Generalization of numerical (im)precision findings to other pragmatic domains

Establish whether the patterns of LLM–human alignment observed for social judgments under numerical (im)precision generalize to other pragmatic domains, including scalar implicature, politeness, and register variation.

Background

The evaluation centers on a single experimental paradigm involving numerical expressions across six scenarios and six social attributes, yielding clear results about structural fidelity and magnitude calibration in LLMs.

The authors explicitly note that it is not yet established whether these findings extend beyond numerical (im)precision to broader pragmatic phenomena such as scalar implicature, politeness, or register variation.