) (tweet_id: MEL_18_2025_0049) Original (IT) “@GiorgiaMeloni Cara fascistona falsa e bugiarda .” Translation (EN) “Dear big fascist, fake and a liar .” Reading: Veracity− ( falsa, bugiarda ) with sanctioning address; Capacity/fitness implicitly impugned. Ex. 5.3d (Meloni, IT; Normality− + Veracity−) (tweet_id: MEL_15_2025_0002) Original (IT) “@GiorgiaMeloni Ma smettila cessa carciofara bugiarda …” Translation (EN) “Oh stop it, ugly artichoke-peddler, liar …” Reading: Normality− (demeaning label) + Veracity− ( bugiarda ), consistent with Italian authenticity/competence frames. In practice, these formulas exhibit high uptake value: users reissue them with minimal lexical editing (e.g., replacing falsa with ipocrita or intensifying corrotta with hashtags), preserving the same compound stance across turns and threads. In the predictive models, Spanish replies showed Propriety- as a key driver of identity-salient stance cues, whereas Italian replies highlighted Normality- and the Capacity×Propriety configuration. These findings indicate that different evaluative paths can lead to the same communicational outcome: delegitimisation of women leaders through identity-salient positioning. In Spanish contexts, hostile users mobilise moral sanction and abnormalisation, echoing cultural scripts of dishonesty and impropriety (Ex. 5.3a). In Italian contexts, users rely more on competence and authenticity critiques, sometimes intensified when paired with moral delegitimation (Ex. 5.3c). Accordingly, the pathway to identity-salience differs by language: in Spanish, Propriety- alone often suffices; in Italian, Normality- and Capacity×Propriety more commonly mark the tipping point. This suggests distinct thresholds for when evaluative talk becomes explicitly identity-salient. These cross-linguistic contrasts refine the claim that appraisal resources are universally available but locally patterned (Thompson & Alba-Juez, 2014 ). While the semantic system of Judgement categories is stable, their deployment reflects culturally specific models of political legitimacy. In Spain, where public scandals are routinely framed in terms of corruption and propriety, moral sanction provides a salient frame for gendered hostility (Ex. 5.3a). In Italy, where public debate often foregrounds questions of authenticity and performativity, competence and veracity become central evaluative pathways (Ex. 5.3c-d). This supports a repertoire-within-system view of Appraisal: the Judgement system is shared, but repertoire activation is locally patterned by public cultures of legitimacy and scandal. From a communicational perspective, these contrasts underscore how digital hostility embeds broader cultural repertoires into online interaction. Evaluative copatterns are not only individual choices but part of a shared resource system that participants draw upon to align stances and make attacks intelligible to their audiences. The recurrent pairings also act as audience-design shortcuts: moralsanction/abnormalisation in Spanish and competence/authenticity in Italian allow posters to index intended publics without elaboration. By showing that gendered delegitimisation is realised through different evaluative bundles across languages, this study contributes to comparative discourse research by situating online aggression within the cultural politics of evaluation. Finally, robustness checks confirm that these contrasts are not artefacts of keyword filtering or event-specific spikes. The 200 non-keyword subsample yielded substantively similar results, and week fixed effects did not alter the direction of key predictors. These checks strengthen the claim that the observed contrasts represent stable evaluative repertoires rather than sampling artefacts. As a boundary condition, cross-language differences are distributional and configurational rather than absolute; all categories are available in both corpora, but activation probabilities and co-pattern strengths differ. 5.4 Implications and Conclusion The analyses across RQ1-RQ3 demonstrate that hostile discourse towards women leaders on X is communicationally structured, configurational in its evaluative deployment, and culturally differentiated in its repertoires. Three overarching implications follow. Theoretically, by focusing on evaluative configurations rather than isolated categories, the study extends Appraisal Theory into a domain of configurational stance analysis. The finding that compound Judgement bundles (e.g., Capacity-+Propriety-) are especially likely to trigger identity-salient cues underscores the need to conceptualise evaluation as a relational system where resources recruit one another to produce emergent meanings. This perspective also refines theories of delegitimisation (Reisigl & Wodak, 2001 ; Van Leeuwen, 2007 ), showing that discursive authority is undermined not through single insults but through patterned co-occurrences that cumulatively erode legitimacy. Such patterned erosion illustrates how language functions communicationally to reframe individual failings as collective truths, enabling hostile publics to transform isolated utterances into a broader narrative of illegitimacy. Methodologically, the corpus-based, cross-linguistic design illustrates how digital discourse analysis can integrate annotation, statistical modelling, and cultural comparison. The robust replication of results in a non-keyword subsample and under temporal controls indicates that these patterns are not artefacts of data selection but represent stable evaluative repertoires. This supports the feasibility of using Appraisal-informed protocols for large-scale, multilingual annotation, even when inter-coder reliability is moderate due to the interpretive nature of evaluative language. Beyond linguistics, this methodological approach offers a template for computational social science, political communication, and psychology, where questions of hostility, stance, and legitimacy similarly hinge on the patterned aggregation of microevaluations. From the communicational perspective, hostility directed at women leaders functions as a public stance practice that draws on shared repertoires of evaluation. In Spain, moral sanction and abnormalisation dominate; in Italy, competence and authenticity critiques prevail. These contrasts show that gendered aggression is not only linguistic but also culturally anchored, reflecting national models of political legitimacy. In both contexts, however, hostility achieves the same communicational outcome: delegitimisation through identity-salient positioning that questions not just performance but the right to lead. Together, these findings demonstrate that digital political hostility is best understood as an organised communicational practice rather than incidental incivility. Our contribution lies in linking the microanalysis of evaluative resources to broader questions of authority, identity, and cultural repertoire. This intersection underscores how linguistic choices function within wider communicational ecologies, integrating insights from discourse analysis, sociolinguistics, and political communication. In doing so, the study contributes to a common theoretical framework for understanding digital hostility, where insights from rhetoric, pragmatics, and semiotics converge on the analysis of how communicational practices shape authority and identity in networked publics. Future research can extend this work by examining multimodal contributions (e.g., memes, images) within the same appraisal framework, tracing cross-platform dynamics, or exploring how male politicians are evaluated using comparable repertoires. Such extensions would test the generalisability of configurational stance analysis and further situate digital hostility within the interdisciplinary study of communication. Ethics declaration This research analyses publicly accessible posts from X (Twitter). No direct interaction with human participants took place, and no private or identifiable data were collected. According to the policies of the Universitat Politècnica de València, formal ethical approval was not required as the study uses only publicly available data. The research complies with institutional guidelines and the platform's terms of service. Declarations Author Contribution S.S. designed the study, built the corpus, developed the annotation protocol, performed statistical analyses/ visualisations, and drafted the manuscript. G.G. curated the Italian dataset, validated annotations, and contributed to analysis and writing. M.L.C.-P. and G.G. jointly supervised the work and provided critical revisions. All authors approved the final manuscript and agree to be accountable for all aspects of the work. Data Availability The study analyses public posts from X (Twitter). Full text cannot be redistributed under the platform’s Terms of Service. The tweet/reply ID lists, derived annotation files (Judgement labels, polarity, salience), and rehydration/analysis code are available from the corresponding author on reasonable request for research use. References Agha, A. (2006). Language and Social Relations (1st ed.). Cambridge University Press. https://doi.org/10.1017/CBO9780511618284 Bednarek, M. (2008). Emotion Talk Across Corpora. Palgrave Macmillan UK . https://doi.org/10.1057/9780230285712 Bednarek, M., & Caple, H. (2017). The Discourse of News Values: How News Organizations Create Newsworthiness . https://www.amazon.com/Discourse-News-Values-Organizations-Newsworthiness/dp/0190653949 Combei, C. R., & Reggi, V. (2024). Appraisal, sentiment and emotion analysis in political discourse: A multimodal, multi-method approach . Routledge. del Saz-Rubio, M. M. (2024). You are not empowered, you have neither character nor pride’: Assessing aggressive language against Spanish female politicians in high-profile positions. Journal of Language Aggression and Conflict . https://doi.org/10.1075/jlac.00109.del Du Bois, J. W. (2007). The stance triangle. In R. Englebretson (Ed.), Pragmatics & Beyond New Series (Vol. 164, pp. 139–182). John Benjamins Publishing Company. https://doi.org/10.1075/pbns.164.07du Erikson, J., Håkansson, S., & Josefsson, C. (2023). Three Dimensions of Gendered Online Abuse: Analyzing Swedish MPs’ Experiences of Social Media. Perspectives on Politics , 21 (3), 896–912. https://doi.org/10.1017/S1537592721002048 Esposito, E., & Breeze, R. (2022). Gender and politics in a digitalised world: Investigating online hostility against UK female MPs. Discourse & Society , 33 (3), 303–323. https://doi.org/10.1177/09579265221076608 Fuoli, M. (2018). A stepwise method for annotating: Appraisal. Functions of Language , 25 , 229–258. https://doi.org/10.1075/fol.15016.fuo Habermas, J., & Rehg, W. (2010). Between Facts and Norms: Contributions to a Discourse Theory of Law and Democracy . Polity. Reprinted. Håkansson, S. (2021). Do Women Pay a Higher Price for Power? Gender Bias in Political Violence in Sweden. The Journal of Politics , 83 (2), 515–531. https://doi.org/10.1086/709838 Håkansson, S. (2024). The Gendered Representational Costs of Violence against Politicians. Perspectives on Politics , 22 (1), 81–96. https://doi.org/10.1017/S1537592723001913 Hayes, A. F., & Krippendorff, K. (2007). Answering the Call for a Standard Reliability Measure for Coding Data. Communication Methods and Measures , 1 (1), 77–89. https://doi.org/10.1080/19312450709336664 Hood, S. (2010). Appraising research: Evaluation in academic writing (1. publ). Palgrave Macmillan. Jane, E. A. (2014). Back to the kitchen, cunt’: Speaking the unspeakable about online misogyny. Continuum , 28 (4), 558–570. https://doi.org/10.1080/10304312.2014.924479 KhosraviNik, M. (Ed.). (2023). Social media critical discourse studies . Routledge. Laclau, E. (2005). On populist reason (1. publ). Verso. Mantilla, K. (2015). Gendertrolling: How misogyny went viral. Praeger . https://doi.org/10.5040/9798400656026 Martin, J. R., & White, P. R. R. (2005). The Language of Evaluation. Palgrave Macmillan UK . https://doi.org/10.1057/9780230511910 Meriläinen, J. M. (2024). The Role of Gender in Hate Speech Targeting Politicians: Evidence from Finnish Twitter. International Journal of Politics Culture and Society . https://doi.org/10.1007/s10767-024-09476-3 Moffitt, B. (2016). The global rise of populism: Performance, political style, and representation . Stanford University Press. Moreno-Ortiz, A., & García-Gámez, M. (2023). Strategies for the Analysis of Large Social Media Corpora: Sampling and Keyword Extraction Methods. Corpus Pragmatics , 7 (3), 241–265. https://doi.org/10.1007/s41701-023-00143-0 Mouffe, C. (2011). On the Political (0 edn). Routledge. https://doi.org/10.4324/9780203870112 Norris, P. (2011). Democratic deficit: Critical citizens revisited . Cambridge University Press. Ochs, E. (1992). Indexing gender. In A. Duranti, & C. Goodwin (Eds.), Rethinking Context: Language as an Interactive Phenomenon (pp. 335–358). Cambridge University Press. Papacharissi, Z. (2014). Affective Publics: Sentiment, Technology, and Politics . Oxford University Press. https://doi.org/10.1093/acprof:oso/9780199999736.001.0001 Pierri, F. (2024). Drivers of hate speech in political conversations on Twitter: The case of the 2022 Italian general election. EPJ Data Science , 13 (1), 63. https://doi.org/10.1140/epjds/s13688-024-00501-1 Reisigl, M., & Wodak, R. (2001). Discourse and discrimination: Rhetorics of racism and antisemitism . Routledge. Silverstein, M. (2003). Indexical order and the dialectics of sociolinguistic life. Language & Communication , 23 (3–4), 193–229. https://doi.org/10.1016/S0271-5309(03)00013-2 Southern, R., & Harmer, E. (2021). Twitter, Incivility and Everyday Gendered Othering: An Analysis of Tweets Sent to UK Members of Parliament. Social Science Computer Review , 39 (2), 259–275. https://doi.org/10.1177/0894439319865519 Thompson, G., & Alba-Juez, L. (Eds.). (2014). Evaluation in Context (Vol. 242). John Benjamins Publishing Company. https://doi.org/10.1075/pbns.242 Trifiro, B. M., Paik, S., Fang, Z., & Zhang, L. (2021). Politics and Politeness: Analysis of Incivility on Twitter During the 2020 Democratic Presidential Primary. Social Media + Society , 7 (3), 20563051211036939. https://doi.org/10.1177/20563051211036939 Van Leeuwen, T. (2007). Legitimation in discourse and communication. Discourse & Communication , 1 (1), 91–112. https://doi.org/10.1177/1750481307071986 Zappavigna, M. (2012). The discourse of Twitter and social media . Continuum International Pub. Group. Zappavigna, M. (2018). Searchable talk: Hashtags and social media metadiscourse . Bloomsbury Academic, an imprint of Bloomsbury Plc. Additional Declarations No competing interests reported. Cite Share Download PDF Status: Published Journal Publication published 18 Dec, 2025 Read the published version in Corpus Pragmatics → Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7820069","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":533890310,"identity":"c7272352-9629-449e-8994-db83f9573664","order_by":0,"name":"Sergei Sikorskii","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAsElEQVRIiWNgGAWjYFACxmYQKUe6FmOSrGEGEYkNRKuX7z/cbPBxT136hts9Bgwf/hChxeBGYnPijGeHczfcOWPAOLONGC0SjM2HeQ4cyN1wIy2BmZcY58n3HwRpqUs3AGn5Q4zDGA4kNifzHGBOMLiRfICZgY0YhwH9YjjjwGHDmXcOHzjYS4xf5PuPP5b4cKBOnu92Y+ODH0Q5DA4kgI4kSQNYyygYBaNgFIwCrAAAyAw7atFfRskAAAAASUVORK5CYII=","orcid":"","institution":"Universitat Politècnica de València","correspondingAuthor":true,"prefix":"","firstName":"Sergei","middleName":"","lastName":"Sikorskii","suffix":""},{"id":533890311,"identity":"934aec3a-4246-447f-9e46-c43a1d4c556d","order_by":1,"name":"Maria Luisa Carrió-Pastor","email":"","orcid":"","institution":"Universitat Politècnica de València","correspondingAuthor":false,"prefix":"","firstName":"Maria","middleName":"Luisa","lastName":"Carrió-Pastor","suffix":""},{"id":533890312,"identity":"4724a181-4904-49ca-8249-b55075e6567e","order_by":2,"name":"Giovanni Garofalo","email":"","orcid":"","institution":"University of Bergamo","correspondingAuthor":false,"prefix":"","firstName":"Giovanni","middleName":"","lastName":"Garofalo","suffix":""}],"badges":[],"createdAt":"2025-10-09 17:38:13","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7820069/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7820069/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1007/s41701-025-00221-5","type":"published","date":"2025-12-18T15:58:19+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":94454632,"identity":"1c8056da-2dff-45f0-89aa-7ba1a70fb58d","added_by":"auto","created_at":"2025-10-27 14:43:32","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":511479,"visible":true,"origin":"","legend":"","description":"","filename":"RevisedManuscriptv3.docx","url":"https://assets-eu.researchsquare.com/files/rs-7820069/v1/5469b6fdbf4228b7fcd695ac.docx"},{"id":94456704,"identity":"5fe75b39-8238-4f0b-9820-708550ee7674","added_by":"auto","created_at":"2025-10-27 14:45:02","extension":"json","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":5063,"visible":true,"origin":"","legend":"","description":"","filename":"2c859401923a4a988443eeb24b83ba9f.json","url":"https://assets-eu.researchsquare.com/files/rs-7820069/v1/079839f2da034fcfd761a6ac.json"},{"id":94456618,"identity":"f5a8be70-51fe-4a0a-8259-e660c64b3a86","added_by":"auto","created_at":"2025-10-27 14:44:51","extension":"xml","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":148648,"visible":true,"origin":"","legend":"","description":"","filename":"2c859401923a4a988443eeb24b83ba9f1enriched.xml","url":"https://assets-eu.researchsquare.com/files/rs-7820069/v1/73efc0f054c2f55b10dd9141.xml"},{"id":94456703,"identity":"1e9592a8-136b-4f01-80c6-aedf82465b41","added_by":"auto","created_at":"2025-10-27 14:45:01","extension":"png","order_by":3,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":78338,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-7820069/v1/04ba404eb251b72aa46141b0.png"},{"id":94456587,"identity":"e8ed67f4-e353-4232-bfe2-a11cc99d3bda","added_by":"auto","created_at":"2025-10-27 14:44:43","extension":"png","order_by":4,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":90913,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-7820069/v1/5ae80b35d188edb2e3bc4bb4.png"},{"id":94455801,"identity":"454dc8d4-6c99-4e7e-a704-94500a21179f","added_by":"auto","created_at":"2025-10-27 14:43:57","extension":"jpeg","order_by":5,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":19837,"visible":true,"origin":"","legend":"","description":"","filename":"groupimage1.