A Computational Analysis of the Indus Script: Identifying Sign Functions in Logo-Syllabic Writing Systems

A key challenge in deciphering any logo-syllabic writing system is to distinguish signs that represent complete lexical units (logograms) from signs that represent syllables.The Indus script is widely assumed to be logo-syllabic, an intermediate system in which some signs encode whole words or concepts, others encode syllables, and some function polyvalently depending on context.This ambiguity complicates decipherment: a single grapheme may serve as a word in one inscription and as a syllable in another.This study proposes methods for separating these classes.To begin, one basic premise is assumed: signs that appear alone in inscriptions must be capable of expressing a complete semantic value and hence are the strongest candidates for logograms.Building on this "singleton" premise, the candidate logogram set was expanded by contextual co-occurrence, associations were validated statistically, and distributional relationships were then mapped among signs and sign-pairs.A second component develops an exclusivity-based procedure for identifying likely syllabic pairs.The results demonstrate several key categories of Indus signs.