What motivates heads of state to engage with international organizations (IOs), and do IOs benefit from leaders’ participation? On the one hand, leaders’ involvement in IOs could indicate the importance of international cooperation to their respective countries, bringing prestige and prominence to multilateralism. On the other hand, leaders’ appearances can politicize IOs and detract from meaningful cooperation. Some of the most memorable visuals of the United Nations General Assembly (UNGA), for example, are not those of diplomatic experts seeking consensus, but rather, for example, Libya’s Muammar Gaddafi ripping up a page of the United Nations charter in 2009, Soviet Premier Nikita Khrushchev allegedly banging his shoe in 1960, or Hugo Chávez of Venezuela in 2006 crossing himself at the rostrum after declaring that “the Devil came here yesterday, right here. It still smells of sulphur today. Yesterday on this rostrum the President of the United States, whom I refer to as the Devil, talked as if he owned the world”(A/61/PV.12, p. 10). Such examples suggest that leaders’ presence may be more disruptive than cooperative.

Leaders’ rhetoric is especially puzzling in the context of the UNGA, which was designed as a forum for parliamentary-style deliberation conducted by diplomats, not heads of state (Gordenker, 1962). When political leaders take the microphone at a venue intended for ministers of foreign affairs and permanent representatives, how do they deliver their messages? Do leaders’ addresses contribute to the quality of debate and communication in the most important global forum, or are they a distraction? The extensive literature on leaders and foreign policy tends to center on conflict, not cooperation, and thus is of limited applicability to this question.Footnote 1 Conceptualizing leader behavior in the world’s most prominent IO can illuminate whether leaders — and individuals more broadly — can influence the practical dynamics and vitality of international cooperation (Gray, 2020).

This paper explores how leaders’ incentives shape their speech and whether their participation in multilateral debate contributes to, or detracts from, discussion on issues requiring global solutions. Unlike many bilateral initiatives, which can bring targeted policy benefits that make credit-claiming easy for leaders (Cruz & Schneider, 2017), the rewards of multilateral cooperation are often incremental and diffuse. On balance, leaders may have an incentive to use multilateral fora for their own political ends. We argue that leaders’ speeches tend to center on their more particularistic agendas, not in service of broader cooperation. Newly collected data on the identities of almost 10,000 country representatives at the UNGA from 1946 to 2019, as well as semi-supervised and non-supervised text analyses of speeches at the assembly, reveal that in contrast to other delegates, leaders’ speeches tend to be removed from the core policy discussion. We argue, and show, that consistent with theories of leader communication, leaders’ speech is more self-serving; they overuse personal pronouns and speak in a less technical manner than do other types of representatives. This sets their rhetoric apart from that of delegates who engage more directly with global policy.

The UNGA, the most prominent multilateral forum, is well placed for testing our argument. All countries have a speaking slot at the beginning of each annual session, but heads of state can choose whether to use it. National representatives, including leaders, also choose the extent of their attention to global policy priorities while addressing the assembly. Leader attendance at the UNGA is also puzzling more broadly, since the original aim of the general assembly was meant to be technocratic, not political; the original designers of the Assembly meant it to be a forum for ministers and permanent delegates.

Our findings suggest that leaders, on balance, conduct themselves in the UNGA in service primarily of their own agenda, implying that the star power that leaders bring may not translate into policy gains for international cooperation, or assist in revitalising the United Nations,Footnote 2 let alone in boosting its legitimacy.Footnote 3 If leaders bring their own motivations and rhetorical tactics to the international stage, it could contribute to the forum’s politicization (Zurn et al., 2012; De Vries et al., 2021). Although leaders can focus citizens’ attention onto international matters (Guisinger & Saunders, 2017), our findings cast doubt on the efficacy of their appearances at IOs to revitalize the UNGA.

This paper unites several important literatures. The first centers on leaders and foreign policy (Horowitz & Stam, 2014; Saunders, 2011), with a particular focus on their rhetoric in cooperation. A substantial literature explores leader communication (Baturo & Tolstrup, 2024; Benoit et al., 2019; Dewan et al., 2014) as well as leader behavior and reputations (Renshon, 2017; Renshon et al., 2018). To our knowledge, ours is the first study to examine the speech of leaders across regime type in an IO rather than in a conflict setting, bringing in insights from the burgeoning literature on diplomacy (Gertz, 2018; Lebovic & Saunders, 2016), with a focus on quantitative text analysis in analyzing leaders’ rhetoric.

We also contribute to theories of elite communication (Dellmuth & Tallberg, 2021) and practice within IOs (Adler & Pouliot, 2011),Footnote 4 particularly in the UN. Most studies of the UN focus on its more prominent agencies, such as the Security Council (Patz & Thorvaldsdottir, 2021; Voeten, 2005; Vreeland & Dreher, 2014). But the UNGA is the most prominent public-facing aspect of the UN, and as such it is a meaningful forum for incorporating models of rhetoric, particularly given the particular importance of speech in diplomacy (Jönsson & Hall, 2003).

