{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,7,24]],"date-time":"2026-07-24T06:02:44Z","timestamp":1784872964859,"version":"3.55.0"},"reference-count":44,"publisher":"Frontiers Media SA","license":[{"start":{"date-parts":[[2026,7,24]],"date-time":"2026-07-24T00:00:00Z","timestamp":1784851200000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":["frontiersin.org"],"crossmark-restriction":true},"short-container-title":["Front. Artif. Intell."],"abstract":"<jats:p>Social media sentiment analysis has become one of the most significant instruments for understanding the opinion of the population in the spheres of healthcare, politics, and education. Yet, large language models (LLMs) remain unevenly distributed in their linguistic coverage, failing to adequately serve a large portion of the world's languages. This study evaluates five state-of-the-art LLMs: GPT-4o, Gemini 2.0 Flash, DeepSeek-V3, Mistral Large, and Claude 3.7 Sonnet on three-class sentiment classification across 36 datasets spanning 36 languages, with emphasis on low- and medium-resource settings, using zero-shot and few-shot prompting without task-specific fine-tuning. In addition to the traditional measures of performance per language, the study presents a hierarchical analysis of languages based on a genealogical tree of Indo-European, Afro-Asiatic, Niger-Congo, Turkic, Austronesian, and English Creole language families, so that it is possible to identify the systematic patterns of performance superiority and inferiority among the language families. The findings show that few-shot prompting improves the results of a vast majority of languages, with several models approaching or surpassing the performance of the state-of-the-art benchmark of task-specific models. The GPT-4o and Claude achieved the highest performance in the high-resource and medium-resource settings, and Gemini is a competent trade-off that allows balancing the performance and the computational cost. Although it has lower zero-shot performance, Mistral benefits the most from few-shot prompting and becomes highly competitive in the few-shot setting. Despite these developments, the level of performance on low-resource languages, such as Oromo, Xitsonga, Azerbaijani, and Twi, remains significantly lower, underscoring that progress in multilingual LLMs requires moving beyond English-centric evaluation toward genuinely representative and globally inclusive benchmarks.<\/jats:p>","DOI":"10.3389\/frai.2026.1854873","type":"journal-article","created":{"date-parts":[[2026,7,24]],"date-time":"2026-07-24T05:34:55Z","timestamp":1784871295000},"update-policy":"https:\/\/doi.org\/10.3389\/crossmark-policy","source":"Crossref","is-referenced-by-count":0,"title":["Do language families matter? Evaluating LLMs for sentiment analysis through a hierarchical cross-lingual lens"],"prefix":"10.3389","volume":"9","author":[{"given":"Muhamet","family":"Kastrati","sequence":"first","affiliation":[{"name":"Department of Computer Science, University of New York Tirana","place":["Tiran\u00eb, Albania"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Abdul","family":"Manaf","sequence":"additional","affiliation":[{"name":"Department of Computer Science, Sukkur IBA University","place":["Sukkur, Pakistan"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Ali Shariq","family":"Imran","sequence":"additional","affiliation":[{"name":"Department of Computer Science, Norwegian University of Science and Technology","place":["Gj\u00f8vik, Norway"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Zenun","family":"Kastrati","sequence":"additional","affiliation":[{"name":"Department of Informatics, Linnaeus University","place":["V\u00e4xj\u00f6, Sweden"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Sher Muhammad","family":"Daudpota","sequence":"additional","affiliation":[{"name":"Department of Computer Science, Sukkur IBA University","place":["Sukkur, Pakistan"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Marenglen","family":"Biba","sequence":"additional","affiliation":[{"name":"Department of Computer Science, University of New York Tirana","place":["Tiran\u00eb, Albania"]}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1965","published-online":{"date-parts":[[2026,7,24]]},"reference":[{"key":"B1","doi-asserted-by":"publisher","first-page":"353","DOI":"10.14569\/IJACSA.2024.0150437","article-title":"Comparative analysis of transformer models for sentiment analysis in low-resource languages","volume":"15","author":"Aliyu","year":"2024","journal-title":"Int. J. Adv. Comput. Sci. Appl"},{"key":"B2","doi-asserted-by":"publisher","first-page":"e64723","DOI":"10.2196\/64723","article-title":"Evaluating large language models for sentiment analysis and hesitancy analysis on vaccine posts from social media: qualitative study","volume":"9","author":"Annan","year":"2025","journal-title":"JMIR Form. Res"},{"key":"B3","first-page":"52","article-title":"\u201cMultilingual