Understanding Anticolonial Nationalism

By “anticolonial nationalism”, we refer to a form of nationalism that emerges in a dependent territory (a colony) and that seeks to create an independent nation-state. This can be differentiated from three other types of nationalisms. The first is core nationalism, the nationalism that first emerged in Europe within powerful metropoles such as England, France, Portugal, Spain, Norway, Germany, Denmark, etc. In contrast to this, anticolonial nationalism comes from the former colonies of these powers (e.g. the nationalism of the American revolution or the Haitian revolution). The second distinction is nationalism without states: that is, nationalism in colonies that fall short of seeking an independent nation-state. Nationalism projects an imagined community of citizens who are deemed equal in some way or another, but it does not have to equate the boundaries of that community with the boundaries of an independent nation-state. Historically and in the present, there have been many such cases of nationalism without states. Elites in colonial fields sometimes spoke of their colony as a “nation” without advocating for or even envisioning independent nation-states. Today, strands of nationalism in territories like Puerto Rico, Quebec or Catalunya—among many others—qualify for this category [Barreto Reference Barreto1998; Duany Reference Duany2001]. By contrast, when we refer to anticolonial nationalism, we refer to movements for political independence. We mean nation-statism or political nationalism.

The third difference is with secessionist movements within already established nation-states. To understand this we must define what a “colony” is. We define a colony (or a “formal colony” to be more precise) as a territory that is under the sovereign power of a state but which is classified and treated as juridically and administratively inferior to other areas and peoples under the sovereign [Osterhammel Reference Osterhammel1999]. Colonies are not equal to other territories in the system. The territory and its inhabitants are colonial subjects as opposed to citizens. They are subjected to the “rule of colonial difference” [Chatterjee Reference Chatterjee1993]. Colonies are thus different from the territories of federalist-type systems (sometimes thought as “land empires”) such as the Russian empire or parts of 19th century United States; and from territories in monarchical systems like the Hapsburg empires. In these systems, the territories are equal in the sense that they are each equally subject to the control and power of the center. Everyone except the king or group of rulers is a subject; the citizen-subject binary does not apply [Go Reference Go2011: 5-8; Howe Reference Howe2002]. This means that anticolonial nationalism is different from separatist or secessionist movements that emerge within an already existing national state or part of a federal system.

The difference is two-fold. First, separatist or secessionist movements emerge within equal territories of a nation, not empires. When the South tried to secede from the United States during the American Civil War, for instance, it was not a colony. Southern states were officially equal to the other states of the Union, and Southern peoples were equal citizens. While they claimed unequal treatment, they were not juridically inferior colonial subjects.^{Footnote 2} Similarly, the secessionist movements in the former Soviet Union were not strictly anticolonial in our definition. As historians and specialists show, while they may have discursively rendered themselves to be “colonies”, and while they have functioned within the Soviet Union as economic colonies, our conceptualization of “colonial” would not apply because these territories were all equally part of a federal, not imperial, system [Lieven Reference Lieven1995 and Reference Lieven2002]. The same goes for other cases like Bangladesh. The movement to create an independent national state called Bangladesh was a secessionist movement, not an anticolonial nationalist movement in our definition. It had been an equal part of Pakistan rather than its dependent colony.

A related difference between anticolonialism and separatist or secessionist movements is that the latter are not the first nationalist acts in the territory, while anticolonial nationalist movements are. Anticolonial nationalism refers to the first time that a group of people who had previously been treated as not part of a nation began to see themselves as equal members of a community. Secessionist movements had already seen themselves as equal members of a nation; they had been part of an independent national state. Secessionist movements simply aimed to break away from that national state and shift their loyalty to a different spatial unit. When the South tried to secede, their territory had already been part of the United States and Southerners had already envisioned themselves to be members of that national state. Alternatively, when the thirteen colonies united and declared an independent “United States,” it was the first time in that territory that a separate and new nation was envisioned. Therefore, what makes anticolonial nationalist movements significant is that they mark the first time that subject peoples envisioned their territory to be a nation and attempted to make themselves equal members of an entirely new nation rather than colonial subjects.

