# Time-Out: Temporal Referencing for Robust Modeling of Lexical Semantic Change

## TL;DR

This paper introduces Temporal Referencing, a method for detecting lexical semantic change that avoids vector space alignment, leading to more robust results demonstrated through empirical testing on synthetic and manual datasets.

## Contribution

It presents the Temporal Referencing approach, which improves lexical semantic change detection by eliminating alignment noise, and provides a systematic way to simulate and control for biases.

## Key findings

- Temporal Referencing outperforms alignment models on synthetic data
- The method is less affected by noise from vector space alignment
- Systematic simulation of lexical semantic change enhances evaluation

## Abstract

State-of-the-art models of lexical semantic change detection suffer from noise stemming from vector space alignment. We have empirically tested the Temporal Referencing method for lexical semantic change and show that, by avoiding alignment, it is less affected by this noise. We show that, trained on a diachronic corpus, the skip-gram with negative sampling architecture with temporal referencing outperforms alignment models on a synthetic task as well as a manual testset. We introduce a principled way to simulate lexical semantic change and systematically control for possible biases.