Pith. sign in

REVIEW 4 major objections 5 minor 1 cited by

Reading between the Lines: Can LLMs Identify Cross-Cultural Communication Gaps?

T0 review · 4 major / 5 minor · reviewed 2026-08-08 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read Most book reviews contain culture-specific details that baffle readers from other cultures, and GPT-4o currently catches them with limited accuracy.

desk verdict New human-annotated dataset of cross-cultural comprehension gaps, but the 83% headline and precision/recall numbers lean on model-selected reviews and GPT-4o-generated cultural labels; the qualitative finding survives. read the letter →

arxiv 2502.09636 v2 pith:2MYQPQ3S submitted 2025-02-09 cs.CL cs.AI

Sougata Saha , Saurabh Kumar Pandey , Harshit Gupta , Monojit Choudhury This is my paper · ORCID
classification cs.CLcs.AI
keywords cross-culturalcommunicationculture-specificitemsLLMevaluationbookreviewsGoodreadsGPT-4ouserstudyculturalmediator
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

Online book reviews are full of culture-specific references that readers from other cultures cannot decode. Running a user study in which 50 readers from India, Mexico, and the USA highlighted unfamiliar spans in 57 Goodreads reviews of books from Ethiopia, India, and the USA, the paper reports that 83% of those reviews contained at least one culture-specific difficult element. The paper then asks whether GPT-4o, prompted as a cultural mediator for a reader of a given country and genre taste, can flag the same spans. It finds the model is roughly even-handed across the three cultures but weak overall: precision around 0.49 and recall around 0.65 against the human-highlighted cultural spans, meaning a real assistant would both miss and over-report cultural hurdles. The broader point the paper wants to establish is that cross-cultural comprehension gaps are measurable, common, and still a live problem for current LLM-based assistive tools.

What carries the argument

The central object is the Culture-Specific Item (CSI), a text span whose meaning depends on a source culture's ecology, material life, social organization, customs, habits, or language. The paper operationalizes CSIs by having readers highlight any span they found hard to understand, then uses GPT-4o to classify each highlighted span as cultural or non-cultural and cluster semantically similar spans, with two human experts checking 120 annotations. To benchmark the assistant, GPT-4o is prompted with the reader's country and genre preference and asked to identify, categorize, explain, and reformulate CSIs; the model's outputs are matched to the human span clusters via embedding cosine similarity at a threshold of 0.5. That overlap measurement, precision, and recall against the human cultural spans is what carries the headline quantitative claims.

What would settle it

Take the same 668 cleaned human responses and have independent annotators from each participant country label every highlighted span as cultural or non-cultural without LLM involvement; if the human labels disagree materially with GPT-4o's taxonomy assignments, the reported 83% rate and the precision and recall scores for GPT-4o would no longer be supported.

Watch

Extended reading notes

Core claim

The paper's central claim is that cross-cultural communication gaps in ordinary English book reviews are pervasive rather than rare, and that GPT-4o acting as a cultural mediator is equitable across the studied readerships but not yet accurate enough to close the gap. When readers highlighted spans they did not understand, more than two-thirds of those spans were judged cultural, and the cultural difficulty was lowest for readers reading books from their own country, evidence that culture, not just vocabulary, drives the barrier. GPT-4o's detected CSIs overlapped with 60% of the human-identified cultural spans while also flagging phrases the humans treated as merely difficult, producing low precision. The authors conclude that LLMs can serve as a starting point for cross-cultural reading assistance, but their alignment with actual reader difficulty is too low to call them reliable cultural mediators.

Load-bearing premise

The load-bearing premise is that GPT-4o's own classification of the human-highlighted spans into cultural versus non-cultural is accurate enough to serve as the gold standard, even though only 120 of the thousands of spans were checked by human experts.