To obtain an empirically-grounded understanding of anticolonial nationalism, we created a database of anticolonial nationalist “events” in the modern era for each ex-colony of the Anglo-European empires. By “event”, we mean either the establishment of a political party, organization, or association that declares national independence as a goal; the beginning of a revolt, rebellion, or political movement that declares national independence as a goal; or a declaration of independence or constitution establishing an independent national government. As yet, no such comprehensive list exists. The closest systematic data is Wimmer and Feinstein’s [Reference Wimmer and Feinstein2010] list of the establishment of the first “national political organization” in every country in the world today, but this list in insufficient for our purposes. Wimmer and Feinstein’s list refers only to “modern political organizations” such as political parties. It does not include anticolonial nationalist revolts or declarations of independence. Furthermore, Wimmer and Feinstein’s variable only lists the first national organization, not nationalist movements. Their data are lists of organizations that proclaimed the existence of a “nation” but did not necessarily demand national independence or a separate nation-state. As noted, many colonies thought of themselves as a nation but did not pursue independent status as a nation-state. They sought reforms or policies that worked within the rubric of the political system and did not necessarily seek independence.^{Footnote 3} Our data is meant to capture the more explicit and arguably bolder act of not only imagining a “nation” but imagining and seeking an independent sovereign nation-state.

To find these events, we first had to determine the territorial units; that is, the colonies. To do this we used existing lists of past colonies and colonial territories from Henige’s list of colonial governors [1970] and the Correlates of War Territorial Changes dataset [Henige Reference Henige1970; Tir et al. Reference Tir, Schafer, Diehl and Goertz1998]. We then used a variety of sources to establish the first anti-colonial nationalist event in each of these territories. We paid careful attention to the fact that many of these colonies do not match present-day nations: some colonies were amalgamated into larger units or broken down into smaller units. For instance, present day Australia was first a series of different colonies before those colonies were all amalgamated into Australia. The same goes for Latin American countries such as Peru, which had been created by combining two separate colonies: La Paz and Cuzco. Nonetheless we were able to locate, for the vast majority of past colonies, the date of the initial anti-colonial event for these territories, and this sometimes yielded different events within the same present-day country (if the present-day country had originally been different colonies). For example, the first anticolonial nationalist event in Cuzco, presently the lower part of Peru, was in 1814, with the Pumacachua rebellion; but the first anticolonial nationalist event in La Paz, presently upper Peru, was in 1809 when Pedro Domingo Murillo led a movement declaring La Paz independent from Spain. Each of these is recorded as a separate event.

One potential problem with this measure is that outright declarations of independence were typically repressed or at least discouraged by colonial officials. For example, in Northern Rhodesia, officials responded to the formation of various “welfare associations” by telling them that they could not advocate for political interests, only for economic issues or related welfare issues [Rothberg Reference Rothberg1965: 132].There might have been political parties or organizations formed during colonialism which did not explicitly declare independence as their goal for fear of repression. However, we take the explicit and public declaration of independence as a goal to be a significant and robust measure of anticolonial nationalism. If groups are willing to publicly declare independence, nationalism must have been a powerful enough idea for them.

The sources included secondary histories of the various colonies and nations, including histories of nationalism in the relevant regions and empires (full list available from authors). This information was supplemented by various other lists, handbooks and compilations of nationalisms or related political movements around the world, including the “Country Studies” series by the Federal Research Division of the Library of Congress (US Library of Congress n.d.), The Oxford Handbook of the History of Nationalism edited by Breuilly [Reference Breuilly2013], The Encyclopedia of Nationalism edited by Motyl [Reference Motyl2001] and Nations and Nationalism: A Global Historical Overview [Herb and Kaplan Reference Herb and Kaplan2008], among others. To ensure validity, we sought out at least two sources for each event. For instance, if we found events, organizations or movements listed in existing compilations of nationalist or secessionist movements, we also used secondary histories to verify this or to determine whether an earlier anticolonial nationalist event had occurred.