We also contribute to the growing literature on IO vitality (Gray, 2018), examining how individuals can impact an organization. Leaders have only recently become a prominent feature at the UNGA, a body long-claimed to be in need of revitalizaiton. Scholars have examined the role of differing levels of delegation in IO staff and how it can impact cooperative outcomes (Gray & Baturo, 2021; Hawkins et al., 2006; Vaubel et al., 2007). In these accounts, individual diplomats can have discrete influence on policy outcomes (Jordan & Tuman, 2018), and professional technocrats face differing incentives than do political actors (Johnson, 2013; Poulsen & Aisbett, 2016). When applied to the UN, these theories show how, for example, changes in the appointment process of the Secretary General have influenced policy outcomes (Wiseman, 2015). Although votes are highly politicized (Dreher et al., 2009), scholars have also shown that national interests can also prevail when countries take the helm of various sites of UN operation, including the Security Council (Kuziemko & Werker, 2006) and the position of the Secretary General (Novosad & Werker, 2019). This paper suggests that leader politicization can affect the discursive environment of the UNGA, which also sheds light on how different branches of a broader IO can vary in their vitality. Although this paper focuses on the UN, the scope conditions extend to other general-purpose IOs (Hooghe & Marks, 2016), particularly those with parliaments (Schimmelfennig et al., 2021), as well as other fora for leader engagement with international cooperation, such as global summits.

This paper proceeds as follows. After discussing the increasing norm of leaders’ attendance of the UNGA, we then explain why and how leaders engage with and address fora for multilateral cooperation. Next, we introduce the data and methods used, and turn to test our arguments. The final section concludes with implications and further directions.

1 National leaders in the United Nations General Assembly

The General Assembly stands as the UN’s most inclusive realm of diplomatic discussion, and early research on UN politics highlighted the importance of sending permanent representatives and foreign ministers to the UNGA debate (Alger, 1963; Ernst, 1978). Heads of state were not really intended to speak at the UNGA, and for the first several decades of of the UN’s existence, this norm was largely upheld, with the sole exception of the 1960 general debate.Footnote 5

However, the pattern shifted in the 2000s (Baturo & Gray, 2023). From 2014 onward, over half of all speakers were national leaders, as shown in Fig. 1.Footnote 6 The 2000 Millennium Summit and then 2005 World Summit — which were held before the beginning of the UNGA sessions in 2000 and 2005, both attended by an unprecedented number of leaders — have arguably normalised UN appearances by heads of state, such that they were more likely to visit regular sessions as well. Indeed, leaders’ once-rare appearances at the UNGA are now so commonplace that, for the 2020 session — held virtually due to the COVID-19 crisis — one observer lamented that, without leaders there to add luster to the proceedings, “the crucial week-long high-level debate will make very little contribution to advancing the cause of international peace and security.”Footnote 7

Fig. 1
figure 1

Leaders’ UNGA attendance over time. Note: Frequency and percentage of leaders to speak at the UN to all speakers in a given session

Full size image

In those same decades, the UNGA also fielded efforts to, in its words, “revitalize” in the face of its perceived ineffectiveness. IO vitality can vary not just across organizations but also across different administrative units within the same IO (Hosman, 2023). In this instance, precisely because the UNGA was designed to be more representative and inclusive than some of the UN’s other bodies, such as the Security Council (Peterson, 1986), many have criticised the degree to which genuine debate occurs in the UNGA and whether the general assembly effectively moves policy forward (Panke, 2014). Different arms of IOs serve different purposes, and because the Assembly acts as the public-facing arm of the UN, critiques of that body seep into the broader perceptions of the organization, with implications for its legitimacy and reputation (Brazys & Panke, 2017). Examining the factors that may add to, or detract from, the UNGA’s vitality has consequences for the UN as a whole, just as the vitality of any one IO has consequences for the international system (Eilstrup & Verdier, 2023; Schmidt, 2023).

The debate regarding the assembly’s vitality and its broader purpose emerged in the very beginning of the organization. Already by 1949, observers had begun to question the role, authority, effectiveness, and efficiency of the General Assembly. “Mindful of the increasing length of General Assembly sessions, and of the growing tendency towards protracted debates” (General Assembly resolution 271(III), 1949, 17), the Assembly established a Special Committee on Methods and Procedures of the General Assembly. During the Cold War, delegates continually discussed how to improve the procedures and organization of the General Assembly, such as seen from resolution A/RES/2837(XXVI) in 1972, as well as the Assembly’s process of deliberation through debate, in particular (Keens-Soper, 1985, 78). In the decade after the fall of the Berlin Wall, members passed three “revitalisation” resolutions,Footnote 8 and starting in 2001 the practice became annual.Footnote 9 At the 60th session in 2005, the Assembly established an Ad Hoc Working Group on revitalization. Ironically, efforts to revitalize the GA have been so numerous and varied that “many delegates now openly complain of revitalization fatigue” (Swart, 2008, 21).

The tone and quality of the general debate, and the attention that it receives, underpin these revitalization discussions. While many delegates tend to argue that UNGA speeches should be aimed with the view to engage “in fruitful, results-oriented discussions, premised on consensus” (U.N. General Assembly, 2023), and that “the main value of the debate is the opportunity it provides to ensure that the GA is the pre-eminent deliberative body in the UN system” (Swart, 2008, 22), they simultaneously bemoan the fragmentation of the debate, noting that media are “more likely to report on what divides its Member States than on what unites them” (Swart, 2008, 23).

