Introduction

During the COVID-19 pandemic, fluctuations in the semantics of government leaders’ discourses were observed. The fluctuation referred to the changes noticed in the words’ usage and in the variation from positive to negative sentiments and different emotions that leaders revealed among their discourses1,2. This had a significant effect on analytical thinking and emotional processing in the population being addressed. It also had an impact on the speech and concepts developed by the local populations in their role as receptive speakers. Here we are focusing the source of that impact in different populations considering country and leadership. Leaders differ greatly in their social status and engagement style. During the pandemic crisis those differences emerged with empathy variation directly in their communities. Almost all countries had the same health prevention rules but the social prescription (of those rules) differed in how that was proposed to populations. Discourse of nations ’leaders determined the fast adherence of individuals to prevention. Words and emotional intonation are two of several dimensions that can elaborate the success of an intervention plan, in this social and political field concerning public health1,2. The changes that occurred for pandemic evolution and containment between 2020 and 2023 impacted also the modifications observed in the discourses that country leaders displayed for their populations. That impact was not quite understood after the pandemic solved1.

The linguistic changes are prone to be maintained in the post-pandemic period as a result of the extended self-isolation and limited language use during the period of the health emergency1,2. On the one hand, language processing deficits and disturbances in oral production were observed in young children during the pedagogical and development process of learning to read and write, between 2020 and 2022. This has implications for the future mature language development of those children. Reading and writing skills cannot be taught properly using e-learning only3,4,5. On the other hand, the states of emergency (of which there were several) and the rapid transition from classroom to remote learning gave rise to an adaptation to new learning skills. This was a result of more autonomy in language learning from the students’ perspective and as reported by teachers3.

The new features of lexicon, in several populations and their respective languages, were marked by the depression, loneliness and anxiety experienced by different cultures and populations around the world6,7. Even from the beginning of 2023 onwards, the depression and anxiety symptomatology following the pandemic has largely remained. Language is one of the main strategies for self-regulating thought and behaviour. Meaning that language choices reflect how individuals reflect their ‘inner thinking’ organization and how their sentiments and emotions are changing1,4. Externally, each individual’s language usage therefore determines their psycholinguistic identity as well mainly identifying their state of mind at a specific period. Discourses and corpora analyses make it possible to identify changes and biases in language communication, the development of lexicon and the evolution of semantics (meaning of things and referents) during and after the COVID-19 pandemic7,8. We should consider also the impact for the democratic environment, not only perceived as political setting, but changes in languages will have influence in the understanding of new forms of freedom (after lockdowns and restricted measures) and of engagement. Communities around the globe may transform their conceptualization about interaction, both in person and in online contexts. Language is the main instrument to the democratic regulation: rules implementation with consensual opinion and respect. Attending to our line of thought, regarding the democratic behaviours and societal challenges in societies with less liberty for expression and communication, our study focused the engagement wording and the sentiment expressiveness.

Differences among populations that experienced the same pandemic context such as COVID-19 are now struggling with new concepts learned during the restrained periods7. This can be perceived through linguistic analysis – verbal behaviours of individuals, despite of their cultural and social origin. Therefore, the importance of country’ leaders to regulate the communication to orient populations for correct notions of freedom, of inclusion process, and of expressiveness in public settings. The public word changed since 2020 until now, in two ways: we understand that publicly terms are new concepts were the online and digital tools/environments took larger place than expected. Both for young and adult populations. Our main concern is the adolescents and their maturation about the isolation concept. Considering the engagement needs and the low levels of interaction that individuals and families were used during three years. In this article, we focus on the acute period of the pandemic and directly observed changes in specific language traits of the individuals who were responsible for communication in their specific countries and governments. Language as a concept refers to different codes or idioms that a human being uses on a daily basis. Discourse refers to an organized system of segments identified from words to sentences, including intonation and punctuation26,27Discourse analysis is the technique used to code and organize those segments in order to understand the main topics and principal messaging27,28.