Because we differentiate anticolonial nationalism from separatist or secessionist movements that emerge within an already existing national state after anticolonial nationalism had already been established in that territory, the latter are not included in our data. We rule out, for example, the nationalism of the South during the American Civil War, and the secessionist movements leading to the demise of the Soviet Union. We do so for good analytic reason, as noted above: since we are interested in the origins and historical emergence of anticolonial nationalism, we focus upon the earliest instances of nationalist sentiment, and separatist or secessionist nationalisms emerged only after nationalism in a territory had already surfaced.^{Footnote 4}

Figure 1 charts the anticolonial events by year. The data suggest that there were two main waves of anticolonialism. The first occurred in around the turn of the 18th-century, which began with the anticolonial movements in the United States and Haiti and culminated in the early 1800s with the South American movements for independence against Spain and Portugal, giving birth to the Latin American republics [Armitage Reference Armitage2005]. The second wave began in the early 20th-century. While the earlier wave occurred primarily in the Americas, this wave encompassed the other regions of the world: Asia, Africa, and the Middle East. It involved the movement for Indian independence, for instance, as well as the nationalist movements in Africa that led to decolonization there. Roughly speaking, these two waves correspond with the two main waves of imperialism. Historically, Western imperialism began in the early modern period in the Americas, including the Caribbean and Central America. This was when Spain, Portugal, England and France founded plantations and settlements across the coasts of the Americas and moved into the interior [Bayly Reference Bayly1998]. By the end of the 19th-century, however, a second wave of imperialism unfolded. Often called the “new imperialism” or the period of “high imperialism”, this was when European countries and the United States seized territory in Africa, the Middle East, Asia and the Pacific [Ballantyne and Burton Reference Ballantyne and Burton2014; Fieldhouse Reference Fieldhouse1982].

Figure 1 Anticolonial Declarations in the World’s Colonies, 1700-1990

The question is how to explain these trends. The historian Roger Owen notes that anticolonialism followed a “familiar dialectic by which imperial rule cannot help but generate the nationalist forces that will eventually drive it out” [Owen 2000: 20 quoted in Lawrence Reference Lawrence2013: 3-4]. In this view, anticolonial nationalism was inevitable: a diametric reaction to colonial rule. The fact that the two waves of anticolonial nationalism followed from the two waves of imperialism respectively suggests this to be true. However, even a cursory visual analysis of the time from initial colonization to the year of the first anticolonial nationalist event shows that anticolonial nationalism cannot be explained as a straightforward response to colonialism itself (see Figure 2, which shows just a sample of the events). If it could be explained that way, the time between initial colonization and the year of the first anticolonial nationalist event would be more or less equal across the cases. But Figure 2 shows that the variance is wide: in some colonies, the first anticolonial nationalist event came only after centuries of colonial rule, in others, it emerged within decades. This suggests that other explanatory factors must be at play.

Figure 2 Years from Initial Colonization to First Anticolonial Nationalist Event

Theorizing Anticolonial Nationalism

If anticolonial nationalism cannot be explained as a diametric response to colonization, how can we explain it? To answer this question we can look at a number of possible theories. As noted, existing scholarship has not yet offered systematic analyses of the origins and spread of anticolonial nationalism. On the one hand, some notable work has explored the emergence of the nation-state as a dominant political form [Wimmer and Feinstein Reference Wimmer and Feinstein2010]. Wimmer and Feinstein [Reference Wimmer and Feinstein2010] argue that the nation-state emerged from the ashes of empires due to various local and regional factors. However, this work—and subsequent debates about the relative importance of local as opposed to global factors in explaining the rise of the nation-state—is not about the emergence and global spread of anticolonial nationalism [Li and Hicks Reference Li and Hicks2016]. It is about the emergence and spread of national political units. The dependent variable is the actual creation of the nation-state, not the emergence and spread of the nation-state as a political idea, goal and value which then became an alternative to imperial states and imperial loyalties. On the other hand, existing case studies of anticolonial nationalism offer potentially generalizable theories, as do existing theories of nationalism writ large, that can be operationalized and adjudicated through further empirical analysis.

