The constitutional codification of academic freedom over time and space

Abstract

This article explores the global spread of domestic codifications of academic freedom norms by mapping constitutional provisions over time and space. Drawing on the new Academic Freedom in Constitutions dataset, the study evaluates several hypotheses that may explain the norm’s geographically diverse, yet comparatively limited, adoption in 52 per cent of constitutions today. The descriptive analysis of constitutional adoption patterns suggests that the as yet large pockets of absence are a result of the fact that academic freedom was not included as a fundamental right from the early days of constitution-making, combined with its close link to higher education development, thus locking many countries into a path dependency of early constitutions exclusive of academic freedom norms. The availability of relevant models in nearby countries, together with higher education expansion, are key facilitators of academic freedom adoption at the critical time of a constitutional reform process. Diverse countries in different regions acted as norm entrepreneurs, often motivated to domestically protect academic freedom, thus leading to the emergence of regional and other clusters of academic freedom reference types. A sizeable proportion of insincere adopters further suggests that, in some regions, academic freedom serves as a legitimizing international norm.

Introduction

Academic freedom is an inherently liberal ideal, rooted in the Enlightenment notion of individual and collective self-determination and the promise of emancipation through reason and science. ¹ In the late twentieth century, the ‘liberal script’, ² which endorsed a liberal, open and individualistic order of society, ³ spread at a global level, albeit with regional differences. ⁴ During that same period, higher education massively expanded globally, challenging earlier models of society that were fearful of ‘over-educating’ the population. ⁵ These fears were replaced by a developmental model that today strongly promotes higher education as ‘instrumental in fostering growth, reducing poverty, and boosting shared prosperity’. ⁶ That said, in a recent analysis, Schofer, Lerch and Meyer directly link growth in higher education to the ‘liberal world society’ to complement previous economy-focused explanations. ⁷ Altbach further describes the ‘Western idea of the university’ as ‘perhaps the most successful’ of all Western norms in terms of their global influence. ⁸ Seen in this light, one might expect that the liberal science script ⁹ and academic freedom norms took hold in societies around the world alongside other norms of the liberal script such as the freedom of expression or democracy – if not in practice, then at least at a declaratory level. However, only roughly half of today’s constitutions worldwide address the issue of academic freedom, compared with a near-universal embrace of democracy or free press/expression and association in constitutional texts (see Figure 1). ¹⁰

Figure 1. Share of constitutions in force worldwide that include academic freedom provisions compared to other norms.

Sources: academic freedom: Academic Freedom in Constitutions; others: Comparative Constitutions Project.

This comparison of constitutional uptake is very significant as constitutions are the ‘natural place’ for states’ symbolic accession to international norms. ¹¹ They reflect the values and political principles that serve as orientation to a nation state (or that it purports to follow). Boli-Bennett and Meyer argue that ‘national constitutions … reflect legitimating ideas dominant in the world system at the time of their creation’. ¹² Although nearly half the world’s countries have constitutions that are silent on academic freedom, thus not suggesting that it is a really ‘dominant’ idea at the world stage, it is also true that more than half of all countries do reference it. The fact that such countries can be found across all world regions indicates relatively intricate processes of diffusion, and raises the question of what can explain certain states adopting the constitutional norm while others have not – and thus what may overall explain the limited diffusion compared to other liberal norms. The general explanations of ‘global cultural models’ that can largely explain the expansion of constitutional human rights provisions ¹³ is apparently not sufficient when it comes to academic freedom – potentially a reflection of the persistence of alternative, illiberal science scripts. ¹⁴ However, so far none of the global comparative analyses of constitutional rights have looked into the diffusion of academic freedom norms specifically, in part because the existing datasets have so far not adequately captured the concept. ¹⁵

This article seeks to explain the puzzle of comparatively limited, yet geographically outspread academic freedom codification by exploring a new data source. The Academic Freedom in Constitutions (AFC) dataset covers 203 independent countries from 1789 to 2022. ¹⁶ It distinguishes between academic freedom references along three main (semantic) ¹⁷ categories: ‘academic freedom’ (matching the exact phrase), ‘scientific freedom’ (in a broad sense, also including freedom of higher education teaching and similar) and ‘university autonomy’ (in a broad sense, also including independence of higher education institutions and similar). ¹⁸ Given the importance of linguistic nuance, as much as possible coding decisions were taken based on original language texts. ¹⁹ The dataset’s distinction between the three categories is particularly useful for the purpose of this research as these categories allow tracing of different wording traditions between constitutions of different countries, while they can at the same time be summarized into the existence (or not) of any academic freedom provision where the semantic details are not essential. ²⁰

The study seeks to contribute specifically to two sets of literature. By investigating at a global comparative level what factors may explain variations in adoption patterns today and over time, we can learn more about constitutional diffusion processes in general, thus enriching the research in comparative constitutional law with the systematic study of a so far neglected aspect. ²¹ At the same time, the analysis of constitutional diffusion patterns also benefits the growing literature on academic freedom, particularly this special issue, by deepening the understanding of its genesis and significance as an international norm. It should be noted that the focus of the analysis is explicitly on the constitutional codification of academic freedom as a means to explore its global diffusion, and thus not on the de jure protection of academic freedom at legislative or subnational levels, nor on its de facto protection on the ground. Indeed, academic freedom is often respected without being constitutionally guaranteed, and it is often violated despite such constitutional protections, as I will show later in the article.

Drawing on the comparative constitutional literature, I propose five hypotheses that may, together, explain the inclusion and omission of academic freedom provisions in national constitutions across time. The empirical analysis will attempt, in an exploratory and mainly descriptive fashion, to gauge their respective plausibility and relative importance in explaining global patterns of academic freedom codification.

Theoretical expectations

Constitutional provisions can only be introduced as part of a formal process where an existing constitution is amended or a new constitution is drafted. We are therefore interested in finding out what may explain why constitution-makers, at the time of a constitutional reform process, decide to include provisions on academic freedom, decide to exclude such provisions or simply do not consider them. There are two types of explanations for the inclusion or exclusion of academic freedom norms at the time of a constitutional reform process. First, there are contextual factors that make constitution-makers more or less likely to even consider such provisions. Second, constitution-makers can have specific motivations that make them more prone to including academic freedom, whether or not they are supported by an otherwise favourable context.