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7820069/v1/83acb27c9dc59d0a41a4f309.jpeg"},{"id":94456627,"identity":"3f3a3283-6567-4cf0-97ae-2f08abde05d4","added_by":"auto","created_at":"2025-10-27 14:44:52","extension":"png","order_by":6,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":19990,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-7820069/v1/386fa2f3a2cf80920bd0f763.png"},{"id":94455887,"identity":"71d88f98-1b9b-4c56-b2f1-f829f6c34cf3","added_by":"auto","created_at":"2025-10-27 14:44:01","extension":"png","order_by":7,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":23196,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-7820069/v1/bc1ce848377abb4ff1f40135.png"},{"id":94456472,"identity":"201ffc13-72b7-46d3-b9d9-9f43a245aa8a","added_by":"auto","created_at":"2025-10-27 14:44:32","extension":"png","order_by":8,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":5638,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinegroupimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-7820069/v1/865c30ed2a312362e4f22210.png"},{"id":94456591,"identity":"685c349f-49a7-4095-b453-5a454fc723f4","added_by":"auto","created_at":"2025-10-27 14:44:44","extension":"xml","order_by":9,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":145722,"visible":true,"origin":"","legend":"","description":"","filename":"2c859401923a4a988443eeb24b83ba9f1structuring.xml","url":"https://assets-eu.researchsquare.com/files/rs-7820069/v1/533dbcf613bccd6fa56eb37d.xml"},{"id":94455758,"identity":"eaccafd6-20fc-4c5e-a8ff-a13fc3f78431","added_by":"auto","created_at":"2025-10-27 14:43:56","extension":"html","order_by":10,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":158349,"visible":true,"origin":"","legend":"","description":"","filename":"earlyproof.html","url":"https://assets-eu.researchsquare.com/files/rs-7820069/v1/793d743e82e25eefa764825e.html"},{"id":94456317,"identity":"49b4e6c4-9d30-4955-8a65-5b00b95016ea","added_by":"auto","created_at":"2025-10-27 14:44:25","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":46206,"visible":true,"origin":"","legend":"\u003cp\u003eJudgement prevalence by leader with 95% confidence intervals\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-7820069/v1/75075857cfa3106258778cc6.png"},{"id":94456702,"identity":"3d58b793-c117-4492-b724-c5cb98206265","added_by":"auto","created_at":"2025-10-27 14:45:01","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":56600,"visible":true,"origin":"","legend":"\u003cp\u003eForest plot of odds ratios for identity-salient stance predictors\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-7820069/v1/f1b20ff51bbd6482ed5048f3.png"},{"id":94456128,"identity":"5c26db95-bf00-40d8-936b-aec158d44109","added_by":"auto","created_at":"2025-10-27 14:44:12","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":72480,"visible":true,"origin":"","legend":"\u003cp\u003eInteraction effects: Capacity×Propriety combinations by leader\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-7820069/v1/fe7ffb54496fb7b966578c8d.png"},{"id":98814121,"identity":"3e2f92e9-1609-4854-bc18-d28f413b339e","added_by":"auto","created_at":"2025-12-22 16:11:23","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2561745,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7820069/v1/4eeb210c-df8f-4180-9e59-111e4a985260.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Cross-linguistic Delegitimisation of Women Leaders in Online Political Discourse","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003eContemporary digital political communication reveals systematic patterns in how hostile discourse targets women leaders. When political actors engage on platforms like X, women face attacks that are not merely random incivility but patterned through recurring evaluative strategies that question competence, integrity, and legitimacy. Such communicative practices function as delegitimising mechanisms that may discourage participation in public debate (Esposito \u0026amp; Breeze, \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2022\u003c/span\u003e; Southern \u0026amp; Harmer, \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e2021\u003c/span\u003e), while men politicians, though also criticised, tend to face abuse less systematically tied to identity markers (Erikson et al., \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e2023\u003c/span\u003e; Meril\u0026auml;inen, \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e2024\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eCross-national evidence confirms these dynamics across contexts. In the UK, women MPs encounter both overt abuse and \"everyday\" microaggressions that frame them as unqualified or inappropriate for leadership roles (Southern \u0026amp; Harmer, \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). Nordic studies demonstrate that while overall volumes of abuse may not differ by gender, its quality does: women politicians face more personalised, identity-focused attacks, particularly when occupying portfolios traditionally coded as \"masculine\" (Erikson et al., \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e2023\u003c/span\u003e; H\u0026aring;kansson, \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2021\u003c/span\u003e, \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; Meril\u0026auml;inen, \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). Recent analyses of Spanish political discourse show that women leaders like Isabel D\u0026iacute;az Ayuso tends toward moral condemnation/infantilisation, while Italian studies of Giorgia Meloni reveal patterns focused on competence and authenticity (Combei \u0026amp; Reggi, \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; del Saz-Rubio, \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). These patterns suggest that hostile evaluations embed gendered assumptions within apparently neutral assessments of political performance.\u003c/p\u003e\u003cp\u003eFollowing (Du Bois, \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e2007\u003c/span\u003e), we treat evaluations as components of stance (a configuration of evaluation, positioning, and alignment) whose social meanings emerge through indexical processes (Agha, \u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e2006\u003c/span\u003e; Ochs, \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e1992\u003c/span\u003e; Silverstein, \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e2003\u003c/span\u003e). Evaluative language thus functions not merely as opinion expression but as a resource for public positioning: critics who portray a woman leader as 'unable to cope with pressure' or recast policy disagreement as moral failure mobilise evaluations that appear to address performance while potentially invoking identity-based assumptions about legitimacy. We examine how such identity-salient linguistic markers (evaluative patterns that may index social positioning beyond policy critique) cluster with specific Judgement configurations, drawing on work on delegitimisation and legitimation in political discourse (Reisigl \u0026amp; Wodak, \u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e2001\u003c/span\u003e; Van Leeuwen, \u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e2007\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eThis study examines hostile replies on X addressed to Giorgia Meloni (Italy) and Isabel D\u0026iacute;az Ayuso (Spain) over a 40-day window, using the Judgement subsystem of Appraisal Theory to identify systematic co-patterns of evaluation. Rather than treating categories in isolation, we analyse their co-occurrence (e.g., Capacity and Propriety) as compact configurations that index delegitimising stances (Bednarek, \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2008\u003c/span\u003e; Hood, \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e2010\u003c/span\u003e; Martin \u0026amp; White, \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2005\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eThe comparative dimension addresses how evaluative strategies vary across linguistic and contexts. Spanish and Italian replies to these leaders reveal distinct emphases: Spanish discourse tends toward moral sanctioning and behavioural evaluation, while Italian discourse foregrounds competence and authenticity assessment. However, rather than proposing universal cultural mechanisms, we interpret these as contrastive repertoires emergent within our corpus (Thompson \u0026amp; Alba-Juez, \u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e2014\u003c/span\u003e).\u003c/p\u003e\u003cp\u003e\u003cb\u003eResearch questions\u003c/b\u003e\u003c/p\u003e\u003cp\u003e\u003col\u003e\u003cspan\u003e\u003cli\u003e\u003cp\u003eHow do Judgement categories distribute and co-pattern in hostile replies on X to two women leaders across Spanish and Italian?\u003c/p\u003e\u003c/li\u003e\u003c/span\u003e\u003cspan\u003e\u003cli\u003e\u003cp\u003eWhich evaluative configurations cluster with identity-salient linguistic markers in this corpus?\u003c/p\u003e\u003c/li\u003e\u003c/span\u003e\u003cspan\u003e\u003cli\u003e\u003cp\u003eWhat Spanish-Italian contrasts appear in evaluative repertoires, and how do these patterns relate to identity-salient stance cues?\u003c/p\u003e\u003c/li\u003e\u003c/span\u003e\u003c/ol\u003e\u003c/p\u003e\u003cp\u003eOur contribution lies in specifying how evaluative co-patterns function as delegitimisation mechanisms in digital political discourse. We develop an operational protocol for identifying identity-salient stance cues through Judgement category analysis, with empirical claims confined to our dataset of 2,000 annotated replies. By examining evaluative configurations rather than isolated categories, this study advances understanding of how linguistic resources index social positioning in antagonistic digital interaction.\u003c/p\u003e"},{"header":"2. Theoretical Background","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e\u003ch2\u003e2.1. Evaluative Language as Stance Performance\u003c/h2\u003e\u003cp\u003eStance encompasses the simultaneous acts of evaluation, positioning, and alignment that speakers perform through linguistic choices (Du Bois, \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e2007\u003c/span\u003e). In the stance triangle, evaluation involves speakers' subjective assessments of objects or propositions, positioning establishes the speaker's relationship to the evaluation, and alignment creates intersubjective relationships between speakers. This framework moves beyond treating evaluation as isolated opinion expression toward understanding how evaluative choices position speakers socially and create relationships with audiences.\u003c/p\u003e\u003cp\u003eThe social meanings of evaluative language emerge through indexical processes whereby linguistic forms point to social categories, relationships, and identities (Silverstein, \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e2003\u003c/span\u003e). Indexicality operates across multiple orders: direct indexical relationships (form-meaning pairings) become embedded within higher-order indexical fields that connect linguistic features to social personae and cultural models (Agha, 2007). When speakers evaluate political figures, their linguistic choices do not merely convey opinions but index stances toward authority, competence, and legitimacy that carry broader social implications.\u003c/p\u003e\u003cp\u003eIndexical processes are particularly relevant for understanding how gender emerges in discourse. Rather than treating gender as a stable variable, sociolinguistic research demonstrates that gender identity is constructed through iterative linguistic practices that index cultural models of femininity and masculinity (Ochs, \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e1992\u003c/span\u003e). Evaluative language can indirectly index gender through stances, affects, and social activities that are culturally associated with gendered identities, creating what Ochs terms \"constitutive indexicality\" where repeated linguistic practices contribute to identity construction.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec4\" class=\"Section2\"\u003e\u003ch2\u003e2.2. Appraisal Theory and Evaluative Configurations\u003c/h2\u003e\u003cp\u003eAppraisal Theory provides a systematic framework for analysing evaluative meaning in discourse, with the Judgement subsystem specifically addressing evaluations of human behaviour and character (Martin and White, \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2005\u003c/span\u003e). Judgement categories capture two broad evaluative domains: Social Esteem (assessments of normality, capacity, and tenacity) and Social Sanction (assessments of veracity and propriety). These categories function as resources for stance-taking rather than fixed semantic labels, allowing speakers to position themselves and their addressees relative to social norms and expectations.\u003c/p\u003e\u003cp\u003eAppraisal research has increasingly emphasised evaluative configurations\u0026mdash; patterns of co-occurring evaluative meanings that create coherent attitudinal positions (Hood, \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e2010\u003c/span\u003e). Rather than analysing isolated evaluative instances, this approach examines how multiple Appraisal resources combine to create \"prosodies\" of evaluation that extend across texts and social interactions. Such configurations allow for more nuanced stance performances where, for example, capacity assessments combine with propriety judgements to create compound delegitimising effects.\u003c/p\u003e\u003cp\u003eCross-linguistic applications of Appraisal Theory reveal both universal and language-specific patterns in evaluative resource deployment (Thompson \u0026amp; Alba-Juez, \u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e2014\u003c/span\u003e). While the semantic domains of Judgement appear cross-linguistically stable, their realisation patterns, co-occurrence tendencies, and social implications vary across cultural and linguistic contexts. This variation reflects differences in cultural values, social hierarchies, and communicative norms that shape how evaluative meanings are interpreted and deployed.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec5\" class=\"Section2\"\u003e\u003ch2\u003e2.3. Delegitimisation and Identity-Salient Discourse\u003c/h2\u003e\u003cp\u003eDelegitimisation operates through discourse strategies that undermine the authority, credibility, or appropriateness of political actors (Reisigl and Wodak, \u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e2001\u003c/span\u003e). These strategies typically involve negative categorisation, attribution of negative qualities, and questioning of competence or moral standing. Van Leeuwen's (2007) analysis of legitimation strategies reveals how discourse constructs and challenges authority through appeals to different value systems, with particular attention to how evaluative language indexes relationships between power and social acceptance.