While these theories are each distinct in their own right, they collectively vary according to whether they emphasize the cultural, discursive or cognitive bases of nationalism on the one hand or, on the other hand, the colonial economy and politics. Anderson’s famous theory of “imagined communities” [Reference Anderson2006] offers a clear example of the first type of theory. According to Anderson, nationalism was made possible through print capitalism. Print capitalism led to the spread of “mechanically produced print-languages” that spread through global markets. In turn, those print-languages “created unified fields of exchange and communication below Latin and above the spoken vernaculars,” thereby uniting previously distinct linguistic groups and offering, for the first time, the ability to imagine a larger “national” community oriented around shared notions of linear time [Reference Anderson2006: 56]. This “modular” form of nationalism emerged initially in literate “creole” communities in the colonial Americas and in the Philippines. It then spread to Europe and to the remaining colonies in the early 20th century. In the colonies, anticolonial nationalism was further facilitated by colonial schools and traveling by colonial elites to universities in the metropole [Anderson Reference Anderson1992]. It was also facilitated by the colonial census, museums and other technologies [Anderson Reference Anderson2006: 163-185].

Anderson’s theory is nuanced and multifaceted but it ultimately sees nationalism as a cultural development; more precisely, a cognitive or discursive capacity. Education and print capitalism allowed colonized elites to read newspapers, novels and documents like the colonial census which in turn enabled them to imagine postcolonial nations uniting otherwise disparate ethnic, religious and linguistic groups. In this admittedly limited sense, Anderson’s theory shares common ground with another theoretical framework that might be enlisted to explain anticolonial nationalism: the World Society or “world polity” approach. This approach argues that modern political forms, not least those associated with the nation-state, have spread through the world system with the aid of various international organizations and the construction of new global norms to which actors in the system have to adhere. Their studies conclude that the 17th-century marks the beginning of a global modernization process that continued through the 20th century [Meyer et al. Reference Meyer, Boli, Thomas and Ramire1997: 163-174]. Strang [Reference Strang1991], for example, finds that decolonization—i.e. the spread of the nation-state form—was partially dependent upon the spread of a global culture emphasizing national sovereignty since World War II and culminating in the 1960 United Nations declaration against colonization. Li and Hicks [Reference Li and Hicks2016] argue that the founding of the United Nations itself was a crucial moment for spreading the nation-state form around the system. None of these studies have examined anticolonial nationalism or even nationalism specifically; they examine the spread of various political forms and the creation of the nation-state, not of nationalism itself. But they offer up the hypothesis that anticolonial nationalism emerged and proliferated due to diffusion within World Society. World Society offered the cognitive model for imagining the nation to which peoples around the world conformed. This is similar to Anderson’s notion of “modular nationalism.”

A different version of diffusion theory would highlight processes of imitation and emulation more broadly. Whereas the World Society approach emphasizes international organizations such as NGOs or the United Nations in the 20th century, this other diffusion approach would suggest that initial anticolonial nationalist events served as models for others to follow. In this story, once the initial anticolonial movements emerged (for whatever historically contingent reason), they inspired actors in other colonies accordingly. Historians often point to this sort of process. In one among many examples, David Armitage [Reference Armitage2009] shows how the American Declaration of Independence produced by the American Revolution served as a model for peoples around the world [Armitage Reference Armitage2009]. Other scholars might point to the global influence of the Haitian Revolution or the victory of the Japanese over Russia in the 1905 Sino-Japanese War as key events that inspired or emboldened anticolonial nationalists. While all such narratives point to different events as important for diffusion, they implicitly share an underlying explanation: anticolonial nationalism diffused throughout the system due to the catalyzing effect of previous anticolonial events that provided a cognitive model—and perhaps an inspiration for—their own movements [Beck Reference Beck2011].

