Knowledge From the Afrikan World, For the World

PMC Articles

Trends in the texts of national anthems: A comparative study

PMCID: PMC10458337

PMID:

Abstract

In a recent previous investigation of national anthems, preferred topics and their bias (e.g., towards identity, fighting, or well-being) were identified subjectively (Silaghi-Dumitrescu, 2020). The present report aims to verify whether a more objective, automated, comparison of the texts of national anthems across the world can also reveal systematic trends – and to what extent. To this end, the Tropes and Semantria software packages are employed, revealing preferred topics (e.g., state, feeling, body, time, land, religion, family, fight), how their relative weights differ across continents and cultures, and how the conveyed sentiments vary. For instance, “liberty” is more common in Latin-country anthems while almost absent in Asia, “feelings” are less mentioned in Germanic-language anthems, and the first-person singular “I” is essentially absent African anthems. The sentiment scores of the anthems vary from neutral in Latin and Mediterranean anthems to much more positive in Central and Western Asian, Germanic and Slavic countries.

Full Text

Music lends itself to cross-cultural comparisons, having been shown to either signal social/group/race/national identity, or reinforce it, or contribute to its modulation/refinement, or act to modulate intergroup communication (either helping in conflict resolution or in its exacerbation) – ultimately with the potential of a social engineering tool both towards establishing identities and towards negotiating between groups of different/competing identities. Applying the framework of optimal distinctiveness theory (ODT), Abrams points out, based on studies on two groups of ∼50 and ∼2600 young individuals/students, that musical preferences are geared towards neither very individual nor very general choices, but rather intermediate ones, as expected based on ODT – where a balance between identity definition (at the limit, an individualizing trait) and inclusion (at the limit, a collective-related trait) is sought [1,2]. Musical preference-based favoritism within social groups was demonstrated, favoring stereotypies identified by group members as group identity-related [3]. In a study with ∼140 African American college students, Dixon et al. found that preference for rap music was correlated with increased self-esteem. Also, Afrocentrist attitudes were correlated with preferences for music featuring strong Afrocentric characters. Within the same group, preferences for music with misogynistic components was found to correlate with the belief that the respective music was not degrading towards women [4]. Related to this, Reyna et al. found that, in other social groups, negative attitudes toward rap music correlate with negative stereotypes of African Americans (e.g., laziness) [5]. In a study on 80 British young individuals, Rentfrow et al. showed that music preferences do form the basis of judgements/perceptions about psychological and social characteristics of individuals and groups [6]. In another example of identity-related type of music, i.e. the Gypsy-associated Flamenco music in Spain, respondents’ attitudes towards the Gypsy community were more positive if the Flamenco was invoked as a positive defining feature of that community; however, a similar music-identity manipulation could not be performed in relation to another minority – the North Africans [7]. Bensimon, in a study based on interviews with a small number of protesters and with their opposing security forces, reported that protesters were able to elicit empathy from security forces when resorting to slow/quiet/sad songs, but that the opposite effect was obtained with songs flaunting identity (be it national or religious, even though this identity would have been common between the protesters and the security forces) [8].
It has been argued that the lyrics of national anthems would also (if not more so) be suitable for exploring identity/group issues [9]. Cerulo has also showed that the musical tune structure of the national anthems is directly affected by social power relationships within the society at the time when the anthem is adopted [10]. Examinations of the lyrics of national anthems have to date included interpretations/commentaries on their nationalism, family, sexism, suicide rates and others – with cross-cultural implications in several cases. Gloominess, nationalism, gender-bias or other features were described in such analyses as features of some of the national anthems. The perceived relevance of such data has led to proposals going as far as replacing national anthems with more positively-oriented versions so as to reduce the rate of suicide in the respective nation [11]. In comparing 6 national anthems, Vörös et al. find that the anthems of countries with lower suicide rates tend to contain relatively more positive contents, emotions and intentions, while in the anthems of countries with higher suicide rates more ambivalence, denial, loss or even aggressive and self-destructive implications were found [12]. A critical stance towards national anthems per se is also documented (e.g., “glorified triviality”, “ridiculous pathos”) [13]. The need for contextualization and avoidance of absolute engagements especially in educational settings was advocated in case studies of the United States of America (USA) and Great Britain anthems. One may cite in this context for the USA controversies regarding the mention of slavery, or the fact that while the verses were composed to commemorate a victory against British military forces, the musical tune itself was British [[14], [15], [16]]. Winstone et al. pointed out a notable age dependence of the identity-building/modelling effect of the national anthem in a sample of children aged 8–10 in Great Britain [17]. A need for balance between the unifying role of a national anthem and the inherent conflicts engendered by dogmatic treatment/implementation thereof was also pointed out [15,18]. In a study on ∼100 secondary students from choirs in Canada and USA, Gilboa and Bodner found that Canadian students were significantly more proficient in performing the national anthem – suggesting that the social impact of the anthem is also significantly dependent on a range of factors, of which education is a key part [19]. In a study of citizens from various backgrounds within the same country, it was found that the national anthem evoked more national associations than any other songs considered. Interestingly, the degree of this trend did differ when marginal groups of the society were considered [20]. Kyridis et al. provided a comparison of 18 national hymns (Russia, Argentina, Austria, Mexico, Canada, Greece, Cuba, Bulgaria, Germany, F.Y.R.O.M., U.K., Serbia, Romania, Albania, Australia, U.S.A., France and Belgium) with emphasis on nationalistic, propagandistic and occasional chauvinistic features and on the degree of expressing identity/belonging [21]. Lauenstein et al. pointed out a focus of anthems on family as a background for promoting, hierarchical structure, social roles and responsibilities (with gender implications), and positive affective connotations [22]. Liao et al. point out, using the Chinese national anthem as a case study, an active role for national anthems in creating a collective memory – as well as the complex interplay with cognitive and social contexts [23]. A similar analysis was provided by Siska for the Turkish national anthem [24], or by Pavković for the countries originating the former Yugoslavia [25]. Oluga et al. pointed out inherent linguistic sexism in a series of ∼60 West/North Germanic and Romance/Italic language national anthems – but also pointed out how these traits were occasionally altered (either exaggerated, or attenuated) during translation into other languages [26]. In the same vein, Rodriguez identified notable distortions within the official translation of the Venezuelan national anthem from Spanish to the native language of Warao, with social, cultural and political implications [27].