Because leaders have become reliable fixtures at the UNGA, it is important to uncover not just patterns in their speech, but also how their speeches fit in with the broader context of the UNGA and its revitalization. The next section discusses theories of leaders’ communication and the role it might play in the UNGA.

2 Leaders, the global agenda, and the quality of communication

Scholars and practitioners alike have long grappled to understand the consequences of individual participation at IOs. On the one hand, core theories of international collaboration center on the idea that actors in international fora can become socialized into norms of cooperative behavior (Finnemore & Sikkink, 2001). On the other hand, principal-agent theories warn of agency slack when certain competencies are handed off to third parties (Hawkins et al., 2006; Johnson, 2014). These ideas to some extent contradict each other: while technical delegates may better steward complicated matters of international governance, their interests may not always align with those of national governments. This tension — between technocratic remove and political realities — has long been a conundrum in international governance.

The UNGA, in particular, was meant to serve as a forum for inclusion as well as discussion, intended to represent the “deliberations of diplomats gathered in public assembly [to] develop a new kind of esprit de corps based upon more than professional solidarity. … Diplomats would enter the assembly ignorant of everything but the arguments of raison d’etat to graduate as citizens of one world, reasonable men embodying ‘the reason of the whole”’ (Keens-Soper, 1985, 77). Such descriptions seem far removed from the tendencies of leaders, who face rather different incentives in their communications.Footnote 10

Leaders’ attention to cooperative foreign policy, and whether such attention is beneficial to multilateral cooperation, is not well understood, even though the politicization of leaders suggests countervailing effects on IOs. While leader participation could be seen as “diplomacy at the ‘highest level’” (Plischke, 1972, 323), at the same time, leaders’ incentives may result in further politicization of the organization. On the one hand, leaders can act as focal points in communication,Footnote 11and given their incentives to appeal to broader audiences, leaders’ rhetorical strategies can make discussions more inclusive (Spirling, 2016). On the other hand, leaders can also use IOs as an overt pulpit primarily to build personal prestige, and it can impede the quality of debate on world issues, also undermining the legitimacy of the organization. Furthermore, politicization leads IOs to manage international problems on the basis of state interests rather than on technical cooperation (Marieke & Maertens, 2021; Weiss, 2012).

Leaders — in developed and developing countries, and regardless of regime type — often put themselves forward internationally to launch their own foreign policy agendas. Consider how the intricacies of leader behaviour and speech play out in the example of Soviet premier Nikita Khrushchev’s attendance at the UNGA at the height of the Cold War. While travelling on a steamship across the Atlantic to address the UNGA in September of 1959, Khrushchev brainstormed various ideas for his forthcoming statement to the UN together with János Kádár of Hungary, Gheorghe Gheorghui-Dej of Romania, and Todor Zhivkov of Bulgaria. As a result, Khrushchev modified the original ministry’s draft to instead reflect his own personal policy priorities, including a plan to surprise the UN with the proposed reform of the UN Secretary-General office (Volkogonov, 1999, 225–6). One year later, in 1960, Khrushchev went again with the intention of delivering a fiery, anti-West message directed simultaneously at the US, at his own public, and at the leaders of newly independent African nations that would be in attendance.Footnote 12 He substantially amended the speech that was given to him by his foreign minister to make it far more incendiary, as well as improvising substantially over the course of his address. Khrushchev later explained to his top officials that “your voice must impress people with certainty. … Don’t be afraid to bring it to a white heat, otherwise we won’t get anything” (Fursenko & Naftali, 2006, 413).

Like Khrushchev, U.S. President Jimmy Carter also used the UN as a space to announce ambitious foreign policy projects. Carter, known for his commitment to human rights, visited the UN in his first year in office and spoke in service of putting his signature foreign policy issue on the agenda, asking the UN to increase its commitment to human rights, promising to put U.S. power behind this goal. “The United States, my own country, has a reservoir of strength: economic force, which we are willing to share; military strength, which we hope never to use again; and the strength of ideals, which are determined fully to maintain the backbone of our own foreign policy.”Footnote 13 In contrast, decades later, US President Donald Trump announced that “The future does not belong to globalists. The future belongs to patriots” (A/74/PV.3). The latter was in the context of what a former UN ambassador described as “the challenge of how to promote his America-first agenda at a meeting of the UN, a global body that is all about addressing global issues.”Footnote 14

Likewise, Mikhail Gorbachev relied on his UN appearances to promote his own agenda. For example, in his speech to the UN in 1988 — on the eve of the first-ever multiparty legislative election in the USSR slotted for 1989 — he highlighted his ambitious economic reform efforts, then proceeded to announce an unprecedented degree of planned disarmament: “Today I can inform you of the following: the Soviet Union has made a decision on reducing its armed forces. In the next two years, their numerical strength will be reduced by 500,000 persons, and the volume of conventional arms will also be cut considerably” (N88/645/59). The speech was deliberated in advance in the Politburo (Service, 2015, 355) and was prominently broadcast on Soviet television.Footnote 15