Another set of explanations pays less attention to discursive capacities and culture and looks instead towards sociopolitical and economic practices, interests, and conflict. These approaches are diverse, but they share an overarching attention to political-economic processes attendant with colonial rule that in turn generated anticolonial nationalism.^{Footnote 5}

In one version of this approach, the colonial state developed local economies through a variety of sociospatial practices (i.e. infrastructural interventions and economic policies) that created the conditions (i) for imagining the colony as an integrated national territory, and (ii) for local socioeconomic interests to develop around that space. Goswami’s innovative analysis of anticolonial nationalism in India exemplifies this approach [Goswami Reference Goswami2004]. Goswami argues that infrastructure such as the national railroad system and tariff policies of the British colonial state produced the sense of an integrated and bounded national economy—a “territorial nativist conception” of the nation. Upon this conception, Indian middle-class intellectuals and activists mounted their political and economic program for a postcolonial India that would better realize the economic interests of Indians. The difference from Anderson’s approach [Reference Anderson2006] should be clear. In Goswami’s view, anticolonial nationalism did not emerge from colonized elites reading newspapers or novels, as Anderson’s cognitive theory suggests, but from lived experiences of “colonial state space” and associated interests in an autonomous postcolonial economy.

This approach bears some similarity to a variety of other approaches that connect anticolonial nationalism to the rise of colonial elites and their sociopolitical interests. Writing about the emergence of African nationalism in the 1950s, Coleman [Reference Coleman1954] noted that nationalism was strongest in colonies with higher levels of urbanization, agricultural commercialization, and wage labor. Coleman contended that these developments marked the growth of “Western”-like middle-classes, intellectuals and professionals who developed more “modern” orientations that contributed to the rise of nationalist consciousness [Coleman Reference Coleman1954: 414-415]. Others complicate the story by stressing that colonialism required a class of more educated colonial elites to serve as collaborators or lower functionaries in the colonial state. Colonialism, therefore, tended to create a “westernized” group of elites: a “class of western-educated natives” who often led the charge against colonialism [Breuilly Reference Breuilly1982: 157]. This was especially so as the elites’ expectations from their education were thwarted. Boahen [Reference Boahen1985: 567] writes of African anticolonial nationalism that colonialism “raised the hopes of the emergent educated elites all over Africa for greater opportunities […] They expected to be absorbed and accepted as colleagues by the colonial rulers.” But “they found their expectations checkmated or frustrated” by colonial racism, leading ultimately to “resentment, bitterness and agitation against the colonial regimes” [Boahen Reference Boahen1985]. In sum, this story explains anticolonial nationalism as the result of rising, “westernized” and educated but resentful colonial elites who saw nationalism as an “identity” for mobilizing “more widespread resistance to colonial rule” [Breuilly Reference Breuilly1982: 157].

Other theories highlight global political and economic processes that might have impacted colonial elite interests. One variant emphasizes political opportunities due to imperial crises. According to this argument, nationalists in the colonial world seized upon the opportunities posed by inter-imperial war and competition, or other events serving to weaken empires. Amidst these events they could more readily mobilize populations against foreign domination, thereby fomenting anticolonial nationalism. Another variant might point to global economic downturns that intensified the demands of colonized peoples. Here we can connect world-systems analysis to political-economic explanations of anticolonial nationalism [Wallerstein Reference Wallerstein2004; Wallerstein Reference Wallerstein1974]. According to some versions of world-systems analysis, for instance, the world-economic system undergoes waves of economic expansion and contraction. Goldstein [Reference Goldstein1988] finds historical cycles of expansion and contraction that manifest the “Kondratieff waves” of economic growth pinpointed by the Russian economist Nikolai Kondratieff [Goldstein Reference Goldstein1988]. Accordingly, we might expect anticolonial nationalism to emerge in periods of economic contraction. This contraction exacerbates the grievances of the colonized elite, in turn generating anticolonial sentiment [Furedi Reference Furedi1994].

