Agentivit\`a e telicit\`a in GilBERTo: implicazioni cognitive

TL;DR

This study examines whether a Transformer-based neural language model can infer lexical semantics like telicity and agentivity, and use this information for morphosyntactic pattern completion, comparing its performance to native Italian speakers.

Contribution

It investigates the capacity of a Transformer model to understand and utilize semantic properties at the interface of semantics and morphosyntax, focusing on telicity and agentivity in Italian.

Findings

The model captures some aspects of telicity and agentivity.

Comparison shows partial alignment with native speaker judgments.

Insights into the semantic competence of neural language models.

Abstract

The goal of this study is to investigate whether a Transformer-based neural language model infers lexical semantics and use this information for the completion of morphosyntactic patterns. The semantic properties considered are telicity (also combined with definiteness) and agentivity. Both act at the interface between semantics and morphosyntax: they are semantically determined and syntactically encoded. The tasks were submitted to both the computational model and a group of Italian native speakers. The comparison between the two groups of data allows us to investigate to what extent neural language models capture significant aspects of human semantic competence.