\u003c/p\u003e\u003cp\u003eIdentity-salient delegitimisation occurs when attacks on political figures deploy evaluative strategies that invoke cultural stereotypes or expectations associated with social identities. This process differs from general political criticism by embedding identity-based assumptions within apparently neutral performance assessments. For instance, questioning a woman politician's emotional stability or decision-making capacity may invoke cultural models of gender and leadership while ostensibly addressing professional competence.\u003c/p\u003e\u003cp\u003eThe digital context intensifies these dynamics through platform affordances that prioritise engagement and emotional response (Papacharissi, \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e2014\u003c/span\u003e). Social media platforms create \"affective publics\" where emotional expressions and evaluative stances circulate rapidly and gain visibility through algorithmic amplification. Within these environments, delegitimising discourse can achieve broader reach and impact while maintaining the appearance of spontaneous user response rather than coordinated attack.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec6\" class=\"Section2\"\u003e\u003ch2\u003e2.4. Cross-Linguistic Evaluation and Digital Discourse\u003c/h2\u003e\u003cp\u003eComparative analysis of evaluative discourse reveals how cultural repertoires shape the deployment of linguistic resources across different contexts (Thompson and Alba-Juez, \u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e2014\u003c/span\u003e). These repertoires encompass not only lexical choices but also structural patterns, pragmatic conventions, and indexical associations that vary across linguistic communities. Cross-linguistic evaluation research demonstrates that while semantic categories of evaluation may be universal, their social meanings and deployment patterns reflect cultural values and communicative norms.\u003c/p\u003e\u003cp\u003eDigital platforms introduce additional complexity by creating spaces where multiple linguistic and cultural repertoires intersect while operating under shared technological constraints (Zappavigna, \u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e2012\u003c/span\u003e). Platform features such as character limits, threading structures, and visibility algorithms shape how evaluative meanings can be expressed and circulated, potentially favouring certain types of evaluative strategies over others. The resulting discourse represents a hybrid of traditional evaluative repertoires and platform-specific adaptations.\u003c/p\u003e\u003cp\u003eResearch on Twitter discourse reveals how platform constraints influence evaluative expression through compression, hashtag use, and intertextual referencing (Zappavigna, \u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e2018\u003c/span\u003e). These adaptations create new forms of evaluative practice while drawing on existing cultural resources for stance-taking and social positioning. Understanding digital evaluative discourse thus requires attention to both linguistic repertoires and technological mediation of social interaction.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec7\" class=\"Section2\"\u003e\u003ch2\u003e2.5. Analytical Framework: From Categories to Configurations\u003c/h2\u003e\u003cp\u003eOur analytical approach treats Judgement categories as resources for stance performance rather than discrete semantic labels. We focus on evaluative configurations, systematic co-occurrence patterns such as Capacity-Propriety combinations, that create compound evaluative effects beyond individual category meanings. These configurations function as compact strategies for positioning social actors within evaluative frameworks that index broader cultural models of competence, authority, and legitimacy.\u003c/p\u003e\u003cp\u003eThe identity-salient dimension of our analysis examines how evaluative configurations may invoke cultural repertoires associated with social identities. Rather than assuming direct gender indexing, we analyse patterns that cluster with linguistic markers potentially associated with identity-based evaluation, such as infantilising language, appearance focus, or emotional stereotyping. This approach allows for empirical identification of identity-salient patterns without presuming their cultural significance.\u003c/p\u003e\u003cp\u003eCross-linguistic comparison focuses on contrastive repertoires within our corpus rather than universal mechanisms. Spanish and Italian evaluative patterns are analysed as emergent configurations specific to responses to these particular political figures rather than as representative of broader cultural systems. This approach acknowledges the limitations of our data while providing insight into how evaluative resources operate across different linguistic contexts within the constraints of digital political discourse.\u003c/p\u003e\u003c/div\u003e"},{"header":"3. Methods","content":"\u003cp\u003eThis study employs a mixed-methods approach combining corpus linguistics, discourse analysis, and statistical methods to examine evaluative configurations in hostile digital political discourse. The methodology integrates theoretical insights from stance analysis and Appraisal Theory to identify systematic patterns in how Judgement resources index delegitimising positions across Spanish and Italian contexts.\u003c/p\u003e\u003cdiv id=\"Sec9\" class=\"Section2\"\u003e\u003ch2\u003e3.1. Research Design and Corpus Construction\u003c/h2\u003e\u003cp\u003eWe used a two-stage design to balance coverage and analytic depth. Stage 1 built the sampling frame of candidate replies; Stage 2 drew stratified analytical subcorpora.\u003c/p\u003e\u003cp\u003eStage 1.We collected reply objects on X between 1 April and 9 May 2025 using leader-specific queries that combined name variants and verified handles (e.g., \"Isabel D\u0026iacute;az Ayuso\" OR @IdiazAyuso; \"Giorgia Meloni\" OR @GiorgiaMeloni), with language filters (lang:es, lang:it). We excluded retweets and retained posts with \u0026ldquo;is_reply\u0026rdquo;; quote tweets were included but flagged as a subtype. After de-duplication and light hygiene (removal of exact duplicates, promotional spam patterns, and off-language leakage identified via spot checks), the sampling frame comprised 88,872 unique replies (Ayuso 42,950; Meloni 45,922). We preserved conversation IDs and source-post IDs to maintain microcontexts and recorded engagement at collection time (likes/retweets) binned into low/medium/high for later modelling.\u003c/p\u003e\u003cp\u003eStage 2. From this sampling frame we sampled 2,000 replies (1,000 per leader). Following social-media corpus guidelines (Moreno-Ortiz \u0026amp; Garc\u0026iacute;a-G\u0026aacute;mez, \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2023\u003c/span\u003e), we combined a lexicon-triggered stratum with a 10% random stratum to balance evaluative density with coverage. The lexicon targeted replies likely to contain Judgement resources or incivility, including competence \u003cem\u003e(incapaz/incompetente; incapace/inetta)\u003c/em\u003e, morality \u003cem\u003e(corrupta/mentirosa; corrotta/bugiarda)\u003c/em\u003e, and behavioural characterisation \u003cem\u003e(hist\u0026eacute;rica/d\u0026eacute;bil; isterica/debole)\u003c/em\u003e. Sampling was stratified along three axes: (i) time (uniform across the window to cover distinct events), (i) microcontext (multiple source posts per leader; cap on replies per thread to prevent thread dominance), and (iii) post subtype (standard replies vs quote tweets, analysed with a subtype flag). With n\u0026thinsp;=\u0026thinsp;1,000 per leader, proportion estimates carry a\u0026thinsp;\u0026asymp;\u0026thinsp;\u0026plusmn;\u0026thinsp;3pp 95% margin of error at p\u0026thinsp;=\u0026thinsp;.5.\u003c/p\u003e\u003cp\u003eThe final analytical corpus comprised 2,000 replies (1,000 per politician) selected through stratified sampling across three dimensions: temporal distribution to capture responses to different political events, conversational diversity to prevent single-thread dominance, and post-type variation to include both original posts and replies. This sampling framework follows recommendations for maintaining discursive diversity while enabling systematic comparative analysis (KhosraviNik, 2017).\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec10\" class=\"Section2\"\u003e\u003ch2\u003e3.2. Operationalising Identity-Salient Evaluation\u003c/h2\u003e\u003cp\u003eIdentity-salient evaluation was operationalised through a systematic protocol targeting linguistic patterns that potentially invoke cultural repertoires associated with social identity categories. Rather than assuming direct indexical relationships, we developed detection criteria based on convergent evidence from gender and political discourse research (Jane, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2014\u003c/span\u003e; Mantilla, \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e2015\u003c/span\u003e; Southern \u0026amp; Harmer, \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e2021\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eThe protocol employed three detection criteria applied sequentially: (i) Explicit identity markers: lexical items directly referencing identity categories or diminutive constructions that may invoke patronising stances; (ii) Differential targeting: evaluative patterns documented in previous research as disproportionately directed at women politicians relative to male counterparts (emotional instability attributions, appearance focus, maternal role invocation); and (iii) Stereotypical bundling: co-occurrence of multiple evaluative themes that collectively invoke cultural models of gender and authority (competence questioning\u0026thinsp;+\u0026thinsp;emotional characterisation\u0026thinsp;+\u0026thinsp;infantilising language).\u003c/p\u003e\u003cp\u003eEach reply meeting any criterion was flagged as potentially identity-salient, with criterion-specific coding (1\u0026thinsp;=\u0026thinsp;explicit, 2\u0026thinsp;=\u0026thinsp;differential, 3\u0026thinsp;=\u0026thinsp;bundled) to enable sensitivity analysis. Inter-criterion overlap was tracked, with 23% of flagged replies meeting multiple criteria. This approach acknowledges the indirect nature of identity indexing while providing empirically grounded detection methods that can be replicated across contexts.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec11\" class=\"Section2\"\u003e\u003ch2\u003e3.3. Appraisal Annotation and Configuration Analysis\u003c/h2\u003e\u003cp\u003eEvaluative analysis employed Martin and White's (2005) Judgement framework, focusing on systematic identification of category co-patterns rather than isolated instances. The annotation protocol operationalised five Judgement categories: Capacity (ability, competence, skill), Propriety (ethics, morality, legality), Veracity (honesty, authenticity, credibility), Tenacity (resolve, dependability, determination), and Normality (conventionality, expectedness, typicality).\u003c/p\u003e\u003cp\u003eEach reply was coded for all present categories with polarity marking (positive/negative) and salience weighting (primary/secondary) to capture evaluative prominence hierarchies. Configuration analysis identified systematic co-occurrence patterns within individual replies, with particular attention to cross-category combinations (e.g., Capacity\u0026thinsp;+\u0026thinsp;Propriety, Veracity\u0026thinsp;+\u0026thinsp;Tenacity) that create compound evaluative effects. Configurations appearing in \u0026lt;\u0026thinsp;3% of replies were grouped as \"miscellaneous\" to focus analysis on substantive patterns.\u003c/p\u003e\u003cp\u003eLanguage-specific annotation incorporated pragmatic markers documented in Spanish and Italian evaluative discourse research. Spanish coding captured \u003cem\u003e\u0026iexcl;qu\u0026eacute;..!\u003c/em\u003e constructions for ironic intensification, diminutive-evaluative combinations (politiquilla, ministra), and moral-register lexicon (sinverg\u0026uuml;enza, criminal). Italian annotation identified subjunctive doubt constructions (non sia onesta), metaphorical competence frames (maschera, teatrino), and authenticated/performance contrasts (finge, recita). This ensures crosslinguistic comparison reflects genuine differences in evaluative deployment rather than analytical artefacts.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec12\" class=\"Section2\"\u003e\u003ch2\u003e3.4. Annotation Procedures and Reliability Assessment\u003c/h2\u003e\u003cp\u003eTwo annotators, trained on a 200-item pilot, independently coded the corpus in 500-reply batches with periodic calibration. The protocol targeted Judgement at two layers: (i) category selection (presence/absence and polarity, with salience weighting of primary vs. secondary instances), and (ii) configuration identification (co-occurring Judgement bundles) and identity-salience flags (three detection criteria). Disagreements were adjudicated after each batch.\u003c/p\u003e\u003cp\u003eWe report reliability across multiple indices tailored to task granularity and class imbalance. For polarity, Fleiss' k\u0026thinsp;=\u0026thinsp;0.52. For fine-grained Judgement type selection, Fleiss' k\u0026thinsp;=\u0026thinsp;0.32, converging with imbalance-robust metrics (Krippendorff's a\u0026thinsp;=\u0026thinsp;0.31, Gwet's AC1\u0026thinsp;=\u0026thinsp;0.34) and consistent with evaluative annotation facing implicit cues (Bednarek \u0026amp; Caple, \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e2017\u003c/span\u003e; Fuoli, \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e2018\u003c/span\u003e; Martin \u0026amp; White, \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2005\u003c/span\u003e). Raw agreement was 63%. For the category presence/absence\u0026thinsp;+\u0026thinsp;salience layer, overall Krippendorff's a\u0026thinsp;=\u0026thinsp;0.52; per-category a values were Capacity\u0026thinsp;=\u0026thinsp;0.57; Propriety\u0026thinsp;=\u0026thinsp;0.62; Veracity\u0026thinsp;=\u0026thinsp;0.55; Tenacity\u0026thinsp;=\u0026thinsp;0.52; Normality\u0026thinsp;=\u0026thinsp;0.6 (Hayes \u0026amp; Krippendorff, \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e2007\u003c/span\u003e). Configuration identification yielded a\u0026thinsp;=\u0026thinsp;0.58. Identity-salience flagging reached a\u0026thinsp;=\u0026thinsp;0.63, with criterion-specific variation (explicit markers a\u0026thinsp;=\u0026thinsp;0.68; differential targeting a\u0026thinsp;=\u0026thinsp;0.52; stereotypical bundling a\u0026thinsp;=\u0026thinsp;0.58).\u003c/p\u003e\u003cp\u003eConfusion analysis showed non-random proximity between Propriety and Normality, where moral critique blurred with characterological characterisation. To reflect construct adjacency rather than coder error, we adopted compound labels (permitting\u0026thinsp;\u0026gt;\u0026thinsp;1 Judgement per unit); 26.3% of units received co-assignments. Final datasets comprise adjudicated consensus with compound labels preserved for transparency.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec13\" class=\"Section2\"\u003e\u003ch2\u003e3.5. Ethical Considerations and Data Governance\u003c/h2\u003e\u003cp\u003eThis research analysed publicly available social media content directed at public political figures, following established protocols for digital discourse research (Townsend \u0026amp; Wallace, 2016). Data collection utilised X's public API within rate limits and terms of service, with no attempts to access private or protected content.\u003c/p\u003e\u003cp\u003eAnonymisation protocols removed individual user identifiers while preserving sufficient context for discourse analysis. Quoted examples are selected to illustrate systematic patterns rather than individual attacks, with preference for analytically relevant rather than sensationalist content. Research outputs focus on linguistic mechanisms and cultural repertoires rather than individual users or personal characteristics.\u003c/p\u003e\u003cp\u003eHostile content handling follows established practices in political incivility research (KhosraviNik, \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e2023\u003c/span\u003e), prioritising scholarly analysis over reproduction of harmful discourse. The research contributes to understanding of digital political communication dynamics without amplifying or endorsing the delegitimising strategies that form its analytical foundation. All procedures received institutional research ethics approval prior to data collection.\u003c/p\u003e\u003c/div\u003e"},{"header":"4. Results","content":"\u003cp\u003eThis section reports findings from the stratified corpus of 2,000 replies (1,000 directed at Isabel D\u0026iacute;az Ayuso in Spanish; 1,000 directed at Giorgia Meloni in Italian). Results are organized by research question.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e4.1. RQ1. How do Judgement categories configure in hostile replies to these leaders, and what co-patterns emerge?