Seeing Through Words: Recovering Visual Information from Chinese Word Use

Abstract Relations among words may provide access to visual information, as proposed by the Symbol Interdependency Hypothesis. We examined this representational premise by predicting independently constructed visual prototypes from usage-derived representations of 2,571 Chinese lexical items. Text representations were aggregated from up to 500 unmasked occurrences in CCI 3.0, and visual prototypes were constructed from selected web-search images. A ridge mapping predicted each visual target from the text representation of an item held out under nested cross-validation. Full-library mean reciprocal rank was 0.1260, compared with 0.0044 for a part-of-speech (POS) mean baseline. In an exploratory evaluation against twenty same-POS semantic neighbors, mean reciprocal rank was 0.3050, compared with 0.1558 for the POS-mean baseline. We related differences in out-of-fold target specificity to twelve independently collected sensorimotor ratings and textual and visual representation properties. Under mutual adjustment, visual and olfactory associations were positive and the interoceptive association was negative. The eleven specific dimensions added 5.87 percentage points of in-sample explained variance beyond overall embodiment and the measurement covariates. Exploratory interactions indicated a weaker visual association at higher contextual dispersion and a stronger association at higher visual split-half stability. Among the analyzed targets in the sampled Chinese Web environment, lexical representations supported target-specific visual recovery. Its accessibility varied with experiential attributes and the organization of word use and visual representations.

Authors

Institutions

Publication Details

Journal
Research Square
Published
2026-11-12
DOI
https://doi.org/10.21203/rs.3.rs-11445696/v1
Primary Topic
Categorization, perception, and language
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Seeing Through Words: Recovering Visual Information from Chinese Word Use

Xinyue Ren, Esther Asare, Yufeng Wu, Meichun Liu
Research Square
Categorization, perception, and language
preprint

Seeing Through Words: Recovering Visual Information from Chinese Word Use

Xinyue Ren, Esther Asare, Yufeng Wu, Meichun Liu
preprint en

Abstract

Abstract Relations among words may provide access to visual information, as proposed by the Symbol Interdependency Hypothesis. We examined this representational premise by predicting independently constructed visual prototypes from usage-derived representations of 2,571 Chinese lexical items. Text representations were aggregated from up to 500 unmasked occurrences in CCI 3.0, and visual prototypes were constructed from selected web-search images. A ridge mapping predicted each visual target from the text representation of an item held out under nested cross-validation. Full-library mean reciprocal rank was 0.1260, compared with 0.0044 for a part-of-speech (POS) mean baseline. In an exploratory evaluation against twenty same-POS semantic neighbors, mean reciprocal rank was 0.3050, compared with 0.1558 for the POS-mean baseline. We related differences in out-of-fold target specificity to twelve independently collected sensorimotor ratings and textual and visual representation properties. Under mutual adjustment, visual and olfactory associations were positive and the interoceptive association was negative. The eleven specific dimensions added 5.87 percentage points of in-sample explained variance beyond overall embodiment and the measurement covariates. Exploratory interactions indicated a weaker visual association at higher contextual dispersion and a stronger association at higher visual split-half stability. Among the analyzed targets in the sampled Chinese Web environment, lexical representations supported target-specific visual recovery. Its accessibility varied with experiential attributes and the organization of word use and visual representations.

Research Square
City University of Hong Kong (HK)
Categorization, perception, and language
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon