Diachronic Synonymy and Polysemy: Exploring Dynamic Relation Between Forms and Meanings of Words Based on Word Embeddings

Abstract

In recent years, there has been a large number of publications that use distributed methods to track temporal changes in lexical semantics. However, most current researches only state the simple fact that the meaning of words has changed, lacking more detailed and in-depth analysis. We combine linguistic theory and word embedding model to study Chinese diachronic semantics. Specifically, two methods of word analogy and word similarity are associated with diachronic synonymy and diachronic polysemy respectively, and the aligned diachronic word embeddings are used to detect the changes of relationship between forms and meanings of words. Through experiments and case studies, our method achieves the ideal result. We also find that the evolution of Chinese vocabulary is closely related to social development, and there is a certain correlation between the polysemy and synonymy of the word meaning.