Contextual factors

Model constitutions

Much of the literature on comparative constitutionalism is at least to some degree interested in the role of international diffusion in constitutional design choices. ²² One such mechanism of diffusion is external coercion, which relates primarily to the situation of countries emerging from colonial rule or wartime occupation. Under such circumstances, the colonial or occupying power often interferes in the constitution-making process or reserves the right to veto the outcome, thus exerting direct or indirect influence on the constitution’s content. There are a few other circumstances under which a country might be incentivized or threatened under military or economic force to adopt certain constitutional provisions. ²³ At the same time, coercion is associated with costs for the enforcing state, meaning that a ‘dominant state’s interest must be strong enough to make worthwhile the expense’. ²⁴ In the context of our research interest, it seems rather questionable that in the wake of war or postcolonial independence, academic freedom specifically would be very high on the dominant state’s agenda. The hypothesis of direct coercion is therefore not very convincing. That said, it still seems plausible that the influence of the colonial/occupying state would direct dependent states towards emulating their constitutional model and adopting similar provisions on rights and freedoms.

This idea aligns with a second mechanism of constitutional diffusion, which other authors have described as learning processes, asserting that constitution-drafters regularly consult foreign constitutions to inspire their own design choices. The ways in which such learning is conceptualized range from hyper-rational ‘Bayesian learning’, in which constitution-makers take in ‘new data’ to revise their constitutional approaches accordingly, ²⁵ to a rather mindless copying exercise of provisions found elsewhere. In most cases, the reality is likely located somewhere between the two extremes. Some analysts argue that it is useful to conceive of constitutions as contracts, as this approach draws our attention to the fact that, similarly to private contracts, constitutions typically are drafted by legal experts, who are likely to use model documents that guide the text’s organization and the subjects covered, rather than to start completely from scratch. Such ‘boilerplate provisions’ also come with the advantage of lowering the transaction costs of negotiation. ²⁶ Bills of rights sections of constitutions appear particularly prone to such modelling exercises given that they are overall more universal in nature than the specifics of government organization.

Such diffusion processes of learning and imitation do not occur at random, however. Instead, they are more likely between countries that are geographically and culturally close, that share a particular legal tradition or common history, or that interact regularly at the international level. Membership in international institutions may generally facilitate the circulation of norms as such organizations can serve both as arenas of socialization between states and as ‘teachers of norms’; ²⁷ however, as Börzel and Spannagel show in their contribution to this special issue, a top-down norm diffusion seems rather unlikely in the case of academic freedom. Legal systems are commonly regarded as important channels of the diffusion of legal rules, given the easier fit and familiarity within common or civil law systems and the high interaction through the legal networks attached to some international organizations, such the Organisation Internationale de la Francophonie. ²⁸ In their study of constitutional rights diffusion, Goderis and Versteeg find strong empirical evidence that rights diffuse mostly between countries that share the same legal origins, the same religion, or a common colonizer or aid donor. ²⁹ Similarly, Law and Versteeg show that two distinct models of constitutional design have developed over time, distinguishing liberal constitutions (statistically associated with common law traditions) from statist ones (associated with civil law traditions). ³⁰ An important note concerns the fact that where hierarchical learning or imitation takes place, it is likely to occur top-down rather than bottom-up, meaning that less-powerful states frequently take on constitutional provisions from more powerful states, but seldom the other way around. ³¹

In sum, as hypothesis (1), we can therefore expect that constitution-drafters are more likely to include academic freedom provisions if such norms are present in constitutions that they turn to as models, and less likely if they are absent from those models. In particular, we expect such constitutional diffusion via learning and imitation to take place at the time of decolonization, as well as where countries are geographically, culturally or legally close to one another. Moreover, where academic freedom norms circulate in this way, wording choices are likely to be similar and lead to the emergence of geographic and other clusters of academic freedom provision types (or the lack thereof).

Path dependency

Closely connected to the first hypothesis is path dependency, in the sense that when lawmakers embark on drafting a new constitution, they not only look to other countries’ constitutions for inspiration, but they usually draw also (or even primarily) on their own past texts – if there were any. Overall, constitutional systems are rather slow and reluctant to change. Some constitutional scholars describe constitutions as ‘coordination devices’, meaning that they serve to politically and socially organize a society. According to this theory, a successful constitutional order depends on coordination through a capable government and the acquiescence of the population. While the coordination idea mainly relates to the system of government and seems less crucial with regard to academic freedom guarantees, an important lesson from this research is that there typically is a ‘high cost of formal constitutional change’. ³² Galligan and Versteeg highlight that, given those costs, ‘when countries are faced with an opportunity to change their constitution … they often maintain the old constitutional arrangements’. ³³ We can therefore expect, as hypothesis (2), that countries with entrenched constitutional traditions are relatively unlikely to change them substantially over time, and that this also applies to the presence or absence of academic freedom provisions.

Salience of higher education issues

Unlike most other universal human rights and freedoms, academic freedom pertains to a specific context, namely that of higher education research and teaching. Where no university, research facility or other higher education institution exists, it makes little sense to expect that countries include academic freedom provisions into their constitutions. Conversely, as the higher education sector grows, there is more cause – and potentially demand – for the constitutional guarantee of academic freedom. Constitutional texts are written in a ‘social context’ ³⁴ of a country, at a particular time when a given issue might be more or less salient. The inclusion of academic freedom provisions in a constitution may thus generally reflect their increased salience as a result of the development of a country’s higher education sector. Such ‘deeper forces’ of a ‘technological, demographic, economic’ nature can also drive constitutional diffusion at a global scale, a process that Dixon and Posner connect to ‘superstructure theories’. ³⁵ The worldwide expansion of higher education is thus likely to lead to an increased diffusion of academic freedom provisions at the global level.

Connected to the salience of higher education is a further diffusion argument that relates to constitutional competition between peers for material benefits. Typically, this concerns provisions that are thought to facilitate the free flow of capital, such as the security of private property and basic human rights. ³⁶ In the context of higher education, one could argue that where countries compete for the minds of talented scholars and students, they may try to attract them to their country (or persuade them to stay) by guaranteeing academic freedom. However, it is unclear that constitutional guarantees specifically would really play an important role in the international competition for academic labour, especially when compared with the eminent role of more tailored instruments such as university rankings (which do not factor in academic freedom or its constitutional guarantee). ³⁷ While the argument is not overly convincing, it would further strengthen the importance of higher education salience for the constitutional codification of academic freedom.

Hypothesis (3) therefore predicts that a higher level of higher education development at the time of a constitution’s adoption makes the inclusion of academic freedom provisions more likely. Countries with close to no higher education activity, on the other hand, are less likely to have any such provisions.

These three contextual factors – model constitutions, path dependency and higher education salience – can be seen as alternative hypotheses, but in reality they are at least partly interrelated. For instance, as countries develop their higher education systems, they may turn to other countries for models on legislative and constitutional frameworks, or they may begin to participate in international conferences on science and higher education that could serve as arenas of norm diffusion.