sentiment analysis using machine translation?,\u201d","volume-title":"Proceedings of the 3rd workshop on computational approaches to subjectivity and sentiment analysis (WASSA'12)","author":"Balahur","year":"2012"},{"key":"B4","doi-asserted-by":"publisher","first-page":"258","DOI":"10.63317\/4fhqtpvii3vg","article-title":"\u201cXLM-T: multilingual language models in twitter for sentiment analysis and beyond,\u201d","author":"Barbieri","year":"2022","journal-title":"Proceedings of the thirteenth language resources and evaluation conference"},{"key":"B5","unstructured":"Leveraging chatgpt as a text annotation tool for sentiment analysis\n          \n          \n            \n              Belal\n              M.\n            \n            \n              She\n              J.\n            \n            \n              Wong\n              S.\n            \n          \n          arXiv\n          \n          2023"},{"key":"B6","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1109\/SEEDA-CECNSM.2019.8908289","article-title":"\u201cA hybrid method for sentiment analysis of election related tweets,\u201d","author":"Beleveslis","year":"2019","journal-title":"The 4th south-east europe design automation, computer engineering, computer networks and social media conference (SEEDA-CECNSM 2019)"},{"key":"B7","author":"Berggren","year":"2023","journal-title":"Sentiment analysis for Swedish: the impact of emojis on sentiment analysis of Swedish informal texts"},{"key":"B8","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1145\/3641289","article-title":"A survey on evaluation of large language models","volume":"15","author":"Chang","year":"2024","journal-title":"ACM Trans. Intell. Syst. Technol"},{"key":"B9","doi-asserted-by":"publisher","first-page":"e2851","DOI":"10.7717\/peerj-cs.2851","article-title":"Bridging resource gaps in cross-lingual sentiment analysis: adaptive self-alignment with data augmentation and transfer learning","volume":"11","author":"Chen","year":"2025","journal-title":"PeerJ Comput. Sci"},{"key":"B10","doi-asserted-by":"publisher","first-page":"419","DOI":"10.3115\/v1\/P15-1041","article-title":"\u201cLearning to adapt credible knowledge in cross-lingual sentiment analysis,\u201d","author":"Chen","year":"2015","journal-title":"Proceedings of the 53rd annual meeting of the association for computational linguistics and the 7th international joint conference on natural language processing, Vol. 1: Long Papers"},{"key":"B11","unstructured":"XNLI: evaluating cross-lingual sentence representations\n          \n          \n            \n              Conneau\n              A.\n            \n            \n              Lample\n              G.\n            \n            \n              Rinott\n              R.\n            \n            \n              Williams\n              A.\n            \n            \n              Bowman\n              S. R.\n            \n            \n              Schwenk\n              H.\n            \n          \n          arXiv\n          \n          2018"},{"key":"B12","doi-asserted-by":"publisher","first-page":"e51069","DOI":"10.2196\/51069","article-title":"Efficacy of chatgpt in cantonese sentiment analysis: comparative study","volume":"26","author":"Fu","year":"2024","journal-title":"J. Med. Internet Res"},{"key":"B13","doi-asserted-by":"publisher","first-page":"285","DOI":"10.18653\/v1\/2023.semeval-1.38","article-title":"\u201cUMUTeam at semeval-2023 task 12: ensemble learning of LLMs applied to sentiment analysis for low-resource african languages,\u201d","author":"Garc\u00eda-D\u00edaz","year":"2023","journal-title":"Proceedings of the 17th international workshop on semantic evaluation (SemEval-2023)"},{"key":"B14","first-page":"65","article-title":"\u201cSentiment analysis in Czech social media using supervised machine learning,\u201d","volume-title":"Proceedings of the 4th workshop on computational approaches to subjectivity, sentiment and social media analysis","author":"Habernal","year":"2013"},{"key":"B15","unstructured":"Azerbaijani Tweet Emotion Classification Dataset\n          \n          2023"},{"key":"B16","doi-asserted-by":"publisher","DOI":"10.20944\/preprints202508.0753.v1","article-title":"A review of trends and challenges in adopting ai models through cross-lingual transfer learning via sentiment analysis","author":"Jefry","year":"2025","journal-title":"Preprints"},{"key":"B17","first-page":"249","article-title":"\u201cSentiment analysis in Twitter for Macedonian,\u201d","author":"Jovanoski","year":"2015","journal-title":"Proceedings of the international conference recent advances in natural language processing"},{"key":"B18","doi-asserted-by":"publisher","first-page":"577","DOI":"10.1007\/978-981-16-8515-6_44","article-title":"\u201cSocial versus physical distancing: analysis of public health messages at the start of COVID-19 outbreak in malaysia