Can Language Models Encode Perceptual Structure Without Grounding? A Case Study in Color
The result's identifiers
Result code in IS VaVaI
<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F00216208%3A11320%2F21%3A10442292" target="_blank" >RIV/00216208:11320/21:10442292 - isvavai.cz</a>
Result on the web
—
DOI - Digital Object Identifier
—
Alternative languages
Result language
angličtina
Original language name
Can Language Models Encode Perceptual Structure Without Grounding? A Case Study in Color
Original language description
Pretrained language models have been shown to encode relational information, such as the relations between entities or concepts in knowledge-bases - (Paris, Capital, France). However, simple relations of this type can often be recovered heuristically and the extent to which models implicitly reflect topological structure that is grounded in world, such as perceptual structure, is unknown. To explore this question, we conduct a thorough case study on color. Namely, we employ a dataset of monolexemic color terms and color chips represented in CIELAB, a color space with a perceptually meaningful distance metric. Using two methods of evaluating the structural alignment of colors in this space with text-derived color term representations, we find significant correspondence. Analyzing the differences in alignment across the color spectrum, we find that warmer colors are, on average, better aligned to the perceptual color space than cooler ones, suggesting an intriguing connection to findings from recent work on efficient communication in color naming. Further analysis suggests that differences in alignment are, in part, mediated by collocationality and differences in syntactic usage, posing questions as to the relationship between color perception and usage and context.
Czech name
—
Czech description
—
Classification
Type
D - Article in proceedings
CEP classification
—
OECD FORD branch
10201 - Computer sciences, information science, bioinformathics (hardware development to be 2.2, social aspect to be 5.8)
Result continuities
Project
—
Continuities
—
Others
Publication year
2021
Confidentiality
S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů
Data specific for result type
Article name in the collection
Proceedings of the 25th Conference on Computational Natural Language Learning
ISBN
978-1-955917-05-6
ISSN
—
e-ISSN
—
Number of pages
24
Pages from-to
109-132
Publisher name
Association for Computational Linguistics
Place of publication
Stroudsburg
Event location
online
Event date
Nov 10, 2021
Type of event by nationality
WRD - Celosvětová akce
UT code for WoS article
—