The above-discussed studies on national anthem lyrics have focused on narrow sets of anthems or topics. It is thus unclear to which extent their conclusions are generally applicable to all anthems, and/or whether the previously-examined topics are the only ones, or even the dominant ones, in national anthems. Therefore, a more systematic analysis, bound to intrinsically uncover trends across nations/cultures [28,29], may be deemed useful. On the other hand, attempts towards cross-cultural analysis on a larger scale would have to be conservative and to consider the very different backgrounds of anthems in terms of their origins. One may consider for instance that some lyrics were written decades or centuries before becoming anthems, while some were written on purpose to this extent. Also, that some were validated by public popularity before the adoption of anthems, some afterwards, and some less so – especially in less democratic countries. Indeed, writing words for anthems is by no means a general institutionalized practice across the world, and naturally so in view of how rare the event of anthem writing is. If any general trends and correlations are to be identified, one would subsequently need to seek links with the poetic/literary field they emerged from and to the historical context thereof. Fluctuations of literary/poetic language in diverse traditions would then be of importance. Anthems could also be seen as an instance of selection in cultural production by political elites at the specific point in time (as already pointed out for the musical tunes thereof [10]). Thus, anthems may well not be expected to describe cross-cultural variation and any features of “nations” (e.g., elites could be global and copy each other or try to innovate to be different from neighbors), or correlate with current sociological data. An attempt to nevertheless perform a comparison of most of the known official national anthems across the world (∼200) was recently reported, focusing on identification of the pervasive topics, and on possible connections between these topics and some basic societal features [30]. A set of recurring themes was identified subjectively by the authors of the study, as follows: ancestry/past, beauty, build/work, country name, courage, democracy, enemy, ethnicity, family, man, woman, fight, flag/colours, forever/never, future, geographical references, glory, independence/freedom, joy/happiness, home/mother/father-land, law/governance, leader, love, loyalty, peace, poverty/wealth, pride, religion, revolution, sacred, sacrifice, salvation, sorrow, treason, tyrant/chains, unity, win/victory. The number of topics present in a given anthem, as well as their type were found to vary significantly between anthems. Groups of anthems were defined based on these tendencies. Correlations, albeit weak, with more general societal features such as age of country, gross domestic product (GDP) per capita, Gini coefficient, size of armed forces, inequality, inequality-adjusted human development index, and a number of parameters from the World Values Survey (WVS) database (related to religion, gender equality, attitude towards other nationalities/races, attitude towards work, attachment to democratic values etc.) were identified. A shortcoming of the study was not only its subjective character, but also its limited quantitative methodology: it did not differentiate between the number of times a given topic was mentioned in a given anthem. Among other things, a detailed evaluation of the sentiments conveyed by the anthem would be very difficult under these conditions (e.g., it would not differentiate between an anthem that mentions “sorrow” once in a total of 10 phrases, and an anthem that mentions the same notion in each of the 10 phrases).
The English versions of the texts (with caveats pointed out e.g. in ([26,31]) were employed. Data concerning the anthems, including the English version of the text, were from Wikipedia.org (https://en.wikipedia.org/wiki/List_of_sovereign_states_by_date_of_formation, https://en.wikipedia.org/wiki/List_of_countries_and_territories_by_population_density, https://en.wikipedia.org/wiki/List_of_national_anthems, https://en.wikipedia.org/wiki/Category:Lists_of_wars_by_country, https://en.wikipedia.org/wiki/List_of_countries_by_inequality-adjusted_HDI), http://www.worldvaluessurvey.org and http://www.national-anthems.org/. The use of English translations puts a limit on the extent of data which can be meaningfully compared between texts originating in different languages. Detailed syntactic comparisons are hence not possible with this approach. Moreover, the small selected set of parameters which are analyzed in the present study will bear two different imprints: the intrinsic imprint of the historical and sociological context, and a linguistic one which may sometimes be difficult to separate from the latter. The types of analyses considered in the present text are therefore conservative and limit themselves to items that are expected to be conservable in translation – e.g., the types of nouns used, or the temporal setting, or the prevalence of verbs, or the types of verbs but not their exact identity, or the general feeling (positive vs. negative).
Anthems were grouped together as described in text (based on geographical regions, or to some extent on the type of original language, as described below) and analyzed using the Tropes software package, [32]. For each group of anthems, data was retrieved in automated manner (without intervention/bias from the user) about (1) the most abundant notions/references and (2) most abundant word categories – all three as defined within the software package without altering the standard settings. An exception from the standard settings was made for the most abundant notions/references, where the software generally identifies country names as belonging to categories named after geographical areas (e.g., Africa, Middle East, Oceania, Asia etc.); these categories are reunited under the common heading “country name” in Table 2. The Tropes software was developed by Pierre Molette and Agnès Landré on the basis of the work of Rodolphe Ghiglione (http://www.semantic-knowledge.com).
Sentiment scores were attributed for each anthem individually (not per group) using the web interface of the Semantria software package [33] without any further intervention or processing from the user, except for adding a space character at the end of each verse/line when pasting the text into the analysis window, in order to ensure word separation across lines. This automated analysis employs natural language processing (NLP) and machine learning techniques in order to identify sentiment-bearing phrases and/or components, and assign each of them a sentiment score ranging from ∼ −1 to 1 for simple expressions. A total score is assigned for the analyzed document by averaging the values for each of the sentiment-related word/expression. For instance in the case of the Romanian anthem, the automatically-recognized sentiment-related terms were “slaves” (score −0.929), “deadly”, “barbaric”, “death”, “tyrants” (scores of −0.600 each), “weapons”, “enemies” (−0.490 each), “renew” (+0.450), “great” (+0.600), “freedom” (+0.680), “holy” (+1.000), “glory” (+1.145), “victor” (+1.503), “pride” (+1.800) – averaging to a total score of 0.205 for the Romanian anthem. Within the framework of the Semantria feeling analysis, texts with total scores above 0.250 are considered to have a positive overall sentiment; texts with total scores between 0.250 and −0.150 are assigned as neutral, while those below −0.150 are categorized as displaying an overall negative sentiment [33].
Table 1 shows the most common categories of words encountered in national anthems. Verbs expressing actions (“factive” in Table 1) are less common in Central and Western Asian, (32%) vs. the world-average of 40%. Verbs expressing states and possession (“stative” in Table 1, 34% world average) occur less often in Africa (27%) but distinctly more in Central and Western Asia and in Slavic anthems (48–55%).
To sum up, Balkan anthems stand out with higher percentages of comparisons, negations, and second-person “you/thou”, as well as lower incidence of first person plural “we”. All of these may be argued to reflect an aspiration/inclination towards actively seeking differentiation against “others” – in line with the Balkan's XIXth century reputation as “the powder keg of Europe” and with the still extant ethnic and religious diversity [34,35]. These considerations may be illustrated by two examples – the Albanian and the Romanian anthems.