\u003c/strong\u003e\u003c/p\u003e\n\u003cdiv class=\"gridtable\"\u003e\n \u003cdiv class=\"colspec\" align=\"left\"\u003e\n \u003cp\u003eTable 1 summarizes the distribution of Judgement categories across the two corpora with 95% confidence intervals. Propriety judgements were most frequent in Spanish replies to Ayuso (59.6%, CI: 56.5-62.6%), while Capacity judgements dominated Italian replies to Meloni (35.0%, CI: 30.0-40.3%). Normality evaluations showed the largest cross-linguistic contrast, appearing in 28.5% of Spanish replies (CI: 25.8-31.4%) versus only 2.2% of Italian replies (CI: 1.2-3.6%). Tenacity judgements were infrequent in both corpora (Ayuso: 1.9%, Meloni: 1.2%).\u003c/p\u003e\n \u003c/div\u003e\n \u003ctable id=\"Tab1\" border=\"1\"\u003e\n \u003ccaption\u003e\n \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e\n \u003cdiv class=\"CaptionContent\"\u003e\n \u003cp\u003eDistribution of Judgement categories across Spanish and Italian replies with 95% confidence intervals\u003c/p\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eLeader\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eCategory\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003en\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003ek\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eprop\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eci_low\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eci_high\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAyuso\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eCapacity\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e289\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.289\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.2617539675950579\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.3178609836083989\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAyuso\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ePropriety\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e596\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.596\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.5652749017650743\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.6259903337063386\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAyuso\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eVeracity\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e207\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.207\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.1830325088651358\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.23321005370648906\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAyuso\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eTenacity\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e19\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.019\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.012196787737404145\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.029484688702703543\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAyuso\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNormality\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e285\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.285\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.25788529526982207\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.3137602711223259\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMeloni\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eCapacity\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e350\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.35\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.3210621906513813\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.3800858789245359\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMeloni\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ePropriety\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e370\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.37\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.34062614660286095\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.4003688470296006\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMeloni\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eVeracity\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e326\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.326\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.2976606245950845\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.3556711361129795\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMeloni\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eTenacity\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e12\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.012\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.006877579092159562\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.020857473928157594\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMeloni\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNormality\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e22\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.022\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.014572598673771597\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.03308591637481774\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n\u003c/div\u003e\n\u003cp\u003eCo-occurrence analysis reveals systematic category clustering within replies. Table \u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003e presents conditional probabilities for key category combinations. The most striking pattern emerges in competence-to-morality transitions: Spanish replies show (Propriety|Capacity)\u0026thinsp;=\u0026thinsp;0.408, indicating that 40.8% of competence critiques co-occur with moral evaluation. In contrast, Italian replies show P(Propriety|Capacity)\u0026thinsp;=\u0026thinsp;0.209, representing a near 2:1 difference in moralization rates.\u003c/p\u003e\n\u003cdiv class=\"gridtable\"\u003e\n \u003ctable id=\"Tab2\" border=\"1\"\u003e\n \u003ccaption\u003e\n \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e\n \u003cdiv class=\"CaptionContent\"\u003e\n \u003cp\u003eConditional probabilities for major category co-patterns\u003c/p\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eleader\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eA\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eB_given_A\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003enum\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eden\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eP(B|A)\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAyuso\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eCapacity\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ePropriety\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e118\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e289\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.4083044982698962\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAyuso\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eCapacity\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eTenacity\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e289\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.006920415224913495\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAyuso\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ePropriety\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eTenacity\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e16\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e596\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.026845637583892617\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAyuso\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eVeracity\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ePropriety\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e89\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e207\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.42995169082125606\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAyuso\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eVeracity\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eTenacity\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e207\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.00966183574879227\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAyuso\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eTenacity\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ePropriety\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e16\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e19\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.8421052631578947\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAyuso\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNormality\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ePropriety\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e124\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e285\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.43508771929824563\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAyuso\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNormality\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eTenacity\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e285\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.014035087719298246\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMeloni\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eCapacity\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ePropriety\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e73\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e350\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.20857142857142857\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMeloni\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eCapacity\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eTenacity\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e350\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.008571428571428572\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMeloni\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ePropriety\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eTenacity\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e370\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.010810810810810811\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMeloni\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eVeracity\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ePropriety\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e65\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e326\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.19938650306748465\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMeloni\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eVeracity\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eTenacity\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e326\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.018404907975460124\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMeloni\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eTenacity\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ePropriety\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e12\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.3333333333333333\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMeloni\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNormality\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ePropriety\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e7\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e22\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.3181818181818182\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMeloni\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNormality\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eTenacity\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e22\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.045454545454545456\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n\u003c/div\u003e\n\u003cp\u003eAdditional co-patterns include Propriety-Tenacity combinations (P(Tenacity|Propriety)\u0026thinsp;=\u0026thinsp;0.027 for Ayuso, 0.011 for Meloni) and Capacity-Normality linkages, though these occur at lower frequencies.\u003c/p\u003e\n\u003cp\u003eChi-square tests confirm significant associations between categories, with phi coefficients indicating moderate to strong effect sizes for key combinations. We observe asymmetric co-occurrence (e.g., P(Propriety|Capacity)\u0026thinsp;\u0026gt;\u0026thinsp;P(Capacity|Propriety)), consistent with an alignment pathway from competence assessment to moral sanction; our design is associational and does not establish temporal or causal ordering.\u003c/p\u003e\n\u003cp\u003eThe distributions show that Spanish replies lean toward Propriety-, while Italian replies foreground Capacity-. These tendencies are visible in the corpus excerpts. In Spanish, Ayuso is attacked through moralised disgust (Ex. 4.1a), while in Italian Meloni is targeted via explicit reference to failure (Ex. 4.1b).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEx. 4.1a (Ayuso, ES; Propriety\u0026minus;)\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(tweet_id: AYS_16_2025_0057)\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eOriginal (ES)\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u0026ldquo;@IdiazAyuso Qu\u0026eacute; \u003cstrong\u003eascazo de mujer\u003c/strong\u003e!!!!\u0026rdquo;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTranslation (EN)\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u0026ldquo;What a \u003cstrong\u003edisgusting woman\u003c/strong\u003e!!!!\u0026rdquo;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eReading: Propriety\u0026minus;\u003c/strong\u003e (moralised disgust; sanctioning register).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEx. 4.1b (Meloni, IT; Capacity\u0026minus;)\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(tweet_id: MEL_11_2025_0055)\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eOriginal (IT)\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u0026ldquo;@user Aspettiamo un post che commenti anche i tuoi \u003cstrong\u003efallimenti\u003c/strong\u003e.\u0026rdquo;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTranslation (EN)\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u0026ldquo;We\u0026rsquo;re waiting for a post that also comments on your \u003cstrong\u003efailures\u003c/strong\u003e.\u0026rdquo;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eReading: Capacity\u0026minus;\u003c/strong\u003e (incompetence via repeated failure).\u003c/p\u003e\n\u003cdiv id=\"Sec15\" class=\"Section2\"\u003e\n \u003ch2\u003e4.2. RQ2. How do these evaluative configurations function as resources for delegitimising stance performance?\u003c/h2\u003e\n \u003cp\u003eLogistic regression analysis identifies the strongest predictors of identity-salient stance cues (Table \u003cspan class=\"InternalRef\"\u003e3\u003c/span\u003e). Tenacity evaluations emerge as the most powerful predictor (OR\u0026thinsp;=\u0026thinsp;6.45, 95% CI: 2.84\u0026ndash;14.66, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), indicating that vulnerability-based attacks are central to delegitimization strategies. Normality judgements show the second-strongest effect (OR\u0026thinsp;=\u0026thinsp;3.83, 95% CI: 2.85\u0026ndash;5.15, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), reflecting the role of behavioral conformity demands in identity-salient positioning.