Motivational factors

The hypotheses considered above describe contextual factors that account for why academic freedom provisions may or may not come up for consideration during a constitution-drafting process. In addition, it is useful to consider motivational factors that can explain why constitution-makers specifically decide to include those provisions in their text. These factors should be regarded as closely intertwined with the contextual factors, rather than separate from them – for example, the increased salience of higher education can motivate law-makers to make the constitution responsive to this changing context. However, there are two types of incentives or motivations that appear particularly relevant for the inclusion of academic freedom provisions so that they deserve special attention. They both relate to the conception of constitutional texts as expressions of values: ³⁸ first, such an expression can be inward-facing to affirm certain values as part of a national identity; and second, it can be outward-facing to gain international acceptance through formal adherence to an international norm.

Affirmation of national values

A domestically oriented motivation to add academic freedom provisions often connects to the occurrence of regime change. New constitutions are often written at the time of a national crisis or otherwise unusual circumstances. ³⁹ The most significant type of crisis in view of the adoption of new constitutional rights is a revolution or regime collapse. Where countries liberalize their political systems and embrace human rights and freedoms, it seems plausible to expect that they will also liberalize their science scripts and grant greater freedom to academia to decide on their own priorities and methods. ⁴⁰ During this type of liberalizing constitutional reform, constitution-makers have particular incentives to try to remedy the problems of past regimes while producing an aspirational, forward-looking mission statement for the future. ⁴¹

We can thus formulate as hypothesis (4) that countries with a recent history of academic freedom suppression that are in the process of democratization have strong aspirational motivations towards enshrining academic freedom provisions in their constitutions. That said, an inward-facing norm expression may also apply to countries with a history of respect for academic freedom that decide at some point to affirm it as part of their constitutional identity and break away from the path dependency of their constitutional order, although the timing at which such changes may occur is less clear.

Pursuit of international legitimacy

The outward-facing expression of values has variably been termed ‘acculturation’ ⁴² in the comparative constitutional literature or ‘emulation’ ⁴³ by diffusion researchers, but they essentially refer to the same phenomenon: when constitution-makers ‘emulate foreign models to obtain social rewards, even when there are no apparent material benefits and they are not persuaded by the content of these models’. ⁴⁴ Once certain norms become a standardized part of global culture, they carry legitimacy. States are therefore incentivized to formally subscribe to such norms in order to increase their own international recognition, whether or not they intend to implement the norms in reality. ⁴⁵

We can therefore consider as hypothesis (5) that as academic freedom provisions become transnationally more prominent over time, states will overall be more incentivized to include them in their own constitutions to externally signal their adherence. While this gravitational pull applies to all states, it will be empirically most visible in cases for which academic freedom adoptions seem otherwise counter-intuitive: those with no or little higher education structures and those that follow an illiberal science script and thus fall outside the category of states that (aspire to) really protect academic freedom.

In the following sections, the AFC dataset will serve to empirically explore these five hypotheses. It should be noted that without detailed case studies of each constitutional drafting process, the exact motives and circumstances behind individual adoptions are hard or even impossible to determine. However, the dataset allows us to analyse these issues at a macro level, where certain patterns (or their absence) can substantiate different hypotheses and their ability to explain academic freedom codifications over time and space. Insofar as the available material allows, I will use individual country cases to provide additional illustrations or to investigate exceptions to the broader patterns.

The empirical analysis is split into three major sections, with the first two considering contextual and the last considering motivational factors of academic freedom inclusion. To address the first two hypotheses (learning/imitation and path dependency), I will examine historic and geographic patterns and genealogies in the adoption of academic freedom provisions at the global and regional levels. Next, I will link academic freedom adoption patterns to data on higher education expansion to consider the relevance of the third contextual factor (salience). In the last section, I will compare patterns of academic freedom codification in relation to de facto academic freedom levels around the time of adoption, using the Academic Freedom Index time-series data. By distinguishing aspirational, confirmatory and insincere adopters, this analysis enables a consideration of the motivational factors of academic freedom adoption.

Constitutional genealogies: Historic and geographic distributions

Brief overview of academic freedom in national constitutions

We can broadly learn from the dataset that there have so far been 116 countries across the world that have had an academic freedom provision in their constitution, 101 of which are contemporary countries that still have such a provision as of 2022. An overview of all countries with academic freedom provisions, organized by world region ⁴⁶ and sorted by date of first adoption, is provided in Figure 2. ⁴⁷ This timeline chart shows for each country the point in time when it adopted or deleted constitutional academic freedom protections, with distinction between provisions referencing ‘academic freedom’ (yellow), the freedom of science or higher education teaching (blue), or the autonomy of universities (black line). Green indicates constitutions that include both the term ‘academic freedom’ and other provisions coded as freedom of science. Constitutions that were suspended in a given year are shown as not including any provision.

Figure 2. Wordings of countries’ academic freedom provisions over time, by world region.

Sources: provisions: AFC, first constitutions: CCP

We see that, in the nineteenth century, any references to academic freedom are sporadic. The first such mention appears in Germany in the Frankfurt Constitution (Paulskirchenverfassung) of 1848–49, which was born out of a liberal-democratic revolutionary movement seeking to establish a German state. The phrase it uses is ‘Science and its teaching is free’ – a precedent to which most constitutions that later refer to the ‘freedom of science,’ ‘scientific freedom’ or similar can ultimately be traced back. The constitutional concept of autonomy or self-governance of universities emerged only in the twentieth century. Intriguingly, it first appeared almost simultaneously in national constitutions of Uruguay (1918), Finland (1919), Estonia (1920) and Georgia (1921), although they all put it in very different words. Provisions on university autonomy became particularly prominent in Latin America, where almost every constitution guarantees university autonomy today. Following the end of the Cold War, many post-Communist countries also included such provisions in their constitution. ‘Academic freedom’ as a term, on the other hand, first appeared another decade later in the Spanish constitution of 1931 (‘libertad de cátedra’). ⁴⁸ Outside the hispanophone world, the first specific references to ‘academic freedom’ came only in the 1940s, in the Filipino and Japanese constitutions. The phrase later became prominently represented in African constitutions, starting with Liberia in 1986.

Table 1 summarizes, for each world region, the share of countries with constitutional academic freedom provisions today and across time, as well as the dominant type of references used. From this, we can already glean that there are certain provision types that are more common in some regions than in others, pointing to the role of model constitutions. In the following, I will consider this hypothesis in more depth.

