Refining Targeted Syntactic Evaluation of Language Models

TL;DR

This paper critically examines targeted syntactic evaluation of language models, proposing new metrics to better assess their syntactic knowledge and behavior, revealing overestimations in current methods and highlighting the importance of context-aware evaluation.

Contribution

The paper introduces refined metrics for targeted syntactic evaluation, distinguishing between systematic knowledge and likely behavior of language models.

Findings

Current TSE overestimates models' systematic syntactic knowledge.

Models perform up to 40% better on verbs they predict as likely.

Proposed metrics better capture models' syntactic behavior.

Abstract

Targeted syntactic evaluation of subject-verb number agreement in English (TSE) evaluates language models' syntactic knowledge using hand-crafted minimal pairs of sentences that differ only in the main verb's conjugation. The method evaluates whether language models rate each grammatical sentence as more likely than its ungrammatical counterpart. We identify two distinct goals for TSE. First, evaluating the systematicity of a language model's syntactic knowledge: given a sentence, can it conjugate arbitrary verbs correctly? Second, evaluating a model's likely behavior: given a sentence, does the model concentrate its probability mass on correctly conjugated verbs, even if only on a subset of the possible verbs? We argue that current implementations of TSE do not directly capture either of these goals, and propose new metrics to capture each goal separately. Under our metrics, we find that TSE overestimates systematicity of language models, but that models score up to 40% better on verbs that they predict are likely in context.