After the first lockdown, linguistic markers changed and new forms of words and sentences emerged with meanings suggesting high levels of negative mental health. This, in turn, impacted language and cognitive elaboration2,9,10. Linguistic markers are determinant predictors used to understand how discourse was affected by COVID-19 isolation. Cognitive elaboration is linked to emotional and analytical thinking processing. Between 2020 and 2022, populations were immersed in negative states of mind (depression, loneliness and anxiety), which had a clear and direct effect on language development in adults and on language acquisition in children7,8. We found that the pandemic had negative effects on children’s vocabulary and phonological awareness9. We believe that this results from the impact of the pandemic on the lexicon used within households as a consequence of the periods of social isolation over the three years. This adversely affected linguistic frequency — the development of a rich vocabulary. Children’s lexicon was therefore not as well developed compared to the way that language learning evolves when children interact with their peers in a school environment, something which is crucial. This reflects the alternation between face-to face and remote learning modalities, alongside other constraints and confinement measures that affected people’s freedom of expression and mobility during the pandemic.

Homeschooling and the emotional tone of family discourse seriously affected the cognitive elaboration and language evolution of children, especially in relation to the critical period for phonological awareness9. In the case of adults, their discourse and language were influenced (and often manipulated) by the government discourse in each country. Besides the lockdowns and remote working (full-time online environment), societies self-regulated differently according to personality traits29,30,31.

For that reason, this commentary examines the psychological traits of discourses. Messaging and speaking turned to be in the spot of social media, most of the times with no control8,10. This leaded to uncontrolled perceived stress that is recently being investigated, still with scarce evidence. Child, adolescent and adult stress indicators are clearly different in what respected the manifestation (of stress). Differences regarding stressful symptoms and impact from traumatic events are moderated by the culture and political factor. The last point referred is important – political factor – because political discourses were the only form, between 2020 and 2022, to endorse the behaviour of communities worldwide; to manage stress and anxiety; to respond to emergent scenarios that we lived in; to generate or mitigate other symptoms such as the cognitive perception about the lockdown measures and health priorities. COVID-19 transformed the way of living, the way of thinking and the way of feeling. Feeling and thinking can be observed through the language profiles9.

On the other hand, the language profiles of individuals were not so able to generate specificities – considering linguistic properties that define each speaker and his/her message – because the online interaction at the time of COVID-19 created some type of unique linguistic model. Not always positive for resilience and for coping strategies. Our research shed light on the narratives that populations worldwide shared online on different platforms, with the main focus being themes related to the harsh conditions of the pandemic and its negative effects on the well-being of individuals. Social media became much more individual, from our perspective, than group sharing of information. The narratives revealed similarity in topics (top words) relating to negative sentiments and biased analytical processes of thinking9. That similarity needs to be examined to perceive how sentiments like empathy, clout and analytic thinking (dimensions explored in this study) are managed by individuals in their interactions and daily tasks. The probability tends to point out for lower levels of empathy after the lockdown periods. A new language activation is needed to reinforce the human being with more embracing skills and less negative thoughts about the future.

Perceptions of the pandemic and the persistence of negative emotions after the COVID-19 lockdowns were directly reflected in language expressions and in the discursive novelties created within communities. Hence, a specific psycholinguistic analysis is important to understand and describe differences in emotional tone and analytical thinking across authentic discourses, and their consequences for public health1,11,12. By authentic discourses we refer to the governmental sites where they appeared in real time. Analytical thinking is one of the categories calculated by LIWC using the respective score from the dictionary. The analytical thinking is captured across several dimensions that words may present: from logical to hierarchical dimensions. Speakers more analytical use high number of personal pronouns and reference to himself/herself, as well focus the “here-and now” context. In this psycholinguistic analysis, it is important to examine the misinformation disseminated by social media and by government leaders from the beginning of the COVID-19 crisis13,14,15. During lockdown, parts of the discourse of government leaders were misinterpreted and fed into the misinformation disseminated on social media. On social media platforms, digital communities developed their own interpretations of the information and guidelines issued by the leadership at specific periods of the lockdowns.

In this article, the authors examined the variation and typology of the changes in emotional tone, analytical thinking and clout, specifically in the discourses of government leaders in two different periods of 2020. To provide rigorous word counting, encoding and discourse analysis, we used the Linguistic Inquiry Word Count (LIWC) program32. From here, we designed two questions for this study: (1) how the dimensions of emotional tone, analytical thinking and clout appeared in the speeches of different nations’ leaders during COVID-19?; (2) how language changes (if they are) in the three dimensions between periods 1 and 2 (February and May 2020)?