Table 1. Share and type of academic freedom references in constitutions by world region

Constitutional learning: The role of model constitutions

I formulated the hypothesis of constitutional learning as a contextual factor because it is directly dependent on the availability of academic freedom provisions in relevant model constitutions. In the following, I will first consider the role of constitutional models in relation to former colonial or occupying powers, and then more broadly by world region.

Former colonial/occupying powers

Of the major colonial (or occupying) powers of the nineteenth and twentieth centuries – such as Spain, Portugal, the United Kingdom, France, the Netherlands, Italy or the United States – none had any academic freedom provision in its own constitution at the time it ceded independence to its colonies. The only exception is Italy, which had enshrined the autonomy of universities and freedom of science in its constitution by the time its last colony, Somalia, gained independence from its trusteeship in 1960. Indeed, we find that the Somali independence constitution guarantees the autonomy of universities – in a formulation that is nearly identical to the Italian constitution of 1949. The reverse conclusion also seems to largely hold true: the vast majority of independence constitutions in Latin America (nineteenth century) and in the Caribbean, North Africa, Sub-Saharan Africa and Asia (twentieth century) did not include any academic freedom reference.

Research on former French colonies shows that, for inspiration on their constitutional rights, many constitution-drafters looked at the French Declaration of the Rights of Man and of the Citizen of 1789, ⁴⁹ which does not include academic freedom. The British colonial office did at first not consider the idea of constitutional rights for its colonies seeking independence, but eventually supported and propagated the Nigerian model of a bill of rights that drew directly on the European Convention on Human Rights of 1950, ⁵⁰ which also lacks an academic freedom clause. The influence of these models thus account for the general lack of academic freedom provisions in post-independence Africa, Asia and the Caribbean – regions where adoption rates remain generally low until today. Few notable exceptions concern several independence constitutions in the Middle East, which appear to have defied the British legacy and instead sought models in their own region (see below).

Two further noteworthy cases are those of the Filipino and Japanese constitutions of 1935 and 1946, which were both drafted under US supervision. Although the concept is absent from the US Constitution, both of these constitutions included a provision on ‘academic freedom’ at a time when none of their Asian neighbours did. The Philippines’ constitution of 1935 (which eventually became operative at independence in 1946) was drafted by Filipino constitution-makers under US colonial rule. It included the academic freedom provision not in the bill of rights section, which largely drew on the US constitution, ⁵¹ but under ‘General Provisions’ in the context of educational stipulations. It is possible that Filipino law-makers also consulted the 1931 constitution of their former colonizer Spain, which included a provision on the ‘libertad de cátedra’. Reports suggest, however, that they were particularly well versed in US constitutional law, and some had in fact studied at US universities. ⁵² The use of the ‘academic freedom’ term specifically had a long tradition in US higher education by that time (less so in Europe). ⁵³ The idea of ‘academic freedom’ was thus likely influenced by the US context – but actively included by local scholars when designing their own constitution.

The case of Japan in 1946 is a very different one. It is broadly accurate to say that US occupants under the leadership of General MacArthur drafted and imposed Japan’s post-war constitution, although the details are murkier. Several analysts remark upon the fact that the constitution includes a guarantee of academic freedom despite its absence in the US text, ⁵⁴ but its actual origins are unclear. The initial US draft included a provision that read ‘Freedom of academic teaching and study, and lawful research are guaranteed to all adults. Any teacher who misuses his academic freedom and authority shall be subject to …’, which does not bear resemblance to any other constitutional text in force at the time. Over the course of February 1946, the provision was edited at least twice to eventually read ‘Academic freedom and choice of occupation are guaranteed’ in the version submitted to the Japanese government. ⁵⁵ The term ‘academic freedom’ therefore seems to be a clear imprint of US terminology. The idea of including this topic, however, may have come from other constitutions that were consulted, or from drafts proposed by Japanese academics and civil society, although none of those submitted before February 1946 used a formulation close to the initial or final American one. ⁵⁶

Western Europe & North America

Constitutional learning and imitation effects were present – but very limited – in Western Europe, where the initial German provision ‘Science and its teaching is free’ was taken up in identical form in Austria two decades later. Similarly, the updated formulation of the Weimar constitution of 1919 can be found in the Greek constitution of 1925 (‘Art, science and their teaching are free’). In 1975, Greece drew again on the German Basic Law for its own constitution and also added the sentence that was included there after World War II (1949): ‘freedom of teaching shall not exempt anyone from his duty of allegiance to the Constitution’ – a phrase that is otherwise unique in the world. Meanwhile Spain had developed its own model of an ‘academic freedom’ (libertad de cátedra) provision in 1931, which likely influenced constitution-makers in the New World but not in Europe. Starting with Finland in 1919, various Western European countries also came to embrace constitutional references to the autonomy of universities. The limited overlap and diffusion within Europe match other findings on European constitutionalism. Elkins notes that ‘the lack of a single successful model in the early years of constitutionalization led to increased experimentation and the development of subregional clusters of constitutionalism’. ⁵⁷

Latin America

Latin America (excluding the Caribbean, see above) provides a very compelling case of regional diffusion, although the constitutional manifestations of academic freedom and university autonomy may partly have been indirect effects from general policy changes and not only a direct result of the consultation of foreign constitutions. ⁵⁸ Its academic freedom adoption wave in the 1930s and 1940s has its roots in the higher education reform movement that is often dubbed the Córdoba movement. Led by students in countries across the subcontinent, the movement sought to democratize universities, to make them more accessible to the wider society and to improve student welfare. ⁵⁹ The movement had very significant and lasting effects on Latin American higher education, ⁶⁰ and inspired the gradual constitutional codification of academic freedom in countries across the region. The prominence of university autonomy clauses, included in almost every Latin American constitution today, can at least partly be attributed to the influence on the movement of Uruguay ⁶¹ – the country that had pioneered such a reference as early as 1918. It appears further likely that some Latin American constitution-makers took inspiration from the Spanish constitution of 1931 when including the ‘libertad de cátedra’ in their own drafts; the term first appears in Peru in 1933 and in multiple other countries thereafter, although such provisions have remained less prominent than the very typical autonomy clauses.

