semantic-features: A User-Friendly Tool for Studying Contextual Word Embeddings in Interpretable Semantic Spaces

TL;DR

semantic-features is a user-friendly library that enables researchers to analyze contextual word embeddings in interpretable semantic spaces, demonstrated through an experiment on dative constructions affecting semantic interpretation.

Contribution

The paper introduces semantic-features, an extensible tool for studying contextualized embeddings in interpretable spaces, with an application to linguistic analysis of dative constructions.

Findings

Embeddings show sensitivity to dative construction variations.

The tool effectively captures semantic nuances in contextualized embeddings.

Results support the usefulness of semantic-features for linguistic analysis.

Abstract

We introduce semantic-features, an extensible, easy-to-use library based on Chronis et al. (2023) for studying contextualized word embeddings of LMs by projecting them into interpretable spaces. We apply this tool in an experiment where we measure the contextual effect of the choice of dative construction (prepositional or double object) on the semantic interpretation of utterances (Bresnan, 2007). Specifically, we test whether "London" in "I sent London the letter." is more likely to be interpreted as an animate referent (e.g., as the name of a person) than in "I sent the letter to London." To this end, we devise a dataset of 450 sentence pairs, one in each dative construction, with recipients being ambiguous with respect to person-hood vs. place-hood. By applying semantic-features, we show that the contextualized word embeddings of three masked language models show the expected sensitivities. This leaves us optimistic about the usefulness of our tool.