Analysing Lexical Semantic Change with Contextualised Word Representations
This work addresses the challenge of tracking word meaning shifts over time for linguists and NLP researchers, representing an incremental advance by applying BERT to an existing problem.
The paper tackles the problem of lexical semantic change by introducing the first unsupervised method using contextualized word representations from BERT, clustering them into usage types and measuring change over time with new metrics, and shows positive correlation with human judgments on a new evaluation dataset.
This paper presents the first unsupervised approach to lexical semantic change that makes use of contextualised word representations. We propose a novel method that exploits the BERT neural language model to obtain representations of word usages, clusters these representations into usage types, and measures change along time with three proposed metrics. We create a new evaluation dataset and show that the model representations and the detected semantic shifts are positively correlated with human judgements. Our extensive qualitative analysis demonstrates that our method captures a variety of synchronic and diachronic linguistic phenomena. We expect our work to inspire further research in this direction.