\u003c/p\u003e\n \u003cdiv class=\"gridtable\"\u003e\n \u003ctable id=\"Tab3\" border=\"1\"\u003e\n \u003ccaption\u003e\n \u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e\n \u003cdiv class=\"CaptionContent\"\u003e\n \u003cp\u003eLogistic regression results: Predictors of identity-salient stance (ORs, 95% CIs, p-values)\u003c/p\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003ePredictor\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eOR\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eCI_low\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eCI_high\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003ep_value\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eIntercept\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.1456602376181486\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.10966486103553034\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.19347040266709906\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e2.2445312240528477e-40\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eCapacity\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.8751150260502529\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.62133743604804\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1.2325449335386953\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.4452090941524517\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ePropriety\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e2.1096318561888068\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1.5908175824884663\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e2.7976473340738286\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e2.1769011435196684e-07\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eVeracity\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1.2099033323092494\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.9305163823786222\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1.5731760356448905\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.15491822270940414\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eTenacity\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e6.4496850040381535\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e2.837034150416943\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e14.662649247701562\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e8.646946902848037e-06\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNormality\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e3.8284395825972344\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e2.8478334966990855\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e5.1467017487455315\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e5.995285189933957e-19\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eCapxProp\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1.8670563554474753\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1.159597995865985\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e3.0061275087092123\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.01019053885020874\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eES\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1.6532440488181457\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1.313399687350815\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e2.081023706093331\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.8531048351199657e-05\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n \u003c/div\u003e\n \u003cp\u003eFigure \u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003e presents these results as a forest plot, visualizing the relative strength of different evaluative configurations. Model diagnostics indicate adequate fit (McFadden pseudo-R\u0026sup2; = 0.191, classification accuracy\u0026thinsp;=\u0026thinsp;73.2%).\u003c/p\u003e\n \u003cp\u003eFigure \u003cspan class=\"InternalRef\"\u003e3\u003c/span\u003e illustrates the Capacity\u0026times;Propriety interaction by leader, demonstrating how competence-morality configurations vary across linguistic contexts. The interaction shows divergent patterns: Spanish replies exhibit steep increases in identity-salient probability when both categories co-occur, while Italian replies show more moderate effects.\u003c/p\u003e\n \u003cp\u003eThe co-pattern analysis shows that Spanish replies frequently bundle Capacity- with other Judgements, often escalating into moral sanction, while Italian replies combine incompetence and dishonesty in compound attacks. This is exemplified by a Spanish tweet that links dishonesty and incompetence (Ex. 4.2a) and an Italian tweet that pairs incapacity with dishonesty (Ex. 4.2b).\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003eEx. 4.2a (Ayuso, ES; Capacity\u0026minus; + Veracity\u0026minus;)\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003e(tweet_id: AYS_16_2025_0004)\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003eOriginal (ES)\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003e\u0026ldquo;@Morcego_Man @IdiazAyuso Ahora dilo pero sin las \u003cstrong\u003ementiras\u003c/strong\u003e, que eres una \u003cstrong\u003ein\u0026uacute;til\u003c/strong\u003e.\u0026rdquo;\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003eTranslation (EN)\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003e\u0026ldquo;Now say it but without the \u003cstrong\u003elies\u003c/strong\u003e, because you\u0026rsquo;re \u003cstrong\u003euseless\u003c/strong\u003e.\u0026rdquo;\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003eReading: Veracity\u0026minus;\u003c/strong\u003e (\u003cem\u003ementiras\u003c/em\u003e\u0026thinsp;=\u0026thinsp;dishonesty)\u0026thinsp;+\u0026thinsp;\u003cstrong\u003eCapacity\u0026minus;\u003c/strong\u003e (\u003cem\u003ein\u0026uacute;til\u003c/em\u003e\u0026thinsp;=\u0026thinsp;incompetence).\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003eEx. 4.2b (Meloni, IT; Capacity\u0026minus; + Propriety\u0026minus;)\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003e(tweet_id: MEL_14_2025_0002)\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003eOriginal (IT)\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003e\u0026ldquo;@user \u0026lsquo;Onoriamo\u0026rsquo;? Ma che devi onorare se pensi solo a te stessa, \u003cstrong\u003eincapace\u003c/strong\u003e e \u003cstrong\u003edisonesta\u003c/strong\u003e.\u0026rdquo;\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003eTranslation (EN)\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003e\u0026ldquo;\u0026lsquo;We honor\u0026rsquo;? What do you have to honor if you only think of yourself, \u003cstrong\u003eincompetent\u003c/strong\u003e and \u003cstrong\u003edishonest\u003c/strong\u003e.\u0026rdquo;\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003eReading: Capacity\u0026minus;\u003c/strong\u003e (\u003cem\u003eincapace\u003c/em\u003e)\u0026thinsp;+\u0026thinsp;\u003cstrong\u003ePropriety\u0026minus;\u003c/strong\u003e (\u003cem\u003edisonesta\u003c/em\u003e); compound configuration.\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003e4.3. RQ3. What Spanish-Italian contrasts appear in evaluative repertoires, and how do these patterns relate to identity-salient stance cues?\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003eCross-linguistic analysis reveals substantial effect sizes for key categories (Table \u003cspan class=\"InternalRef\"\u003e4\u003c/span\u003e). Normality judgements show the largest contrast (Cram\u0026eacute;r\u0026apos;s V\u0026thinsp;=\u0026thinsp;0.365, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), confirming Spanish discourse\u0026apos;s emphasis on behavioral policing. Propriety evaluations demonstrate a large effect size (V\u0026thinsp;=\u0026thinsp;0.226, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), while Veracity shows medium-sized differences (V\u0026thinsp;=\u0026thinsp;0.135, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), with Italian discourse featuring more credibility-focused attacks.\u003c/p\u003e\n \u003cdiv class=\"gridtable\"\u003e\n \u003ctable id=\"Tab4\" style=\"width: 447px;\" border=\"1\"\u003e\n \u003ccaption\u003e\n \u003cdiv class=\"CaptionNumber\"\u003eTable 4\u003c/div\u003e\n \u003cdiv class=\"CaptionContent\"\u003e\n \u003cp\u003e\u003cem\u003eCross-linguistic effect sizes (Cram\u0026eacute;r\u0026apos;s V) for category contrasts\u003c/em\u003e\u003c/p\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth style=\"width: 60px;\" align=\"left\"\u003e\n \u003cp\u003eCategory\u003c/p\u003e\n \u003c/th\u003e\n \u003cth style=\"width: 136px;\" align=\"left\"\u003e\n \u003cp\u003eCramer\u0026acute;s_V\u003c/p\u003e\n \u003c/th\u003e\n \u003cth style=\"width: 194.139px;\" align=\"left\"\u003e\n \u003cp\u003ep_value\u003c/p\u003e\n \u003c/th\u003e\n \u003cth style=\"width: 10px;\" align=\"left\"\u003e\n \u003cp\u003eAyuso\u003c/p\u003e\n \u003c/th\u003e\n \u003cth style=\"width: 40px;\" align=\"left\"\u003e\n \u003cp\u003eMeloni\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 60px;\" align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eCapacity\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\" align=\"char\"\u003e\n \u003cp\u003e0.06541093292775553\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 194.139px;\" align=\"left\"\u003e\n \u003cp\u003e0.0034416206734931088\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\" align=\"char\"\u003e\n \u003cp\u003e289\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 40px;\" align=\"char\"\u003e\n \u003cp\u003e350\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 60px;\" align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003ePropriety\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\" align=\"char\"\u003e\n \u003cp\u003e0.2261307413636883\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 194.139px;\" align=\"left\"\u003e\n \u003cp\u003e4.84418793725353e-24\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\" align=\"char\"\u003e\n \u003cp\u003e596\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 40px;\" align=\"char\"\u003e\n \u003cp\u003e370\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 60px;\" align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eVeracity\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\" align=\"char\"\u003e\n \u003cp\u003e0.13457621205415032\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 194.139px;\" align=\"left\"\u003e\n \u003cp\u003e1.7611557411151298e-09\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\" align=\"char\"\u003e\n \u003cp\u003e207\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 40px;\" align=\"char\"\u003e\n \u003cp\u003e326\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 60px;\" align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eTenacity\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\" align=\"char\"\u003e\n \u003cp\u003e0.028333115696007258\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 194.139px;\" align=\"left\"\u003e\n \u003cp\u003e0.20512116079234521\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\" align=\"char\"\u003e\n \u003cp\u003e19\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 40px;\" align=\"char\"\u003e\n \u003cp\u003e12\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 60px;\" align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eNormality\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\" align=\"char\"\u003e\n \u003cp\u003e0.36480273166894667\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 194.139px;\" align=\"left\"\u003e\n \u003cp\u003e7.787392189432449e-60\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\" align=\"char\"\u003e\n \u003cp\u003e285\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 40px;\" align=\"char\"\u003e\n \u003cp\u003e22\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n \u003c/div\u003e\n \u003cdiv class=\"gridtable\"\u003e\n \u003cdiv class=\"colspec\" align=\"left\"\u003e\u0026nbsp;\u003c/div\u003e\n \u003cdiv class=\"colspec\" align=\"left\"\u003eLeader-specific analysis reveals distinct cultural pathways for identity-salient stance performance (Table\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e5\u003c/span\u003e). In Spanish contexts, Propriety judgements strongly predict identity-salient stance (OR\u0026thinsp;=\u0026thinsp;3.62, 95% CI: 2.41\u0026ndash;5.41), while the same category shows nonsignificant effects in Italian contexts (OR\u0026thinsp;=\u0026thinsp;1.08, 95% CI: 0.71\u0026ndash;1.65). Conversely, Normality evaluations predict identity-salient stance in both contexts but with different magnitudes (Ayuso: OR\u0026thinsp;=\u0026thinsp;4.54; Meloni: OR\u0026thinsp;=\u0026thinsp;5.68).\u003c/div\u003e\n \u003ctable id=\"Tab5\" border=\"1\"\u003e\n \u003ccaption\u003e\n \u003cdiv class=\"CaptionNumber\"\u003eTable 5\u003c/div\u003e\n \u003cdiv class=\"CaptionContent\"\u003e\n \u003cp\u003eLeader-specific odds ratios for identity-salient stance predictors\u003c/p\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eLeader\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003ePredictor\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eOR\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eCI_low\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eCI_high\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003ep_value\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eAyuso\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003econst\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.15423637510495014\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.10173182398208255\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.23383891563474454\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.321685001577727e-18\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eAyuso\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eCapacity\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1.001460051903562\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.6011986725178501\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.6682043414340828\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.995528829094319\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eAyuso\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003ePropriety\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e3.6149401264576655\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e2.4147827684633394\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e5.411580821487112\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e4.3018810100922885e-10\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eAyuso\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eVeracity\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1.2615017957069514\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.8661385727546785\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.8373350761998686\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.2259430498484959\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eAyuso\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eTenacity\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e29335490064.573086\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003einf\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.9988590384664778\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eAyuso\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eNormality\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e4.542449305475346\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e3.2174902807763477\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e6.413025026398749\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e7.870942260837633e-18\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eAyuso\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eCapxProp\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1.633461336374365\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.8452535247433561\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e3.1566812315157984\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.1443402590231145\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eMeloni\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003econst\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.20717622682060402\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.14561666985633484\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.29476013290215414\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e2.1246657345483944e-18\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eMeloni\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eCapacity\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.7688707820468026\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.4885754237442349\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.209971379556599\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.25591361179501493\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eMeloni\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003ePropriety\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1.079652768195933\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.7068680298957951\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.6490349691511414\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.7228547555635039\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eMeloni\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eVeracity\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1.1322662290345151\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.7801486732335287\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.643310893676648\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.5133552827202998\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eMeloni\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eTenacity\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1.1995658248902334\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.30613689944510414\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e4.700374802425989\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.7939845064298733\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eMeloni\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eNormality\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e5.683590576594332\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e2.3903640098087653\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e13.513925791133468\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e8.426322155890154e-05\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eMeloni\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eCapxProp\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e2.441562066432025\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1.170692585882572\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e5.0920501215489615\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.017303868255399178\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n \u003c/div\u003e\n \u003cp\u003eThese patterns indicate culturally specific evaluative repertoires: Spanish replies systematically moralize political critique through Propriety-Normality combinations, while Italian replies foreground competence-authenticity challenges through Capacity-Veracity configurations. The interaction model confirms that category effects vary significantly by linguistic context (Category \u0026times; Leader interactions, p\u0026thinsp;\u0026lt;\u0026thinsp;0.05 for Propriety and Normality), supporting the hypothesis of culturally embedded evaluative pathways.