Did the Cat Drink the Coffee? Challenging Transformers with Generalized Event Knowledge

TL;DR

This paper evaluates whether Transformer language models can understand and predict the typicality of entire events in language, comparing their performance to a specialized event knowledge framework and analyzing their limitations.

Contribution

It introduces a benchmark for dynamic estimation of thematic fit and compares TLMs with SDM, revealing strengths and weaknesses in event knowledge understanding.

Findings

TLMs perform comparably to SDM in thematic fit estimation.

TLMs often rely on surface linguistic features rather than true event knowledge.

Event understanding in TLMs is limited by superficial cues, affecting generalization.

Abstract

Prior research has explored the ability of computational models to predict a word semantic fit with a given predicate. While much work has been devoted to modeling the typicality relation between verbs and arguments in isolation, in this paper we take a broader perspective by assessing whether and to what extent computational approaches have access to the information about the typicality of entire events and situations described in language (Generalized Event Knowledge). Given the recent success of Transformers Language Models (TLMs), we decided to test them on a benchmark for the \textit{dynamic estimation of thematic fit}. The evaluation of these models was performed in comparison with SDM, a framework specifically designed to integrate events in sentence meaning representations, and we conducted a detailed error analysis to investigate which factors affect their behavior. Our results show that TLMs can reach performances that are comparable to those achieved by SDM. However, additional analysis consistently suggests that TLMs do not capture important aspects of event knowledge, and their predictions often depend on surface linguistic features, such as frequent words, collocations and syntactic patterns, thereby showing sub-optimal generalization abilities.