Word-color association from large-scale online data: Methodology and applications
- 1Graduate School of Culture Technology, Korea Advanced Institute of Science and Technology, Daejeon, Republic of Korea
- 2School of Computational Sciences, Korea Institute for Advanced Study, Seoul, Republic of Korea
Abstract
Words and colors are intimately associated in humans, a sign of the close relationship between verbal, linguistic expression and cognition. Word-color association therefore plays an important role in areas of communication such as marketing and literature. Traditional efforts to build a general, scalable, and quantitative mapping of word-color association have often relied on surveys, experiencing difficulties in curating large-scale data sets, enrolling sufficient participants, and excising potential inherent researcher biases in the choice of the words and the colors to be mapped. Even the most recent computational efforts based on Large Language Models (LLMs) do not overcome these problems due to their near-complete lack of transparency in data and methodology that results in low confidence, consistency, and reliability. Here we propose a framework that overcomes these limitations via the use of freely accessible large-scale web image data, and fully transparent and reproducible methods that allow one to build color associations for almost any given word. We also showcase the methodology in action through analyses of color usage in corporate branding and 20th-century Anglo-American poetry. These show that our framework can help form new answers to and perspectives on long-standing questions in communication. Additionally, our framework by design can be readily adopted for use in any language supported by a web image search engine, and its code is freely available in open source; we therefore expect it to contribute towards achieving a universal, multilingual understanding of word-color association.
Related
- The Multi-scale Copula Index (MCI): a new information-theoretic measure of association and its applications in functional cancer genomics
- A winter wheat genome-wide association study identifies a large effect Fusarium head blight resistance haplotype using diverse lines from US breeding programs
- Taylor's Law, Smith's Law, and Diversity Power Laws: A Novel Triple Power Law Methodology for Scaling Diversity and Heterogeneity
- Pseudorabies virus as vaccine vectors for delivering foreign antigens: methodology and application
- A novel methodology for assessing reinforcement choice in primates under human care
BibTeX
@article{wang2026wordcolor,
title = {{Word-color association from large-scale online data: Methodology and applications}},
author = {Sungpil Wang and Juyong M. Park},
journal = {PLOS One},
year = {2026},
month = oct,
doi = {10.1371/journal.pone.0360407},
url = {https://doi.org/10.1371/journal.pone.0360407},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - Word-color association from large-scale online data: Methodology and applications
AU - Sungpil Wang
AU - Juyong M. Park
PY - 2026
DA - 2026/10/16
JO - PLOS One
DO - 10.1371/journal.pone.0360407
UR - https://doi.org/10.1371/journal.pone.0360407
AB - Words and colors are intimately associated in humans, a sign of the close relationship between verbal, linguistic expression and cognition. Word-color association therefore plays an important role in areas of communication such as marketing and literature. Traditional efforts to build a general, scalable, and quantitative mapping of word-color association have often relied on surveys, experiencing difficulties in curating large-scale data sets, enrolling sufficient participants, and excising potential inherent researcher biases in the choice of the words and the colors to be mapped. Even the most recent computational efforts based on Large Language Models (LLMs) do not overcome these problems due to their near-complete lack of transparency in data and methodology that results in low confidence, consistency, and reliability. Here we propose a framework that overcomes these limitations via the use of freely accessible large-scale web image data, and fully transparent and reproducible methods that allow one to build color associations for almost any given word. We also showcase the methodology in action through analyses of color usage in corporate branding and 20th-century Anglo-American poetry. These show that our framework can help form new answers to and perspectives on long-standing questions in communication. Additionally, our framework by design can be readily adopted for use in any language supported by a web image search engine, and its code is freely available in open source; we therefore expect it to contribute towards achieving a universal, multilingual understanding of word-color association.
SN - 1932-6203
ER -