\u003c/p\u003e\n \u003cp\u003eCross-linguistic contrasts are clear in the excerpts: Spanish users often frame Ayuso as abnormal or \u0026quot;out of line\u0026quot; (Ex. 4.3a), while Italian users depict Meloni as politically mad (Ex. 4.3b). These examples reflect the broader repertoires identified in the statistical analysis.\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003eEx. 4.3a (Ayuso, ES; Normality\u0026minus;)\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003e(tweet_id: chemay59_2025-04-11_11:30:43)\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003eOriginal (ES)\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003e\u0026ldquo;@user Dudo que seas Ayuso oficial\u0026hellip; \u003cstrong\u003eTe has ido de la olla\u003c/strong\u003e.\u0026rdquo;\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003eTranslation (EN)\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003e\u0026ldquo;I doubt you\u0026rsquo;re the official Ayuso\u0026hellip; \u003cstrong\u003eYou\u0026rsquo;ve lost it\u003c/strong\u003e.\u0026rdquo;\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003eReading: Normality\u0026minus;\u003c/strong\u003e (abnormal behaviour/being \u0026ldquo;out of line\u0026rdquo;).\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003eEx. 4.3b (Meloni, IT; Normality\u0026minus;)\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003e(tweet_id: MEL_18_2025_0047)\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003eOriginal (IT)\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003e\u0026ldquo;@GiorgiaMeloni Prospero? Siamo ad una soglia di \u003cstrong\u003efollia\u003c/strong\u003e politica.\u0026rdquo;\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003eTranslation (EN)\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003e\u0026ldquo;Prospero? We\u0026rsquo;re at a threshold of political \u003cstrong\u003emadness\u003c/strong\u003e.\u0026rdquo;\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003eReading: Normality\u0026minus;\u003c/strong\u003e (abnormalisation via \u003cem\u003efollia\u003c/em\u003e).\u003c/p\u003e\n \u003cp\u003eThe Spanish pattern emphasizes moral sanctioning and behavioral conformity as delegitimization strategies, while the Italian pattern prioritizes competence questioning and authenticity challenges. These contrasts persist after controlling for temporal effects, consistent within this corpus and window; broader stability is a hypothesis for future work.\u003c/p\u003e\n\u003c/div\u003e"},{"header":"5. Discussion","content":"\u003cp\u003e\u003cb\u003e5.1 RQ1. How do Judgement categories configure in hostile replies to women leaders, and what co-patterns emerge?\u003c/b\u003e\u003c/p\u003e\u003cp\u003eThe distributional results show that hostile replies are not scattered individual attacks but are structured by recurring evaluative resources. This confirms the premise that digital hostility functions communicationally, as patterned stance performance rather than idiosyncratic abuse (cf. Du Bois, \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e2007\u003c/span\u003e; Silverstein, \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e2003\u003c/span\u003e). The dominance of Propriety- judgements in Spanish replies and Capacity-/ Veracity- judgements in Italian replies-visible in the aggregate (Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e; Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e) \u0026mdash;is also instantiated in typical excerpts: Spanish moral sanction with abnormalisation (corrupta\u0026thinsp;+\u0026thinsp;emphatic capitals) in Ex. 5.1a, and Italian competence/ authenticity framing (hai sempre fallito) in Ex. 5.1b. Taken together with the cooccurrence counts (Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e), these examples indicate that categories of moral sanction and competence assessment are mobilised as collective repertoires of delegitimisation. These findings extend prior research on gendered political aggression, which identified a shift from overtly sexist invective to subtler, performance-framed delegitimation (Esposito \u0026amp; Breeze, \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2022\u003c/span\u003e; Southern \u0026amp; Harmer, \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e2021\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eCrucially, these recurrent configurations are recognisable to participants themselves: the pairing of incompetence with corruption or of fakeness with failure functions as a discursive shorthand for stance alignment, inviting uptake and circulation within hostile publics.\u003c/p\u003e\u003cp\u003e\u003cb\u003eEx. 5.1a \u0026mdash; Ayuso (ES; Propriety\u0026minus; + Normality\u0026minus;)\u003c/b\u003e\u003c/p\u003e\u003cp\u003e(\u003cb\u003etweet_id\u003c/b\u003e: PepBausset_2025-04-11_08:04:06)\u003c/p\u003e\u003cp\u003e\u003cstrong\u003eOriginal (ES)\u003c/strong\u003e\u003cp\u003e\u0026ldquo;@user Ayuso \u003cb\u003es\u0026uacute;per corrupta\u003c/b\u003e, a los tribunales \u003cb\u003ePOR FIN\u003c/b\u003e.\u0026rdquo;\u003c/p\u003e\u003c/p\u003e\u003cp\u003e\u003cstrong\u003eTranslation (EN)\u003c/strong\u003e\u003cp\u003e\u0026ldquo;Ayuso \u003cb\u003esuper corrupt\u003c/b\u003e\u0026mdash;to the courts \u003cb\u003eAT LAST\u003c/b\u003e.\u0026rdquo;\u003c/p\u003e\u003c/p\u003e\u003cp\u003e\u003cb\u003eReading: Propriety\u0026minus;\u003c/b\u003e (\u003cem\u003ecorrupta\u003c/em\u003e); \u003cb\u003eNormality\u0026minus;\u003c/b\u003e (social deviance frame); \u003cb\u003eGraduation (force)\u003c/b\u003e via capitals.\u003c/p\u003e\u003cp\u003e\u003cb\u003eEx. 5.1b \u0026mdash; Meloni (IT; Capacity\u0026minus; + Veracity\u0026minus;)\u003c/b\u003e\u003c/p\u003e\u003cp\u003e(\u003cb\u003etweet_id\u003c/b\u003e: MEL_18_2025_0069)\u003c/p\u003e\u003cp\u003e\u003cstrong\u003eOriginal (IT)\u003c/strong\u003e\u003cp\u003e\u0026ldquo;@GiorgiaMeloni Basta cazzate\u0026hellip; \u003cb\u003ehai sempre fallito\u003c/b\u003e.\u0026rdquo;\u003c/p\u003e\u003c/p\u003e\u003cp\u003e\u003cstrong\u003eTranslation (EN)\u003c/strong\u003e\u003cp\u003e\u0026ldquo;Enough nonsense\u0026hellip; \u003cb\u003eyou\u0026rsquo;ve always failed\u003c/b\u003e.\u0026rdquo;\u003c/p\u003e\u003c/p\u003e\u003cp\u003e\u003cb\u003eReading: Capacity\u0026minus;\u003c/b\u003e (repeated failure), \u003cb\u003eVeracity\u0026minus;\u003c/b\u003e (misrepresentation implied).\u003c/p\u003e\u003cp\u003eThe co-occurrence analysis highlights that evaluative meanings rarely appear in isolation. Instead, they are bundled into compact configurations-such as Capacity- \u0026rarr; Propriety- in Spanish and Capacity-\u0026raquo;Veracity- in Italian-that intensify delegitimisation. In the Spanish set, this escalation is instantiated in Ex. 5.1c (cobarde recruiting sinverg\u0026uuml;enza), while the Italian route is instantiated in Ex. 5.1d (falsa, bugiarda alongside a fitness attack). These pairings correspond to the significant asymmetries detected by McNemar tests and the conditional probabilities reported in Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e (e.g., higher P(Propriety-|Capacity-) in Spanish; higher P(Veracity-|Capacity-) in Italian). From an Appraisal perspective, such pairings operate as evaluative prosodies (Hood, \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e2010\u003c/span\u003e), whereby one Judgement resource recruits another to consolidate stance and shift footing\u0026mdash;from an apparently technical competence assessment to moral or authenticity condemnation. In interactional terms, this escalation provides a portable template for attack sequences, enabling users to move from local critique to categorical exclusion in one turn or across turns.\u003c/p\u003e\u003cp\u003e\u003cb\u003eEx. 5.1c (Ayuso, ES)\u003c/b\u003e\u003c/p\u003e\u003cp\u003e(tweet_id: Fran_Toro_2025-04-10_12:12:55)\u003c/p\u003e\u003cp\u003e\u003cstrong\u003eOriginal (ES)\u003c/strong\u003e\u003cp\u003e\u0026ldquo;@user \u003cb\u003eCobarde\u003c/b\u003e, \u003cb\u003esinverg\u0026uuml;enza\u003c/b\u003e, te largas a Ecuador\u0026hellip;\u0026rdquo;\u003c/p\u003e\u003c/p\u003e\u003cp\u003e\u003cstrong\u003eTranslation (EN)\u003c/strong\u003e\u003cp\u003e\u0026ldquo;\u003cb\u003eCoward\u003c/b\u003e, \u003cb\u003eshameless\u003c/b\u003e, you\u0026rsquo;re off to Ecuador\u0026hellip;\u0026rdquo;\u003c/p\u003e\u003c/p\u003e\u003cp\u003e\u003cb\u003eReading: Capacity\u0026minus;\u003c/b\u003e (\u003cem\u003ecobarde\u003c/em\u003e) recruiting \u003cb\u003ePropriety\u0026minus;\u003c/b\u003e (\u003cem\u003esinverg\u0026uuml;enza\u003c/em\u003e).\u003c/p\u003e\u003cp\u003e\u003cb\u003eEx. 5.1d (Meloni, IT)\u003c/b\u003e\u003c/p\u003e\u003cp\u003e(tweet_id: MEL_18_2025_0049)\u003c/p\u003e\u003cp\u003e\u003cstrong\u003eOriginal (IT)\u003c/strong\u003e\u003cp\u003e\u0026ldquo;@GiorgiaMeloni Cara fascistona \u003cb\u003efalsa\u003c/b\u003e e \u003cb\u003ebugiarda\u003c/b\u003e.\u0026rdquo;\u003c/p\u003e\u003c/p\u003e\u003cp\u003e\u003cstrong\u003eTranslation (EN)\u003c/strong\u003e\u003cp\u003e\u0026ldquo;Dear big fascist, \u003cb\u003efake\u003c/b\u003e and a \u003cb\u003eliar\u003c/b\u003e.\u0026rdquo;\u003c/p\u003e\u003c/p\u003e\u003cp\u003e\u003cb\u003eReading: Veracity\u0026minus;\u003c/b\u003e (\u003cem\u003efalsa, bugiarda\u003c/em\u003e) recruited alongside \u003cb\u003eCapacity\u0026minus;\u003c/b\u003e (leadership fitness).\u003c/p\u003e\u003cp\u003eThese results also refine theories of delegitimisation (Reisigl \u0026amp; Wodak, \u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e2001\u003c/span\u003e; Van Leeuwen, \u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e2007\u003c/span\u003e) by demonstrating that the undermining of authority in digital environments is not a matter of single evaluative acts but of recurrent configurational patterns. What is delegitimised is not just a leader's individual action but their broader eligibility to hold authority, achieved through the patterned layering of Judgement categories. Such findings underscore the communicational function of hostility: it organises alignment among users by drawing on shared evaluative repertoires, making the discourse legible and resonant in wider publics.\u003c/p\u003e\u003cp\u003eFinally, the cross-linguistic contrasts in distribution (Propriety-/Normalitydominance in Spanish vs. Capacity-Neracity- in Italian) demonstrate that evaluative repertoires are culturally inflected. This aligns with Thompson and AlbaJuez\u0026rsquo;s (2014) claim that while the semantic system of Judgement is stable across languages, its deployment patterns vary according to cultural norms of political critique. In our data, Spanish users more readily mobilise abnormalisation and infantilisation as delegitimising resources, as in Ex. 5.1e, while Italian users foreground authenticity and competence challenges, often coupling abnormalisation with veracity attacks, as in Ex. 5.1f. These contrasts suggest that digital political communication not only reflects but amplifies culturally specific models of what counts as legitimate leadership. These repertoires also differ in how they naturalise gendered delegitimisation. Spanish abnormalisation/infantilisation frames women as intrinsically unsuited to politics, whereas Italian competence/authenticity critiques cast exclusion as ostensibly rational performance evaluation.\u003c/p\u003e\u003cp\u003e\u003cb\u003eEx. 5.1e (Ayuso, ES)\u003c/b\u003e\u003c/p\u003e\u003cp\u003e(tweet_id: FerminYugueros_2025-04-11_07:12:34)\u003c/p\u003e\u003cp\u003e\u003cstrong\u003eOriginal (ES)\u003c/strong\u003e\u003cp\u003e\u0026ldquo;@user \u003cb\u003eEres rid\u0026iacute;cula\u003c/b\u003e a l\u0026iacute;mites insospechados.\u0026rdquo;\u003c/p\u003e\u003c/p\u003e\u003cp\u003e\u003cstrong\u003eTranslation (EN)\u003c/strong\u003e\u003cp\u003e\u0026ldquo;You are \u003cb\u003eridiculous\u003c/b\u003e beyond belief.\u0026rdquo;\u003c/p\u003e\u003c/p\u003e\u003cp\u003e\u003cb\u003eReading: Normality\u0026minus;\u003c/b\u003e (abnormalisation/infantilisation).\u003c/p\u003e\u003cp\u003e\u003cb\u003eEx. 5.1f (Meloni, IT)\u003c/b\u003e\u003c/p\u003e\u003cp\u003e(tweet_id: MEL_15_2025_0002)\u003c/p\u003e\u003cp\u003e\u003cstrong\u003eOriginal (IT)\u003c/strong\u003e\u003cp\u003e\u0026ldquo;@GiorgiaMeloni Ma smettila \u003cb\u003ecessa\u003c/b\u003e carciofara \u003cb\u003ebugiarda\u003c/b\u003e\u0026hellip;\u0026rdquo;\u003c/p\u003e\u003c/p\u003e\u003cp\u003e\u003cstrong\u003eTranslation (EN)\u003c/strong\u003e\u003cp\u003e\u0026ldquo;Oh stop it, \u003cb\u003eugly\u003c/b\u003e artichoke-peddler, \u003cb\u003eliar\u003c/b\u003e\u0026hellip;\u0026rdquo;\u003c/p\u003e\u003c/p\u003e\u003cp\u003e\u003cb\u003eReading: Normality\u0026minus;\u003c/b\u003e (demeaning label) coupled with \u003cb\u003eVeracity\u0026minus;\u003c/b\u003e (\u003cem\u003ebugiarda\u003c/em\u003e).\u003c/p\u003e\u003cdiv id=\"Sec17\" class=\"Section2\"\u003e\u003ch2\u003e5.2 RQ2. How do evaluative configurations function as resources for delegitimising stance performance?\u003c/h2\u003e\u003cp\u003eThe modelling shows that Propriety- and Normality- are the most reliable predictors of identity-salient stance cues across the corpus, with an additional Capacity\u0026times;Propriety compound effect (significant in Italian). In communicational terms, delegitimising force is intensified when technical incompetence is coupled with moral sanction, and when abnormalisation/infantilisation frames are invoked. These patterns are visible in the corpus excerpts: in Spanish, a Capacity-Propriety- bundle escalates from cowardice to shamelessness (Ex. 5.2a) and a stand-alone Normality- evaluation targets behavioural deviance (Ex. 5.2b). In Italian, Capacity\u0026times;Propriety compounds crystallise incompetence together with corruption (Ex. 5.2c), and Normality- invocations (often cooccurring with sanction) act as direct identity-salient triggers (Ex. 5.2d). These examples instantiate the configurational operation of evaluative meaning documented in the statistical results.\u003c/p\u003e\u003cp\u003eOperationally, these configurations function as reusable attack scripts: once competence and sanction are paired, subsequent turns can repeat or vary the pairing with minimal linguistic work, sustaining a stable delegitimising frame across replies and threads.\u003c/p\u003e\u003cp\u003e\u003cb\u003eEx. 5.2a (Ayuso, ES; Capacity\u0026minus; \u0026rarr; Propriety\u0026minus;)\u003c/b\u003e\u003c/p\u003e\u003cp\u003e(tweet_id: Fran_Toro_2025-04-10_12:12:55)\u003c/p\u003e\u003cp\u003e\u003cstrong\u003eOriginal (ES)\u003c/strong\u003e\u003cp\u003e\u0026ldquo;@user \u003cb\u003eCobarde\u003c/b\u003e, \u003cb\u003esinverg\u0026uuml;enza\u003c/b\u003e, te largas a Ecuador\u0026hellip;\u0026rdquo;\u003c/p\u003e\u003c/p\u003e\u003cp\u003e\u003cstrong\u003eTranslation (EN)\u003c/strong\u003e\u003cp\u003e\u0026ldquo;\u003cb\u003eCoward\u003c/b\u003e, \u003cb\u003eshameless\u003c/b\u003e, you\u0026rsquo;re off to Ecuador\u0026hellip;\u0026rdquo;\u003c/p\u003e\u003c/p\u003e\u003cp\u003e\u003cb\u003eReading: Capacity\u0026minus;\u003c/b\u003e (\u003cem\u003ecobarde\u003c/em\u003e, lack of resolve) recruiting \u003cb\u003ePropriety\u0026minus;\u003c/b\u003e (\u003cem\u003esinverg\u0026uuml;enza\u003c/em\u003e).\u003c/p\u003e\u003cp\u003e\u003cb\u003eEx. 5.2b (Ayuso, ES; Normality\u0026minus;)\u003c/b\u003e\u003c/p\u003e\u003cp\u003e(tweet_id: FerminYugueros_2025-04-11_07:12:34)\u003c/p\u003e\u003cp\u003e\u003cstrong\u003eOriginal (ES)\u003c/strong\u003e\u003cp\u003e\u0026ldquo;@user \u003cb\u003eEres rid\u0026iacute;cula\u003c/b\u003e a l\u0026iacute;mites insospechados.