These examples illustrate leaders’ dramatic cast on the debate, departing from the messages that professional diplomats tend to deliver. In contrast to ministers of foreign affairs and other diplomatic representatives, leaders not only have stronger personal political concerns in mind when addressing IOs,Footnote 16 but as principals they also have more leeway over the contents and style of their messages (Baturo & Tolstrup, 2024). Although leaders’ constituents tend to prioritize domestic mattersFootnote 17 over international ones, executives can use the stage of IOs to project their messages to multiple audiences.Footnote 18

When heads of state delegate diplomatic interactions to agents, they may be intending their messages to be framed to be congruent with the norms and ideals of the host (Lindsey, 2017), and indeed diplomats have been shown to have similar speech patterns among themselves (Gray & Baturo, 2021). By contrast, when leaders choose to supplant their diplomatic agents, they may deliver messages that are closer to their own agendas; they should be unconcerned about sticking to the goals of the meeting itself, focusing instead on messages consistent with their own motivations. We therefore propose the following core hypothesis:

  • H1: Agendas in Leaders’ Speech: Leaders’ speech will be more particularistic, less focused on key global issues, and more dissimilar from the core agenda, than will be the speech of other delegates.

We also propose a second, supplementary hypothesis regarding leaders’ rhetoric in IOs. Rhetoric is about not only content but also tone and affect.Footnote 19 Leaders’ motivations to attend the UNGA should be reflected in the quality of their communication, specifically in stylistic and tonal differences. If leaders visit the UNGA out of a desire to participate in global debate, we might expect them to adapt their speeches — or to have their speechwriters adapt those speeches — to more closely resemble those of diplomats or ministers of foreign affairs. Attendantly, a motivation for global policy would likely result in leaders not only hewing to the assembly’s agenda, as described in H1, but also deploying rhetoric that is more specific, resembling the speech of professional diplomats.

Indeed, diplomats and other foreign policy professionals have been shown to speak in a manner targeted at international audiences (Bayram & Ta, 2019) rather than at the public. This rhetorical conduct reflects diplomats’ own career motivations as well as their mandate to engage in multilateral policymaking (Pouliot, 2016). A shared style of communication also enables more efficient negotiation over policy issues. Leaders, however, have distinct styles of communication (Dewan & Myatt, 2008). We know that political leaders speak simply in political communication (Teten, 2003), shying away from technical language (Charteris-Black, 2011, 225, 246). In fact, simplicity is a hallmark of political rhetoric (Gustainis, 1990, 159). This tendency has been demonstrated qualitatively as well as quantitatively (Benoit et al., 2019). Furthermore, political leaders are known to practice credit claiming (Baturo & Tolstrup, 2024)—consistent with our argument about bringing their own agendas to boost personal standing—which in turn is reflected in their frequent usage in speech of the first-person singular pronoun, ‘I’ in particular (Liu, 2022).

Leaders’ general communicative strategies are likely to carry over into their rhetoric on the international stage. Rather than adapting their speech to be congruent with the host — and thus more effectively contributing to multilateral debate — we argue that leaders stay true to their regular rhetorical tactics. Attendantly, the rhetoric of political leaders will be more abstract, self-centred, and targeted at a simpler audience than what might be expected in addressing a crowd of experts (Dewan & Myatt, 2008). Leaders will be also less likely to use the diplomatic jargon espoused by permanent representatives and ministers of foreign affairs.

  • H2: The Quality of Communication: Consistent with their role as politicians, leaders are likely to speak more generally, using more personal pronouns, and with less diplomatic jargon.

H2 is in line with our general argument centred on leaders’ advancing their own policy agendas, not multilateral policy, in IOs. If leaders were concerned about policy issues relating to a particular global event, we would expect their rhetorical style to be more specific and more complex. Instead, the generality of their speech (H2), along with its dissimilarity to those who outline the policy goals of the session (H1), serve as observable implications of leaders’ motivations to commandeer the stage to advance their personal policy priorities.

3 Data and methods

We test our arguments on originally collected data on leader attendance and speeches at the UNGA. The UN Dag Hammarskjold Library stores the names of the attendees, which we retrieved through the UN Bibliographic Information System (UNBIS). Altogether, excluding non-national delegates, such as speakers representing the UN and other organizations, from 1946 to 2019 there are 9,959 total speakers, of which 2,179 individuals, or 22 percent, are effective national political leaders.Footnote 20 Because leaders systematically appear during the most publicised and high-profile part of the regular session, the UN General Debate, we focus on the logic of individual leaders’ attendance and speech during this first substantive item on the session agenda. We then match the type of attendees with the texts of the UN General Debate, sourced from Baturo et al. (2017). We further expand the text corpus by adding the statements of non-country delegates, such as assembly presidents (PGA), as well as those of the UN Secretary-General.Footnote 21

H1 proposed that leaders’ speech will be more particularistic in terms of policies and less focused on key global issues than that of other delegates. Measuring the degree of particularistic policy agendas, represented by very different policy-particular terms across leaders’ speeches— such as, for instance, the specifics of Argentine debt restructuring in Cristina Kirchner’s 2015 speech or Robert Mugabe’s 2004 anti-sanctions’ diatribe for Zimbabwe—is not feasible at scale. Instead, we can measure the extent of more universalistic concerns across speeches, so that a lower degree of such concerns proxy for leaders’ propensity to deviate from the global agenda and instead to advance their more particularistic messages. Attendantly, we rely on four empirical approaches: word embeddings; the degree of similarity between leaders’ speech and that of the assembly president; also, that of the UN Secretary-General; and topic analysis.