African anthems display higher percentages of plural first person pronouns – and lower percentages of first person singular pronouns. Of all the groups of anthems, the African ones may be interpreted to most clearly reflect an aspiration towards group identity and unity, perhaps in a collectivistic sense [36]. The anthems of Zimbabwe and Mali (the former listed below), illustrate this state of things – zero use of “I” but repeated use of “we”:
Latin European anthems display more time modalities – and fewer manner and place modalities. An apparent relative high incidence of “I” seen in Table 1 is in fact an artifact due to the small size of the analyzed set of anthems (the Andorran anthem has an 80% prevalence of “I”, while the other ones have 0% prevalence). An example of Latin European anthem is the Romanian one, listed above. There, one may indeed see repeated use of time modalities/adverbs (“now”, “still”, “once again” – at 50% of all adverbs vs. a 34% world-average), and less manner and place modalities (only one mention – “in full glory”). A second example may be taken that of the Spanish anthem, which does highlight its time modalities (“long live”) – though in this case the place and manner ones are rather on the average than below the world-average:
Overall, the above considerations reveal clear differences in the manners in which various cultures address the anthems. These differences can be further explored if one examines the correlation coefficients between the percentages for each of the categories in Table 1, across the various groups of anthems, as shown in Supporting Information. Considering the small differences seen between most of the parameters in Table 1, these coefficients are all at ∼0.9 and above. The weakest correlations (0.81–0.89) involve Central and Western Asia vs. English-speaking countries, Slavic vs. Latin European, and the Balkans vs. several other groups (Africa, SE Asia, Germanic, Latin European, Western European, English). Most of these differences can be explained by the inherently different grammar/syntax structures of the respective languages.
The most common terms encountered in anthems across the world are shown on Table 2, alongside differences in frequency of usage of the respective terms in the anthems pertaining to the various groups. The nature of these terms on average across the world is expected considering the purpose of the anthems – i.e., the state/nation/country and its contents/definition (land, people, social structures/groups, religion, as well as liberty when asserting one's country in relation to the rest), allegiance/feelings towards it (this includes the engagement to “fight” as well as the term “body” in Table 2 – where the chest, arms, blood are often invoked).
Feelings are the second most common term in anthems across the world. The frequency of this term is smallest in Germanic anthems (by ∼70% compared to the world average) – perhaps in line with general public perceptions about the respective nations as well as in line with numerical data based on more general analyses of literature in a small set of languages [31]. To illustrate these trends one may look at the two Germanic anthems cited above (Germany and Iceland) – where feelings (e.g., love, admiration, pain, horror, as seen in Latin American anthems for instance) are not explicitly mentioned. The ratio between the use of state and of feeling is only slightly larger than 1 in the world average (0.6 vs. 0.5); intuitively, this may be expected since most anthems would logically speak of love/pride for the country. This trend is mirrored in several regional categories, but there are also severe exceptions. A slight reversal is seen in Africa (0.4:0.5), and a major one in English-speaking countries (0.5:1.0); both of these categories are dominated by relatively younger countries, where arguably the state is/was not yet an established a notion as it was in the older or more traditionally established countries. Indeed, a large state:feeling ratio is seen in Germanic anthems (0.9:0.2) and in Central and Western Asia (1.0:0.4).
Correlation coefficients between the groups of anthems for the data shown in Table 2 (cf. Supporting Information Table S2) vary widely, from 1 to −0.2 – significantly more so than for the types of words discussed in relation to Table 1, and expectedly so given the wider differences between percentages in Table 2 (again, expectedly, since these are more specific/detailed categories than those in Table 1).
Reasonable correlations with the world average are expected for all of the regions/categories of anthems; this is indeed the case especially for Asian and Mediterranean anthems (r = 0.9 cf. Table S2), and reasonably so for Europe and Africa (r = 0.7). At the opposite end, English-language anthems only show an r = 0.3 vs. the world average, followed closely by Latin Europe at 0.4. Moreover, the English-language anthems show negligible correlations towards any of the other groups of anthems in Table S2.
Latin American anthems also show their weakest correlations against Latin European and Western European anthems (0.0–0.2, i.e. essentially absent), despite the historical and cultural connections between the two groups of nations. Against the background of a 0.95 correlation coefficient based on the structural grammar data in Table 1, these numbers suggest that a common language or even common history can still leave significant room for cultural differences that may outgrow the differences towards cultures of other languages.
In a previous analysis [30], a list of topics was identified subjectively/manually in the same set of national anthems as analyzed here. The topics were ranked based on the number of anthems where they occurred, but the number of occurrences in each anthem was not counted. By contrast in the present study the topics were identified in automated manner and the number of occurrences within the text was also counted. With these methodological differences in mind, one may nevertheless compare the topics ranked in Table 2 with those from the previous study, noting the general similarities but also some differences. The previous study listed (in this order) the following as most common topics: land, future, country name, independence/freedom, pride, loyalty, geography, religion, unity, forever/never, ancestors/past, family, love. Indeed, “land” is also found among the most common topic in the present quantitative analysis cf. Table 2; “geography” does not appear as individual term in Table 2 but may be viewed as a more general proxy of “land”. “Future”, “forever/never” and “past” are not found by themselves in Table 2, but are arguably still encompassed by the 4th most common term there – “time”. “Country name” is identifiable in Table 2 as the top notion – “state”. “Independence/freedom” is identified in Table 2 as “liberty”. “Pride” and “love” are not present per se in Table 2, but may be taken as part of the second most common notion there, “feelings”. “Religion” and “family” are present in exactly the same form in Table 2. “Loyalty” and “unity” are notably absent from Table 2, unless they are deemed to be convoluted into other terms (e.g., “feelings”). There is thus an overall agreement between the results of the two methodologies. The case of the two terms, “loyalty” and “unity”, which appear to be among the top 10 notions in anthems according to the previous study [30] in terms of a binary count (are present vs. are not present), but are not identified among the top 10 notions in the present statistical study where the number of occurrences within each text is also counted (i.e., non-binary), may be instructive to note. On one hand, may interpret this difference as a manifestation of the subjective methodology in the previous study. However, one may also use this example to raise the question of the intrinsic relative weights of the words – and point out that by no means these are (or should be taken as) uniform, neither within the same culture and especially nor across cultures.
Table 3 summarizes the salient features identified for each class of anthems in comparison with the others.
While this body of data may offer grounds for further socio-cultural comparisons, with theoretical frameworks ranging from philosophy to exact sciences [[36], [37], [38]], further discussion about the cultural relevance of the anthem analyses must take into account the at-most-indirect representability of these texts for the respective nations. Though most of them were validated at least tacitly in time by the people of the respective countries, the anthems were generally adopted by a very small group of people (“elites”), often entirely unrepresentative, from a statistical point of view, of the social/demographic fabric of the country – and in many cases not even elected democratically to do so. One must thus consider in each case whether the texts chosen by these elites were describing the fabric of their nations, or the future goals thereof, or if in fact there was a hiatus between the respective elites and nations on these topics. The latter may be particularly relevant for anthems in countries that have more recently gained their independence (and hence where popular confirmation of the anthem over time is not a factor to be considered yet). Last but not least, in interpreting statistical data on the presence of types of words and notions in anthems belonging to various cultures, one must also point out that the intrinsic weights of various words/notions are not inherently uniform within the same culture and much less so across cultures. If one mentions “unity” and “time” with a similar frequency in a text, it should not follow that the importance received by the two terms in the mind of the citizen/speaker is equal – nor that their relative weights would be identical regardless of language, culture, or historical context. Because of this, the discussion of the present data is for the moment left at a rather basic level, since follow-up studies would need to entail a much more interdisciplinary approach – e.g., psychological, historical, social beyond the obvious linguistic and literary components.