Data collection and conceptual modelling

A corpus of 39,073 words was collected from eight leaders’ discourses (both oral and written) available online in authentic settings. These settings were government authorized platforms and the English translation was available as official source also (whose content was spread in other information websites and also on social media) and the discourses were encoded in real time for this paper. The sixteen discourses of the country leaders were produced during two distinct states of emergency in 2020. The discourses specifically set out measures and provided updates during the COVID-19 health emergency. The LIWC was used to analyse the psychological variables and language functions, particularly in terms of emotional tone, analytical thinking and clout. As eligibility criteria, texts were encoded only if they were produced in the same periods of 2020, in English official translation, with origin in government sources, and if texts are related to prevention for health and social measures in the pandemic context.

LIWC is a dictionary that examines large databases and discourses placed online, computing and scoring the texts and words according to several dimensions. The LIWC algorithm provides scores with basis in corpora (of words extracted from natural sources/languages to gather the psychological and language traits properly). Toward the algorithm the more recent version of LIWC follows the Meaning Extraction Method and also enable simultaneously the topic modelling. The proper dictionary of LIWC has several subdictionaries to able the codes for groups of words and respective semantic fields. “Crying” is a word that appears frequently in our corpora, so that word is decoded by LIWC with the following designations or hierarchical categories of the algorithm: “affect, tone_pos, emotion, emo_neg, emo_sad, verbs, focuspast, communication, linguistic, and cognition.” (nodes of LIWC algorithm). As the proper program defines, this software is a dictionary computerized for classification of language in specific categories: psychological, content and grammar. In the psychological dimension we found subcategories like emotional tone. The main advantage of LIWC is to help to predict behaviours through the language used by individuals. Several concepts can be explored with LIWC and they involve psychological and personality traits very important to prevent dynamics among demographics in specific contexts. Its algorithm uses a sequence to sequence model to encode words.

For this study, we focused the following LIWC functions: analytical thinking, the clout and the emotional tone. The data extracted refers to the period between February and May 2020 and is all in English to allow for psycholinguistic comparison. The linguistic samples came from leaders in the following countries: Germany (period 1: March 19 2020; period 2: April 15 2020), China (period 1: February 3 2020; period 2: March 26 2020); Spain (period 1: March 10 2020; period 2: April 18 2020); France (period 1: March 16 2020; period 2: April 14 2020); Italy (period 1: March 11 2020; period 2: April 21 2020); UK (period 1: March 16 2020; period 2: April 12 2020); Russia (period 1: March 25 2020; period 2: April 20 2020); USA (period 1: March 13 2020; period 2: April 19 2020).

In terms of conceptual modelling for the data analysis and interpretation, the authors used qualitative thematic analysis to identify the topic modelling and sentiment analysis. This qualitative approach is defined toward the spectrum of emotions informed directly by wording (words and sentences in discourses), Specific emotions and attitudinal tendencies would vary according to the case (leaders speech, the context of the country and pandemic situation, the culture of the nation, the language of emotions as allowed in each case – from West to East the cultural factor is determining the speech and the emotion regulation out of the inner speech because we are examining the extern speech)4,7,9. The discourses examined had the assumption of divergent segments (identified in words and sentences) appearing for the same emotions addressed by the speakers. The Meaning Extraction Method from the quantitative insight displayed by LIWC, in this report, helped for the sentiment analysis in the qualitative perspective10,11. This triangulation of qualitative and quantitative approaches assured the interpretation of large data (texts from the discourses) according to specific core of emotions as they appear along the two periods in analysis. Negative emotions appeared more evident in acute phases of the crisis and moderated by the cultural characteristics of the leaders9,10,11. On the other hand, a quantitative method supports the analysis by considering that LIWC is based on the computations of scores previously determined for each psycholinguistic dimension. The LIWC algorithm provides specific cut-off points for variables such as emotional tone (cut-off: 0.50), and also the word counting and distribution of words and topics according to language functions and psychological dimensions (Fig. 1).

Fig. 1
figure 1

Top words in the eight discourses: periods 1 and 2 of the state of emergency.