Editorial extensions

If this is right

  • A deployed cultural reading assistant built on current GPT-4o prompting would miss roughly a third of the cultural references a human reader stumbles on.
  • It would also raise about twice as many false leads as true hits, since precision is around 0.49.
  • The assistant's performance does not favor any one of the three studied countries enough to call it culturally biased in that direction.
  • Readers from all three countries showed some cultural difficulty even with books from their own country, implying culture is not uniform within a country.
  • Same-country agreement is higher for cultural spans than for general difficulty, so CSIs are a viable cold-start signal for personalization systems.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Because the cultural versus non-cultural labels were assigned by GPT-4o itself at scale, the 83% rate and the precision/recall figures are best read as measuring the model's own consistency; a fully human-labeled gold set could shift them.
  • The same measurement recipe could be run on product reviews, social media, or news comments, where the cultural gap may be larger because the content is less edited than book reviews.
  • A testable extension would be to evaluate explanations and reformulations, not just span identification: even when GPT-4o finds the right CSI, readers may not accept its explanation, which would change how alignment should be measured.
  • The findings suggest cultural assistance should be personalized beyond country, since within-country variation in highlighted spans implies reader-specific factors such as genre taste and education matter as much as nationality.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 5 minor

Summary. The paper investigates cross-cultural comprehension gaps in Goodreads book reviews through a user study with 50 participants from India, Mexico, and the USA, who highlighted spans they found difficult to understand. The authors report that 83% of the 57 studied reviews contain at least one culture-specific difficult-to-understand element, and they benchmark GPT-4o as a cultural mediator, reporting precision 0.49 and recall 0.65 against human-identified cultural spans, concluding that GPT-4o is roughly equitable across the studied cultures but with limited alignment. The paper also releases a dataset of human and LLM annotations.

Significance. The paper tackles a genuinely important and under-studied problem: measuring and mitigating cross-cultural comprehension gaps in user-generated text. Its strengths include a careful multi-stage piloting process, a real crowdsourced user study with participants from three countries, and an openly released dataset of human span-level annotations, which is a valuable resource for future work. The qualitative finding that cross-cultural comprehension gaps are common and that GPT-4o is not yet a reliable cultural mediator is plausible and useful. However, the central quantitative claims (the 83% prevalence estimate and the precision/recall evaluation) are weakened by two model-mediated steps: reviews were preselected to contain unfamiliar CSIs, and the cultural/non-cultural labels used as ground truth were themselves generated by GPT-4o. These issues do not invalidate the qualitative direction, but they require the claims to be substantially reframed and the evaluation to be strengthened before the results can be taken at face value.

major comments (4)
  1. [Section 3.1] The headline finding that 83% of reviews had at least one culturally difficult element is computed on a set of 57 reviews that were explicitly preselected by GPT-3.5 to contain at least one unfamiliar culture-specific item (CSI). This is not a random sample of Goodreads reviews, so the 83% figure is a conditional rate within a CSI-positive subset, not an estimate of the prevalence of CSIs in book reviews generally. The abstract and Section 7 state the result as if it applied to Goodreads reviews at large. The paper should either report the prevalence on an unfiltered random sample or clearly and consistently frame the 83% as a property of the preselected set. This is load-bearing because it is the paper's first quantitative contribution.
  2. [Section 3.3.1 and Section 4.2] The evaluation of GPT-4o is circular in an important respect. The human participants highlighted spans they did not understand, which is independent evidence. However, the distinction between cultural and non-cultural spans is assigned by GPT-4o using a taxonomy that GPT-4o itself helped generate, and the same model is then benchmarked against these GPT-4o-assigned labels in Section 4.2. Only 120 annotations were expert-checked (with 54 overlapping), and the paper reports a 10% error margin for span type and 15% for cluster assignment. A 10% misclassification rate on the 116 cultural spans is enough to shift the reported precision/recall numbers by several points and could alter the equitability conclusion if the errors are uneven across countries. The authors should either obtain human labels for the full set of spans or provide a sensitivity analysis showing that their conclusions are stable under the reported error margins. As written, the numbers in Section 4.2 partly measure GPT-4o's agreement with itself.
  3. [Section 4.2, Figure 6] The claim that GPT-4o 'performs equitably across the studied cultures' is supported only by point estimates of precision and recall without confidence intervals or any statistical test. The per-country and per-country-plus-genre cells are small (for example, only 8 participants from India participated in the study, as reported in Section 3.2.2), and the differences between countries could easily be within sampling noise. Given that the equitability claim is one of the paper's three main contributions, the authors should provide uncertainty estimates or at minimum state the sample sizes per cell and avoid making a strong no-difference claim. This is particularly important because the label noise discussed in the previous comment could differentially affect the cells.
  4. [Section 3.4] The inter-annotator agreement analysis is used to support the claim that CSIs are a consistent and useful target for personalization. However, the raw span-level Krippendorff alpha values are very low (near zero for 'overall' level in several cases), and the increase to values around 0.11–0.14 for cultural spans, while directionally interesting, is still extremely weak agreement. With such low values, the statement that 'there is a level of consensus on CSIs' overstates the evidence. The paper should soften this claim and discuss the implications of near-zero span-level agreement for the reliability of the gold labels derived from the clustering process.
minor comments (5)
  1. [Section 1] The phrase 'perform outdoor activities, household chores, or surfing the internet' is grammatically uneven; 'surfing' should be parallel to the other verbs.
  2. [Section 2] "Inglhart-Welzel's world cultural map" contains a typo: 'Inglehart' is the correct spelling.
  3. [Section 4.2] The text says 'As depicted in Figure 1' when referring to the overlap between human-identified and GPT-4o-identified spans; this should be Figure 4. Similar figure cross-reference errors may exist elsewhere and should be checked.
  4. [Section 3.4] The phrase 'Mexcian participants' and 'Hemmingway' are typos; also, the table captions and figure labels should be checked for consistent capitalization of 'Krippendorf's alpha' (the standard spelling includes two 'f's and one 'd').
  5. [Table 1] The table note says '1 | 2' but does not explain what the two numbers in each cell represent (presumably mean and standard deviation). This should be clarified.