Eastern Europe and Central Asia

In this second region with near-universal adoption rates, the beginnings of the constitutional history of academic freedom provisions date back to the aftermath of World War I. Four of the constitutions adopted at that time included academic freedom provisions (Estonia 1920, Georgia, Poland and Yugoslavia 1921). They were differently worded, although in combining the freedom of art and science, the Estonian, Georgian and Yugoslavian versions are all reminiscent of Germany’s Weimar constitution of 1919. Yugoslavia then adopted a modified clause in its 1963 constitution (‘scientific and artistic creativity shall be unrestricted’), which likely influenced the relevant provision ⁶² in the Soviet constitution adopted in 1977 and the near-identical texts passed across the fifteen member states of the union in 1978. ⁶³ It appears that the phrase ‘scientific creativity’ or ‘creation,’ which was subsequently adopted in many socialist-inspired constitutions ⁶⁴ around the world, can indeed be traced back to this model. Moreover, the uniform Soviet constitutions also help to explain the serial adoption of academic freedom guarantees in the region in the early 1990s, which almost all refer to the freedom of science, and many of which actually still include the specific reference to the ‘freedom of artistic, scientific, and technical creativity’ or similar.

However, about half of those 1990s constitutions in the region also incorporated an (additional) reference to the autonomy of universities, which had not been included in the Soviet or Yugoslavian models. The only two prior examples of such clauses in the region were Estonia (1920–40) and Georgia (1921). Judging from the relatively uniform or even identical wording choices across different countries (‘the autonomy of universities is guaranteed’), it seems a likely result of the circulation of model draft constitutions. Law-makers in the region were at the time supported by various overlapping networks of transnational jurists. ⁶⁵ The earliest draft text containing the autonomy of universities that I could find was put forward by the Slovenian Writers’ Association in 1988; however, it sounded rather different: ‘universities and other higher education institutions determine their internal organization autonomously within the limits of the law’. ⁶⁶ Instead, it may ultimately have been the Estonian text of 1921 that served as an example to post-communist constitution-makers, as it similarly stated that ‘higher education institutions are guaranteed autonomy’. It may further be interesting to some readers that the original Russian draft constitution proposed by Boris Yeltsin in 1990 did also include such a clause on university autonomy, which was later dropped in the constitution adopted in 1993.

Middle East and North Africa

Another concrete case of regional state-to-state diffusion is the case of the Middle East and North Africa (MENA), where in the 1960s and early 1970s a series of states adopted academic freedom provisions in practically identical wording: starting with Egypt in 1956, almost all states in the region that adopt relevant provisions in subsequent years use a slight variation ⁶⁷ of the formulation ‘freedom of opinion and scientific research are safeguarded’, namely North Yemen (1962), Kuwait (1962), Iraq (1963) and Bahrain (1973) – some, but not all, of which are the countries’ first constitutions. The only divergences in wording are Turkey (1961), which follows an altogether different constitutional tradition, ⁶⁸ and South Yemen (1970), which was influenced by the socialist constitutions of Eastern bloc countries. In contrast, the constitutions adopted by North African countries emerging from French colonial rule at that time (Tunisia 1959, Morocco 1962, Algeria 1963) did not include any academic freedom clauses, nor did that of Qatar (1970) or the UAE (1971), although Qatar adopted a formulation reminiscent of the Egyptian one in 2004. Algeria’s constitution of 1976 drew on the socialist-inspired formulation. ⁶⁹ The region’s recently adopted constitutions of 2011 (Jordan, Morocco, Libya), 2012 (Syria) and 2014 (Tunisia) do also include academic freedom, but vary greatly in their specific formulations. It is generally interesting to note that most of these countries in the MENA region had very low records on academic freedom at the time they included those provisions in their constitutions – pointing to cases of insincere emulation of norms, something to which I will return below.

Asia and the Pacific

The Asia Pacific region appears to be one with very little constitutional cross-fertilization. I already discussed the peculiar cases of the Philippines’ and Japan’s constitution drafted under US influence above. Moreover, we find several communist- or socialist-inspired constitutions that include similar academic freedom provisions in China, North Korea and (historically) Vietnam. A curious case is that of Thailand, where a tumultuous constitutional history produced numerous texts, variably with or without references to university autonomy or academic freedom. I could not ascertain how the first provision on university autonomy was introduced in 1949, except that Thai higher education was at the time already quite well developed and to some degree Western-influenced, ⁷⁰ which would speak more directly to the salience hypothesis (3) than that of constitutional learning. The first provision on ‘academic freedom’ was included in the Thai constitution of 1974 after a student-led uprising had overthrown the military regime of 1971–73. Following their revolutionary success, students were closely involved in political life and this was presumably also the case during the constitutional drafting process, ⁷¹ which might have played a role in enshrining ‘academic freedom’ in the new (though also short-lived) text – a case pointing to affirmative motivation.

Sub-Saharan Africa

In contrast to the Asia Pacific region, the constitutions of Sub-Saharan Africa show very strong, albeit diverging, patterns of constitutional academic freedom adoptions. As noted above, apart from Somalia none of the African independence constitutions of the late 1950s through the early 1970s referred to academic freedom at all. One pattern starts to emerge with the three lusophone countries of São Tomé and Principe (1975), Cape Verde (1980) and Guinea-Bissau (1984), which all adopted identical provisions (‘intellectual, artistic and scientific creation are free’), later followed by Mozambique (1990) and Angola (2010). What all these countries had in common were liberation movements that were strongly influenced by socialism, ⁷² which explains the wording choice borrowed from the Soviet model. ⁷³ As a side note, it is curious that Portugal itself, in its constitution of 1976, uses a similarly worded provision. Drafted during an interim period after a leftist military coup against the authoritarian regime of the Estado Novo and in the midst of its process of decolonization, it is unclear whether it was directly inspired by the Soviet model or whether this may be a rare example of ‘bottom-up’ influence from a former colony. In any case, over the following years, several other African countries follow the same model: Ethiopia (1987), Mauritania (1991), Burkina Faso (1991), Comoros (1996), Eritrea (1997) and DRC (2003) were all using socialist-inspired references to the freedom of scientific ‘creativity,’ ‘creation’ or ‘creative activities in science’.

A distinct second model is followed by numerous other countries in the region, starting with Liberia in 1986, using the exact phrase ‘academic freedom’. Outside of hispanophone contexts, this term had rarely been used in prior constitutions and may suggest that – similarly to the cases in Asia – notions from the US higher education tradition informed the wording choice. This is particularly plausible given the strong historic connection between the United States and Liberia, ⁷⁴ paired with US efforts in the 1980s to push back Soviet influences on the African continent. Some of the African countries following this model use a variation of the formula ‘freedom of expression includes academic freedom’ (Liberia 1986, Sierra Leone 1991, Kenya 2010), which is congruent with US jurisprudence that sees academic freedom protected under the freedom of expression. However, more frequent in this group is a different version, whose ideological origins I could not determine: ‘freedom of thought, conscience and belief include academic freedom’ (Namibia 1990, Ghana 1992, South Africa 1993, Malawi 1994, Uganda 1995, Gambia 1996).

