Social media platforms, particularly TikTok, have become primary arenas for linguistic experimentation among adolescents, yet systematic analyses of how platform-specific affordances shape lexical and semantic innovation remains limited. This study investigated lexical and semantic variations in adolescent digital communication on TikTok, addressing three research questions concerning the types of lexical innovations, processes of semantic change, and the role of platform affordances in shaping language evolution. Methods: A mixed-methods design integrated quantitative corpus linguistics with qualitative discourse analysis. A corpus of 2,848 TikTok comments was compiled across four major trends (September–December 2024). Lexical analysis identified neologisms, graphical variations, and acronyms; semantic analysis documented broadening, narrowing, metaphoric extension, and pejoration/amelioration; platform affordances analysis examined meme-driven language and intertextual policing. Analysis revealed 15 lexical innovations with 63 occurrences across semantic categories. Neologisms (fr, bestie, delulu) and graphical variations (tryna, cuz, ion) served dual functions of efficiency and identity performance. Semantic shifts included ameliorative broadening (slay, fire), pejoration (basic, cringe), metaphoric extension (era, main character), and reclamatory usage (ghetto). Platform analysis identified 11 meme-driven phrases generating 2,848 occurrences with near-neutral sentiment, and 347 policing instances (12.2%) concentrated during rising and peak trend phases, demonstrating active semantic negotiation through definition, debate, and correction. TikTok functions as an accelerated laboratory for language change where adolescents deploy multiple mechanisms of linguistic innovation simultaneously. Platform affordances fundamentally reshape traditional sociolinguistic processes, with intertextual policing serving as the mechanism by which communities enforce emerging semantic norms. The findings extend communities of practice frameworks to algorithmically-mediated digital environments. Educators should recognize digital language as systematic innovation; lexicographers should develop protocols for documenting ephemeral platform-specific terms; platform designers should account for in-group reclamation practices; and researchers should prioritize cross-platform longitudinal studies to track whether observed innovations represent enduring change or age-graded phenomena.