Circularity Check

2 steps flagged · score 6.0 of 10

The precision/recall benchmark is partially self-scoring: GPT-4o generates the cultural-span gold labels and is then evaluated against them; the 83% prevalence figure is inflated by preselection on model-identified CSIs.

  1. other [Section 3.1 (Dataset) and Section 3.4 (Analysis, item 1)]
    "Next, to ensure enough cultural diversity in our study, for each review text, we prompted GPT 3.5 to identify CSIs... The model identified CSIs in almost all review texts... We randomly sampled 57 reviews containing at least one unfamiliar CSI as the final dataset for our study. ... all 60 reviews had at least one difficult-to-understand element, where 50 reviews (83%) had culturally difficult-to-understand elements."

    The 83% review-level cultural-difficulty rate is presented as a finding of the user study, but the sample was explicitly preselected to contain reviews with at least one GPT-3.5-identified unfamiliar CSI. The reported percentage is therefore a conditional rate inside a set already filtered for the phenomenon being measured, not an estimate for Goodreads reviews generally. This is not a full by-construction equivalence, because GPT-4o could still label some preselected spans as non-cultural, but the selection materially forces a high prevalence and undercuts the headline claim as an independent discovery.

  2. self definitional [Section 3.3.1 (LLM-assisted post processing) and Section 4.2 (Analysis, benchmarking)]
    "Since the study only asked participants to identify spans they found hard to understand, we processed the results using GPT-4o to categorize them as CSIs (cultural spans) or non-CSIs. ... Considering the user-identified cultural spans as the ground truth, we compute the precision and recall scores of GPT-4o-identified CSIs at different levels of country and genre-preference combinations."

    The gold 'cultural' labels are produced by GPT-4o classifying the human-highlighted spans, and the model being benchmarked, GPT-4o, is then scored against those same labels. Agreement between GPT-4o's CSI identification and GPT-4o's cultural/non-cultural classification can reflect internal consistency of one model rather than correct cultural grounding. The paper acknowledges the fragility: Section 3.3.2 reports only 120 expert-checked annotations (54 overlapping) with ~10% span-type and ~15% cluster error margins. The reported precision (≈0.49) and recall (≈0.65) and the equitability conclusion therefore partly reduce to GPT-4o's own outputs rather than to independent human gold.

full rationale

The paper has genuinely independent raw material: 50 human participants marked spans they found difficult to understand, and that marking does not depend on the LLM being evaluated. However, the two central quantitative claims are each mediated by the same model family. First, the 83% figure is measured on a sample preselected by GPT-3.5 for containing unfamiliar CSIs, conditioning on the outcome. Second, the cultural versus non-cultural split of the human spans is assigned by GPT-4o, and GPT-4o is then benchmarked against these model-assigned labels; the expert check of 120 annotations with substantial overlap and admitted error margins does not eliminate the self-scoring component. The equitability conclusion inherits this problem. Thus the paper is partially circular rather than fully self-contained: the human difficulty data are independent, but the key aggregating labels and the evaluation target share the same model. Score 6 reflects one central evaluation reducing in part to its own inputs, while acknowledging the independent human span-marking and the expert validation as mitigating factors.

