Age, Gender, and Class Heterogeneity in Marital Status Stereotypes in Chinese Natural Language
(2025) SIMZ51 20251Graduate School
- Abstract
- This study employs large-scale Chinese pre-trained language models (BERT series) and the Fill-Mask Association Test (FMAT) to systematically measure gender- and age-based, as well as class-based, heterogeneity in marital status stereotypes embedded in natural language.
Four sub-studies were designed to examine the effects of age (22–40 years), residence (urban vs. rural), and education (high vs. low educational attainment) on marital status stereotypes and their gender differences. The standardized log probability ratio (LPR) was used as the dependent variable, and linear mixed-effects models (LMM) were applied for inferential analysis.
The results show that: (1) in public language, there is no significant stereotypical difference... (More) - This study employs large-scale Chinese pre-trained language models (BERT series) and the Fill-Mask Association Test (FMAT) to systematically measure gender- and age-based, as well as class-based, heterogeneity in marital status stereotypes embedded in natural language.
Four sub-studies were designed to examine the effects of age (22–40 years), residence (urban vs. rural), and education (high vs. low educational attainment) on marital status stereotypes and their gender differences. The standardized log probability ratio (LPR) was used as the dependent variable, and linear mixed-effects models (LMM) were applied for inferential analysis.
The results show that: (1) in public language, there is no significant stereotypical difference between unmarried and married individuals, while divorced individuals are even perceived as warmer and more competent than the married; (2) single women (both unmarried and divorced) are perceived as warmer and more competent than their male counterparts; (3) young divorced men are viewed as warmer than young divorced women, whereas young unmarried men are seen as more competent than young unmarried women; (4) rural singles are perceived as less competent and less warm than urban singles, although the gender differences are not significant; (5) low-educated singles are considered less competent and less warm than highly educated ones, especially among divorced women.
By systematically comparing stereotypes along the dimensions of "warmth" and "competence" across gender, age, urban-rural status, and education, this study reveals the multidimensional evaluative frameworks applied to those who deviate from the marital norm. It provides new empirical evidence and methodological insight into how structural inequalities are embedded in everyday language. (Less)
Please use this url to cite or link to this publication:
http://lup.lub.lu.se/student-papers/record/9199872
- author
- Ma, Ziheng LU
- supervisor
- organization
- course
- SIMZ51 20251
- year
- 2025
- type
- H2 - Master's Degree (Two Years)
- subject
- keywords
- stereotypes, gender, natural language processing, large language models, FMAT
- language
- English
- id
- 9199872
- date added to LUP
- 2025-06-25 11:19:52
- date last changed
- 2025-07-29 09:28:43
@misc{9199872,
abstract = {{This study employs large-scale Chinese pre-trained language models (BERT series) and the Fill-Mask Association Test (FMAT) to systematically measure gender- and age-based, as well as class-based, heterogeneity in marital status stereotypes embedded in natural language.
Four sub-studies were designed to examine the effects of age (22–40 years), residence (urban vs. rural), and education (high vs. low educational attainment) on marital status stereotypes and their gender differences. The standardized log probability ratio (LPR) was used as the dependent variable, and linear mixed-effects models (LMM) were applied for inferential analysis.
The results show that: (1) in public language, there is no significant stereotypical difference between unmarried and married individuals, while divorced individuals are even perceived as warmer and more competent than the married; (2) single women (both unmarried and divorced) are perceived as warmer and more competent than their male counterparts; (3) young divorced men are viewed as warmer than young divorced women, whereas young unmarried men are seen as more competent than young unmarried women; (4) rural singles are perceived as less competent and less warm than urban singles, although the gender differences are not significant; (5) low-educated singles are considered less competent and less warm than highly educated ones, especially among divorced women.
By systematically comparing stereotypes along the dimensions of "warmth" and "competence" across gender, age, urban-rural status, and education, this study reveals the multidimensional evaluative frameworks applied to those who deviate from the marital norm. It provides new empirical evidence and methodological insight into how structural inequalities are embedded in everyday language.}},
author = {{Ma, Ziheng}},
language = {{eng}},
note = {{Student Paper}},
title = {{Age, Gender, and Class Heterogeneity in Marital Status Stereotypes in Chinese Natural Language}},
year = {{2025}},
}