It may be noted that the averages listed in Fig. 1 have relatively large standard deviations (generally around 0.3, cf. Supporting Information Fig. S1 and Table 2). In Africa, the most notable outliers are Senegal and Central African Republic (which display negative values in a group where the average is distinctly positive), as well as Ghana and Niger (standing at twice the average positive score of the group). In Latin America, Peru is the sole notable exception, some 0.6 units more negative than any of the other anthems. In Asia, three countries -China, Japan and Vietnam – stand out with distinctly lower scores than the rest of their group. Central and Western Asia is on the other hand the most homogeneous group. By contrast, the Balkans offer one of the most heterogeneous set of values – spread across a full 1.5 units, twice as large compared to Central and Western Asia (with Bosnia & Herzegovina at the higher end at ∼ 1, and Albania at the lower one at ∼ -0.3). Among the Germanic anthems, Germany itself is and exception, with a score almost twice as positive compared to the average of its congeners from the same group (and in fact the third largest in the world, surpassed by only Georgia and Bosnia & Herzegovina). Among the Slavic anthems, the two southernmost Balkan nations of Montenegro and Northern Macedonia are by far outliers, with lower values compared to the rest. Western Europe is slightly more heterogeneous than the Balkans, with Ireland providing the most negative value against Germany's most positive. The Mediterranean region, more than others, appears split across two main relatively homogeneous subgroups: one with distinct positive values (e.g., Croatia, Egypt, Israel, Jordan, Morocco, Spain) and one with neutral and negative values (Albania, Algeria, France, Greece, Italy, Northern Macedonia, Montenegro).
The herein reported general positive-sentiment trend in national anthems is in line with the previous observation that human (written) language has a positive bias [31]. In the same previous study, it was also noted that among the ten languages studied, Spanish and Portuguese appear to have a higher positive bias than others – with Chinese, Korean and Russian at the opposite end. This is in contrast with the opposite finding in Fig. 1 for Latin-American (Spanish and Portuguese-speaking nations) and Russian, but in reasonable agreement for SE-Asian group. The case of English language is also illustrative: Dodds and co-workers find English books to be at the higher end of the positive bias scale defined by all languages studied, while English music lyrics are at the opposite end. Indeed, the anthem data of Fig. 1 correlate with this latter finding – and this may also offer at least a partial explanation for the Latin-American and Russian cases [31].
When examining the correlations between the averages of the respective values per group of anthems, rather than between all individual values, higher correlation coefficients are found. The correlation coefficients between averages of sentiment scores and averages of WVS data (counted as percentages of citizens who choose a given response in WVS questionnaires) were performed for countries where such data were available. Negative correlations were found with the percentage of people declaring they are very happy or rather happy (−0.8), or for whom leisure is very important (−0.7), or for whom democracy is very good or fairly good (−0.6) – and positive correlations with the percentage of people who agree that being a housewife is just as fulfilling as working for pay (0.7), or who agree or agree strongly that men are better leaders than women (0.8), or who would prefer not to have people of other races as neighbors (0.6), or who think that having a strong leader who does not have to bother with parliament and elections is very good (0.7). These data paint a picture of positively-oriented anthems as pertaining to more conservative countries – which indeed is in line with the fact that Central/Western Asian and Eastern-European anthems are at the higher end of the positive feeling score in Fig. 1.

Sections

"[{\"pmc\": \"PMC10458337\", \"pmid\": \"\", \"reference_ids\": [\"bib1\", \"bib2\", \"bib3\", \"bib4\", \"bib5\", \"bib6\", \"bib7\", \"bib8\"], \"section\": \"Introduction\", \"text\": \"Music lends itself to cross-cultural comparisons, having been shown to either signal social/group/race/national identity, or reinforce it, or contribute to its modulation/refinement, or act to modulate intergroup communication (either helping in conflict resolution or in its exacerbation) \\u2013 ultimately with the potential of a social engineering tool both towards establishing identities and towards negotiating between groups of different/competing identities. Applying the framework of optimal distinctiveness theory (ODT), Abrams points out, based on studies on two groups of \\u223c50 and \\u223c2600 young individuals/students, that musical preferences are geared towards neither very individual nor very general choices, but rather intermediate ones, as expected based on ODT \\u2013 where a balance between identity definition (at the limit, an individualizing trait) and inclusion (at the limit, a collective-related trait) is sought [1,2]. Musical preference-based favoritism within social groups was demonstrated, favoring stereotypies identified by group members as group identity-related [3]. In a study with \\u223c140 African American college students, Dixon et al. found that preference for rap music was correlated with increased self-esteem. Also, Afrocentrist attitudes were correlated with preferences for music featuring strong Afrocentric characters. Within the same group, preferences for music with misogynistic components was found to correlate with the belief that the respective music was not degrading towards women [4]. Related to this, Reyna et al. found that, in other social groups, negative attitudes toward rap music correlate with negative stereotypes of African Americans (e.g., laziness) [5]. In a study on 80 British young individuals, Rentfrow et al. showed that music preferences do form the basis of judgements/perceptions about psychological and social characteristics of individuals and groups [6]. In another example of identity-related type of music, i.e. the Gypsy-associated Flamenco music in Spain, respondents\\u2019 attitudes towards the Gypsy community were more positive if the Flamenco was invoked as a positive defining feature of that community; however, a similar music-identity manipulation could not be performed in relation to another minority \\u2013 the North Africans [7]. Bensimon, in a study based on interviews with a small number of protesters and with their opposing security forces, reported that protesters were able to elicit empathy from security forces when resorting to slow/quiet/sad songs, but that the opposite effect was obtained with songs flaunting identity (be it national or religious, even though this identity would have been common between the protesters and the security forces) [8].