Assumptions & free parameters 2 free parameters · 4 assumptions · 0 invented entities

The central empirical claims rest on domain assumptions about what culture is, how it is proxied by nationality and genre preference, and the validity of LLM-generated labels for cultural difficulty. The most load-bearing choice is that GPT-4o's taxonomy and label assignments are treated as ground truth for cultural difficulty.

free parameters (2)
  • Cosine similarity threshold for span-cluster overlap = 0.5
    Section 4.2: GPT-4o spans are matched to human span clusters using Sentence-BERT embeddings with a hand-chosen threshold of 0.5; changing this threshold changes the reported precision and recall.
  • Review preselection thresholds = at least 50 reviews; 50 to 200 words
    Section 3.1: sampling criteria used to select candidate reviews from Goodreads; these choices shape the distribution of review content and the final dataset.
assumptions (4)
  • domain assumption Country of upbringing is a valid proxy for a reader's culture.
    Section 3.2.2 uses self-reported location before age 18 to define India, Mexico, and USA cultures; within-country cultural variation is acknowledged but not modeled.
  • ad hoc to paper The GPT-4o-generated taxonomy and clustering in Section 3.3.1 correctly separates cultural from non-cultural difficulty.
    Only 120 spans were expert-validated (with 89% span-type agreement on 54 overlapping cases); all other cultural labels are model outputs.
  • domain assumption Genre preference (fiction vs. non-fiction) is the only semantic proxy of cultural background considered.
    Stated in the Limitations section; the equitability analysis is conditioned on this single proxy, ignoring other demographic and semantic dimensions.
  • domain assumption Inglehart-Welzel's world cultural map supports Ethiopia, India, and the USA as distinct cultures.
    Section 3.1 uses this map to justify the three book-origin countries; no cultural-distance measure is used in the analysis.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Reading between the Lines: Can LLMs Identify Cross-Cultural Communication Gaps?." pith.science (2026). https://pith.science/paper/2MYQPQ3S

@misc{pith2026250209636,
  author       = {Pith},
  title        = {Pith review of: Reading between the Lines: Can LLMs Identify Cross-Cultural Communication Gaps?},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/2MYQPQ3S}},
  note         = {Machine review of arXiv:2502.09636}
}
read the original abstract

In a rapidly globalizing and digital world, content such as book and product reviews created by people from diverse cultures are read and consumed by others from different corners of the world. In this paper, we investigate the extent and patterns of gaps in understandability of book reviews due to the presence of culturally-specific items and elements that might be alien to users from another culture. Our user-study on 57 book reviews from Goodreads reveal that 83\% of the reviews had at least one culture-specific difficult-to-understand element. We also evaluate the efficacy of GPT-4o in identifying such items, given the cultural background of the reader; the results are mixed, implying a significant scope for improvement. Our datasets are available here: https://github.com/sougata-ub/reading_between_lines

Figures

Figures reproduced from arXiv: 2502.09636 by the authors.

Figure 1
Figure 1. Participant location-wise difficulty in understanding book reviews by country of origin of the book. [PITH_FULL_IMAGE:figures/full_fig_p005_1.png] view at source ↗
Figure 2
Figure 2. Inter Annotator Agreement at Review Level and Span Level across Countries. [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗
Figure 3
Figure 3. Top 15 Span Clusters identified by participants across countries. Corresponding span text in Table [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: Overlap between Human-identified difficult [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]
Figure 5
Figure 5. Figure 5: Overlap percentage of fiction and non-fiction [PITH_FULL_IMAGE:figures/full_fig_p008_5.png]
Figure 6
Figure 6. Figure 6: Precision and recall of the overlap between user-identified culturally difficult spans and GPT-4o-identified [PITH_FULL_IMAGE:figures/full_fig_p009_6.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. XCR-Bench: Benchmarking Cross-Cultural Reasoning in LLMs via Culture-Specific Items and Hall's Triad

    cs.CL 2026-01 conditional novelty 6.0 of 10

    XCR-Bench provides 4,100+ parallel sentences with 1,098 culture-specific items mapped to Hall's Triad, and shows LLMs struggle most with deeper, semi-visible cultural norms.