using natural language processing,\u201d","author":"Juan","year":"2022","journal-title":"Proceedings of the 8th international conference on computational science and technology: ICCST 2021"},{"key":"B19","doi-asserted-by":"publisher","first-page":"13","DOI":"10.1007\/978-3-031-16564-1_2","article-title":"\u201cSentiment polarity and emotion detection from tweets using distant supervision and deep learning models,\u201d","author":"Kastrati","year":"2022","journal-title":"Proceedings of the international symposium on methodologies for intelligent systems"},{"key":"B20","doi-asserted-by":"publisher","first-page":"110136","DOI":"10.1016\/j.engappai.2025.110136","article-title":"Unlocking language barriers: assessing pre-trained large language models across multilingual tasks and unveiling the black box with explainable artificial intelligence","volume":"149","author":"Kastrati","year":"2025","journal-title":"Eng. Appl. Artif. Intell"},{"key":"B21","doi-asserted-by":"publisher","first-page":"1045","DOI":"10.1007\/s10844-024-00845-0","article-title":"Leveraging distant supervision and deep learning for twitter sentiment and emotion classification","volume":"62","author":"Kastrati","year":"2024","journal-title":"J. Intell. Inf. Syst"},{"key":"B22","doi-asserted-by":"publisher","first-page":"1133","DOI":"10.3390\/electronics10101133","article-title":"A deep learning sentiment analyser for social media comments in low-resource languages","volume":"10","author":"Kastrati","year":"2021","journal-title":"Electronics"},{"key":"B23","doi-asserted-by":"publisher","first-page":"26541","DOI":"10.1109\/ACCESS.2023.3257283","article-title":"Soaring energy prices: understanding public engagement on twitter using sentiment analysis and topic modeling with transformers","volume":"11","author":"Kastrati","year":"2023","journal-title":"IEEE Access"},{"key":"B24","unstructured":"Zero-shot sentiment analysis in low-resource languages using a multilingual sentiment lexicon\n          \n          \n            \n              Koto\n              F.\n            \n            \n              Beck\n              T.\n            \n            \n              Talat\n              Z.\n            \n            \n              Gurevych\n              I.\n            \n            \n              Baldwin\n              T.\n            \n          \n          arXiv\n          \n          2024"},{"key":"B25","doi-asserted-by":"publisher","first-page":"e50150","DOI":"10.2196\/50150","article-title":"A comparison of ChatGPT and fine-tuned open pre-trained transformers (opt) against widely used sentiment analysis tools: sentiment analysis of COVID-19 survey data","volume":"11","author":"Lossio-Ventura","year":"2024","journal-title":"JMIR Ment. Health"},{"key":"B26","unstructured":"Large multimodal models for low-resource languages: a survey\n          \n          \n            \n              Lupascu\n              M.\n            \n            \n              Rogoz\n              A.-C.\n            \n            \n              Stupariu\n              M. S.\n            \n            \n              Ionescu\n              R. T.\n            \n          \n          arXiv\n          \n          2025"},{"key":"B27","doi-asserted-by":"publisher","first-page":"9603","DOI":"10.1038\/s41598-024-60210-7","article-title":"A multimodal approach to cross-lingual sentiment analysis with ensemble of transformer and LLM","volume":"14","author":"Miah","year":"2024","journal-title":"Sci. Rep"},{"key":"B28","doi-asserted-by":"publisher","first-page":"e0155036","DOI":"10.1371\/journal.pone.0155036","article-title":"Multilingual twitter sentiment classification: the role of human annotators","volume":"11","author":"Mozeti\u010d","year":"2016","journal-title":"PLoS ONE"},{"key":"B29","unstructured":"SemEval-2023 task 12: sentiment analysis for african languages (AfriSenti-SemEval)\n          \n          \n            \n              Muhammad\n              S. H.\n            \n            \n              Abdulmumin\n              I.\n            \n            \n              Yimam\n              S. M.\n            \n            \n              Adelani\n              D. I.\n            \n            \n              Ahmad\n              I. S.\n            \n            \n              Ousidhoum\n              N.