Conventional research on phoney speech in legal circumstances typically views deception as a moral or cognitive defect at the individual level that can be recognised by consistent linguistic indicators. By suggesting that deception is an institutionally created discourse practice that results from the interplay of cognitive load, procedural limitations, and power imbalances in courtroom communication, this research proposes a theoretical reorientation. The study uses a mixed-methods strategy that combines qualitative forensic analysis with natural language processing techniques applied to specific Indian criminal court rulings, drawing on forensic linguistics, discourse analysis, and computational language modelling. Patterns of strategic ambiguity, evasive coherence, emotional modulation, and pragmatic indeterminacy are found in the testimonies of witnesses and accused individuals. To track how institutional forces influence communication behaviour, computational methods such as sentiment trajectory mapping, stance identification, and lexical dispersion metrics are combined with careful language reading. The analysis shows that deceitful discourse in legal contexts functions more as an adaptive, situationally sensible tactic conditioned by juridical norms and interpretive authority than as a sign of personal dishonesty. This study offers a reusable analytical framework for forensic linguistics and legal discourse studies by modelling deception as a situated, procedural, and culturally mediated phenomenon. This framework has implications for judicial interpretation, evidentiary evaluation, and the moral use of computational tools in legal contexts.