The two broader models in the African region cannot be tied just to socialist vs liberal influences, but can also be mapped against differences in their legal system (which may be the result of similar influences): ‘academic freedom’ provisions are representative of a common law influence, while ‘freedom of science’ provisions can mostly be found in civil law systems. As discussed above, the respective legal networks may constitute very important channels of constitutional diffusion and explain the relative homogeneity of formulations within the two groups. That said, we should not overlook the fact that about half of the nations in Sub-Saharan Africa have so far not included any provision relating to higher education freedoms or autonomy in their constitutions, which stresses the fact that there is no single determinant factor of academic freedom inclusion.

In recent years, similarly to the MENA region, the models in the African region have become more blurred: Zimbabwe’s constitution of 2013 includes both ‘academic freedom’ (as part of freedom of expression) and a reference to the freedom of ‘scientific research and creativity’, whereas South Sudan in 2011 uses a new formula, also combining the two. ⁷⁵ Moreover, Madagascar in 2010 only refers to the autonomy of universities.

Summary observations

In sum, we can firstly see that the constitutional models of former colonial powers likely played a very significant role in terms of constitutional academic freedom provisions of independence constitutions – although largely to favor their absence, with very few exceptions.

The clusters of reference types – and often identical wordings – used by countries within the same region further clearly evidence that states are looking to fundamental texts in neighbouring countries for inspiration when drafting their constitution. Apart from the Latin American case, the extent to which these academic freedom provisions were explicitly and deliberately included in this process remains largely unclear, as opposed to being emulated without particular attention as part of lists of rights and freedoms taken up from elsewhere. It should be noted, however, that academic freedom provisions are not always included in the ‘bill of rights’ section of constitutions but often under a section dealing with educational policies, which limits this ‘boilerplate’ type reproduction. Whether explicitly rationalized or not, the mechanism of learning and imitation appears to be a key explanatory factor for why states adopt academic freedom provisions and why we can find regionally contained ‘waves’ of the concept’s constitutional uptake.

It is important to acknowledge that just because countries use the same constitutional wording doesn’t mean that they apply the same interpretation in practice. ⁷⁶ That said, it could still be quite interesting to investigate further the extent to which the different wording models may correspond to substantial differences in higher education design and practice, such as between the use of US-inspired ‘academic freedom’ provisions, as opposed to provisions reflective of a European or socialist wording, and a separate emphasis on university autonomy. ⁷⁷

However, at the constitutional level we can also observe that the formerly distinct models of either referencing the ‘freedom of science’ or ‘academic freedom’ have started to blur in recent years. Outside of Latin America, where ‘libertad de cátedra’ has always been used alongside freedom of science provisions, for a long time only Greece (1975) had a constitution including both phrases, and then South Africa in 1996. Since the 2000s, these instances have multiplied: Armenia (2005), Sudan (2005), Kenya (2010), South Sudan (2011), Fiji (2013), Zimbabwe (2013), Tunisia (2014) and Algeria (2016). This development is probably a reflection of the increasing availability of constitutional texts and may indicate a decline in significance of particular regional or ideological models – and thus actually point to a decrease in the polarization of legal systems that Law and Versteeg diagnosed in their analysis. ⁷⁸

Path dependency

When considering the global diffusion of academic freedom constitutional provisions, a striking observation is that many of the liberal democracies that endorse the liberal script and practise academic freedom do not have a relevant provision in their constitution. This group includes, for example, the United States, France, Belgium, the Netherlands and Norway. ⁷⁹ What these countries have in common is that they all had formulated their first constitution already in the eighteenth century – except Belgium, whose constitution of 1831 nevertheless came well before the world’s first constitutional academic freedom provision. All these constitutions already had enumerations of rights in them, such as on the freedom of expression or of the press. Moreover, none of these countries has adopted a new constitution over the past five decades.

The fact that academic freedom was not conceived of as a fundamental right from the early days of constitution-making has thus significantly hampered transnational diffusion up to the present day. Most of the countries named above have enshrined some elements of academic freedom in their legislation governing higher education, or the courts have – as in the case of the United States – derived it from existing constitutional provisions. As a consequence, even where constitutions are regularly amended, there has apparently been no sense of urgency for law-makers to include academic freedom in their constitutions, thus locking them in a path dependency of constitutional silence on academic freedom. In turn, as we have seen above, the lack of such provisions in prominent model constitutions has also meant that the diffusion in other world regions has been rather slow compared with more universally established provisions such as the freedom of expression. Colonial legacies have further created a similar path dependency of an absence of academic freedom provisions in many Asian, African and Caribbean constitutions.

However, there are signs that the views of European lawmakers may be changing: Switzerland amended its constitution in 1999 to include the freedom of science (and again in 2006 to add university autonomy); Austria inserted both elements for the first time in its post-war constitution in 2008; Sweden added a provision on scientific freedom in 2010; and Luxembourg has recently adopted a similar reference as part of its constitutional reform that entered into force in July 2023. ⁸⁰ These changes can be read in conjunction with rising contestations of academic freedom in liberal democracies ⁸¹ – which could provide a compelling explanation for such affirmative adoptions in the absence of regime change.

Salience: Linking adoptions to higher education expansion

The third essential contextual factor is that of higher education expansion. Although higher education institutions exist in virtually all countries around the world today, this was not always the case for all independent countries over the course of the nineteenth and twentieth centuries. If we consider the foundation year of the first higher education institution, ⁸² we find that most countries already had at least one such institution in place when they adopted their first constitution. ⁸³ Among those that did not, many added academic freedom provisions later on (27 out of 53), in most cases after a higher education institution had been established. However, often more than 25 years passed between the creation of the first higher education institution and the eventual inclusion of academic freedom provisions, which does not hint at a general causal relationship between the two. Yet there are a handful of countries where the temporal connection is quite close. One is Panama, where one of the co-founders of the University of Panama in 1935, José Dolores Moscone, took part in the drafting process of the 1946 constitution and is reported as having been the driving force behind the inclusion of unusually detailed academic freedom and autonomy provisions. ⁸⁴ This particular example illustrates how the prominence of the higher education sector, and the involvement of domestic legal scholars in the drafting process, can directly shape the constitutional outcome in terms of academic freedom.

