Sanskrit texts reveal shifting meanings over centuries
Summary and headline written by AI from the source article. How we work
Researchers developed a computational method to trace how the meanings of words changed in Sanskrit literature over time.
The technique, tested on a corpus of 2.7 million tokens spanning four historical periods of the ancient Indian language, overcomes challenges posed by its complex grammar. The system accurately identifies shifts in word meaning by comparing changes in word embeddings, mathematical representations of words, to established historical linguistic data.
The analysis successfully predicted the direction of 19 out of 21 known semantic changes, confirming the method’s effectiveness for a low-resource language with features like sandhi (phonological fusion) and extensive inflection. The research indicates that Sanskrit forces specific configurations within the computational model, offering insights into the language’s structure. This work demonstrates the potential of computational linguistics to study ancient texts and opens avenues for exploring semantic shifts in other historically significant, complex languages.
Further refinement of the sandhi splitter and lemmatizer could improve the accuracy of the system.
