Exploration of the Usage of Color Terms by Color-blind Participants in Online Discussion Platforms

Ella Rabinovich, Boaz Carmeli

Abstract

This study investigates the extent to which sensory experience shapes linguistic systems by analyzing the natural language use of red-green color-blind (CB) individuals on Reddit. By examining millions of sentences, researchers found that CB individuals use color terms like "red" and "green" in less conventional contexts and with nouns that evoke weaker mental images compared to normal-sighted (NS) peers. These findings provide large-scale empirical support for the theory of embodied cognition, suggesting that sensory deficits manifest in subtle but measurable linguistic differences in spontaneous digital communication.

Introduction

A central debate in psycholinguistics concerns how physical sensory experience influences semantics. The theory of embodied cognition posits that word meanings are grounded in sensory perception. If an individual cannot perceive the difference between red and green, does their internal representation of these concepts diverge from the norm? While previous research has focused on controlled laboratory settings or total blindness, this study leverages a massive dataset of spontaneous online discourse to observe how color-blindness affects everyday language use.

Methodology

The researchers analyzed a novel dataset of Reddit posts from 2,523 self-reported color-blind authors (10.1M sentences) and 14,055 balanced normal-sighted authors (45.7M sentences). To ensure the analysis focused on natural language, all "meta-talk" regarding color blindness was filtered out. The study employed three primary analytical lenses:

  1. Frequency Analysis: Comparing the relative usage rates of specific color terms.
  2. Noun Imageability: Identifying nouns modified by color terms (e.g., "red shirt") and measuring their "imageability"—the ease with which they evoke a mental image. A beta regression model was used to infer imageability scores for over 17,000 nouns.
  3. Contextual Predictability (CWP): Using the BERT masked language model to determine how "predictable" a color term was within its specific sentence context. High rank (low probability) indicates unconventional usage.

Key Results

The Imageability Gap

The study found that normal-sighted individuals typically use "red" and "green" to describe highly concrete, visual objects. In contrast, color-blind users use these terms significantly more often with abstract or less "imageable" nouns.

Mean imageability scores for nouns modified by color terms. Significant differences (marked with *) are observed for red, green, and brown.

As shown in the figure above, the mean imageability of nouns modified by "red" and "green" was significantly lower for CB users. Interestingly, a "compensatory" effect was noted for the color brown; CB users showed higher imageability scores for this term, possibly because brown often represents the visual reality of red/green stimuli for this population.

Contextual Predictability Divergence

BERT analysis revealed that the usage of "red" and "green" by CB individuals was significantly less predictable than that of NS individuals. This suggests that without sensory grounding, CB users employ these terms in more idiosyncratic or less conventional ways.

BERT predicted rank distributions for green (left) and red (right). CB usage (light lines) results in higher average ranks, indicating lower predictability compared to NS usage (dark lines).

Control tests using non-color adjectives (e.g., "hot," "tiny") showed no significant differences between the groups, confirming that the observed linguistic shifts are specific to the sensory deficit.

Discussion and Conclusion

The results demonstrate that while CB individuals learn the linguistic "rules" of color (e.g., "apples are red"), this exposure does not fully replace the nuances of direct sensory experience. The divergence in noun imageability and contextual predictability suggests that sensory grounding is a critical component of human linguistic competence. These findings have implications for the development of Large Language Models (LLMs); if machines lack sensory input, can they ever achieve true human-like grounding, or will they always exhibit the same "unpredictability" seen in populations with sensory deficits?

Summary updated

Abstract

Prominent questions about the role of sensory vs. linguistic input in the way we acquire and use language have been extensively studied in the psycholinguistic literature. However, the relative effect of various factors in a person's overall experience on their linguistic system remains unclear. We study this question by making a step forward towards a better understanding of the conceptual perception of colors by color-blind individuals, as reflected in their spontaneous linguistic productions. Using a novel and carefully curated dataset, we show that red-green color-blind speakers use the "red" and "green" color terms in less predictable contexts, and in linguistic environments evoking mental image to a lower extent, when compared to their normal-sighted counterparts. These findings shed some new and interesting light on the role of sensory experience on our linguistic system.
BibTeX
@inproceedings{rabinovich2022exploration,
  author = {Ella Rabinovich and Boaz Carmeli},
  title = {{Exploration of the Usage of Color Terms by Color-blind Participants in Online Discussion Platforms}},
  booktitle = {Conference on Empirical Methods in Natural Language Processing},
  year = {2022},
  url = {https://openreview.net/forum?id=08EHRPW2Qv},
}
Figure