This study investigates the linguistic characteristics, pragmatic functions, and legal implications of formal and humorous speech across digital media by employing a forensic linguistic and multimodal analytical framework. Drawing on data from the Indonesian comedy program Lapor Pak! and the Korean film Vincenzo, the research examines how formal and humorous utterances are constructed, interpreted, and potentially contested within online and audiovisual environments. The analysis focuses on the interplay between linguistic form, speaker intention, and contextual cues, as well as the extent to which multimodal elements—such as visual cues, sound effects, gestures, and cinematography—shape or intensify meaning. Findings reveal that formal speech, although structurally explicit, may still generate legal vulnerabilities when embedded in parody or institutional impersonation. Conversely, humorous speech, characterized by ambiguity, irony, and linguistic deviation, is prone to misinterpretation as defamation, hate speech, or harassment, particularly in digital settings lacking shared contextual knowledge. The study further argues that multimodality amplifies interpretive risks by reinforcing or altering the illocutionary force of an utterance. Overall, this research demonstrates the value of integrating pragmatic, discourse, and multimodal perspectives within forensic linguistic analysis to better understand how digital communication practices may trigger social and legal consequences. The findings contribute to the development of digital literacy, ethical communication guidelines, and more context-sensitive approaches to evaluating public speech in online environments.