Cyberbullying among Chinese social media users has been described in relation to hostile words, pragmatic meanings, group alignment, identity positioning, and platform-mediated amplification. This scoping review mapped the linguistic and behavioral mechanisms reported in English-language studies indexed in Scopus and Web of Science concerning target-directed harm in Chinese social media contexts. Following the Arksey and O’Malley framework, Levac refinements, and PRISMA-ScR guidance, searches of Scopus and Web of Science identified 592 records. The 2015–2025 publication-date criterion was applied during screening. After deduplication, screening, and full-text eligibility assessment, 19 sources were included. Data were charted and synthesized thematically across lexical and lexico-semantic features, pragmatic strategies, discursive processes, and platform conditions. Four patterns emerged: evaluative labeling and semantic degradation; indirect aggression conveyed through irony, metaphor, insinuation, and culturally situated meanings; participant alignment and identity construction associated with legitimized hostility; and repetition, visibility, and collective participation associated with amplified harmful communication. Within this indexed English-language evidence base, cyberbullying was recurrently represented as a cumulative communicative process in which linguistic actions acquired harmful force through interpretation, social positioning, and social and technological circulation. Effective identification therefore requires attention to contextual meaning, interactional sequences, target relations, and platform dynamics, rather than reliance on isolated offensive terms. The review offers a provisional behavioral–linguistic framework derived from this indexed English-language evidence base while identifying methodological, platform, language, and database gaps for future empirical research.

Cyberbullying is a media-psychological phenomenon enacted through language, interpreted within relationships, and conditioned by platform affordances. Yet linguistic and pragmatic research remains fragmented across psychology, communication, media studies, pragmatics, and computational detection. This bibliometric review maps the development and intellectual structure of this research from 2010 to 2025 using 345 English-language records from the Web of Science Core Collection. Annual output increased from 1 publication in 2010 to 51 in 2025, and 218 records (63.2%) were published after 2020. The mapping identified prominent psychosocial and computational knowledge structures, while explicitly pragmatic constructs remained comparatively less consolidated in high-frequency keyword and burst patterns. To interpret these mapped patterns, the second half of the review is organized through four non-mutually exclusive analytical lenses: psychosocial victimization, lexical-semantic harm, pragmatics and discourse, and computational detection. Drawing on the bibliometric findings and theory-informed interpretation, we propose a conceptual language-to-harm framework linking platform affordances, linguistic cues, pragmatic interpretation, psychological appraisal, behavioral and social responses, and cumulative harm. The framework is presented as a theory-informed synthesis and future research model rather than as a causal structure empirically validated by the bibliometric data.