Full size image

Topic Modelling is not here addressed as a computational tool, but a qualitative technique to complete the sentiment analysis. Once 16 long discourses and respective topics, words and segments were settled and organized by the LIWC, the topic modelling add value for a specific and detailed analysis by interpreting which are the main topics and, subsequently, the topwords. Those are the ‘guidance’ in the text to understand how changes occurred among discourses in the different settings but addressing the same periods, respectively. Here, topic modelling followed intentionally with no algorithm or online tool needed. In sum, we have both qualitative and quantitative methods, being LIWC the interested part for the quantitative examination that assure tasks such as to compute, select and score massive information gathered from the online settings. Thus, the topic modelling was grounded in the main theme — management of the COVID-19 crisis in the perspective of country leaders; the sentiment analysis discriminated the emotional tone (supported by the LIWC scores), the analytical thinking and the clout, the principal dimensions or functions that we select from the LIWC computation. With the topic modelling, latent concepts and meanings were achieved during the analysis and top words were identified. By top words, we mean the words that identify the principal theme of the discourse and the most frequent segments. The sentiment analysis, as part of our conceptual modelling, identified dominant words and sentences for feelings of anger, fear and sadness. These are related to the three psycholinguistic levels analyzed (emotional tone, analytical thinking and clout). Emotional tone refers to the positive or negative emotions captured across the type of words (mainly adjectives, intonation and punctuation); analytical thinking refers to formal discourse with psychological distancing between the speakers and their audience (using the 3rd person frequently; complex words; narrative manner; abstract concepts; direct orders with less empathy). The clout dimension refers to the level of empathy and engagement (words and sentences organized to generate proximity with the public to achieve personal specific goals) that populations may feel in different level considering a combined social status and leadership type of the leader. The combination is important because the confidence of leaders are not necessarily correlated to positive empathy.

Results

Empathy: how can discourse generate engagement for prevention?

With regard to the corpus of our analysis, the top words identified were, for example, “COVID-19”, “disease”, “death”, but also sentences and expressions such as stay home”, “use masks”, “this is about people”, “every individual can do to help” and “transparent communication”. This linguistic structure enhanced the empathy sought by the leaders and decreased analytical thinking. There was a positive correlation between emotional tone and clout (engagement with the public), and conversely a negative correlation between emotional tone and analytical thinking. The analytic discourse used the following words and expressions: “I” and “me”, “the state”, “the health professionals”, “social”. Figure 1 presents the distribution of different words (the main segments observed in the speeches – attending to the frequency and to sentiment analysis representation) in the eight speeches delivered in two periods of emergency (during the pandemic). Figure 1 presents a chart illustrating the distribution of the linguistic data extracted from each speech. Words appear in the x axis, and score (frequency of words said) in the y axis. Colours distinguish the 8 countries across the two periods. The words refer to the frequency of each segment in the discourse of leaders considering the two phases of pandemic crisis. Despite of the colours’ similarity, our intention with the graphic is to focus on the discrepancy (and low number of words in some of the cases) between countries and periods when using the same words.

In the Fig. 1, we made it possible to explore the sentiment analysis in more detail by using the relationship between words x country x period.

Are we just counting words? Discourse and LIWC

The scoring to obtain emotional tone indices, clout and analytical thinking evidence was based on the imputation method of the LIWC program. That scoring is based in talkativeness and verbal fluency across all excerpts of speeches from the 2 periods. By selecting the word list (by segments: for example, “crying, grief, sad”), in the chart we can access the scores and the respective speaker who produced the discourse concerning COVID-19 safety measures and scenario reasoning (sharing of thoughts with audience) in his/her respective country. Therefore, the negative words – “hate”, “kill”, “annoyed”, “grief”, “hurt”, “ugly” among similar other words – were more prominent in the discourses of period 2 (worst wave of the pandemic when the number of deaths and proportion of the population infected were at a peak) for the USA and the United Kingdom (UK). A lexicon of a utilitarian nature was also observed more for the USA in period 1 and period 2 (“eat”, “carry”). The utilitarian use of words identifies more analytical thinking than clout. These words – negative emotions and analytical typology – had a lower rate of occurrence than in the other six countries. However, the UK demonstrated a simultaneously high frequency of positive words relating to emotional tone (“love”, “sweet”), which contrasted with the low clout behavior (frequently using pronouns “you” and “your” more than other countries, when the scores are compared).