First, we measure attention to the global agenda in leaders’ speech. We rely on a semi-supervised machine learning method, Latent Semantic Scaling (LSS) (Watanabe, 2021). This method allows us to measure the intensity of the “global agenda” rhetoric, by computing similarity between selected seed words and all other terms in each text.Footnote 22 As seed words, we include “cooperat,” “global,” “agenda,” “united nations,” “organization” and “assembly.” We then compute average scores for each speech, where the more positive values stand for a higher propensity to engage in rhetoric centred on global issues.

Figure 2 displays the terms with the highest estimated coefficients, as computed by the LSS algorithm, on our text corpus. The results from Fig. 2 have strong face validity: the terms with estimated higher coefficients signify more attention to global issues and to the organization’s agenda. The algorithm has estimated that among a hundred of the most lexically similar terms to chosen seed words, there are—all terms are stemmed—such terms as “ambiti agenda,” “transform,” “risk reduct,” “disaster risk,” “climat summit,” “ebola,” “virus,” “develop framework,” “global econom,” “global problem” and other “global” terms.Footnote 23

Fig. 2
figure 2

Global agenda terms. Note: Figure includes 100 top terms with the highest estimated coefficients, that is, most lexically similar to “cooperation”, “global”, “agenda”, “united nations” “organization” and “assembly” used as seed words, following Latent Semantic Scaling analysis, as explained in text. Font size depends on the size of coefficients, from 0.04 to 0.13

Full size image

Operationalizing the global agenda requires a benchmark against which to which to compare speech. We next turn to a more specific test, and measure the degree of similarity between leaders’ speeches and that of the UNGA president (PGA). No member-state can be regarded as a suitable candidate for a quasi-median legislator (Proksch & Slapin, 2014) with policy preferences that stand for global policy preferences, in every year.Footnote 24 In contrast, PGAs, elected for one year, are usually ministers of foreign affairs or diplomats, and attendantly their speech is both stylistically and topically closer to the language of international diplomacy and the goals of the UN. Furthermore, they specifically determine the focus of each year’s debate.Footnote 25 As a benchmark for comparison, we utilize speeches by the PGA, who is the only speaker at the UNGA who speaks on behalf of the assembly and not her national government (Peterson, 2006, 50–55).Footnote 26 Upon election, the newly elected PGA vacates the task of representing their own country to another diplomat, and makes an address to the assembly outlining the main issues facing the world and the assembly (Peterson, 2006, 50). This makes the PGA’s address an appropriate benchmark against which to evaluate whether leaders’ speeches are distinct from the IO policy agenda, compared with speeches made by other types of national representatives.

To illustrate, Table 1 includes short examples from the 49th session. While ministers of foreign affairs and diplomats raise matters germane to the agenda, and reflect on similar themes to those that the PGA discusses, leaders tend to speak out of tune, as seen from their lower LSS scores, discussed above, as well their lower similarity scores to the PGA (explained below).Footnote 27

Table 1 Illustrative examples from the UNGA 49 session (1994)
Full size table

As a test of face validity for whether PGA speech can be used as a benchmark for global agenda, we also fit several simple keyness analyses, below. The keyness approach isolates words that frequently appear in a given text, capturing the degree to which a given word is “key” overall, in contrast to another text(s). Simply put, we want to show whether changes in speech are revealed when the same individual shifts roles from nation-state delegate to that of the PGA, and whether the PGA’s terms tend to reflect the UNGA’s agenda, in contrast to other speakers. This approach allows us to isolate whether a UNGA speech varies as a function of the office or the officeholder (Baturo & Elkink, 2014).

Consider the case of Abdelaziz Bouteflika, the foreign minister of Algeria (1963–79) who subsequently served as the country’s president (1999–2019). On 17 September 1974 Bouteflika was elected president of the 29th UNGA session and made a speech outlining his views of the international situation and the assembly priorities.Footnote 28 The left subplot of Fig. 3 shows that when Bouteflika speaks on behalf of Algeria as its foreign minister, his speech apparently centers on colonialism, Arab-Israel issues, African security, and development—the issues of concern to Algeria as a North African nation. In contrast, when Bouteflika — the same individual — speaks for the assembly as PGA, the top most distinct terms suggest that he turns to global and UN affairs such as questions of new membership and procedural questions in the organization. As a more general illustration of PGAs’ speeches as distinct from those of country representatives, we also compare the speeches of all PGAs with the speeches of their own countries’ national delegates in a given year, in the right subplot of Fig. 3. It clearly shows that PGAs tend to emphasize the UN affairs and global norms, in contrast to their national counterparts who focus on international and regional security issues, in the same year. In summary, Fig. 3 underlines differences in speech across delegates depending on their role, and that PGA’s speech can indeed be a suitable proxy for global policy priorities.