The simple existence of higher education institutions does overall not seem a good predictor for the inclusion of academic freedom provisions. However, we would expect that academic freedom as a concern becomes more salient as the higher education system expands to assume a gradually more prominent role in a country’s education system, economy, and cultural and political life. The relative size of the higher education sector can be captured by countries’ tertiary school enrolment ratio. This measure indicates what share of the age group that officially corresponds to tertiary education is actually enrolled in a higher education institution in a given year. ⁸⁵ Figure 3 shows boxplots ⁸⁶ indicating since 1970 the distribution of higher education enrolments for three groups: countries with new constitutions during that time that did not include academic freedom provisions; countries with new constitutions that did include such provisions, and (a subset of the latter) those that did so for the first time. Note that each data point represents one new constitution, meaning that countries that adopted several constitutions since 1970 are included several times in the chart.

Figure 3. Boxplots of tertiary school enrolment ratios in countries with new constitutions since 1970, using decade mean of enrolment ratio per country to improve data availability.

Sources: enrolment data: World Bank, constitutional events: CCP, academic freedom provisions: AFC.

We can see that countries with the highest enrolment ratios (>50 per cent) have without fail included academic freedom provisions in new constitutions since 1970, two of which (Fiji and the Dominican Republic) have recently done so for the very first time. It is further worth noting that nearly ⁸⁷ all ‘no provision’ countries with enrollment ratios higher than 20 per cent have actually had such a provision in prior constitutions (Georgia, Kazakhstan, Philippines, Poland, Thailand) and/or have included one since then (Argentina, Azerbaijan, Dominican Republic, Latvia, Lithuania, Sweden, Tunisia) – some of them via amendments, which are not captured by this chart. For the group that had earlier academic freedom provisions, the non-inclusion cannot be explained by path dependency, nor by a lack of salience. Instead, it appears to be a deliberate choice of omission. When considering instances over the course of the AFC dataset’s time frame when countries completely dropped academic freedom provisions from their constitutions, we find that most fall in the context of authoritarian regression, typically when constitutions are abrogated and replaced after a coup d’état. Even though the number of constitutions that include academic freedom has grown over the past decades, such instances of deletion have actually become less frequent, which points us to the increasing role of outward-facing value-signaling discussed further below.

In terms of higher education salience, the most obvious conclusion from Figure 3 relates to the fact that countries that adopt constitutions without academic freedom provisions tend to have significantly lower tertiary enrolment ratios (2 per cent median enrolment ratio) compared with those that adopt constitutions inclusive of such guarantees (16 per cent median). Very small higher education sectors therefore seem to be a relatively good predictor of non-inclusion, although there are exceptions. Kuwait and Yemen, for instance, both adopted academic freedom provisions in the 1960s before they had established their first higher education institution. In the case of São Tomé and Príncipe, seventeen years passed between the constitutional academic freedom adoption and the first such institution. Such counter-intuitive cases can serve as evidence for the symbolic outward-facing expression of values (hypothesis 5) – what Goderis and Versteeg call ‘structural decoupling’, where countries are acculturated into adopting certain norms into their constitutions that do not make much sense in that context, ⁸⁸ such as adopting academic freedom in a country with little to no higher education structures. Similarly, Simmons et al note that ‘world polity theorists have found that countries embrace new norms for symbolic reasons,’ even when they ‘are not developmentally capable of implementing them’. ⁸⁹ That said, in countries where higher education establishments are in planning or where small structures already exist, constitutional academic freedom provisions may also take an aspirational, inward-facing meaning. In the next and last section, I will take a closer look at these two motivational factors.

Motivational factors: Adoptions and de facto academic freedom levels

A useful empirical approach to the question of motivational factors is to compare the inclusion of academic freedom provisions to countries’ de facto academic freedom levels at and after the time of adoption. While this cannot provide us with a full picture of states’ motivations, it allows us to broadly distinguish three ideal-type groups of countries: (1) those that have been practising academic freedom at a relatively high level before adding it to their constitution (confirmatory adopters); (2) those where the adoption of academic freedom provisions coincides with or shortly precedes a marked increase in academic freedom levels (aspirational adopters); and (3) those where academic freedom is and remains low before and after the constitutional inclusion (insincere adopters). Next to those, we will find countries that are somewhere ‘in between’ these categories or whose academic freedom levels jitter as a function of tumultuous political events that happen around the constitutional reform. Of particular interest to us are the countries that fit in the ideal-type groups, since the first two can roughly be linked to hypothesis (4) of affirmative domestic value expression and the third to hypothesis (5) of international value expression, in their case with no apparent intention of implementation.

Using the Academic Freedom Index (AFI) as an indicator of de facto academic freedom levels, ⁹⁰ Figure 4 shows ten-year trends in the freedom at universities for countries that included academic freedom provisions in their constitution for the first time. Included are all countries for which AFI data is available during the relevant period, which is the case for nearly all twentieth-century first-time adopters. The figure is split into three graphs, the first of which shows countries with significant improvement after adoption (green), those with similar levels after ten years (blue) and those with significant deterioration (red). ⁹¹

Figure 4. Ten-year trends in academic freedom levels for first-time academic freedom adopters. Diamonds represent date of first adoption.

Sources: Academic Freedom Index: V-Dem v13, academic freedom adoptions: AFC.

The top chart serves as a remarkable illustration of how the aspirational promises that countries inscribe in their constitutions do in many cases translate into significant real-world changes in the levels of academic freedom. After World War II, as well as during the so-called third wave of democratization, numerous countries significantly improved their academic freedom levels in the years after constitutional adoption. In fact, most of the countries in the third chart that display severe deterioration, such as Spain (1931) or Chile (1971), also fall in the categories of countries that constitutionally adopted academic freedom provisions as expression of an aspirational ideal, which was then quashed shortly afterwards by a civil war, coup or other authoritarian turn.

Another interesting pattern relates to countries in the second chart, where academic freedom levels remained largely unchanged after constitutional academic freedom adoptions. At the upper end of the AFI scale, we find confirmatory adopters whose constitutions eventually catch up after long periods of high de facto academic freedom (e.g. Switzerland 1999, Sweden 2010), as well as those that adopted their constitution a few years into their democratic transition process at which point academic freedom levels had already improved (e.g. Portugal 1976, various Eastern European and Central Asian countries in 1990s, Argentina 1994 or Tunisia 2014).