Excerpts from the full public discourses from leaders, at the time, can illustrate part of these contrasts. For example, in Spain, the second period presents more clout dimension:

those who have lost a loved one, often without being able to offer confort or say goodbye. To them, I convey a message of deep sorrow on behalf of the entire Spanish society and the Government of Spain. And to all those people we have lost, as soon as possible, we will pay the tribute they all deserve (…) to our healthcare professionals (…) our compatriots who have given an exemplary response, I cannot say it enough (…) we have saved (…) we have managed to contain the virus from spreading (…). [sic]

On the contrary we found less clout dimension in the same period, for Germany:

This will have to happen in smaller groups, there must be a school bus concept, there must be a concept for breaks, so it will require a high logistic effort to carry out and therefore it needs intensive preparation. i do know how much doing without this still means for parents, but I think it is simply necessary when we say that we have to live with this virus in a pandemic. Hairdressers companies must also design hygiene concepts will then be able to resume their work from 04 May, provided the protective measures are sufficient. [sic]

LIWC algorithm calculate the clout through the word function and frequency referring to empathy from leader to the populations. The emotions aside, clout differs from negative or positive state of mind, because the leadership is being calculated in the basis os relationship skills and psychological proximity between who leads and who is leaded. [sic]

On other different case, the speech in Italy in period 1 was filled with hope (positive emotional tone and clout) by using a high frequency of words like “happy” and “free” when referring to the lockdown measures and the immediate future. This changed drastically in period 2, which considers the wave in April for the Italian context.

An excerpt from the discourse emitted in period 1, in Italy regarding pandemic prevention showed hope and positive resistance,

And I have a deep conviction - and I’d like to share it with you - tomorrow not only will they look at us again,, they will admire us, they will take us as a positive example of a country that, thanks to its sense of community, has managed to win its battle against this pandemic. We are, I remember, country that was first in Europe be hit hardest by coronavirus, but we are also the ones that are reacting with the greatest force and with the utmost precaution, becoming day after day a model for everyone else too. [sic]

Here the discourse also changed mainly in terms of emotional tone – negative, below 0.50 score according to the LIWC criteria. The words “a”, “an”, “the” (more utilitarian than those mentioned above) were very strongly represented in the discourses from Russia, the USA, Germany and other countries in a larger sample and mostly in period 2. Conversely, if we examine the analysis of the clout dimension in the Russian context, the use of pronouns like “we” was more prominent, thus outlining less analytical thinking and enhancing clout (proximity to audience).

Emotional tone and the acute phase of pandemic: differences among leaders in speech

Still on the topic of sentiment analysis, we identified proximity sentiment (less analytical thinking and high emotional tone) in the frequent use of 1st person pronouns in several cases (several speech samples), specifically in the second state of emergency. Emotions run high when use of words such as “love”, “happy” is increased (1st period), as opposed to “hurt” and “ugly” (2nd opposition). Through LIWC it was possible to distinguish positive and negative emotional tone because the algorithm is prepared to recognize words, in English, that are informing negative or positive emotions (anger, sadness, happiness, optimism). Below 0.50 negative emotions are identified. The sentiments became more negative in period 2 considering the less controlled COVID-19 scenario and the respective measures. Therefore, the words and expressions used focused on order and the obligation to follow the measures and be aware of the threat. In period 2, Germany and Italy revealed more austere and narrative discourses reflecting analytical thinking, and thus less proximity between government and audience. The clout dimension was parallel to the evolution and changes in the analytical thinking, reflecting a negative correlation whereby more clout meant less analytical thinking. The clout dimension only ‘arrived’ in speech in the China context in period 2 (when the crisis in China was more controlled compared to the other countries analysed).