Fig. 3
figure 3

Lexical differences. Note: The keyness analyses isolate words that frequently appear in a given text(s), capturing the degree to which a given word is “key” overall, in contrast to another text(s). Left subplot compares Abdelaziz Bouteflika’s speech as the PGA (1974), in grey, and as the representative of Algeria (1971–73, 1975, 1977–78), in black. Right subplot compares speeches of all PGAs (in grey) with speeches of PGAs’ own countries’ delegates that year (in black), in the whole corpus

Full size image

Finally, alongside word embeddings and text similarities, we also compare the most distinct policy topics found in speeches of PGAs with those in speeches of leaders, as well as other types of speakers, as explained below.

4 Empirical tests

We now turn to the question of how leaders address the global forum, and whether they are more distant from the global priorities than other national delegates (H1).

Table 2 includes specifications estimated as country-fixed effects regression models with time period (annual dummies) controls. First, in Columns 1–5, we include average country-year “global agenda” LSS scores (explained above) as the dependent variable, followed by similarity scores (discussed below), in 6–12.

Table 2 Global agenda and similarity to PGA and UN secretary-general
Full size table

The first, baseline model includes a binary indicator for whether the speaker is the political leader. Second, we test for the possibility that it is not leaders’ priorities that differ from the global agenda, but differences within the leader category that are important. Therefore, we distinguish between democratic and nondemocratic leaders, as well as nondemocratic other (nonleader) speakers, in contrast to democratic other speakers (the baseline). Third, we include additional controls, such as a democracy indicator (Coppedge et al., 2019), a logarithm of income per capita, from the World Development Indicators, and whether a country holds a UNSC seat and therefore may opt to reflect global priorities because of its membership in this important security forum. Also, because speakers may be more likely to address global problems in the post-Cold War war period, when such issues as climate change or sustainable development are prominent, in contract to the earlier period dominated by the East–West rivalry (Bailey et al., 2017), in Columns 4–5 we split the sample accordingly.

The results strongly suggest that leaders’ speeches, regardless of regime type, have a lower propensity to engage with the global agenda, compared with other representatives (H1). Furthermore, leaders are less likely to discuss the global agenda even in the Cold War period (Column 4). In turn, all democratic speakers, as well as those from more developed nations, have a higher affinity to the global agenda, but the UNSC seat has no effect. As revealed in Fig. 4, which plots marginal effects following the second and third models, on average all leaders — and particularly nondemocratic leaders — are less likely to discuss the global agenda issues in their speech, compared with other delegates. In magnitude, this is about a quarter of a standard deviation of this measure (or half of the total effect of GDP pc), and the difference is statistically significant.Footnote 29

Fig. 4
figure 4

Global agenda in speech. Note: The plot displays marginal effects of world leaders’ propensity to engage with the global agenda in speech (latent semantic scaling), based on Models 2 and 3, Table 2

Full size image

Next we turn to another measure to test whether leaders are more likely to bring their own agendas, in contrast to other delegates. As a test of whether leaders’ speech diverges from the debate’s policy goals (H1), we compute cosine similarities to measure distances between different types of speakers and PGAs, separately for each session.Footnote 30 A lower degree of similarity to the PGA’s speech will indicate that leaders speak more distinctly and cover different issues than those covered in the president’s speech. A higher cosine indicates a higher degree of similarity between texts. On average, the cosine similarity between PGAs’ and leaders’ speeches has a score of 0.547 (standard deviation of 0.106), compared with 0.602 (0.097) for other types of speakers; the difference is statistically significant (t-value of 21.098).

Columns 6–10 include the same specifications as in 1–5, with the similarity to the PGA’s speech as the dependent variable. Positive coefficients indicate higher similarity to the assembly’s chief officer. In Columns 8–10, additionally, we add an indicator for speakers from the same country as the one that nominated the president that year. The results obtained from this alternate measure corroborate those obtained earlier: on average, leaders’ speeches are more distant from the PGA’s, than other delegates. In turn, democratic and nondemocratic leaders alike are more distant, while those from more democratic and developed nations have stronger similarity.

The specifications in columns 9–10, Table 2 additionally account for a change in terms of how the UNGA has arranged the debates. Even though newly elected PGAs have always tended to speak on international issues facing the world and the United Nations in their address to the assembly, according to resolution 58/126, from 2004 on the president in her or his speech was also required to suggest the issues of global concern, more formally.Footnote 31 Splitting the sample between speeches made before 2004, and from 2004 onwards, we therefore examine whether this change in procedure has influenced the degree of text similarity between speeches of country representatives and that of PGA that year. Results in columns 9–10 clearly indicate that leaders tended to have lower similarity to speeches of PGAs, whether before 2004 or after.

While the PGA is the only speaker who speaks on behalf of the assembly and not their government, as explained earlier, the UN Secretary-General (UNSG) is another speaker who in principle speaks on behalf of the UN at the Assembly—but this practice was adopted from 1997 only, while we have annual statements of the PGA for the whole time period. Nonetheless, we perform an auxillary test on those speeches: the dependent variable in Columns 11–12 is based on a measure of cosine similarity with the UNSG, available for 1997–2019 only, estimated in the same manner as that for PGA. We include the same additional predictors. Results in 11–12 show that leaders tend to have lower similarity to speeches made by Kofi Annan (1997–2006), Ban Ki-moon (2007–16) and António Guterres (2017–), as they do to the PGA speeches.

