The study explores how blindness influences conceptual representations through language modeling. It introduces a novel approach to analyzing semantic shifts in blind individuals compared to sighted individuals across various verb categories. The findings suggest that blindness leads to significant changes in the way certain verbs are associated with semantic dimensions.
The research uses supervised pruning and probing tasks to identify feature subsets that optimize prediction accuracy for human similarity judgments. By comparing retained features for blind and sighted individuals, the study uncovers distinct patterns in semantic representation. The results indicate that blind individuals exhibit different associations with certain verb types compared to sighted individuals, highlighting the impact of sensory experience on conceptual organization.
Through detailed analyses and computational models, the study provides insights into how language models can be used to understand interindividual differences in word meaning. The methodology employed offers a new perspective on studying semantic knowledge across different populations, shedding light on the complex relationship between language, experience, and conceptual representation.
To Another Language
from source content
arxiv.org
Thông tin chi tiết chính được chắt lọc từ
by Wanqian Bao,... lúc arxiv.org 03-12-2024
https://arxiv.org/pdf/2403.06204.pdfYêu cầu sâu hơn