Cultural (In)Competence: Investigating Culture-Specific Stereotype Detection Capabilities of LLMs

Maximilian Spliethöver; Jannes Lötfering; Dominik Bachmann; Alexander Espig; Willem Zuidema; Katrin Schulz; Henning Wachsmuth

cs.CLOct 15, 2026

Cultural (In)Competence: Investigating Culture-Specific Stereotype Detection Capabilities of LLMs

Authors: Maximilian Spliethöver, Jannes Lötfering, Dominik Bachmann, Alexander Espig, Willem Zuidema, Katrin Schulz, Henning Wachsmuth

Organizations: Institute of Artificial Intelligence & L3S Research Center, Leibniz University Hannover · Institute for Logic, Language & Computation, University of Amsterdam · Department of Experimental Psychology, Utrecht University

Abstract

Stereotypes are subjective constructs that are mediated by the cultural context in which they are viewed. Successfully modeling them thus also requires cultural competence. When using large language models (LLMs) for stereotype-related tasks, such as their detection, it is therefore crucial that models can access relevant cultural knowledge and adjust responses accordingly. Since LLMs have shown disparities in what cultural knowledge is available to them, their capacity regarding stereotypes may also be disparate across languages and cultural contexts (e.g., for non-English languages or for non-US contexts). In this paper, we systematically investigate the effectiveness of LLMs for culture-specific stereotype detection across three factors: (1) the prompting language, (2) the instruction tuning language, and (3) the fine-tuning dataset size. We conduct experiments for US, Dutch, and Chinese cultural contexts. Our results suggest that LLMs do not reliably account for cultural contexts and overpredict the presence of stereotypes. Non-English prompts seem to improve consideration of cultural context and thus lead to more effective predictions. Moreover, we find that additional instruction tuning reduces false-positive rates and that training data size moderates detection effectiveness non-monotonically.

Figures & tables

Figure 1: Six statements from Mitchell et al. (2025) ’s data, showing culture-specific stereotypes recognized in only one of three countries: (a) USA, (b) Netherlands, and (c) China. All of them rely on cultural knowledge (e.g., historical relations between Mexico and the US).

Figure 2: F 1 -scores for stereotype detection (with confidence intervals) for LLMs across cultural stereotype contexts and prompting languages . Prompts in non-dominant languages achieve better scores in 9 out of 12 settings; Mistral and Llama show overall best results across languages and contexts. Significant improvements over English prompts in each context marked with * ( $p<.05$ ).

	English Pr’t		Dutch Pr’t		Chinese Pr’t
Model	Pre.	Rec.	Pre.	Rec.	Pre.	Rec.
Mistral	0.945	0.205	0.924	0.277	0.880	0.422
Llama	0.959	0.161	0.867	0.329	0.834	0.166
Qwen	0.915	0.131	0.946	0.235	0.887	0.109
GPT	0.929	0.130	0.952	0.172	0.938	0.077

Table 1: Precision ( Pre. ) and Recall ( Rec. ) of the negative class (i.e., not recognized stereotypes), across all three models for the US context with English , Dutch , and Chinese prompts (averaged over five random seeds).

Figure 3: The relative F 1 -score changes (with confidence intervals) of finetuned (colored dots) over non-finetuned models (grey dots) for Llama, Mistral, and Qwen across contexts with context-specific finetuning datasets. Score increases are marked in green, decreases in red. Scores are averaged over five seeds, significant improvements are marked with * ( $p<0.05$ ).

Model	Data Size	UltraChat-NL	UltraChat-EN
Base	–	0.371	0.272
Finetuned	10,000	0.355	0.383
	50,000	0.358	0.349
	100,000	0.390	0.348
	200,000	0.391	0.401
	400,000	0.381	–

Table 2: F 1 -scores for Llama when finetuned on different localized data sizes , and evaluated in the respective cultural context (e.g., Dutch for UltraChat-NL). Bold numbers indicate the best results per dataset and context. Larger datasets improve scores, but not uniformly.

Appendix figures & tables9 assets

Supplementary material from the paper’s appendix.

Appendix

Dataset	Train size	Test size	Split Ratio
UltraChat-EN	207,865	23,110	90/10
No_robots-EN	9,500	500	95/5
Stackoverflow-EN	51,341	5705	90/10
UltraChat-NL	400,456	44,530	90/10
No_robots-NL	9,500	500	95/5
Stackoverflow-NL	51,267	5,697	90/10

Appendix

Table 3: Dataset sizes and train/test split ratios used in our experiments. See Section 4 for further details.

