Constructing identities in institutional impersonation fraud: self-styling and other-styling practices through stances

The paper addresses institutional impersonation fraud (IIF) conducted via phone calls, a prevalent form of telecom fraud in China where scammers impersonate institutional actors (e.g., banks, law enforcement) to coerce victims into payments. Official statistics highlight the scale of the problem, with billions of fraudulent calls intercepted and impersonation scams among the most common. Prior discourse research has focused largely on email or romance fraud, leaving phone-based IIF underexplored, particularly the dynamic, real-time nature of interaction and the construction of multiple participant identities (fraudsters, targets, non-present others). Adopting a discourse-analytic, identity-foregrounded approach, the study combines stance and style to examine how fraudsters project self-, target-, and opponent-identities through indirect, multiple indexical paths and how these practices affect rapport. Research questions: (Q1) How do fraudsters perform self-styling and other-styling via stances in IIF interaction? (Q2) What interactive patterns emerge from these stance-based styling practices?

Phone fraud discourse differs from email fraud in its highly dynamic, interactive nature; fraudsters adapt in real-time to targets’ responses, constructing fluid yet patterned false institutional identities. Prior work on phone fraud has examined threatening language, modus operandi, and fraudsters’ self-identity construction, but has underexplored the indexical pathways by which semiotic resources project identities and how targets and non-present others are styled. The study draws on sociolinguistic theories of style/styling and stance: styles are coherent bundles of signs derived from repertoires of stance moves; repeated stance-taking creates locally recognizable styles and identities via multiple indexical paths. Intertextuality and interdiscursivity underpin stance-taking through de/recontextualization of prior texts and discourses. Rapport management theory (Spencer-Oatey) is used to interpret how linguistic choices manage face, sociality rights/obligations, and interactional goals, shaping relational dynamics among fraudsters, targets, and third parties in criminal contexts.

Data: 26 authentic audio/video recordings of IIF interactions sourced from mainstream media (e.g., Tencent, CCTV), bilibili, and crowdsourced via students/colleagues/friends (with permission). Personally identifying information was removed or altered. Recordings were transcribed using IFLYREC and manually checked; total word count: 87,834. Meta-information (sources, durations, links) is in Appendix 1. Across cases, typical moves include: initial impersonation to construct a troublesome scenario implicating the target; invoking information leakage and proposing to report to police; transferring the call to additional impersonators; isolating the target; obtaining account information and executing transfers. Analysis: A discourse-analytic, identity-foregrounded approach. Authors independently annotated stance instances for self-, target-, and opponent-styling, reconciled discrepancies by majority agreement, and then examined linguistic features within each stance and cross-stance configurations. Coding followed Dörnyei’s qualitative protocol. The analysis is informed by rapport management (face, sociality rights/obligations, interactional goals) and by intertextuality/interdiscursivity (de/recontextualization of institutional discourse, laws, media reports) to track indexical paths from linguistic resources to stances, styles, and identities.

• Representative stances identified (occurrence across 26 cases): self-styling—professional authority (26/26), target avidity (19/26), community integrity (10/26); target-styling—being troublesome (26/26), being innocent (21/26), needing help (17/26); opponent-styling—negative stances toward culprits/obstacles (21/26).
• Self-styling via stances: • Professional authority: enacted through institutional titles and pronouns (e.g., “our police”), technical/legal jargon, formulaic self-introductions (names, badge numbers), identity verification routines, and didactic “educational talks.” Co-constructed with rapport challenges (e.g., casting targets as suspects, orders/threats) that damage quality/identity face and equity rights to secure compliance. • Target avidity: fraudsters display eagerness to help/protect and align with targets’ interactional goals via commissives, grounding moves, respect for apparent choice (interrogatives seeking consent), and in-group markers (inclusive “we”), thereby building solidarity while maintaining authority. • Community integrity: self-presentation of veracity and propriety (e.g., denials of deceptive intent; appeals to fairness and just law enforcement), upholding fraudsters’ quality/identity face.
• Target-styling via stances: • Being troublesome: targets framed as responsible for trouble (e.g., suspects), often via reported speech to create objectivity/authority; reinforced with threats (lawsuits, deductions) and stern tones; imposition is strategically alleviated with solutions/politeness to motivate cooperation. • Being innocent: negotiated through a “probability duel” about information leakage, heavily leveraging intertextual references (news, precedent cases). Acceptance enables alignment on reporting the case and shared pursuit of the “real culprits.” Innocent and Troublesome stances are dynamically alternated to propel action. • Needing help: targets positioned as ignorant and in need of guidance; frequent suggestions/instructions align with presumed goals; linguistic bombardment can shift targets from passive to active help-seekers.
• Opponent-styling via negative stances: • Culprits: labeled with derogatory, criminalizing terms (e.g., wrongdoers, criminal gangs), constructing an out-group and consolidating targets’ innocence while fostering a “we-ness” with fraudsters. • Obstacles: humans (family, roommates, friends) and non-humans (anti-fraud messages, app prompts) framed as threats to valid recording or case handling; accusations (e.g., coaching false testimony) and authoritative pretexts used to justify isolation and information checks; promotes egocentric decision-making and blocks alternative information.
• Interactive patterns configured through stances: • Antagonizing pattern: initial out-grouping of targets via Troublesome stance against impersonated professionals. • Pseudo-allied pattern: in-group formation between fraudsters and targets (Innocent + Needing help + Target avidity), while culprits/obstacles are out-grouped to enable cooperative problem-solving. • Selective Stance-accentuation pattern: see-saw alternation between Troublesome and Innocent stances based on perceived compliance, leveraging fear (threat) and hope (relief) to reduce deeper information processing and drive targets to follow directives (e.g., “reporting,” credential sharing, disguised transfers).

The stance repertoire exhibits patterned deployment aligned with shifting local goals across the fraud timeline. Fraudsters begin by antagonizing targets (imposing Troublesome stance), then pivot to a pseudo-alliance (Innocent/Needing help for targets; Target avidity/Community integrity/Professional authority for self; negative stances for opponents) to foster a cooperative in-group aimed at “solving” the fabricated problem. Professional authority and community integrity operate throughout to accrue symbolic capital, enhancing credibility and compliance. The selective accentuation of Troublesome vs. Innocent stances functions as affective control—invoking fear and hope to curtail deliberative processing and induce action—thereby operationalizing the broader manipulative strategy via rapport management and intertextual legitimation.

The study identifies recurrent stance-based styling practices in IIF: self-styling (professional authority, target avidity, community integrity), target-styling (being troublesome, being innocent, needing help), and opponent-styling (negative stances toward culprits and obstacles). It also delineates three interactive patterns—Antagonizing, Pseudo-allied, and Selective Stance-accentuation—that configure how these stances interplay to progress scams. These insights offer practical markers for anti-fraud detection and education, such as scripted negotiations around information leakage and oscillation between accusation and exoneration. Future research should extend beyond linguistic resources to multimodal features and examine stance/styling practices across other telecom/online fraud types (e.g., romance fraud).

The study focuses on linguistic representation of stance construction within IIF and does not analyze multimodal features (e.g., prosody) or other fraud genres. Findings are based on 26 cases and may not generalize to all settings; recordings may not cover complete scam processes. Future work should incorporate multimodal analysis and explore additional fraud types and contexts.