DO UNHAPPY REVIEWERS MIX IN MORE RUSSIAN? STAR RATINGS AND KAZAKH-RUSSIAN CODE-SWITCHING IN 175,002 ONLINE REVIEWS

Abstract

Bilingual people often say that they switch languages when they are emotional, but no measurement of this for Kazakh and Russian was found. This study asks whether Kazakh-speaking reviewers mix Russian words into their reviews more often when they are dissatisfied (1-2 stars) than when they are satisfied (4-5 stars). The data are all 175,002 non-empty reviews of the public KazSAnDRA dataset of Kazakh online reviews, each with a star rating from 1 to 5. For every review, the share of Russian words among all words was counted with an automatic rule that the author's mentoring team built from a boundary and twenty example words she wrote before the count: an old, established loanword such as телефон or автобус counts as Kazakh, while a Russian word used instead of an existing Kazakh one, such as скидка or доставка, counts as Russian. A word-by-word check of all 1,763 words in 160 randomly drawn reviews (60 of them drawn from reviews the rule marked as mixed), made by an AI assistant, found 90% precision and 83% recall (92% and 83% in the fully random part). The mean share of Russian words falls with the rating, from 3.85% in 1-star reviews to 2.49% in 5-star reviews (ratio 1.55, 95% CI 1.50 to 1.60), and 1-2-star reviews contain 1.49 times as much Russian as 4-5-star reviews. The direction predicted by the author's hypothesis is confirmed, but her forecast of "three times" was about twice the observed ratio. The difference comes almost entirely from how often a review contains any Russian (23.5% of 1-star against 15.5% of 5-star reviews); among reviews that do, the share of Russian words is only slightly higher (1.05 times in short and 1.18 times in long reviews). The difference holds in every additional check (smallest ratio 1.17, at 10 per million) and is larger in long reviews (1.79 times) than in short ones (1.43). Relative to their frequency, the largest excess (3.15 times; a grouping made after a first review) is in Russian emphatic particles and degree adverbs such as вообще, уже, еще and даже; they are about a tenth of the extra Russian words, most of which are content words such as реклама and обновление.

Keywords: code-switching, Kazakh-Russian bilingualism, emotion, sentiment, online reviews, star ratings, corpus linguistics, KazSAnDRA.

Список литературы

1. Komekova S. Testing classic code-switching constraints on Kazakh-Russian social-media text: an insertional corpus study // Central Asian Scientific Journal. 2026. No. 2(30), Vol. 3. P. 50-58.
2. Smagulova J. Language policies of Kazakhization and their influence on language attitudes and use // International Journal of Bilingual Education and Bilingualism. 2008. Vol. 11, No. 3-4. P. 440-475. DOI: 10.1080/13670050802148798.
3. Yeshpanov R., Varol H.A. KazSAnDRA: Kazakh Sentiment Analysis Dataset of Reviews and Attitudes // Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024). Torino: ELRA and ICCL, 2024. P. 9657-9667. URL: https://aclanthology.org/2024.lrec-main.844/ (accessed: 11.10.2026).
4. Dewaele J.-M. The emotional force of swearwords and taboo words in the speech of multilinguals // Journal of Multilingual and Multicultural Development. 2004. Vol. 25, No. 2-3. P. 204-222. DOI: 10.1080/01434630408666529.
5. Rudra K., Rijhwani S., Begum R., Bali K., Choudhury M., Ganguly N. Understanding language preference for expression of opinion and sentiment: what do Hindi-English speakers do on Twitter? // Proceedings of the 2016 Conference on Empirical Methods in Natural Language Processing. Austin: Association for Computational Linguistics, 2016. P. 1131-1141. DOI: 10.18653/v1/D16-1121.
6. Acuña Ferreira A.V. Code-switching and emotions display in Spanish/Galician bilingual conversation // Text & Talk. 2017. Vol. 37, No. 1. P. 47-69. DOI: 10.1515/text-2016-0036.
7. Karimpour S., De Costa P., Nazari M. Emotions in switching the code: contributions for English language teachers' identity construction // Journal of Multilingual and Multicultural Development. 2026. Vol. 47, No. 3. P. 1528-1540 (published online 2024). DOI: 10.1080/01434632.2024.2429413.
8. Gazez A.A. Emotional expression in two languages: bilingual perspectives on language choice and emotional authenticity // Central Asian Scientific Journal. 2026. No. 1(29), Vol. 2. P. 213-217.
9. Lee S.Y.M., Wang Z. Emotion in code-switching texts: corpus construction and analysis // Proceedings of the Eighth SIGHAN Workshop on Chinese Language Processing. Beijing: Association for Computational Linguistics, 2015. P. 91-99. DOI: 10.18653/v1/W15-3116.
10. Tang T., Tang X., Yuan T. Fine-tuning BERT for multi-label sentiment analysis in unbalanced code-switching text // IEEE Access. 2020. Vol. 8. P. 193248-193256. DOI: 10.1109/ACCESS.2020.3030468.
11. Vilares D., Alonso M.A., Gómez-Rodríguez C. EN-ES-CS: an English-Spanish code-switching Twitter corpus for multilingual sentiment analysis // Proceedings of the Tenth International Conference on Language Resources and Evaluation (LREC'16). Portorož: ELRA, 2016. P. 4149-4153. URL: https://aclanthology.org/L16-1655/ (accessed: 11.10.2026).
12. Hu N., Zhang J., Pavlou P.A. Overcoming the J-shaped distribution of product reviews // Communications of the ACM. 2009. Vol. 52, No. 10. P. 144-147. DOI: 10.1145/1562764.1562800.
13. Савирдинова А. Сленг в современном языке молодёжи: влияние социальных сетей // Central Asian Scientific Journal. 2026. № 2(30), Т. 1. С. 569-572.
14. Anuarova S.A., Konarova A.B. Language, psyche, and society: cognitive and social interrelations in a multilingual context // Central Asian Scientific Journal. 2026. No. 2(30), Vol. 1. P. 140-151.
15. Assan R. KazSwitch: a controlled benchmark for Kazakh-Russian code-switching in multilingual language models // Central Asian Scientific Journal. 2026. No. 2(30), Vol. 3. P. 65-72.
16. Institute of Smart Systems and Artificial Intelligence (ISSAI). KazSAnDRA: Kazakh Sentiment Analysis Dataset of Reviews and Attitudes. GitHub repository. URL: https://github.com/IS2AI/KazSAnDRA (accessed: 11.10.2026).
17. Korobov M. Morphological analyzer and generator for Russian and Ukrainian languages // Analysis of Images, Social Networks and Texts (AIST 2015). Communications in Computer and Information Science. Cham: Springer, 2015. P. 320-332. DOI: 10.1007/978-3-319-26123-2_31.
18. Wikimedia Foundation. Wikipedia (Kazakh), dump of 01.11.2023, dataset "wikimedia/wikipedia", configuration 20231101.kk. URL: https://huggingface.co/datasets/wikimedia/wikipedia (accessed: 11.10.2026).

Как цитировать

Komekova S. DO UNHAPPY REVIEWERS MIX IN MORE RUSSIAN? STAR RATINGS AND KAZAKH-RUSSIAN CODE-SWITCHING IN 175,002 ONLINE REVIEWS // Central Asian Scientific Journal. – 2026. – No. 4(32), Vol. 1. – Pp. 291–303.