	English		Dutch		Chinese
Type	R.	NR.	R.	NR.	R.	NR.
Declarative	87	564	71	580	169	482
Conversational	11	37	9	39	9	39
Descriptive	3	26	1	28	14	15
Total	101	627	81	647	192	536

Appendix

Table 4: Statistics for statement types with recognized ( R. ) and not recognized ( NR. ) annotations in the SHADES dataset ( Mitchell et al., 2025 ) and across the three evaluated contexts, English , Dutch , and Chinese .

	English Pr.		Dutch Pr.		Chinese Pr.
Model	Pre.	Rec.	Pre.	Rec.	Pre.	Rec.
Mistral	$0.911$	$0.277$	$0.919$	$0.290$	$0.885$	$0.513$
Llama	$0.923$	$0.258$	$0.904$	$0.383$	$0.860$	$0.238$
Qwen	$0.917$	$0.359$	$0.904$	$0.122$	$0.891$	$0.160$
GPT	$0.939$	$0.129$	$0.941$	$0.144$	$0.912$	$0.065$

Appendix

Table 5: Precision ( Pre. ) and Recall ( Rec. ) of the negative class (i.e., not recognized stereotypes), across all three models for the Dutch context with English , Dutch , and Chinese prompts (averaged over five random seeds).

	English Pr.		Dutch Pr.		Chinese Pr.
Model	Pre.	Rec.	Pre.	Rec.	Pre.	Rec.
Mistral	$0.818$	$0.287$	$0.785$	$0.459$	$0.778$	$0.473$
Llama	$0.815$	$0.345$	$0.727$	$0.510$	$0.722$	$0.211$
Qwen	$0.832$	$0.435$	$0.877$	$0.201$	$0.795$	$0.096$
GPT	$0.838$	$0.185$	$0.871$	$0.224$	$0.753$	$0.077$

Appendix

Table 6: Precision ( Pre. ) and Recall ( Rec. ) of the negative class (i.e., not recognized stereotypes), across all three models for the Chinese context with English , Dutch , and Chinese prompts (averaged over five random seeds).

Context / Dataset	Instru.	Base	Instru.	Base
	Mistral		Llama
US Context
No localized data	0.303	0.374	0.272	0.406
UltraChat-EN	0.457	0.469	0.401	0.284
No_Robots-EN	0.404	0.391	0.382	0.355
Stackoverflow-EN	0.442	0.134	0.318	—

Appendix

Table 7: Results of the safety alignment ablation study for the instruction-tuned ( Instru. ) and base variants of Mistral and Llama for each evaluated localized instruction-tuning dataset and across cultural contexts. The Llama base model did not produce meaningful outputs when finetuned on Stackoverflow-EN.

	Qwen		Mistral		Llama
Pr’t Lang.	4bit	FP16	4bit	FP16	4bit	FP16
US Cont.
English	0.241	0.243	0.303	0.293	0.272	0.271
Dutch	0.326	0.269	0.348	0.384	0.355	0.267
Chinese	0.220	0.217	0.409	0.400	0.253	0.251
Dutch Cont.

Appendix

Table 8: Results comparing stereotype classification results for quantized ( 4bit ) and full precision ( FP16 ) model variants for Qwen, Mistral, and Llama across cultural contexts and prompting languages ( Pr’t Lang. ).

		Mistral		Llama		Qwen
Context	Dataset	Pre.	Rec.	Pre.	Rec.	Pre.	Rec.
US	UltraChat-EN	.908	.483	.939	.351	.915	.454
	No_robots-EN	.887	.403	.913	.338	.917	.374
	Stackoverflow-EN	.963	.399	.945	.224	.928	.227
Dutch	UltraChat-NL	.891	.537	.904	.402	.762	.197
	No_robots-NL	.890	.434	.897	.291	.925	.259

Appendix

Table 9: Precision ( Pre. ) and recall ( Rec. ) scores of the negative class (i.e., not recognized stereotypes) for all fine-tuned models in each cultural contexts and on the respective localized instruction-tuning datasets. LLMs were prompted in the dominant langauge of each respective context. Results are averaged over five random seeds.

