Who says “larger” and who says “smaller”? Individual differences in the language of comparison

Abstract

Abstract When comparing a pair of attribute values, English speakers can use a “larger” comparative (“A is larger/longer/higher/more than B”) or a “smaller” comparative (“B is smaller/shorter/lower/less than A”). This choice matters because it affects people’s inferences about the absolute magnitudes of the compared items, and influences the perceived truthfulness of the comparative sentence itself. In 4 studies (total N = 2335), we investigated the language that people use to describe ordinal relations between attributes. Specifically, we examined whether demography, emotion, and personality predict the tendency to use “larger” comparatives rather than “smaller” ones. Participants viewed pairs of items differing in a single attribute and indicated the word they would use to describe the relationship between them; they also completed a series of self-report measures. Replicating previous work, we found a robust tendency to use “larger” comparatives, both when people chose between two adjectives and when they freely produced their own words in a sentence completion task. We also found that this tendency was more pronounced in older participants, those with positive mood or outlook, and among people high in agreeableness, conscientiousness, and emotional stability. However, these effects were very small, with meta-analytic effect sizes indicating they explain less than 1% of the variance. We conclude that, although people’s use of comparative adjectives is influenced by properties of the items that are being compared, the way that people describe magnitude relations is relatively stable across variation in a range of important traits and dispositions, protecting decision-makers from a potentially undesirable source of bias in their inferences and representations of described options.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 4 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.