We also draw from another, more general approach to text differences, that of classification. We present the results of a structural topic model, in Fig. 5.Footnote 32

Fig. 5
figure 5

What topics leaders tend to focus on. Note: Results are conditional expectations of topic prevalence given that the speeches are made by PGAs versus all other types of speakers (left subplot); by leaders versus other speakers (right subplot). Based on the results of a structural topic model with 23 topics and structural covariates (details are in the appendix). Topics that make PGAs’ speeches distinct (to the right of a zero-line, left subplot) are additionally marked in black for leaders and non-leaders (in the right sublpot)

Full size image

The left subplot in Fig. 5 validates our assumption that the PGAs address the global agenda and organizational priorities: they are more likely to speak about the UN and its role in global affairs, norms and principles, human rights and climate change (five topics to the right of the zero-line). The right subplot indicates that in contrast to other representatives, leaders tend to cover five distinct topics (right of the zero-line), such as those related to colonialism and racial discrimination, trade and economic reforms, terrorism, international norms and principles, and global cooperation.Footnote 33 Notably, leaders and PGAs have only one topic in common: that of norms and principles (associated with lofty rhetoric and such terms as “concept”, “true”, “ideolog-”, “moral-”, “digniti-”). Certainly, some leaders engage with the global agenda, but their engagement, on average, is much lower than that of other delegates. Indeed, and in contrast, nonleaders “share” with PGAs all four other topics that make the PGAs’ speeches distinct, which gives additional support for our argument that leaders tend to deviate from the global agenda, as set by PGAs.

For additional validation, Table 3 includes the results based on the same model specifications as in Columns 1–5 of Table 2, except for alternative dependent variables. First, we estimate the prevalence of the most spoken topic in a given session in leader’s speech, in Columns 1–5. Second, we estimate the average prevalence of the top three ranked topics per session, in columns 6–10. We find the top session’s topic by calculating the average gammas of each topic, out of 23 in total, in a given session, including all speakers, and then we rank the resulting values from the highest to the lowest. For example, in 1979, the year of the invasion of Afghanistan, “Self-determination and sovereignty” was ranked as number one topic (with the average proportion of 0.16) that year, while following the war in Iraq in 1991 the most spoken topic was “UN and security cooperation” (0.16) and in 2015, the year of Paris Accords, it was “Climate change and MDGs” (0.25). Results displayed in Table 3 underline that political leaders, whether democratic or nondemocratic, tend to engage much less with topics that occupy the most significant attention among other delegates in a given session, and the results are statistically significant across all specifications. Speakers from democracies, except during the Cold War, are more engaged with the most prevalent topics during each session, as generally expected.

Table 3 Most prevalent topics per session and leaders’ speech
Full size table

4.1 The quality of communication

The results above serve as a bridge for a closer examination of how leaders speak. Any observed differences also serve as proof of the agency that leaders exercise over their own speeches. If national addresses to the UN are drafted by the same foreign service officials to reflect the underlying country position on issues — irrespective of whether these addresses are delivered by national leaders or other types of speakers — we should not notice any difference in terms of observed text across different speakers. Leaders tend to deviate from the global agenda (H1)— more so than other speakers— and their presence also diminishes the quality of the overall debate, as proposed by (H2), and as we show below. Due to space constraints, we keep the present discussion succinct and relegate further details to supplementary appendix.

To analyse leaders’ quality of speech, we account for its level of sophistication, its use of diplomatic terminology, and its self-centeredness (H2). We fit the same specifications as in Table 2, this time measuring text style scores as the dependent variables (Table 4).

Table 4 The quality of leaders’ communication
Full size table

First we compare whether leaders’ speech is less sophisticated than that of other speakers. We rely on the Flesch-Kincaid readability test where higher scores stand to a higher sophistication. The Flesch reading-ease test measures the difficulty of a particular text (Flesch, 1948).

As can be seen from Table 4 (columns 1–3), leaders’ speech is much more ‘easy’ to understand and listen to, because of its relative lack of complexity.

We next turn to an even more direct test of speech complexity and examine whether leaders are less likely to use specific diplomatic terms. Leaders may be less likely to adopt the prevailing rhetorical styles of diplomacy, speaking less like artful negotiators and more like politicians. Specifically, we estimate the share of terms that belong to a diplomatic dictionary (such as bona fide, aide memoire or nota verbale, as described in the appendix) to all terms in speech. As also evidenced by results from columns 4–6 in Table 4, leaders indeed deploy a much lower percentage of diplomatic terminology than nonleaders.

Additionally, many leaders are known to claim credit and bolster their standing through their frequent usage of personal pronouns in speech (Liu, 2022). Do they employ such rhetoric in the international arena as well? We extract personal pronouns such as “I”, “my”, “me” and “myself” using a named-entity recognition (NER) process, which tokenizes all words in a text and classifies them into parts of speech, while seeking to identify named entities into categories. We then estimate the percentage of personal pronouns among all terms in each text. Higher values indicate a greater degree of self-centeredness in a given text.Footnote 34 The last three columns in Table 4 reveal that leaders are indeed more likely to speak from a more personal perspective using personal pronouns in their addresses to the UNGA—by over a half of one standard deviation of this variable—than any other country representatives.