\"}, {\"pmc\": \"PMC10458337\", \"pmid\": \"\", \"reference_ids\": [\"bib9\", \"bib10\", \"bib11\", \"bib12\", \"bib13\", \"bib14\", \"bib15\", \"bib16\", \"bib17\", \"bib15\", \"bib18\", \"bib19\", \"bib20\", \"bib21\", \"bib22\", \"bib23\", \"bib24\", \"bib25\", \"bib26\", \"bib27\"], \"section\": \"Introduction\", \"text\": \"It has been argued that the lyrics of national anthems would also (if not more so) be suitable for exploring identity/group issues [9]. Cerulo has also showed that the musical tune structure of the national anthems is directly affected by social power relationships within the society at the time when the anthem is adopted [10]. Examinations of the lyrics of national anthems have to date included interpretations/commentaries on their nationalism, family, sexism, suicide rates and others \\u2013 with cross-cultural implications in several cases. Gloominess, nationalism, gender-bias or other features were described in such analyses as features of some of the national anthems. The perceived relevance of such data has led to proposals going as far as replacing national anthems with more positively-oriented versions so as to reduce the rate of suicide in the respective nation [11]. In comparing 6 national anthems, V\\u00f6r\\u00f6s et al. find that the anthems of countries with lower suicide rates tend to contain relatively more positive contents, emotions and intentions, while in the anthems of countries with higher suicide rates more ambivalence, denial, loss or even aggressive and self-destructive implications were found [12]. A critical stance towards national anthems per se is also documented (e.g., \\u201cglorified triviality\\u201d, \\u201cridiculous pathos\\u201d) [13]. The need for contextualization and avoidance of absolute engagements especially in educational settings was advocated in case studies of the United States of America (USA) and Great Britain anthems. One may cite in this context for the USA controversies regarding the mention of slavery, or the fact that while the verses were composed to commemorate a victory against British military forces, the musical tune itself was British [[14], [15], [16]]. Winstone et al. pointed out a notable age dependence of the identity-building/modelling effect of the national anthem in a sample of children aged 8\\u201310 in Great Britain [17]. A need for balance between the unifying role of a national anthem and the inherent conflicts engendered by dogmatic treatment/implementation thereof was also pointed out [15,18]. In a study on \\u223c100 secondary students from choirs in Canada and USA, Gilboa and Bodner found that Canadian students were significantly more proficient in performing the national anthem \\u2013 suggesting that the social impact of the anthem is also significantly dependent on a range of factors, of which education is a key part [19]. In a study of citizens from various backgrounds within the same country, it was found that the national anthem evoked more national associations than any other songs considered. Interestingly, the degree of this trend did differ when marginal groups of the society were considered [20]. Kyridis et al. provided a comparison of 18 national hymns (Russia, Argentina, Austria, Mexico, Canada, Greece, Cuba, Bulgaria, Germany, F.Y.R.O.M., U.K., Serbia, Romania, Albania, Australia, U.S.A., France and Belgium) with emphasis on nationalistic, propagandistic and occasional chauvinistic features and on the degree of expressing identity/belonging [21]. Lauenstein et al. pointed out a focus of anthems on family as a background for promoting, hierarchical structure, social roles and responsibilities (with gender implications), and positive affective connotations [22]. Liao et al. point out, using the Chinese national anthem as a case study, an active role for national anthems in creating a collective memory \\u2013 as well as the complex interplay with cognitive and social contexts [23]. A similar analysis was provided by Siska for the Turkish national anthem [24], or by Pavkovi\\u0107 for the countries originating the former Yugoslavia [25]. Oluga et al. pointed out inherent linguistic sexism in a series of \\u223c60 West/North Germanic and Romance/Italic language national anthems \\u2013 but also pointed out how these traits were occasionally altered (either exaggerated, or attenuated) during translation into other languages [26]. In the same vein, Rodriguez identified notable distortions within the official translation of the Venezuelan national anthem from Spanish to the native language of Warao, with social, cultural and political implications [27].\"}, {\"pmc\": \"PMC10458337\", \"pmid\": \"\", \"reference_ids\": [\"bib28\", \"bib29\", \"bib10\", \"bib30\"], \"section\": \"Introduction\", \"text\": \"The above-discussed studies on national anthem lyrics have focused on narrow sets of anthems or topics. It is thus unclear to which extent their conclusions are generally applicable to all anthems, and/or whether the previously-examined topics are the only ones, or even the dominant ones, in national anthems. Therefore, a more systematic analysis, bound to intrinsically uncover trends across nations/cultures [28,29], may be deemed useful. On the other hand, attempts towards cross-cultural analysis on a larger scale would have to be conservative and to consider the very different backgrounds of anthems in terms of their origins. One may consider for instance that some lyrics were written decades or centuries before becoming anthems, while some were written on purpose to this extent. Also, that some were validated by public popularity before the adoption of anthems, some afterwards, and some less so \\u2013 especially in less democratic countries. Indeed, writing words for anthems is by no means a general institutionalized practice across the world, and naturally so in view of how rare the event of anthem writing is. If any general trends and correlations are to be identified, one would subsequently need to seek links with the poetic/literary field they emerged from and to the historical context thereof. Fluctuations of literary/poetic language in diverse traditions would then be of importance. Anthems could also be seen as an instance of selection in cultural production by political elites at the specific point in time (as already pointed out for the musical tunes thereof [10]). Thus, anthems may well not be expected to describe cross-cultural variation and any features of \\u201cnations\\u201d (e.g., elites could be global and copy each other or try to innovate to be different from neighbors), or correlate with current sociological data. An attempt to nevertheless perform a comparison of most of the known official national anthems across the world (\\u223c200) was recently reported, focusing on identification of the pervasive topics, and on possible connections between these topics and some basic societal features [30]. A set of recurring themes was identified subjectively by the authors of the study, as follows: ancestry/past, beauty, build/work, country name, courage, democracy, enemy, ethnicity, family, man, woman, fight, flag/colours, forever/never, future, geographical references, glory, independence/freedom, joy/happiness, home/mother/father-land, law/governance, leader, love, loyalty, peace, poverty/wealth, pride, religion, revolution, sacred, sacrifice, salvation, sorrow, treason, tyrant/chains, unity, win/victory. The number of topics present in a given anthem, as well as their type were found to vary significantly between anthems. Groups of anthems were defined based on these tendencies. Correlations, albeit weak, with more general societal features such as age of country, gross domestic product (GDP) per capita, Gini coefficient, size of armed forces, inequality, inequality-adjusted human development index, and a number of parameters from the World Values Survey (WVS) database (related to religion, gender equality, attitude towards other nationalities/races, attitude towards work, attachment to democratic values etc.) were identified. A shortcoming of the study was not only its subjective character, but also its limited quantitative methodology: it did not differentiate between the number of times a given topic was mentioned in a given anthem. Among other things, a detailed evaluation of the sentiments conveyed by the anthem would be very difficult under these conditions (e.g., it would not differentiate between an anthem that mentions \\u201csorrow\\u201d once in a total of 10 phrases, and an anthem that mentions the same notion in each of the 10 phrases).