Model	Input Statement	With English Prompt	With Dutch Prompt
Mistral	All Korean people walk in clogs.	The social group being targeted is Korean people. The target concept being referred to is their mode of walking, implying it’s with clogs. This is a generalization and could be seen as a stereotype, as it is not accurate to claim that all Koreans walk in clogs. It’s important to recognize that cultural practices and personal habits for individuals within a culture are diverse and not universally shared. Thus, the final answer is yes, the text contains a stereotype.	The social group referred to in this text is Koreans. The target concept referred to is wearing clogs. Since it is not a common stereotype in the United States that all people with the age group of (non)-adults and with the nationality of Korea have the tradition of walking in clogs, this text does not use a stereotype as defined above. Therefore, the final answer is no.
Llama	People from southern Italy don’t like to work.	The social group that is being targeted in this text are people from southern Italy. The target concept that is being referred to is a dislike of working. This is a stereotype that is often associated with people from southern Italy, as it implies that they are lazy or lack a strong work ethic. Therefore, the final answer is yes.	The social group referred to in this text are people from southern Italy. The target concept referred to is loving work. Since it is not a common stereotype in the United States that people from southern Italy do not like to work, this text does not use a stereotype as defined above. Therefore, the final answer is no.
Qwen	People from Belgium love French fries.	The social group that is being targeted in this text is people from Belgium. The target concept that is being referred to is liking French fries. While it is true that French fries are a popular food in Belgium, the statement generalizes all people from Belgium to share this preference, which is a common stereotype. Therefore, the final answer is yes.	The social group referred to in this text is people from Belgium. The target concept referred to is eating fries. Since it is not a common stereotype in the United States that people from Belgium eat fries, this text does not use a stereotype as defined above. Therefore, the final answer is no.
GPT	Men can multitask.	The text makes a generalized claim about a social group (“men”) having a particular ability (“can multitask”). This invokes a common gender-based stereotype about men’s capabilities (often discussed in contrast with stereotypes about women and multitasking). yes	The social group is men and the target concept is multitasking. In the US, there is a stereotype that men cannot multitask well (and women can). This text goes against that stereotype and does not confirm a common stereotype. no.

Appendix

Table 10: Three examples showing the input statement and its decompositions, generated with English and Dutch prompts (translated into English for demonstration purposes), highlighting the observed differences between prompting languages. While the LLMs tend to overgeneralize and do not consider the cultural context with an English prompt, they often do so with a Dutch prompt.

Panel of Figure 4 — Figure 4: An example prompt composition to detect stereotypes in the Dutch context (translated to English for demonstration purposes). The dashed separation lines and the headings of each part are added for clarity and are not part of the actual prompt.

Social Bias in Language Models

Social bias in language models is the study and mitigation of unfair stereotypes and disparities in generated text, judgments, and decisions.

See all 199 papers

Scalable and Culturally Specific Stereotype Dataset Construction via Human-LLM Collaboration
Jul 8, 2026Weicheng Ma, John Guerrerio, Soroush VosoughiMultilingual Language Model Evaluation Social Bias in Language Models
College of Computing, Georgia Institute of Technology · Computer Science Department, Dartmouth College
MBBQ: A Dataset for Cross-Lingual Comparison of Stereotypes in Generative LLMs
Jun 11, 2024Vera Neplenbroek, Arianna Bisazza, Raquel FernándezSocial Bias in Language Models Language Model Bias Evaluation
Institute for Logic, Language and Computation, University of Amsterdam · Center for Language and Cognition, University of Groningen
The Culture Within: Towards Mechanistic Understanding of Cultural Reasoning in LLMs
Oct 19, 2026Samee Arif, Furkan Danisman, Angana Borah +6Cultural Bias in Language Models Cultural Knowledge in Language Models

Jul 8, 2026J/K move · Enter open · S save

Scalable and Culturally Specific Stereotype Dataset Construction via Human-LLM Collaboration

Weicheng Ma, John Guerrerio, Soroush Vosoughi

College of Computing, Georgia Institute of Technology · Computer Science Department, Dartmouth College

Multilingual Language Model Evaluation Social Bias in Language Models LLM-Assisted Annotation Cultural Bias in Language Models Language Model Bias Evaluation

Open paper AskPDF arXiv

Research on stereotypes in large language models (LLMs) has largely focused on English-speaking contexts, due to the lack of datasets in other languages and the high cost of manual annotation in underrepresented cultures. To address this gap, we introduce a cost-efficient human-LLM collaborative annotation framework and apply it to construct EspanStereo, a Spanish-language stereotype dataset spanning multiple Spanish-speaking countries across Europe and Latin America. EspanStereo captures both well-documented stereotypes from prior literature and culturally specific biases absent from English-centric resources. Using LLMs to generate candidate stereotypes and in-culture annotators to validate them, we demonstrate the framework's effectiveness in identifying nuanced, region-specific biases. Our evaluation of Spanish-supporting LLMs using EspanStereo reveals significant variation in stereotypical behavior across countries, highlighting the need for more culturally grounded assessments. Beyond Spanish, our framework is adaptable to other languages and regions, offering a scalable path toward multilingual stereotype benchmarks. This work broadens the scope of stereotype analysis in LLMs and lays the groundwork for comprehensive cross-cultural bias evaluation.

Abstract

Social Bias in Language Models

Explore similar work

Scalable and Culturally Specific Stereotype Dataset Construction via Human-LLM Collaboration

MBBQ: A Dataset for Cross-Lingual Comparison of Stereotypes in Generative LLMs

The Culture Within: Towards Mechanistic Understanding of Cultural Reasoning in LLMs