Revisiting the Systematicity in Negation in the Era of In-Context Learning
2026-06-15 • Computation and Language
Computation and Language
AI summaryⓘ
The authors studied how well large language models understand sentences with negation, like "not" or "never." They found that models can often identify negation words and the parts of the sentence they affect, but not perfectly. The difficulty changes depending on how the answer is expected to be given. They also looked at how internal representations, called function vectors, help models understand negation cues versus negation scope, finding the latter harder to represent reliably.
negationlarge language modelsin-context learningnegation scopefunction vectorsbehavioral systematicityrepresentational systematicitynegation cue extraction
Authors
Hitomi Yanaka, Taisei Yamamoto
Abstract
Understanding the meaning of negated sentences remains one of the challenges for language models, even in the era of large language models (LLMs). We analyze systematicity regarding LLM understanding of negation from two perspectives: behavioral systematicity and representational systematicity. For behavioral systematicity, we confirm that through demonstrations and in-context learning, LLMs can recognize negation expressions and scope within sentences to some extent, but they fail to achieve perfect performance. In particular, the difficulty of the negation scope recognition for models varies depending on the output format. For representational systematicity, we analyze the extent to which function vectors can be robustly constructed from in-context examples for tasks that are essential to understanding negation. The experiments suggest that while function vectors can be composed for negation cue extraction tasks, extracting function vectors for recognizing scope is more challenging.