However, most intriguing are those countries that adopt academic freedom provisions at a low point of de facto academic freedom, which subsequently remains low – just like the countries that lack a higher education sector, these are counterintuitive cases of academic freedom adoptions that point to the inclusion of such provisions in pursuit of international legitimacy. For them to be used in this way, academic freedom needs to be relatively well accepted as an international or regional norm in the first place. Even though academic freedom norms have been slow to broadly assert themselves as part of a ‘liberal canon’ of constitutional rights, it is fitting that, other than Iran, these types of disingenuous adoptions first appear in the 1930s in Latin America (Brazil, Nicaragua and later Paraguay), where such a norm had indeed emerged by that time. Other cases follow in the late 1940s, when academic freedom adoptions were globally on the rise (see Figure 1). After that, it is striking that seven of the academic freedom adopters with AFI scores that were and remained below 0.2 are from the MENA region: Egypt, Iraq, Bahrain, Sudan, Algeria, Qatar and Syria. In fact, before 2010 there was not a single country from this region whose AFI patterns suggest a ‘sincere’ constitutional adoption of academic freedom. Moreover, several countries from Central Asia follow the trend of constitutional academic freedom adoptions in the 1990s without following through with their promises (Uzbekistan, Turkmenistan, Tajikistan, Azerbaijan and Kazakhstan, which took a downturn). Given that academic freedom is still trailing behind other constitutional freedoms in asserting itself at an international level (see Figure 1), it is in fact very plausible that its acculturation/emulation effects have been focused on those regions where adoption rates are comparatively high.

These adoptions by low-freedom countries have meant that the average AFI scores of first-time adopters has at times been significantly under the global AFI average of all countries. Combined with deteriorations in academic freedom levels in some countries with existing constitutional provisions, this has meant that the average AFI scores has periodically been lower for countries with than those without relevant guarantees, including in recent years (in 2022: mean AFI of 0.59 vs 0.65). Elkins noted in 2010 that ‘illiberal constitutions are exceedingly rare’ and that the ‘overwhelming majority of constitutions written by authoritarians formally espouse democratic principles and provide for democratic institutions’. ⁹² The academic freedom data underscore this point even more vividly than his comparison of traditional civil rights, which were at the time represented more often in constitutions of democratic countries than authoritarian ones. In contrast, constitutional academic freedom provisions were in 2022 at 71 per cent most strongly represented in the group scoring lowest on the AFI (<0.2), compared with 55 per cent in all other countries for which AFI data are available (51 per cent among high performers at >0.8). Paradoxically, it thus appears that those countries that least respect academic freedom most often invoke its legitimacy as a global norm.

Conclusion

The history of constitutional codifications of academic freedom stretches over some 175 years and diverges in important ways from that of other liberal rights and freedoms. Academic freedom provisions – in the form of references to the freedom of science, of higher education teaching, of the autonomy or self-governance of universities, or the exact phrase of ‘academic freedom’ – have overall become more frequent over time in all world regions, though the progression has not been a linear one and coverage still remains limited to 52 per cent of today’s constitutions. This article set out to explain these global patterns of constitutional codification – and omission – of academic freedom, using the Academic Freedom in Constitutions dataset, combined with data on higher education enrolment and de facto academic freedom levels. The analysis found evidence for all five hypotheses considered, but they are deeply intertwined and none can by itself explain global codification developments.

A first important observation is the relatively late and only temporary inclusion of the first academic freedom provision in a national constitution in 1848, compared with other liberal rights, such as the freedom of expression, freedom of religion or the right to property and equality before the law, which were all present since the early years of constitution-making in the late eighteenth and early nineteenth centuries. The path dependency of those early texts has meant that various prominent constitutions dating back that far – including the US or Belgian constitutions – do not include any academic freedom guarantee to the present day. However, there are other liberal constitutional provisions that appeared late but spread more quickly and widely than academic freedom norms. References to democracy, for example, also only emerged during the revolutionary movements of the mid-nineteenth century yet, unlike academic freedom, immediately spread across numerous constitutions, were consolidated after World War II and are today present in some 90 per cent of constitutions. The freedom of association spread even more linearly (see Figure 1).

A second key explanatory factor that accounts for this comparatively reluctant diffusion of academic freedom guarantees is their close connection to the existence and development of a higher education sector, which in many countries only became significant after the adoption of their first constitution. The data analysis suggests a clear connection, also in recent decades, between a country’s relative size of the higher education sector and the likelihood that a new constitution includes academic freedom provisions.

These two factors that delayed academic freedom adoptions also meant that, at key moments of constitution-making history, not many model constitutions were available that included such provisions. For instance, almost none of the constitutions of colonial powers did so at the time they ceded independence to their colonies. Most independence constitutions thus omitted academic freedom provisions not only for lack of salience of higher education issues, but also for lack of appearance in the model constitutions they considered. A particular historic contingency in the global constitutional codification of academic freedom was its omission from the European Convention on Human Rights of 1950, ⁹³ on which many former British colonies modelled their constitutional bills of rights. These independence constitutions, in turn, have locked many (although not all) countries in Africa, Asia and the Caribbean into a path dependency of constitutional academic freedom exclusion: over the past two decades, all countries that adopted new constitutions that still did not include a relevant provision were located in those three regions.

Yet, despite these constraining factors, academic freedom provisions have appeared and spread across the world. This fact can be attributed to a few pioneering countries that have acted as norm entrepreneurs when adopting academic freedom provisions, most of them for idealistic reasons to affirm national values, but some partly in pursuit of outside legitimacy. This group of countries includes, among others, Germany (1848 and 1919), Uruguay (1918), Peru (1933), Estonia (1920), Yugoslavia (1921 and 1963), Egypt (1956) and Liberia (1986). These regional leaders were instrumental in the diffusion of academic freedom norms by providing influential constitutional models for their regional neighbors, legal relatives and ideological peers. Although these processes were partly interlinked, the finding that clusters of different types of academic freedom references emerged across the world clearly shows that their spread cannot be described by a simple centre-to-periphery movement.

Overall, both the slow adoption at a global level and the dependency on higher education development have meant that academic freedom has until now eluded global constitutional convergence and not become a fixed part of the ‘liberal canon’ or of a ‘global cultural model’ of constitutional rights. Only half of all constitutions currently include it, a share that has lately remained relatively stable. Compared with other international norms, academic freedom thus appears to have less of a universal gravitational pull that leads countries to emulate it simply to tap into its legitimacy. The data analysis suggests, in fact, that this effect exists but that it has been concentrated on those regions where adoption rates are generally high: Latin America, Eastern Europe and Central Asia, and the MENA region. That said, the overall significant proportion of insincere adopters of academic freedom provisions, combined with countries where academic freedom has deteriorated, has led to the curious situation where constitutional protections are today most frequent among states that severely violate academic freedom in practice. While their own illiberal science script seems unaffected by this liberal constitutional commitment, these countries are also the ones that indicate most clearly a perceived legitimacy of academic freedom as a constitutional norm. On the other hand, the rising contestations of academic freedom in recent years have also appeared to provide an impetus in some liberal democracies to overcome their constitutional legacy and newly incorporate academic freedom guarantees into their fundamental texts.