\"}, {\"pmc\": \"PMC10458337\", \"pmid\": \"\", \"reference_ids\": [\"bib26\", \"bib31\"], \"section\": \"Data collection\", \"text\": \"The English versions of the texts (with caveats pointed out e.g. in ([26,31]) were employed. Data concerning the anthems, including the English version of the text, were from Wikipedia.org (https://en.wikipedia.org/wiki/List_of_sovereign_states_by_date_of_formation, https://en.wikipedia.org/wiki/List_of_countries_and_territories_by_population_density, https://en.wikipedia.org/wiki/List_of_national_anthems, https://en.wikipedia.org/wiki/Category:Lists_of_wars_by_country, https://en.wikipedia.org/wiki/List_of_countries_by_inequality-adjusted_HDI), http://www.worldvaluessurvey.org and http://www.national-anthems.org/. The use of English translations puts a limit on the extent of data which can be meaningfully compared between texts originating in different languages. Detailed syntactic comparisons are hence not possible with this approach. Moreover, the small selected set of parameters which are analyzed in the present study will bear two different imprints: the intrinsic imprint of the historical and sociological context, and a linguistic one which may sometimes be difficult to separate from the latter. The types of analyses considered in the present text are therefore conservative and limit themselves to items that are expected to be conservable in translation \\u2013 e.g., the types of nouns used, or the temporal setting, or the prevalence of verbs, or the types of verbs but not their exact identity, or the general feeling (positive vs. negative).\"}, {\"pmc\": \"PMC10458337\", \"pmid\": \"\", \"reference_ids\": [\"bib32\", \"tbl2\"], \"section\": \"Data analysis\", \"text\": \"Anthems were grouped together as described in text (based on geographical regions, or to some extent on the type of original language, as described below) and analyzed using the Tropes software package, [32]. For each group of anthems, data was retrieved in automated manner (without intervention/bias from the user) about (1) the most abundant notions/references and (2) most abundant word categories \\u2013 all three as defined within the software package without altering the standard settings. An exception from the standard settings was made for the most abundant notions/references, where the software generally identifies country names as belonging to categories named after geographical areas (e.g., Africa, Middle East, Oceania, Asia etc.); these categories are reunited under the common heading \\u201ccountry name\\u201d in Table 2. The Tropes software was developed by Pierre Molette and Agn\\u00e8s Landr\\u00e9 on the basis of the work of Rodolphe Ghiglione (http://www.semantic-knowledge.com).\"}, {\"pmc\": \"PMC10458337\", \"pmid\": \"\", \"reference_ids\": [\"bib33\", \"bib33\"], \"section\": \"Data analysis\", \"text\": \"Sentiment scores were attributed for each anthem individually (not per group) using the web interface of the Semantria software package [33] without any further intervention or processing from the user, except for adding a space character at the end of each verse/line when pasting the text into the analysis window, in order to ensure word separation across lines. This automated analysis employs natural language processing (NLP) and machine learning techniques in order to identify sentiment-bearing phrases and/or components, and assign each of them a sentiment score ranging from \\u223c \\u22121 to 1 for simple expressions. A total score is assigned for the analyzed document by averaging the values for each of the sentiment-related word/expression. For instance in the case of the Romanian anthem, the automatically-recognized sentiment-related terms were \\u201cslaves\\u201d (score \\u22120.929), \\u201cdeadly\\u201d, \\u201cbarbaric\\u201d, \\u201cdeath\\u201d, \\u201ctyrants\\u201d (scores of \\u22120.600 each), \\u201cweapons\\u201d, \\u201cenemies\\u201d (\\u22120.490 each), \\u201crenew\\u201d (+0.450), \\u201cgreat\\u201d (+0.600), \\u201cfreedom\\u201d (+0.680), \\u201choly\\u201d (+1.000), \\u201cglory\\u201d (+1.145), \\u201cvictor\\u201d (+1.503), \\u201cpride\\u201d (+1.800) \\u2013 averaging to a total score of 0.205 for the Romanian anthem. Within the framework of the Semantria feeling analysis, texts with total scores above 0.250 are considered to have a positive overall sentiment; texts with total scores between 0.250 and \\u22120.150 are assigned as neutral, while those below \\u22120.150 are categorized as displaying an overall negative sentiment [33].\"}, {\"pmc\": \"PMC10458337\", \"pmid\": \"\", \"reference_ids\": [\"tbl1\", \"tbl1\", \"tbl1\"], \"section\": \"Relative frequency of word categories\", \"text\": \"Table 1 shows the most common categories of words encountered in national anthems. Verbs expressing actions (\\u201cfactive\\u201d in Table 1) are less common in Central and Western Asian, (32%) vs. the world-average of 40%. Verbs expressing states and possession (\\u201cstative\\u201d in Table 1, 34% world average) occur less often in Africa (27%) but distinctly more in Central and Western Asia and in Slavic anthems (48\\u201355%).\"}, {\"pmc\": \"PMC10458337\", \"pmid\": \"\", \"reference_ids\": [\"bib34\", \"bib35\"], \"section\": \"Relative frequency of word categories\", \"text\": \"To sum up, Balkan anthems stand out with higher percentages of comparisons, negations, and second-person \\u201cyou/thou\\u201d, as well as lower incidence of first person plural \\u201cwe\\u201d. All of these may be argued to reflect an aspiration/inclination towards actively seeking differentiation against \\u201cothers\\u201d \\u2013 in line with the Balkan's XIXth century reputation as \\u201cthe powder keg of Europe\\u201d and with the still extant ethnic and religious diversity [34,35]. These considerations may be illustrated by two examples \\u2013 the Albanian and the Romanian anthems.\"}, {\"pmc\": \"PMC10458337\", \"pmid\": \"\", \"reference_ids\": [\"bib36\"], \"section\": \"Relative frequency of word categories\", \"text\": \"African anthems display higher percentages of plural first person pronouns - and lower percentages of first person singular pronouns. Of all the groups of anthems, the African ones may be interpreted to most clearly reflect an aspiration towards group identity and unity, perhaps in a collectivistic sense [36]. The anthems of Zimbabwe and Mali (the former listed below), illustrate this state of things \\u2013 zero use of \\u201cI\\u201d but repeated use of \\u201cwe\\u201d:\"}, {\"pmc\": \"PMC10458337\", \"pmid\": \"\", \"reference_ids\": [\"tbl1\"], \"section\": \"Relative frequency of word categories\", \"text\": \"Latin European anthems display more time modalities - and fewer manner and place modalities. An apparent relative high incidence of \\u201cI\\u201d seen in Table 1 is in fact an artifact due to the small size of the analyzed set of anthems (the Andorran anthem has an 80% prevalence of \\u201cI\\u201d, while the other ones have 0% prevalence). An example of Latin European anthem is the Romanian one, listed above. There, one may indeed see repeated use of time modalities/adverbs (\\u201cnow\\u201d, \\u201cstill\\u201d, \\u201conce again\\u201d \\u2013 at 50% of all adverbs vs. a 34% world-average), and less manner and place modalities (only one mention \\u2013 \\u201cin full glory\\u201d). A second example may be taken that of the Spanish anthem, which does highlight its time modalities (\\u201clong live\\u201d) \\u2013 though in this case the place and manner ones are rather on the average than below the world-average:\"}, {\"pmc\": \"PMC10458337\", \"pmid\": \"\", \"reference_ids\": [\"tbl1\", \"tbl1\"], \"section\": \"Relative frequency of word categories\", \"text\": \"Overall, the above considerations reveal clear differences in the manners in which various cultures address the anthems. These differences can be further explored if one examines the correlation coefficients between the percentages for each of the categories in Table 1, across the various groups of anthems, as shown in Supporting Information. Considering the small differences seen between most of the parameters in Table 1, these coefficients are all at \\u223c0.9 and above. The weakest correlations (0.81\\u20130.89) involve Central and Western Asia vs. English-speaking countries, Slavic vs. Latin European, and the Balkans vs. several other groups (Africa, SE Asia, Germanic, Latin European, Western European, English). Most of these differences can be explained by the inherently different grammar/syntax structures of the respective languages.