\n            \n          \n          arXiv\n          \n          2023"},{"key":"B30","doi-asserted-by":"publisher","first-page":"100124","DOI":"10.1016\/j.nlp.2024.100124","article-title":"Evaluation of open and closed-source LLMs for low-resource language with zero-shot, few-shot, and chain-of-thought prompting","volume":"10","author":"Nazi","year":"2025","journal-title":"Nat. Lang. Process. J"},{"key":"B31","author":"Oye","year":"2015","journal-title":"Sentiment analysis of Norwegian twitter messages"},{"key":"B32","first-page":"460","article-title":"\u201cDANLP: an open-source toolkit for danish natural language processing,\u201d","author":"Pauli","year":"2021","journal-title":"Proceedings of the 23rd Nordic conference on computational linguistics (NoDaLiDa)"},{"key":"B33","doi-asserted-by":"publisher","first-page":"5993","DOI":"10.3390\/app10175993","article-title":"Zero-shot learning for cross-lingual news sentiment classification","volume":"10","author":"Pelicon","year":"2020","journal-title":"Appl. Sci"},{"key":"B34","unstructured":"A systematic survey of prompt engineering in large language models: techniques and applications\n          \n          \n            \n              Sahoo\n              P.\n            \n            \n              Singh\n              A. K.\n            \n            \n              Saha\n              S.\n            \n            \n              Jain\n              V.\n            \n            \n              Mondal\n              S.\n            \n            \n              Chadha\n              A.\n            \n          \n          arXiv\n          \n          2024"},{"key":"B35","doi-asserted-by":"publisher","first-page":"214","DOI":"10.1109\/FIT60620.2023.00047","article-title":"\u201cExploring the potential of large-language models (LLMs) for student feedback sentiment analysis,\u201d","author":"Shaikh","year":"2023","journal-title":"Proceedings of the 2023 international conference on frontiers of information technology (FIT)"},{"key":"B36","doi-asserted-by":"publisher","DOI":"10.24432\/C58325","author":"Sharf","year":"2017","journal-title":"Roman Urdu Data Set"},{"key":"B37","doi-asserted-by":"publisher","first-page":"1426","DOI":"10.1109\/ICMLA61862.2024.00221","article-title":"\u201cLeveraging LLMs for integrated sentiment and topic analysis on african social media,\u201d","author":"Sibitenda","year":"2024","journal-title":"Proceedings of the 2024 international conference on machine learning and applications (ICMLA)"},{"key":"B38","doi-asserted-by":"publisher","DOI":"10.3233\/FAIA200622","article-title":"\u201cWhat can we learn from almost a decade of food tweets,\u201d","year":"2020","journal-title":"Proceedings of the 9th conference human language technologies"},{"key":"B39","unstructured":"Resolution Adopted by the General Assembly on 25 September 2015\n          \n          2015"},{"key":"B40","doi-asserted-by":"publisher","first-page":"121","DOI":"10.1080\/24751839.2023.2173843","article-title":"Exploring zero-shot and joint training cross-lingual strategies for aspect-based sentiment analysis based on contextualized multilingual language models","volume":"7","author":"Van Thin","year":"2023","journal-title":"J. Inf. Telecommun"},{"key":"B41","doi-asserted-by":"publisher","first-page":"5731","DOI":"10.1007\/s10462-022-10144-1","article-title":"A survey on sentiment analysis methods, applications, and challenges","volume":"55","author":"Wankhade","year":"2022","journal-title":"Artif. Intell. Rev"},{"key":"B42","doi-asserted-by":"publisher","first-page":"8079","DOI":"10.1007\/s13042-025-02711-z","article-title":"Evaluating zero-shot multilingual aspect-based sentiment analysis with large language models","volume":"16","author":"Wu","year":"2025","journal-title":"Int. J. Mach. Learn. Cybernet"},{"key":"B43","unstructured":"Sentiment analysis in the era of large language models: a reality check\n          \n          \n            \n              Zhang\n              W.\n            \n            \n              Deng\n              Y.\n            \n            \n              Liu\n              B.\n            \n            \n              Pan\n              S. J.\n            \n            \n              Bing\n              L.\n            \n          \n          arXiv\n          \n          2023"},{"key":"B44","unstructured":"The model arena for cross-lingual sentiment analysis: a comparative study in the era of large language models\n          \n          \n            \n              Zhu\n              X.\n            \n            \n              Gardiner\n              S.\n            \n            \n              Rold\u00e1n\n              T.\n            \n            \n              Rossouw\n              D.\n            \n          \n          arXiv\n          \n          2024"}],"container-title":["Frontiers in Artificial Intelligence"],"original-title":[],"link":[{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/frai.2026.1854873\/full","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,7,24]],"date-time":"2026-07-24T05:34:58Z","timestamp":1784871298000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/frai.2026.1854873\/full"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,7,24]]},"references-count":44,"alternative-id":["10.3389\/frai.2026.1854873"],"URL":"https:\/\/doi.org\/10.3389\/frai.2026.1854873","relation":{},"ISSN":["2624-8212"],"issn-type":[{"value":"2624-8212","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,7,24]]},"article-number":"1854873"}}