Dynamic Language DNA: A Universal Method for Context-Dependent Translation Key Discovery

This theoretical paper develops Language DNA (L-DNA) as a hypothesis about the discovery of translation relationships across human languages. Its principal proposition is that universality may reside not in a fixed dictionary or permanent language-pair mapping, but in a reusable inference procedure that constructs a context-dependent translation key for each utterance. The proposed method forms multiscale representations of source expressions, integrating acoustic modulation when speech is available, temporal recurrence, frequency of linguistic usage, morphosyntactic dependencies, semantic relations and discourse context. It searches for alignments with target-language representations under limited, independently grounded semantic constraints and reconstructs candidate meanings before producing a target expression. A bounded recursive diagnostic mechanism tests structural coherence, locates suspected analytical failures and retains evidence-qualified corrections in persistent memory. Six falsifiable hypotheses address conditional-key benefits, multiscale representations, partial relational invariance, anchor-efficient inference, error localisation and transfer across linguistically diverse language pairs. The paper distinguishes formal correspondence from semantic equivalence: internal coherence and back-reconstruction can reveal errors but cannot themselves certify an intended meaning. The universal discovery procedure is thus proposed as an empirically testable research programme, not an accomplished translation technology. Initial validation is confined to human languages with independently evaluable meanings; extension to other communicative systems is left for later theoretical and empirical work. No experimental findings are reported.