\u0026rdquo;\u003c/p\u003e\u003c/p\u003e\u003cp\u003e\u003cstrong\u003eTranslation (EN)\u003c/strong\u003e\u003cp\u003e\u0026ldquo;You are \u003cb\u003eridiculous\u003c/b\u003e beyond belief.\u0026rdquo;\u003c/p\u003e\u003c/p\u003e\u003cp\u003e\u003cb\u003eReading: Normality\u0026minus;\u003c/b\u003e (abnormalisation/infantilisation).\u003c/p\u003e\u003cp\u003e\u003cb\u003eEx. 5.2c (Meloni, IT; Capacity\u0026times;Propriety)\u003c/b\u003e\u003c/p\u003e\u003cp\u003e(tweet_id: MEL_15_2025_0058)\u003c/p\u003e\u003cp\u003e\u003cstrong\u003eOriginal (IT)\u003c/strong\u003e\u003cp\u003e\u0026ldquo;@GiorgiaMeloni \u0026hellip; patriota? \u003cb\u003eIncapace\u003c/b\u003e e \u003cb\u003ecorrotta\u003c/b\u003e.\u0026rdquo;\u003c/p\u003e\u003c/p\u003e\u003cp\u003e\u003cstrong\u003eTranslation (EN)\u003c/strong\u003e\u003cp\u003e\u0026ldquo;Patriot? \u003cb\u003eIncompetent\u003c/b\u003e and \u003cb\u003ecorrupt\u003c/b\u003e.\u0026rdquo;\u003c/p\u003e\u003c/p\u003e\u003cp\u003e\u003cb\u003eReading: Capacity\u0026minus; \u0026times; Propriety\u0026thinsp;\u0026minus;\u003c/b\u003e\u0026thinsp;compound (configuration significant in IT models).\u003c/p\u003e\u003cp\u003e\u003cb\u003eEx. 5.2d (Meloni, IT; Normality\u0026minus; \u0026plusmn; Propriety\u0026minus;)\u003c/b\u003e\u003c/p\u003e\u003cp\u003e(tweet_id: MEL_15_2025_0041)\u003c/p\u003e\u003cp\u003e\u003cstrong\u003eOriginal (IT)\u003c/strong\u003e\u003cp\u003e\u0026ldquo;@GiorgiaMeloni \u003cb\u003eVERGOGNATI\u003c/b\u003e FASCIO\u0026hellip;!\u0026rdquo;\u003c/p\u003e\u003c/p\u003e\u003cp\u003e\u003cstrong\u003eTranslation (EN)\u003c/strong\u003e\u003cp\u003e\u0026ldquo;\u003cb\u003eShame on you\u003c/b\u003e, fascist\u0026hellip;!\u0026rdquo;\u003c/p\u003e\u003c/p\u003e\u003cp\u003e\u003cb\u003eReading: Normality\u0026minus;\u003c/b\u003e (abnormalisation via \u003cem\u003evergognati\u003c/em\u003e) with \u003cb\u003ePropriety\u0026minus;\u003c/b\u003e (moral sanctioning epithet).\u003c/p\u003e\u003cp\u003eIn interactional terms, such scripts exhibit high uptake value: users can cite, echo, or minimally edit prior formulations (e.g., swapping \"incapace\" for \"fallito\") while preserving the compound stance, which facilitates cross-turn propagation in hostile publics.\u003c/p\u003e\u003cp\u003eThe significance of Normality- is particularly notable. While often under-analysed in Appraisal Theory, this category captures accusations of being abnormal, inappropriate, or lacking conventional comportment. Its predictive weight in both corpora suggests that abnormalisation is a salient pathway for identity-salient delegitimation: women leaders are constructed as \"out of place,\" thereby undermining their legitimacy (see Ex. 5.2b, 5.2d). Critically, Normality- provides a shortcut to identity-salience: charges of being \"out of place\" render performance failings constitutive, not contingent. This converts what might appear as episodic critique into status-based exclusion, aligning with indexical accounts of stance and personhood. This finding supports recent observations that gendered aggression frequently operates through the language of behavioural deviance and infantilisation (Esposito \u0026amp; Breeze, \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2022\u003c/span\u003e; Southern \u0026amp; Harmer, \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e2021\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eThe Capacity\u0026times;Propriety interaction confirms that delegitimisation is most powerful when technical incompetence and moral failure are combined. The odds ratios show that such combinations are over twice as likely to trigger identity-salient cues than either category alone. The interaction remains directionally stable under week fixed effects and in a 200reply non-keyword subsample, indicating it is not an artefact of temporal spikes or trigger words. Marginal effects show the largest step change occurs when both Capacity- and Propriety- are present, consistent with a threshold (stacking) dynamic rather than linear additivity. This supports the theoretical claim that evaluative meaning operates configurationally, with bundled resources producing emergent stances (Hood, \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e2010\u003c/span\u003e; Bednarek, \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2008\u003c/span\u003e). From a communicational perspective, these compound configurations facilitate escalation: a competence critique (\"unable to govern\") readily recruits moral sanction (\"corrupt\"), thereby shifting the discourse from conditional critique to categorical exclusion (cf. Ex. 5.2a, 5.2c).\u003c/p\u003e\u003cp\u003eImportantly, these results bridge appraisal theory with broader discourse studies of delegitimisation. They show that gendered hostility is not reducible to lexical items or overt insults but emerges from systematic co-patterning of evaluative resources. Delegitimisation thus functions as a discursive strategy that is recognisable to participants, portable across contexts, and resonant in digital publics. By grounding delegitimising force in configurations rather than isolated categories, this study demonstrates how evaluation operates as a communicational mechanism for stance alignment and authority contestation in online interaction. Taken together, this points to a template-based communicational mechanism shaped by platform affordances (brevity, reply threading): compact evaluative bundles travel efficiently, organise stance alignment, and scale delegitimisation beyond single turns.\u003c/p\u003e\u003cp\u003e\u003cb\u003e5.3 RQ3. What Spanish-Italian contrasts appear in evaluative repertoires, and how do these patterns relate to identity-salient stance cues?\u003c/b\u003e\u003c/p\u003e\u003cp\u003eThe comparative analysis demonstrates that hostile replies to women leaders are structured by culturally differentiated evaluative repertoires. Spanish discourse was characterised by higher frequencies of Propriety- and Normality-, whereas Italian discourse foregrounded Capacity- and Veracity-. These distributional contrasts were reinforced by effect sizes (Cram\u0026eacute;r\u0026acute;s V\u0026thinsp;\u0026ge;\u0026thinsp;.27 for Propriety- and Normality-), indicating substantive differences in the evaluative resources recruited across languages. In keeping with these distributional tendencies, Spanish examples commonly mobilise moral sanction and abnormalisation (see Ex. 5.Зa-b), whereas Italian examples foreground competence/authenticity challenges, often with abnormalising invective (see Ex. 5.3c-d). These contrasts indicate not only different inventories of evaluative resources but different mechanisms of portability: Spanish moral-sanction/abnormalisation pairings circulate as ready-made formulas, whereas Italian competence/authenticity pairings circulate as performance-assessment scripts.\u003c/p\u003e\u003cp\u003e\u003cb\u003eEx. 5.3a (Ayuso, ES; Propriety\u0026minus; + Normality\u0026minus;)\u003c/b\u003e\u003c/p\u003e\u003cp\u003e(tweet_id: PepBausset_2025-04-11_08:04:06)\u003c/p\u003e\u003cp\u003e\u003cstrong\u003eOriginal (ES)\u003c/strong\u003e\u003cp\u003e\u0026ldquo;@user Ayuso \u003cb\u003es\u0026uacute;per corrupta\u003c/b\u003e, a los tribunales \u003cb\u003ePOR FIN\u003c/b\u003e.\u0026rdquo;\u003c/p\u003e\u003c/p\u003e\u003cp\u003e\u003cstrong\u003eTranslation (EN)\u003c/strong\u003e\u003cp\u003e\u0026ldquo;Ayuso \u003cb\u003esuper corrupt\u003c/b\u003e\u0026mdash;to the courts \u003cb\u003eAT LAST\u003c/b\u003e.\u0026rdquo;\u003c/p\u003e\u003c/p\u003e\u003cp\u003e\u003cb\u003eReading: Propriety\u0026minus;\u003c/b\u003e (\u003cem\u003ecorrupta\u003c/em\u003e); \u003cb\u003eNormality\u0026minus;\u003c/b\u003e (social deviance frame; graduation via capitals).\u003c/p\u003e\u003cp\u003e\u003cb\u003eEx. 5.3b (Ayuso, ES; Normality\u0026thinsp;\u0026minus;\u0026thinsp;with infantilisation)\u003c/b\u003e\u003c/p\u003e\u003cp\u003e(tweet_id: FerminYugueros_2025-04-11_07:12:34)\u003c/p\u003e\u003cp\u003e\u003cstrong\u003eOriginal (ES)\u003c/strong\u003e\u003cp\u003e\u0026ldquo;@user \u003cb\u003eEres rid\u0026iacute;cula\u003c/b\u003e a l\u0026iacute;mites insospechados.\u0026rdquo;\u003c/p\u003e\u003c/p\u003e\u003cp\u003e\u003cstrong\u003eTranslation (EN)\u003c/strong\u003e\u003cp\u003e\u0026ldquo;You are \u003cb\u003eridiculous\u003c/b\u003e beyond belief.\u0026rdquo;\u003c/p\u003e\u003c/p\u003e\u003cp\u003e\u003cb\u003eReading: Normality\u0026minus;\u003c/b\u003e (abnormalisation/infantilisation).\u003c/p\u003e\u003cp\u003e\u003cb\u003eEx. 5.3c (Meloni, IT; Capacity\u0026minus; + Veracity\u0026minus;, with sanctioning register)\u003c/b\u003e\u003c/p\u003e\u003cp\u003e(tweet_id: MEL_18_2025_0049)\u003c/p\u003e\u003cp\u003e\u003cstrong\u003eOriginal (IT)\u003c/strong\u003e\u003cp\u003e\u0026ldquo;@GiorgiaMeloni Cara fascistona \u003cb\u003efalsa\u003c/b\u003e e \u003cb\u003ebugiarda\u003c/b\u003e.\u0026rdquo;\u003c/p\u003e\u003c/p\u003e\u003cp\u003e\u003cstrong\u003eTranslation (EN)\u003c/strong\u003e\u003cp\u003e\u0026ldquo;Dear big fascist, \u003cb\u003efake\u003c/b\u003e and a \u003cb\u003eliar\u003c/b\u003e.\u0026rdquo;\u003c/p\u003e\u003c/p\u003e\u003cp\u003e\u003cb\u003eReading: Veracity\u0026minus;\u003c/b\u003e (\u003cem\u003efalsa, bugiarda\u003c/em\u003e) with sanctioning address; \u003cb\u003eCapacity/fitness\u003c/b\u003e implicitly impugned.\u003c/p\u003e\u003cp\u003e\u003cb\u003eEx. 5.3d (Meloni, IT; Normality\u0026minus; + Veracity\u0026minus;)\u003c/b\u003e\u003c/p\u003e\u003cp\u003e(tweet_id: MEL_15_2025_0002)\u003c/p\u003e\u003cp\u003e\u003cstrong\u003eOriginal (IT)\u003c/strong\u003e\u003cp\u003e\u0026ldquo;@GiorgiaMeloni Ma smettila \u003cb\u003ecessa\u003c/b\u003e carciofara \u003cb\u003ebugiarda\u003c/b\u003e\u0026hellip;\u0026rdquo;\u003c/p\u003e\u003c/p\u003e\u003cp\u003e\u003cstrong\u003eTranslation (EN)\u003c/strong\u003e\u003cp\u003e\u0026ldquo;Oh stop it, \u003cb\u003eugly\u003c/b\u003e artichoke-peddler, \u003cb\u003eliar\u003c/b\u003e\u0026hellip;\u0026rdquo;\u003c/p\u003e\u003c/p\u003e\u003cp\u003e\u003cb\u003eReading: Normality\u0026minus;\u003c/b\u003e (demeaning label)\u0026thinsp;+\u0026thinsp;\u003cb\u003eVeracity\u0026minus;\u003c/b\u003e (\u003cem\u003ebugiarda\u003c/em\u003e), consistent with Italian authenticity/competence frames.\u003c/p\u003e\u003cp\u003eIn practice, these formulas exhibit high uptake value: users reissue them with minimal lexical editing (e.g., replacing \u003cem\u003efalsa\u003c/em\u003e with \u003cem\u003eipocrita\u003c/em\u003e or intensifying \u003cem\u003ecorrotta\u003c/em\u003e with hashtags), preserving the same compound stance across turns and threads.\u003c/p\u003e\u003cp\u003eIn the predictive models, Spanish replies showed Propriety- as a key driver of identity-salient stance cues, whereas Italian replies highlighted Normality- and the Capacity\u0026times;Propriety configuration. These findings indicate that different evaluative paths can lead to the same communicational outcome: delegitimisation of women leaders through identity-salient positioning. In Spanish contexts, hostile users mobilise moral sanction and abnormalisation, echoing cultural scripts of dishonesty and impropriety (Ex. 5.3a). In Italian contexts, users rely more on competence and authenticity critiques, sometimes intensified when paired with moral delegitimation (Ex. 5.3c). Accordingly, the pathway to identity-salience differs by language: in Spanish, Propriety- alone often suffices; in Italian, Normality- and Capacity\u0026times;Propriety more commonly mark the tipping point. This suggests distinct thresholds for when evaluative talk becomes explicitly identity-salient.\u003c/p\u003e\u003cp\u003eThese cross-linguistic contrasts refine the claim that appraisal resources are universally available but locally patterned (Thompson \u0026amp; Alba-Juez, \u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e2014\u003c/span\u003e). While the semantic system of Judgement categories is stable, their deployment reflects culturally specific models of political legitimacy. In Spain, where public scandals are routinely framed in terms of corruption and propriety, moral sanction provides a salient frame for gendered hostility (Ex. 5.3a). In Italy, where public debate often foregrounds questions of authenticity and performativity, competence and veracity become central evaluative pathways (Ex. 5.3c-d). This supports a repertoire-within-system view of Appraisal: the Judgement system is shared, but repertoire activation is locally patterned by public cultures of legitimacy and scandal.\u003c/p\u003e\u003cp\u003eFrom a communicational perspective, these contrasts underscore how digital hostility embeds broader cultural repertoires into online interaction. Evaluative copatterns are not only individual choices but part of a shared resource system that participants draw upon to align stances and make attacks intelligible to their audiences. The recurrent pairings also act as audience-design shortcuts: moralsanction/abnormalisation in Spanish and competence/authenticity in Italian allow posters to index intended publics without elaboration. By showing that gendered delegitimisation is realised through different evaluative bundles across languages, this study contributes to comparative discourse research by situating online aggression within the cultural politics of evaluation.\u003c/p\u003e\u003cp\u003eFinally, robustness checks confirm that these contrasts are not artefacts of keyword filtering or event-specific spikes. The 200 non-keyword subsample yielded substantively similar results, and week fixed effects did not alter the direction of key predictors. These checks strengthen the claim that the observed contrasts represent stable evaluative repertoires rather than sampling artefacts. As a boundary condition, cross-language differences are distributional and configurational rather than absolute; all categories are available in both corpora, but activation probabilities and co-pattern strengths differ.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec18\" class=\"Section2\"\u003e\u003ch2\u003e5.4 Implications and Conclusion\u003c/h2\u003e\u003cp\u003eThe analyses across RQ1-RQ3 demonstrate that hostile discourse towards women leaders on X is communicationally structured, configurational in its evaluative deployment, and culturally differentiated in its repertoires. Three overarching implications follow.\u003c/p\u003e\u003cp\u003eTheoretically, by focusing on evaluative configurations rather than isolated categories, the study extends Appraisal Theory into a domain of configurational stance analysis. The finding that compound Judgement bundles (e.g., Capacity-+Propriety-) are especially likely to trigger identity-salient cues underscores the need to conceptualise evaluation as a relational system where resources recruit one another to produce emergent meanings. This perspective also refines theories of delegitimisation (Reisigl \u0026amp; Wodak, \u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e2001\u003c/span\u003e; Van Leeuwen, \u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e2007\u003c/span\u003e), showing that discursive authority is undermined not through single insults but through patterned co-occurrences that cumulatively erode legitimacy. Such patterned erosion illustrates how language functions communicationally to reframe individual failings as collective truths, enabling hostile publics to transform isolated utterances into a broader narrative of illegitimacy.\u003c/p\u003e\u003cp\u003eMethodologically, the corpus-based, cross-linguistic design illustrates how digital discourse analysis can integrate annotation, statistical modelling, and cultural comparison. The robust replication of results in a non-keyword subsample and under temporal controls indicates that these patterns are not artefacts of data selection but represent stable evaluative repertoires. This supports the feasibility of using Appraisal-informed protocols for large-scale, multilingual annotation, even when inter-coder reliability is moderate due to the interpretive nature of evaluative language. Beyond linguistics, this methodological approach offers a template for computational social science, political communication, and psychology, where questions of hostility, stance, and legitimacy similarly hinge on the patterned aggregation of microevaluations.\u003c/p\u003e\u003cp\u003eFrom the communicational perspective, hostility directed at women leaders functions as a public stance practice that draws on shared repertoires of evaluation. In Spain, moral sanction and abnormalisation dominate; in Italy, competence and authenticity critiques prevail. These contrasts show that gendered aggression is not only linguistic but also culturally anchored, reflecting national models of political legitimacy. In both contexts, however, hostility achieves the same communicational outcome: delegitimisation through identity-salient positioning that questions not just performance but the right to lead.