Regarding the quantitative approach of LIWC, it was possible also to understand the emotions’ awareness that different leaders presented during their speeches. LIWC provides a score that inform the BIG FIVE traits, in terms of personality evaluation, considering the word counting. For these 8 leaders and respective two speeches in both periods, we had concluded differences that are a new contribute in this field. UK and Germany showed the higher values regarding the emotions (proper emotions and others’) awareness and how to express them. The percentages were extremely high (in LIWC this means that they are above 0.75) compared to all other countries. In this analysis, scores should be higher than 0.50 to inform positive awareness of emotions and how to communicate them. Any score above the mean of 0.5 indicates a greater than average tendency for a characteristic. On the other hand, in the properties related to clout dimension, the consciousness about the respect toward rules and safety differs greatly between China leadership and other nations, China revealed, at the time, the lowest score which means, in the context of analysis, that China leadership, despite of higher levels for the value attributed for facts and not hypotheses during pandemic crisis, prefer to follow and individual caution model (differing from world rules for pandemic situation). We denoted also that China leadership was also correlating negatively (lower levels, below 0.25) for emotions’ awareness and preoccupation with the expression of them toward people. In sum, agreeableness (one of the big five personality traits) is not highly scored for China speeches during the pandemic context. Parallelly, China confirmed, through the speeches evidence, that they prefer to maintain traditional rules and not stability as priority.

Discussion

In this commentary paper, we can inform that less controlled contexts in relation to the acute phase of the COVID-19 pandemic (mostly identified in the first semester of 2020 with two waves of lockdown and restraining measures) showed more analytical thinking in the leaders’ discourses (more formal discourse, more narratives, less proximity and careful selection of words, especially descriptive words) attending the health and social crisis lived between 2020 and 2022. The changes observed from less to high analytical dimension corroborates previous and recent literature in the field1,2.

In terms of the psychological functions assessed by the LIWC, we can comment that the sentiment analysis for this line of thought — negative emotional tone (consequent negative emotional reasoning and feelings engaging the audience) and low level of clout (lower engagement with public) — were more evident when the formal discourses appeared, essentially in period 2 (second trimester of 2020). Considering a biopsychosocial approach, populations – including the leaders here referred – changed completely (and in a rapid way) their habits, interactions, sleep routines, language forms and sentiments. A rapid change is not positive considering the non-controlled stress. We need to discuss our results using a humanistic perspective1,7,8,16. In terms of implications, of course we are aware that stress and depression feelings increased correlating with huge analytical thinking of leaders and mainly negative emotional tone in leaders and also among populations.

We notice the top words were frequently associated with disease, death, danger, grief; but also, with happiness and hope/love (those were less frequently observed in the repertoire). The sentiment analysis, after the computation provided by the LIWC, attested to variations and changes in the linguistic forms used in the discourses at different moments in time. This was supported by previous mentioned literature1,2. Precisely, those variations detected an emotional deterioration in the discourses from period 1 to period 2. This mainly related to the lower level of emotions and fragile engagement with others (as well as empathy and resilience). Italy and Spain were the most evident examples of a rapid change of speech (from period 1 to period 2) with wording pointing to a negative state of mind, pessimism and lack of control regarding the information about the severity of the country’s situation. Positive wording decreased in these discourses and countries and had a direct impact on the emotions and empathy of the audience. Further investigation is needed to determine if that also generated major levels of stress and anxiety during the lockdowns and also impacted remote working and school performance9,15,16.

Negative wording and low clout level (engagement and empathy scores in diminished way) suggest, in post-pandemic life, experience of cognition risk and difficulty for face recognition (being part of the language expression) in a short-term20. The COVID context and the confinement long periods limited the continued lifespan learning (in several areas of behaviour), despite of age. One of the simple learning in a daily basis is the facial expressions decoding and the adequation to new pragmatics of language16,20,21. If individuals were confined and with low (or none) level of social and physical interaction for extensive periods (adding the mask use during the short periods of social events) they will be less able to decode new face expressions for new language meaning and messaging22,23,24,25. The face recognition difficulty refers, in this context, to the high odds of disruptive decoding behaviours of individuals toward the face expressions of other speakers and their messages24. Those messages can be distorted when individuals (receptor) have no clear understanding of symbolic language attached to the message20,25. And that distortion, with origin in absence of knowledge blocked by the low level of social interaction, generates more probability of negative decoding, negative language and comprehension, as well low empathy23. Sociocultural and social-cognition theories are emergent in published literature recently exploring the effects of pandemic isolation in early development – in the childhood -, in pubescents period, during emergent adulthood and in mature adults16,21. Sociocultural factors are the only moderators for those negative indicators after concluding poor sentiment analysis in our investigation. One good example was observed in China speech (mainly for the first period) because we found less investment in innovative strategies and low tolerance to challenges in authority. Emotional awareness is low for China when compared to other countries, but this is cultural explained and clearly understood regarding the Chinese system in the political perspective. Sociocultural factors will determine, currently, how individuals are oriented in the new era for more or less engagement behaviours with direct implications in democratic way of living.

