Marsji attempts to separate meaning from sound by treating conceptual identity as the primary layer and pronunciation as a secondary layer. In most natural languages, words are heavily shaped by historical evolution, regional accents, cultural influences, and phonetic traditions. This creates situations where words sound similar but have unrelated meanings, or where the same concept has radically different forms across languages.
Marsji approaches this differently by first defining a stable semantic core for a concept before assigning a structured linguistic form to it. The generated terms are intended to represent conceptual consistency rather than historical phonetics. In practice, this means the language prioritizes semantic mapping over cultural familiarity. The project also explores deterministic patterns where words follow internal structural logic instead of random linguistic evolution.
By doing so, Marsji tries to reduce semantic ambiguity and create a system where meaning remains stable even when translated across different human languages. The ultimate goal is to make conceptual relationships clearer for both humans and AI systems.