Reference graph

Works this paper leans on

30 extracted references · 23 canonical work pages · cited by 1 Pith paper

  1. [1]

    Cultural, Linguistic: Differences related to language use, idioms, dialects, or phrasing specific to a culture

    work page

  2. [2]

    Cultural, Non-linguistic: Cultural references such as items, traditions, customs, or social norms unrelated to language

    work page

  3. [3]

    Non-Cultural, Linguistic: General language issues such as grammar, vocabulary, or syntax that are not culturally specific

    work page

  4. [4]

    Non-Cultural, Non-linguistic: Non-cultural factors like complex concepts, technical jargon, or other knowledge-related gaps

    work page

  5. [5]

    Task 1": {<span number>: {’Type’: <taxonomy number>, ’Explanation’: <explain classification reason in 20 words>}},

    Poor Quality Annotations: Vague or Ambiguous Annotation, Superfluous Annotation, Marked majority span. Task 1: For each span from the review text below, select the appropriate category from the above taxonomy and briefly explain your reasoning behind the classification in 20 words. Task 2: Cluster the spans which are similar. Review text: {text} Spans: {s...

    work page

  6. [6]

    Linguistic: Terms unique to a specific language or dialect, including metaphors, idioms, proverbs, humor, sarcasm, slang, and colloquialisms

    work page

  7. [7]

    In Proceedings of the 28th Conference on Computational Natural Lan- guage Learning, pages 400–418, Miami, FL, USA

    Translating across cultures: LLMs for in- tralingual cultural adaptation. In Proceedings of the 28th Conference on Computational Natural Lan- guage Learning, pages 400–418, Miami, FL, USA. Association for Computational Linguistics. Ami D Sperber, Robert F Devellis, and Brian Boehlecke

    work page

  8. [11]

    the oath of vayuputras

    Does mapo tofu contain coffee? probing llms for food-related cultural knowledge. Preprint, arXiv:2404.06833. A Appendix Identified CSICustomsHabitsLinguisticMaterialotherSocialcrazy damselin distress 4 kickass model 4purple for mytaste 4 gateway book 2paranormal/vampireelement 1 2 2 Harlequins 1 3leave muchroom 1 Minus 1romance forthe ages 1 slow andatmos...

    arXiv

Show all 30 references
  1. [17]

    Ecology: Geographical features, flora, fauna, weather conditions, etc

  2. [18]

    Material: Objects, artifacts, and products specific to a culture, such as food, clothing, houses, and towns

  3. [19]

    Social: Hierarchies, practices, and rituals specific to a culture

  4. [20]

    Customs, activities, procedures, and concepts

    Customs: Political, social, legal, religious, and artistic organizations and practices. Customs, activities, procedures, and concepts

  5. [21]

    Habits: Gestures, non-verbal communication methods, and everyday habits unique to a culture

  6. [23]

    Task 3: For each CSI, identify its familiarity from one of the following four levels:

    Other: Anything not belonging to the above six categories. Task 3: For each CSI, identify its familiarity from one of the following four levels:

  7. [24]

    Familiar: Most people from my culture know and relate to the concept as intended

  8. [25]

    Somewhat familiar: Only some people from my culture know and relate to the concept as intended

  9. [26]

    Unfamiliar: Most people from my culture do not know or relate to the concept

  10. [27]

    Task 4: For each CSI, identify its impact on the readability and understandability of the main point of the entire review text from one of the following three levels:

    Ambiguous: Most people from my culture know the concept, but its interpretation is varied or conflicting. Task 4: For each CSI, identify its impact on the readability and understandability of the main point of the entire review text from one of the following three levels:

  11. [28]

    High: Greatly hinders the readability and comprehension of the review, making it difficult to convey its main points effectively

  12. [29]

    Medium: It somewhat affects the readability and comprehension of the review, leading to only partial conveyance of its content

  13. [30]

    Task 5: Within 50 words, detail your reason for highlighting the span as CSI in Task 1 by correlating it with my background

    Low: The review text’s readability and comprehension will remain unaffected. Task 5: Within 50 words, detail your reason for highlighting the span as CSI in Task 1 by correlating it with my background. Task 6: Explain each CSI span within 20 words to make it more understandabl...