\"}, {\"pmc\": \"PMC10458337\", \"pmid\": \"\", \"reference_ids\": [\"tbl2\", \"tbl2\"], \"section\": \"Most common notions\", \"text\": \"The most common terms encountered in anthems across the world are shown on Table 2, alongside differences in frequency of usage of the respective terms in the anthems pertaining to the various groups. The nature of these terms on average across the world is expected considering the purpose of the anthems \\u2013 i.e., the state/nation/country and its contents/definition (land, people, social structures/groups, religion, as well as liberty when asserting one's country in relation to the rest), allegiance/feelings towards it (this includes the engagement to \\u201cfight\\u201d as well as the term \\u201cbody\\u201d in Table 2 \\u2013 where the chest, arms, blood are often invoked).\"}, {\"pmc\": \"PMC10458337\", \"pmid\": \"\", \"reference_ids\": [\"bib31\"], \"section\": \"Most common notions\", \"text\": \"Feelings are the second most common term in anthems across the world. The frequency of this term is smallest in Germanic anthems (by \\u223c70% compared to the world average) \\u2013 perhaps in line with general public perceptions about the respective nations as well as in line with numerical data based on more general analyses of literature in a small set of languages [31]. To illustrate these trends one may look at the two Germanic anthems cited above (Germany and Iceland) \\u2013 where feelings (e.g., love, admiration, pain, horror, as seen in Latin American anthems for instance) are not explicitly mentioned. The ratio between the use of state and of feeling is only slightly larger than 1 in the world average (0.6 vs. 0.5); intuitively, this may be expected since most anthems would logically speak of love/pride for the country. This trend is mirrored in several regional categories, but there are also severe exceptions. A slight reversal is seen in Africa (0.4:0.5), and a major one in English-speaking countries (0.5:1.0); both of these categories are dominated by relatively younger countries, where arguably the state is/was not yet an established a notion as it was in the older or more traditionally established countries. Indeed, a large state:feeling ratio is seen in Germanic anthems (0.9:0.2) and in Central and Western Asia (1.0:0.4).\"}, {\"pmc\": \"PMC10458337\", \"pmid\": \"\", \"reference_ids\": [\"tbl2\", \"appsec1\", \"tbl1\", \"tbl2\", \"tbl1\"], \"section\": \"Most common notions\", \"text\": \"Correlation coefficients between the groups of anthems for the data shown in Table 2 (cf. Supporting Information Table S2) vary widely, from 1 to \\u22120.2 \\u2013 significantly more so than for the types of words discussed in relation to Table 1, and expectedly so given the wider differences between percentages in Table 2 (again, expectedly, since these are more specific/detailed categories than those in Table 1).\"}, {\"pmc\": \"PMC10458337\", \"pmid\": \"\", \"reference_ids\": [\"appsec1\", \"appsec1\"], \"section\": \"Most common notions\", \"text\": \"Reasonable correlations with the world average are expected for all of the regions/categories of anthems; this is indeed the case especially for Asian and Mediterranean anthems (r\\u00a0=\\u00a00.9 cf. Table S2), and reasonably so for Europe and Africa (r\\u00a0=\\u00a00.7). At the opposite end, English-language anthems only show an r\\u00a0=\\u00a00.3 vs. the world average, followed closely by Latin Europe at 0.4. Moreover, the English-language anthems show negligible correlations towards any of the other groups of anthems in Table S2.\"}, {\"pmc\": \"PMC10458337\", \"pmid\": \"\", \"reference_ids\": [\"tbl1\"], \"section\": \"Most common notions\", \"text\": \"Latin American anthems also show their weakest correlations against Latin European and Western European anthems (0.0\\u20130.2, i.e. essentially absent), despite the historical and cultural connections between the two groups of nations. Against the background of a 0.95 correlation coefficient based on the structural grammar data in Table 1, these numbers suggest that a common language or even common history can still leave significant room for cultural differences that may outgrow the differences towards cultures of other languages.\"}, {\"pmc\": \"PMC10458337\", \"pmid\": \"\", \"reference_ids\": [\"bib30\", \"tbl2\", \"tbl2\", \"tbl2\", \"tbl2\", \"tbl2\", \"tbl2\", \"tbl2\", \"tbl2\", \"tbl2\", \"bib30\"], \"section\": \"Most common notions\", \"text\": \"In a previous analysis [30], a list of topics was identified subjectively/manually in the same set of national anthems as analyzed here. The topics were ranked based on the number of anthems where they occurred, but the number of occurrences in each anthem was not counted. By contrast in the present study the topics were identified in automated manner and the number of occurrences within the text was also counted. With these methodological differences in mind, one may nevertheless compare the topics ranked in Table 2 with those from the previous study, noting the general similarities but also some differences. The previous study listed (in this order) the following as most common topics: land, future, country name, independence/freedom, pride, loyalty, geography, religion, unity, forever/never, ancestors/past, family, love. Indeed, \\u201cland\\u201d is also found among the most common topic in the present quantitative analysis cf. Table 2; \\u201cgeography\\u201d does not appear as individual term in Table 2 but may be viewed as a more general proxy of \\u201cland\\u201d. \\u201cFuture\\u201d, \\u201cforever/never\\u201d and \\u201cpast\\u201d are not found by themselves in Table 2, but are arguably still encompassed by the 4th most common term there \\u2013 \\u201ctime\\u201d. \\u201cCountry name\\u201d is identifiable in Table 2 as the top notion \\u2013 \\u201cstate\\u201d. \\u201cIndependence/freedom\\u201d is identified in Table 2 as \\u201cliberty\\u201d. \\u201cPride\\u201d and \\u201clove\\u201d are not present per se in Table 2, but may be taken as part of the second most common notion there, \\u201cfeelings\\u201d. \\u201cReligion\\u201d and \\u201cfamily\\u201d are present in exactly the same form in Table 2. \\u201cLoyalty\\u201d and \\u201cunity\\u201d are notably absent from Table 2, unless they are deemed to be convoluted into other terms (e.g., \\u201cfeelings\\u201d). There is thus an overall agreement between the results of the two methodologies. The case of the two terms, \\u201cloyalty\\u201d and \\u201cunity\\u201d, which appear to be among the top 10 notions in anthems according to the previous study [30] in terms of a binary count (are present vs. are not present), but are not identified among the top 10 notions in the present statistical study where the number of occurrences within each text is also counted (i.e., non-binary), may be instructive to note. On one hand, may interpret this difference as a manifestation of the subjective methodology in the previous study. However, one may also use this example to raise the question of the intrinsic relative weights of the words \\u2013 and point out that by no means these are (or should be taken as) uniform, neither within the same culture and especially nor across cultures.