Across all three measures, we find that democratic and nondemocratic leaders alike tend to display marked differences in speech, in contrast to other delegates. In turn, representatives from the more economically developed nations also tend to use plainer language (Column 3), as well as less self-centred language (Column 9).

In summary, we have shown that leaders are more likely to speak out of step with the assembly, whether measured through their propensity to engage with the global agenda or their dissimilarity to the PGA. Consistent with our hypotheses regarding leaders’ differences from other delegates, they are also more likely to speak as politicians do, paying more attention to themselves in their own speech, using generalized language that is free of diplomatic jargon and targeted at a more basic level than what is common among diplomats and MFAs. This suggests that even though the UN may indeed benefit from various initiatives to revitalise its work, bringing in more leaders, as regards the debate at least, is unlikely to have a positive effect on the overall discursive environment.

The supplementary appendix includes further details about the text data and methods, dictionary terms, different model specifications with additional explanatory variables, alternative dependent variables including those based on specific themes centred on global concerns, pooled and random effects specifications, as well as more details regarding the structural topic model. The additional analyses support the results included in the paper.

5 Implications and conclusion

The UNGA used to be the domain of professional diplomats, but leaders increasingly take the stage. This paper has shown that leaders speak in particularistic ways, deviating from the broader UN policy agenda. If leaders tend to pivot to IOs to advance their own political agendas, then their speeches may detract from the overall vitality of the UNGA. Although direct speech can target a broader class of constituents and ‘democratize’ communication (Spirling, 2016), in the context of multilateral negotiation, plainer speech may also gloss over the complexities of cooperation and undermine discourse. Leaders’ presence may stymie the Assembly’s search for solutions to global challenges, and also impact public perceptions of the institution’s legitimacy: although heads of potentially generate publicity for the UNGA, drawing more attention to the UN, such attention might backfire, turning the debate into “a late-summer version of Davos,” as one observer put it.Footnote 35 As Pouliot (2016, 30–31) describes, “the most visible form of multilateral diplomacy today is that of summitry. … These high-level meetings easily catch headlines when they happen, even though their outcomes usually rest on lower-key and often-informal transgovernmental interactions.” In other words, when leaders steal the spotlight, it can crowd out room for genuine debate, which could stall negotiations. As a result, leaders’ engagement in debate, and their focus on their own, more particularistic, agendas could lead to more centrifugal positions voiced at the UNGA.

These findings shed light on how vitality can vary within agencies in a given IO, but with implications for the broader organization. Particularly when the entity in need of revitalization is its most public-facing and representative one, individuals play an outsized role in the organization’s operation. Although many studies of individuals in IO vitality have focused on the secretary generals or executive heads of those organizations (da Conceicao-Heldt, 2018; Hall & Woods, 2018), this paper suggests that delegates themselves play an important role in IO vitality. As one observer noted, “the spotlight of publicity places undue pressure on a state not on its ability to adjust to a situation, but on its ability to preserve its position” (Hovet, 1963, 32); this implies that any added publicity that a leader brings will entrench country positions rather than encourage consensus. Our evidence also suggests that the politicization brought by leaders may indeed undermine the UNGA debate, adding another layer to research on IO life cycles.

Although this paper focuses on the UNGA, the argument could extend to other IOs, particularly given an increase in the number of IOs that adopt parliamentary-style assemblies. We might expect similarly diffuse policy benefits, and greater prestige benefits, in broad and general-purpose IOs, but smaller, more technical IOs might have more direct policy impacts; leader visits to those IOs might be more directly linked to policy than to personal standing, and their speech may reflect this by being more substantive and more in keeping with the agenda of the particular meeting. To that end, further research could also explore the impact of messages delivered by other types of delegates at IOs (Kaya & Schofield, 2020), even those that constitute the “third UN” of civil-society actors (Weiss et al., 2009).

Future work could also look at whether the rhetoric translates into changes in outcomes. Some scholars argue that heads of state play only a symbolic role in international politics. In this view, political appearances and speeches merely act as manifestations of state interests that find more meaningful expression in votes and in actual policymaking. A full investigation of the effects of leader speech — in terms of media coverage, changes in public opinion both at home and abroad, or other outcomes — is beyond the scope of this paper. But further research could look at whether alliances, trade or aid deals, voting realignments, cosponsorship of UN agenda items, or informal agreements increase as a result of leader visits. Although some find that public speech has less of an effect on public opinion than is often assumed (Simon & Ostrom, 1989), this could be tested in a UN setting as well.

If leaders bring their own politics to the debate, the public perceptions of the efficacy and objectivity of the UN may alter. One view of international cooperation suggests that IOs should be relatively insulated from domestic politics, so as to better facilitate efficient outcomes for cooperation. The early days of the UNGA debate were described as “an international political process of unequaled vigor” (Gordenker, 1962, 525). By contrast, if the concerns of high-profile leaders are the public’s primary imprint of the business of IOs, it may sap confidence in the organization, creating the impression that the UN is a forum for power politics and not for equitable cooperation. Our findings also suggest that leader preferences may be distinct from country preferences overall, which suggests caution in the use of UNGA speeches as a proxy for foreign policy preferences without taking the incentives of the speaker into account.