\u003c/p\u003e\u003cp\u003eTogether, these findings demonstrate that digital political hostility is best understood as an organised communicational practice rather than incidental incivility. Our contribution lies in linking the microanalysis of evaluative resources to broader questions of authority, identity, and cultural repertoire. This intersection underscores how linguistic choices function within wider communicational ecologies, integrating insights from discourse analysis, sociolinguistics, and political communication. In doing so, the study contributes to a common theoretical framework for understanding digital hostility, where insights from rhetoric, pragmatics, and semiotics converge on the analysis of how communicational practices shape authority and identity in networked publics.\u003c/p\u003e\u003cp\u003eFuture research can extend this work by examining multimodal contributions (e.g., memes, images) within the same appraisal framework, tracing cross-platform dynamics, or exploring how male politicians are evaluated using comparable repertoires. Such extensions would test the generalisability of configurational stance analysis and further situate digital hostility within the interdisciplinary study of communication.\u003c/p\u003e\u003cp\u003e\u003cstrong\u003eEthics declaration\u003c/strong\u003e\u003cp\u003eThis research analyses publicly accessible posts from X (Twitter). No direct interaction with human participants took place, and no private or identifiable data were collected. According to the policies of the Universitat Polit\u0026egrave;cnica de Val\u0026egrave;ncia, formal ethical approval was not required as the study uses only publicly available data. The research complies with institutional guidelines and the platform's terms of service.\u003c/p\u003e\u003c/p\u003e\u003c/div\u003e"},{"header":"Declarations","content":"\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003eS.S. designed the study, built the corpus, developed the annotation protocol, performed statistical analyses/ visualisations, and drafted the manuscript. G.G. curated the Italian dataset, validated annotations, and contributed to analysis and writing. M.L.C.-P. and G.G. jointly supervised the work and provided critical revisions. All authors approved the final manuscript and agree to be accountable for all aspects of the work.\u003c/p\u003e\u003ch2\u003eData Availability\u003c/h2\u003e\u003cp\u003eThe study analyses public posts from X (Twitter). Full text cannot be redistributed under the platform\u0026rsquo;s Terms of Service. The tweet/reply ID lists, derived annotation files (Judgement labels, polarity, salience), and rehydration/analysis code are available from the corresponding author on reasonable request for research use.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eAgha, A. (2006). \u003cem\u003eLanguage and Social Relations\u003c/em\u003e (1st ed.). Cambridge University Press. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1017/CBO9780511618284\u003c/span\u003e\u003cspan address=\"10.1017/CBO9780511618284\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eBednarek, M. (2008). Emotion Talk Across Corpora. \u003cem\u003ePalgrave Macmillan UK\u003c/em\u003e. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1057/9780230285712\u003c/span\u003e\u003cspan address=\"10.1057/9780230285712\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eBednarek, M., \u0026amp; Caple, H. (2017). \u003cem\u003eThe Discourse of News Values: How News Organizations Create Newsworthiness\u003c/em\u003e. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.amazon.com/Discourse-News-Values-Organizations-Newsworthiness/dp/0190653949\u003c/span\u003e\u003cspan address=\"https://www.amazon.com/Discourse-News-Values-Organizations-Newsworthiness/dp/0190653949\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eCombei, C. R., \u0026amp; Reggi, V. (2024). \u003cem\u003eAppraisal, sentiment and emotion analysis in political discourse: A multimodal, multi-method approach\u003c/em\u003e. Routledge.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003edel Saz-Rubio, M. M. (2024). You are not empowered, you have neither character nor pride\u0026rsquo;: Assessing aggressive language against Spanish female politicians in high-profile positions. \u003cem\u003eJournal of Language Aggression and Conflict\u003c/em\u003e. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1075/jlac.00109.del\u003c/span\u003e\u003cspan address=\"10.1075/jlac.00109.del\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eDu Bois, J. W. (2007). The stance triangle. In R. Englebretson (Ed.), \u003cem\u003ePragmatics \u0026amp; Beyond New Series\u003c/em\u003e (Vol. 164, pp. 139\u0026ndash;182). John Benjamins Publishing Company. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1075/pbns.164.07du\u003c/span\u003e\u003cspan address=\"10.1075/pbns.164.07du\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eErikson, J., H\u0026aring;kansson, S., \u0026amp; Josefsson, C. (2023). Three Dimensions of Gendered Online Abuse: Analyzing Swedish MPs\u0026rsquo; Experiences of Social Media. \u003cem\u003ePerspectives on Politics\u003c/em\u003e, \u003cem\u003e21\u003c/em\u003e(3), 896\u0026ndash;912. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1017/S1537592721002048\u003c/span\u003e\u003cspan address=\"10.1017/S1537592721002048\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eEsposito, E., \u0026amp; Breeze, R. (2022). Gender and politics in a digitalised world: Investigating online hostility against UK female MPs. \u003cem\u003eDiscourse \u0026amp; Society\u003c/em\u003e, \u003cem\u003e33\u003c/em\u003e(3), 303\u0026ndash;323. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1177/09579265221076608\u003c/span\u003e\u003cspan address=\"10.1177/09579265221076608\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eFuoli, M. (2018). A stepwise method for annotating: Appraisal. \u003cem\u003eFunctions of Language\u003c/em\u003e, \u003cem\u003e25\u003c/em\u003e, 229\u0026ndash;258. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1075/fol.15016.fuo\u003c/span\u003e\u003cspan address=\"10.1075/fol.15016.fuo\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHabermas, J., \u0026amp; Rehg, W. (2010). \u003cem\u003eBetween Facts and Norms: Contributions to a Discourse Theory of Law and Democracy\u003c/em\u003e. Polity. Reprinted.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eH\u0026aring;kansson, S. (2021). Do Women Pay a Higher Price for Power? Gender Bias in Political Violence in Sweden. \u003cem\u003eThe Journal of Politics\u003c/em\u003e, \u003cem\u003e83\u003c/em\u003e(2), 515\u0026ndash;531. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1086/709838\u003c/span\u003e\u003cspan address=\"10.1086/709838\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eH\u0026aring;kansson, S. (2024). The Gendered Representational Costs of Violence against Politicians. \u003cem\u003ePerspectives on Politics\u003c/em\u003e, \u003cem\u003e22\u003c/em\u003e(1), 81\u0026ndash;96. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1017/S1537592723001913\u003c/span\u003e\u003cspan address=\"10.1017/S1537592723001913\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHayes, A. F., \u0026amp; Krippendorff, K. (2007). Answering the Call for a Standard Reliability Measure for Coding Data. \u003cem\u003eCommunication Methods and Measures\u003c/em\u003e, \u003cem\u003e1\u003c/em\u003e(1), 77\u0026ndash;89. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1080/19312450709336664\u003c/span\u003e\u003cspan address=\"10.1080/19312450709336664\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHood, S. (2010). \u003cem\u003eAppraising research: Evaluation in academic writing\u003c/em\u003e (1. publ). Palgrave Macmillan.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eJane, E. A. (2014). Back to the kitchen, cunt\u0026rsquo;: Speaking the unspeakable about online misogyny. \u003cem\u003eContinuum\u003c/em\u003e, \u003cem\u003e28\u003c/em\u003e(4), 558\u0026ndash;570. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1080/10304312.2014.924479\u003c/span\u003e\u003cspan address=\"10.1080/10304312.2014.924479\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKhosraviNik, M. (Ed.). (2023). \u003cem\u003eSocial media critical discourse studies\u003c/em\u003e. Routledge.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLaclau, E. (2005). \u003cem\u003eOn populist reason\u003c/em\u003e (1. publ). Verso.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMantilla, K. (2015). Gendertrolling: How misogyny went viral. \u003cem\u003ePraeger\u003c/em\u003e. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.5040/9798400656026\u003c/span\u003e\u003cspan address=\"10.5040/9798400656026\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMartin, J. R., \u0026amp; White, P. R. R. (2005). The Language of Evaluation. \u003cem\u003ePalgrave Macmillan UK\u003c/em\u003e. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1057/9780230511910\u003c/span\u003e\u003cspan address=\"10.1057/9780230511910\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMeril\u0026auml;inen, J. M. (2024). The Role of Gender in Hate Speech Targeting Politicians: Evidence from Finnish Twitter. \u003cem\u003eInternational Journal of Politics Culture and Society\u003c/em\u003e. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s10767-024-09476-3\u003c/span\u003e\u003cspan address=\"10.1007/s10767-024-09476-3\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMoffitt, B. (2016). \u003cem\u003eThe global rise of populism: Performance, political style, and representation\u003c/em\u003e. Stanford University Press.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMoreno-Ortiz, A., \u0026amp; Garc\u0026iacute;a-G\u0026aacute;mez, M. (2023). Strategies for the Analysis of Large Social Media Corpora: Sampling and Keyword Extraction Methods. \u003cem\u003eCorpus Pragmatics\u003c/em\u003e, \u003cem\u003e7\u003c/em\u003e(3), 241\u0026ndash;265. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s41701-023-00143-0\u003c/span\u003e\u003cspan address=\"10.1007/s41701-023-00143-0\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMouffe, C. (2011). \u003cem\u003eOn the Political\u003c/em\u003e (0 edn). Routledge. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.4324/9780203870112\u003c/span\u003e\u003cspan address=\"10.4324/9780203870112\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eNorris, P. (2011). \u003cem\u003eDemocratic deficit: Critical citizens revisited\u003c/em\u003e. Cambridge University Press.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eOchs, E. (1992). Indexing gender. In A. Duranti, \u0026amp; C. Goodwin (Eds.), \u003cem\u003eRethinking Context: Language as an Interactive Phenomenon\u003c/em\u003e (pp. 335\u0026ndash;358). Cambridge University Press.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003ePapacharissi, Z. (2014). \u003cem\u003eAffective Publics: Sentiment, Technology, and Politics\u003c/em\u003e. Oxford University Press. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1093/acprof:oso/9780199999736.001.0001\u003c/span\u003e\u003cspan address=\"10.1093/acprof:oso/9780199999736.001.0001\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003ePierri, F. (2024). Drivers of hate speech in political conversations on Twitter: The case of the 2022 Italian general election. \u003cem\u003eEPJ Data Science\u003c/em\u003e, \u003cem\u003e13\u003c/em\u003e(1), 63. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1140/epjds/s13688-024-00501-1\u003c/span\u003e\u003cspan address=\"10.1140/epjds/s13688-024-00501-1\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eReisigl, M., \u0026amp; Wodak, R. (2001). \u003cem\u003eDiscourse and discrimination: Rhetorics of racism and antisemitism\u003c/em\u003e. Routledge.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eSilverstein, M. (2003). Indexical order and the dialectics of sociolinguistic life. \u003cem\u003eLanguage \u0026amp; Communication\u003c/em\u003e, \u003cem\u003e23\u003c/em\u003e(3\u0026ndash;4), 193\u0026ndash;229. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/S0271-5309(03)00013-2\u003c/span\u003e\u003cspan address=\"10.1016/S0271-5309(03)00013-2\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eSouthern, R., \u0026amp; Harmer, E. (2021). Twitter, Incivility and Everyday Gendered Othering: An Analysis of Tweets Sent to UK Members of Parliament. \u003cem\u003eSocial Science Computer Review\u003c/em\u003e, \u003cem\u003e39\u003c/em\u003e(2), 259\u0026ndash;275. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1177/0894439319865519\u003c/span\u003e\u003cspan address=\"10.1177/0894439319865519\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eThompson, G., \u0026amp; Alba-Juez, L. (Eds.). (2014). \u003cem\u003eEvaluation in Context\u003c/em\u003e (Vol. 242). John Benjamins Publishing Company. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1075/pbns.242\u003c/span\u003e\u003cspan address=\"10.1075/pbns.242\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eTrifiro, B. M., Paik, S., Fang, Z., \u0026amp; Zhang, L. (2021). Politics and Politeness: Analysis of Incivility on Twitter During the 2020 Democratic Presidential Primary. \u003cem\u003eSocial Media\u0026thinsp;+\u0026thinsp;Society\u003c/em\u003e, \u003cem\u003e7\u003c/em\u003e(3), 20563051211036939. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1177/20563051211036939\u003c/span\u003e\u003cspan address=\"10.1177/20563051211036939\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eVan Leeuwen, T. (2007). Legitimation in discourse and communication. \u003cem\u003eDiscourse \u0026amp; Communication\u003c/em\u003e, \u003cem\u003e1\u003c/em\u003e(1), 91\u0026ndash;112. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1177/1750481307071986\u003c/span\u003e\u003cspan address=\"10.1177/1750481307071986\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eZappavigna, M. (2012). \u003cem\u003eThe discourse of Twitter and social media\u003c/em\u003e. Continuum International Pub. Group.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eZappavigna, M. (2018). \u003cem\u003eSearchable talk: Hashtags and social media metadiscourse\u003c/em\u003e. Bloomsbury Academic, an imprint of Bloomsbury Plc.\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-7820069/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7820069/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eThis study examines evaluative configurations in hostile digital political discourse targeting women leaders across Spanish and Italian contexts. Through systematic analysis of 2,000 replies to Isabel D\u0026iacute;az Ayuso and Giorgia Meloni on X, we identify culturally distinct evaluative repertoires that function as delegitimization mechanisms. Spanish discourse emphasizes moral sanctioning through Propriety-Normality combinations (59.6% vs. 37.0% Propriety in Italian replies), while Italian discourse foregrounds competence-authenticity challenges through Capacity-Veracity configurations. Logistic regression reveals systematic relationships between evaluative configurations and identity-salient stance markers, with Tenacity (OR\u0026thinsp;=\u0026thinsp;6.45) and Normality (OR\u0026thinsp;=\u0026thinsp;3.83) judgements emerging as strongest predictors. The findings advance theoretical frameworks in stance analysis and Appraisal Theory by demonstrating how evaluative copatterns create compound delegitimizing effects beyond individual category contributions, while revealing persistent cultural influences on evaluative practice within shared digital environments.\u003c/p\u003e","manuscriptTitle":"Cross-linguistic Delegitimisation of Women Leaders in Online Political Discourse","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-10-27 11:40:02","doi":"10.21203/rs.3.rs-7820069/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"8330fb51-8df4-4bf9-a93b-75c640babeae","owner":[],"postedDate":"October 27th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2025-12-22T16:05:57+00:00","versionOfRecord":{"articleIdentity":"rs-7820069","link":"https://doi.org/10.1007/s41701-025-00221-5","journal":{"identity":"corpus-pragmatics","isVorOnly":false,"title":"Corpus Pragmatics"},"publishedOn":"2025-12-18 15:58:19","publishedOnDateReadable":"December 18th, 2025"},"versionCreatedAt":"2025-10-27 11:40:02","video":"","vorDoi":"10.1007/s41701-025-00221-5","vorDoiUrl":"https://doi.org/10.1007/s41701-025-00221-5","workflowStages":[]},"version":"v1","identity":"rs-7820069","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7820069","identity":"rs-7820069","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}