\"}, {\"pmc\": \"PMC10458337\", \"pmid\": \"\", \"reference_ids\": [\"tbl3\"], \"section\": \"Most common notions\", \"text\": \"Table 3 summarizes the salient features identified for each class of anthems in comparison with the others.\"}, {\"pmc\": \"PMC10458337\", \"pmid\": \"\", \"reference_ids\": [\"bib36\", \"bib37\", \"bib38\"], \"section\": \"Most common notions\", \"text\": \"While this body of data may offer grounds for further socio-cultural comparisons, with theoretical frameworks ranging from philosophy to exact sciences [[36], [37], [38]], further discussion about the cultural relevance of the anthem analyses must take into account the at-most-indirect representability of these texts for the respective nations. Though most of them were validated at least tacitly in time by the people of the respective countries, the anthems were generally adopted by a very small group of people (\\u201celites\\u201d), often entirely unrepresentative, from a statistical point of view, of the social/demographic fabric of the country \\u2013 and in many cases not even elected democratically to do so. One must thus consider in each case whether the texts chosen by these elites were describing the fabric of their nations, or the future goals thereof, or if in fact there was a hiatus between the respective elites and nations on these topics. The latter may be particularly relevant for anthems in countries that have more recently gained their independence (and hence where popular confirmation of the anthem over time is not a factor to be considered yet). Last but not least, in interpreting statistical data on the presence of types of words and notions in anthems belonging to various cultures, one must also point out that the intrinsic weights of various words/notions are not inherently uniform within the same culture and much less so across cultures. If one mentions \\u201cunity\\u201d and \\u201ctime\\u201d with a similar frequency in a text, it should not follow that the importance received by the two terms in the mind of the citizen/speaker is equal \\u2013 nor that their relative weights would be identical regardless of language, culture, or historical context. Because of this, the discussion of the present data is for the moment left at a rather basic level, since follow-up studies would need to entail a much more interdisciplinary approach \\u2013 e.g., psychological, historical, social beyond the obvious linguistic and literary components.\"}, {\"pmc\": \"PMC10458337\", \"pmid\": \"\", \"reference_ids\": [\"fig1\", \"appsec1\", \"tbl2\"], \"section\": \"Sentiment analysis\", \"text\": \"It may be noted that the averages listed in Fig. 1 have relatively large standard deviations (generally around 0.3, cf. Supporting Information Fig. S1 and Table 2). In Africa, the most notable outliers are Senegal and Central African Republic (which display negative values in a group where the average is distinctly positive), as well as Ghana and Niger (standing at twice the average positive score of the group). In Latin America, Peru is the sole notable exception, some 0.6 units more negative than any of the other anthems. In Asia, three countries -China, Japan and Vietnam - stand out with distinctly lower scores than the rest of their group. Central and Western Asia is on the other hand the most homogeneous group. By contrast, the Balkans offer one of the most heterogeneous set of values \\u2013 spread across a full 1.5 units, twice as large compared to Central and Western Asia (with Bosnia & Herzegovina at the higher end at\\u00a0\\u223c\\u00a01, and Albania at the lower one at\\u00a0\\u223c\\u00a0-0.3). Among the Germanic anthems, Germany itself is and exception, with a score almost twice as positive compared to the average of its congeners from the same group (and in fact the third largest in the world, surpassed by only Georgia and Bosnia & Herzegovina). Among the Slavic anthems, the two southernmost Balkan nations of Montenegro and Northern Macedonia are by far outliers, with lower values compared to the rest. Western Europe is slightly more heterogeneous than the Balkans, with Ireland providing the most negative value against Germany's most positive. The Mediterranean region, more than others, appears split across two main relatively homogeneous subgroups: one with distinct positive values (e.g., Croatia, Egypt, Israel, Jordan, Morocco, Spain) and one with neutral and negative values (Albania, Algeria, France, Greece, Italy, Northern Macedonia, Montenegro).\"}, {\"pmc\": \"PMC10458337\", \"pmid\": \"\", \"reference_ids\": [\"bib31\", \"fig1\", \"fig1\", \"bib31\"], \"section\": \"Sentiment analysis\", \"text\": \"The herein reported general positive-sentiment trend in national anthems is in line with the previous observation that human (written) language has a positive bias [31]. In the same previous study, it was also noted that among the ten languages studied, Spanish and Portuguese appear to have a higher positive bias than others \\u2013 with Chinese, Korean and Russian at the opposite end. This is in contrast with the opposite finding in Fig. 1 for Latin-American (Spanish and Portuguese-speaking nations) and Russian, but in reasonable agreement for SE-Asian group. The case of English language is also illustrative: Dodds and co-workers find English books to be at the higher end of the positive bias scale defined by all languages studied, while English music lyrics are at the opposite end. Indeed, the anthem data of Fig. 1 correlate with this latter finding \\u2013 and this may also offer at least a partial explanation for the Latin-American and Russian cases [31].\"}, {\"pmc\": \"PMC10458337\", \"pmid\": \"\", \"reference_ids\": [\"fig1\"], \"section\": \"Sentiment analysis\", \"text\": \"When examining the correlations between the averages of the respective values per group of anthems, rather than between all individual values, higher correlation coefficients are found. The correlation coefficients between averages of sentiment scores and averages of WVS data (counted as percentages of citizens who choose a given response in WVS questionnaires) were performed for countries where such data were available. Negative correlations were found with the percentage of people declaring they are very happy or rather happy (\\u22120.8), or for whom leisure is very important (\\u22120.7), or for whom democracy is very good or fairly good (\\u22120.6) \\u2013 and positive correlations with the percentage of people who agree that being a housewife is just as fulfilling as working for pay (0.7), or who agree or agree strongly that men are better leaders than women (0.8), or who would prefer not to have people of other races as neighbors (0.6), or who think that having a strong leader who does not have to bother with parliament and elections is very good (0.7). These data paint a picture of positively-oriented anthems as pertaining to more conservative countries \\u2013 which indeed is in line with the fact that Central/Western Asian and Eastern-European anthems are at the higher end of the positive feeling score in Fig. 1.\"}]"

Metadata

"{}"