<![CDATA[Digital communication has expanded cybercrime that relies on language to threaten, extort, and deceive victims. Research in Indonesia has not yet produced validated criminal linguistic parameters that can support an automated language analysis application. This study develops forensic linguistic parameters for identifying online threats, extortion, and fraud and designs a conceptual application based on artificial intelligence and natural language processing. A qualitative descriptive design was applied to 200 digital documents comprising 40 final court decisions, 30 official cybercrime reports, 50 officially published fraudulent messages, and 80 simulated messages derived from authentic case characteristics. Data were analyzed through text preprocessing, pragmatic, semantic, speech act, and discourse analysis, followed by open, axial, and selective coding with NVivo 14. The study identified twelve parameters: urgency, persuasive strategies, imperative sentences, promises of financial gain, emotional manipulation, institutional impersonation, implicit threats, identity fraud, explicit threats, false authority, exploitation of social relationships, and economic pressure. These parameters informed a conceptual system containing preprocessing, parameter extraction, risk classification, and reporting modules. Source triangulation reached 91.7%, theoretical triangulation 92.8%, and expert judgment 91.1%. The findings provide a validated foundation for AI-assisted early detection, cybercrime investigation, and digital security literacy in Indonesia.]]>
