ABSTRACT
Political leaders articulate themselves best via speeches and/or writings across diverse media (print or electronic) for campaigning, pitching their stand, confronting the opponent/s, impressing upon the citizens, penning down their biographies, and the like. While making speeches, politicians evince provocative sentiments themselves that are likely to move the audience-that is the prime objective of any orator. Concomitantly, however, the politicians make speeches charged with emotions to drive home a point. The present study seeks to hinge itself upon the speeches of Volodymyr Oleksandrovych Zelenskyy, the Ukrainian President, who is embroiled in a war with Russia since February 2022. Specifically, sentiment analysis was done to understand the dynamics of emotions that wavered with the progress of war. Computational text analysis of speeches for a specified period (24 February 2022 until 24 July 2022) shows that sentiments appear to increase over a period of time wherein the best predictor, in our Bayesian regression models, for a change in Zelenskyy’s sentiment between today and tomorrow is his “present” sentiment-the sentiment that he evinces “today.” Implicitly, if we detect a high/low positive sentiment “today,” we would expect to see a strong mean regression such that tomorrow’s sentiment should be close to neutral. Findings suggest that in contrast with the general observation that peculiar war events tend to have great power in explaining changes in sentiments, the same was found only to be an ancillary factor in the present study. The study is rounded off with further research pointers with practitioner implications.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Data availability statement
Data and code can be found in https://github.com/felipemaiapolo/zelenskyy_speeches Alternatively, please refer https://doi.org/10.7910/DVN/QLU9CN, reference number UNF:6:ylu9Et7U+Wk6q/6Q2yQXjw== [fileUNF].
Correction Statement
This article has been corrected with minor changes. These changes do not impact the academic content of the article.
Notes
1. See https://www.president.gov.ua/en/news/speeches (last accessed on 06/26/2022).
2. All Ukrainian speeches were translated to English by the presidential office.
3. We are grateful for the code piece written by Yuri Zhukov that helped us filtering rows in the dataset.
4. Municipality level.
5. See https://radimrehurek.com/gensim/auto_examples/howtos/run_downloader_api.html (last accessed on 06/26/2022).
6. It is also worth mentioning that even before standardization, the covariates series do not exhibit deterministic or stochastic trends and seem to be at least mean stationary.
7. Please check Appendix A for the likelihood function derivation. The Bayesian regression Python implementation we used is available in https://github.com/tonyduan/conjugate-bayes..
8. The goal of this analysis is just to check the predictive power of some of the predictors and not to make real forecasts. Given this, we consider that using the whole data to detrend or standardize our variables does not invalidate this analysis.
9. Using the same prior distributions specified earlier.
10. Where the standard deviation for the posterior distribution is .002.
11. From our diagnostics, we see that residuals do not exhibit clear autocorrelation, heteroskedasticity, or non-normal distribution.
12. The greater the better in their formulation of BIC.
13. See https://www.statsmodels.org/devel/examples/notebooks/generated/stationarity_detrending_adf_kpss.html (accessed in 07/26/2022).