Other significant implication to consider is the impact for school aged children and their language development that would be constrained by the self-isolation need, but also by the type of words that they were exposed to. Therefore, word usage is likely limited in this population of children and the general lexicon and semantics that should be achieved until 11 years old. This was already estimated and reflected by previous scholars3,7,9,10. We notice again that ‘democratic’ sense here refers to freedom values related to language expression abilities and attitudes toward communities. And communities should be differentiated according to sociodemographic factors such as the case of gender: nations’ leaders or heads of state speak equally to several genders still ignoring the specificities that target genders need to understand the message17,18,25. Only cross-cultural studies will help to understand how the clout and analytical thinking are present in our time. Survival depends on democratic discourses even when analytical language is episodically present. Lifetime experiences such was COVID-19 with several type of loss associated should be carefully considered by scientific investigation, starting by the analysis of language and languages23,24,25.

Other implication that we should highlight is the suitability of these data and analysis for the future considering global crisis. But also, currently we are living crisis that are not global but they affect in a global manner all societies, such as the case of Russo-Ukrainian conflict that came during pandemic crisis (2022, February) and is still causing all type of damages, mainly for mental health and trauma in displaced people19. With great risk for the hosting countries and the refugees’ stability19. From our study we can prevent for more balanced messaging when delivering public discourse with social and health prescription. That balanced messaging can mitigate less fluctuation of emotional tone, positive clout and generating better levels of empathy23. Moreover, if future crisis may involve lockdown which impacts mainly the empathy and the social interaction.

Conclusion

In conclusion, based on our examination of only four months, verbal behaviour clearly changed as the pandemic evolved in 2020 with repercussion in actuality. Considering the impact of the language fluctuations in the government leaders’ discourses, for example, the way that wording and top words influenced negative mental health and negative sentiments in the period from 2020 to 2022, but also after the pandemic period. The evidence showed differences between country leaders and their speeches, specifically in relation to the usage of terms (top words) and the psychological functions determined (sentiment analysis). The negative emotional tone (below 0.50) and fragile clout gained expression with the advance of COVID-19. The USA context presented sui generis speech in both periods: less psychological distance was perceived in period 2 (speech 2). During period 2, the pandemic scenario was still very problematic for the US; while in case of China (period 2), the proximity to audience increased during the speech (according to the analysis of the words used by the leader). This seems to reflect the context of the low rate of infection compared to the other countries at the time. Topic modelling and sentiment analysis were used as a qualitative method, based on the LIWC algorithm, to understand the psycholinguistic profile (and effect on the audience) of each speaker who managed discourses referring to the acute phases of lockdown and the health emergency situations during two different periods in 2020. Insights into personality traits are also gathered from this type of discourse analysis in different life circumstances. Further reviews and research should focus on the current language and deficit wording of this population (leaders). We also note that psychologists and schoolteachers can play an important role in supporting language programmes with positive wording and by emphasising the collateral effects of face-to-face classes when teaching children to read and write. Finally, it is important to notice that cultural factors and the language/idiom identity appeared to have no effect on the wording observed in every speech examined.

Limitations

This study as being a commentary report with a qualitative component of analysis (besides the quantitative analysis through the calculation provided by LIWC) may generate bias during the interpretation of some data. The interpretation, despite the evidence based, was followed by the authors’ perspective. We used the Sentiment Analysis in the qualitative perspective, further studies may replicate our study with that method in quantitative model (natural language processing context). The number of words for the corpora used in this study may be a limitation also to represent the global fluctuation of emotions and language changes between periods and international leaders. To add, further studies would be interested in the replication of our rationale but specifying for ethnic and migrant minorities and how they handle the health and social prescriptions when they have origin in government leaders’ discourses.