Transformer-Enabled Diachronic Analysis of Vedic Sanskrit: Neural Methods for Quantifying Types of Language Change

TL;DR

This paper introduces a hybrid neural-symbolic approach using weak supervision and ensemble methods to analyze 2000 years of Sanskrit evolution, revealing complex morphological redistribution rather than simplification.

Contribution

It presents a novel scalable, interpretable framework combining symbolic and neural methods for diachronic linguistic analysis of low-resource languages.

Findings

Sanskrit's morphological complexity redistributes over time

Verbal features decline cyclically, replaced by compounding and new terminology

Ensemble achieves 52.4% feature detection rate with well-calibrated uncertainty

Abstract

This study demonstrates how hybrid neural-symbolic methods can yield significant new insights into the evolution of a morphologically rich, low-resource language. We challenge the naive assumption that linguistic change is simplification by quantitatively analyzing over 2,000 years of Sanskrit, demonstrating how weakly-supervised hybrid methods can yield new insights into the evolution of morphologically rich, low-resource languages. Our approach addresses data scarcity through weak supervision, using 100+ high-precision regex patterns to generate pseudo-labels for fine-tuning a multilingual BERT. We then fuse symbolic and neural outputs via a novel confidence-weighted ensemble, creating a system that is both scalable and interpretable. Applying this framework to a 1.47-million-word diachronic corpus, our ensemble achieves a 52.4% overall feature detection rate. Our findings reveal that Sanskrit's overall morphological complexity does not decrease but is instead dynamically redistributed: while earlier verbal features show cyclical patterns of decline, complexity shifts to other domains, evidenced by a dramatic expansion in compounding and the emergence of new philosophical terminology. Critically, our system produces well-calibrated uncertainty estimates, with confidence strongly correlating with accuracy (Pearson r = 0.92) and low overall calibration error (ECE = 0.043), bolstering the reliability of these findings for computational philology.