  14. [1994]

    Journal of cross-cultural psychology , 25(4):501–524

    Cross-cultural translation: methodology and validation. Journal of cross-cultural psychology , 25(4):501–524. Kumar Tanmay, Aditi Khandelwal, Utkarsh Agarwal, and Monojit Choudhury. 2023. Probing the moral de- velopment of large language models through defining issues test. Pr...

    2023 arXiv

  15. [1998]

    Interacting with computers, 9(3):287–309

    Meaning, the central issue in cross-cultural hci design. Interacting with computers, 9(3):287–309. Yong Cao, Li Zhou, Seolhwa Lee, Laura Cabello, Min Chen, and Daniel Hershcovich. 2023. Assessing cross-cultural alignment between ChatGPT and hu- man societies: An empirical stud...

    1987 arXiv

  16. [2007]

    Proceedings of the National Academy of Sciences , 104(35):13868–13874

    Cultural mosaics and mental models of nature. Proceedings of the National Academy of Sciences , 104(35):13868–13874. Susan Bassnett. 2007. Culture and translation. A com- panion to translation studies, pages 13–23. Jacob Beck. 2023. Quality aspects of annotated data: A researc...

    2007

  17. [2009]

    In Proceedings of the NAACL HLT 2009 workshop on active learning for natural language processing, pages 27–35

    Data quality from crowdsourcing: a study of annotation selection criteria. In Proceedings of the NAACL HLT 2009 workshop on active learning for natural language processing, pages 27–35. Yifan Hu, Yehuda Koren, and Chris V olinsky. 2008. Collaborative filtering for implicit fee...

    2009 arXiv

  18. [2018]

    Eurasian Journal of Educational Research , 18(78):139–158

    A study on communication breakdowns: Sources of misunderstanding in a cross-cultural set- ting. Eurasian Journal of Educational Research , 18(78):139–158. Fajri Koto, Rahmad Mahendra, Nurul Aisyah, and Tim- othy Baldwin. 2024. IndoCulture: Exploring geo- graphically influenced...

    2024 arXiv

  19. [2020]

    Nature Hu- man Behaviour, 4(10):1029–1038

    Cultural influences on word meanings revealed through large-scale semantic alignment. Nature Hu- man Behaviour, 4(10):1029–1038. Maxim Tkachenko, Mikhail Malyuk, Andrey Holmanyuk, and Nikolai Liubimov. 2020-

    2020

  20. [2022]

    Open source software available from https://github.com/heartexlabs/label-studio

    Label Studio: Data labeling soft- ware. Open source software available from https://github.com/heartexlabs/label-studio. Harish Trivedi. 2007. Translating culture vs. cultural translation. Benjamins translation library, 71:277. Mengting Wan and Julian J. McAuley. 2018. Item re...

    2007 arXiv

  21. [2023]

    In Findings of the Association for Com- putational Linguistics: EMNLP 2023, pages 13370– 13388, Singapore

    Ethical reasoning over moral alignment: A case and framework for in-context ethical policies in LLMs. In Findings of the Association for Com- putational Linguistics: EMNLP 2023, pages 13370– 13388, Singapore. Association for Computational Linguistics. Nils Reimers and Iryna Gu...

    2023 arXiv

  22. [2024]

    Preprint, arXiv:2410.01811

    Evaluating cultural awareness of llms for yoruba, malayalam, and english. Preprint, arXiv:2410.01811. Arthur T Denzau, Douglass C North, et al. 1994. Shared mental models: ideologies and institutions. KYKLOS- BERNE-, 47:3–3. Benjamin D Douglas, Patrick J Ewell, and Markus Brau...

    1994 arXiv

Pith tools

Reviewed August 8, 2026 · model on record in the stance chip above.