Data & Variables

To see which if any of the foregoing theories best explain the emergence and spread of anticolonial nationalism, we deploy time-series analysis methods. The dependent variable is the year of the anticolonial event from our original database outlined earlier. The independent variables operationalize the various theories discussed above. We test Benedict Anderson’s theory [Reference Anderson2006] of nationalist consciousness in two ways. According to Anderson, creole elites in the colonies came to nationalist consciousness as they read about events occurring around the world and in their colony. These reading practices enabled elites to create a “secular, historically clocked imagined community” [Anderson Reference Anderson2006: 37-46]. The best available measure for these reading practices is the literacy rate in the colony (percentage of adult population that is literate), based on data collected by Wimmer and Feinstein [Reference Wimmer and Feinstein2010]. We predict a positive relationship between literacy rates and anticolonial nationalism.

Print capitalism and the spread of literacy among elites can also be theorized as a feature of the global system, however. Accordingly, we operationalize a global version of Anderson’s theory by the number of universities in the world-system. While the universities typically formed in European metropoles, colonial elites often benefitted from them. As Anderson [Reference Anderson1992] notes, nationalism among colonial elites like Jose Rizal from the Philippines also originated and was emboldened by their travels to imperial metropoles for higher education, leading to the proliferation of “long-distance nationalism” [Anderson Reference Anderson1992]. Relatedly, Goebel [Reference Goebel2015] shows how Paris became a center for anticolonial elites who travelled there initially to study. We expect a positive relationship between the number of universities in the world-system and anticolonial nationalism.

As noted above, World Society theory posits that the spread of the nation-state and other associated forms of political modernity emerges from a global culture rather than from factors endogenous to nations. We operationalize this explanation by the total number of intergovermental organizations (IGOs) in the world system^{Footnote 6}. This is a common measure used by World Society studies, including most recently Li and Hicks’ [Reference Li and Hicks2016] analysis of the emergence of the nation-state form. We predict a positive association between the total number of IGOs and anti-colonial nationalist events. We also measure the number of intergovernmental organizations to which a colony’s metropole belonged. The hypothesis is that, the more integrated into world society the metropole was, the more likely its colonies would be, and therefore the more likely that anticolonial nationalism would have developed in those colonies.

A different version of this cognitive approach alerts us to the role of contagion and emulation: anticolonial events or revolts inspire others in the system similarly. From our dependent variable of anticolonial nationalist events, we created a number of independent variables to measure these two scales. The first is the lagged number of anticolonial nationalist events in the global system. The second is the lagged number of anticolonial nationalist events in the same geographic region. The third is the lagged number of anticolonial nationalist events in the same empire. To extend the time frame to account for the possible delayed influence of events, we also created variables for the number of anticolonial nationalist events within the past five years for each of these scales.

Goswami’s analysis of anticolonial nationalism in India argues that it emerged from the various “sociospatial practices” of the state that produced an integrated economy. To measure this sociospatial transformation, we include a variable capturing the length of railroad tracks in the colony at the time. In Goswami’s analysis, railroad construction was a key part of the state’s sociospatial reconstruction that in turn facilitated anticolonial nationalism. Railroad development facilitated the construction of an integrated territorially bounded colonial economy that embedded elite interests while also facilitating an imagining of the postcolonial “nation” around that economy. We use Wimmer and Feinstein’s [Reference Wimmer and Feinstein2010] measure of the length of railway tracks in the colony per 1,000 km (a continuous variable that runs from 0-113.73), which contains information on many colonies and is based upon the comprehensive information in Mitchell’s International Historical Statistics series [Mitchell Reference Mitchell1988, Reference Mitchell1998 and Reference Mitchell1993].

Another related factor is the strength of the local colonized elite. Goswami [Reference Goswami2004] finds that Indian anticolonial nationalism emerged from middle-class Indian groups. Breuilly [Reference Breuilly1982: 157] suggests that anticolonial nationalism emerged from colonial political elites who had slowly risen to power in the colony through positions within the colonial state but whose ambitions were thwarted. Coleman [Reference Coleman1954] argues that modernization in Africa created a new class of elites who in turn became active anticolonial nationalists. Our subsequent hypothesis is that a more developed and/or stronger colonized elite makes anticolonial nationalism more likely.

Unfortunately it is difficult to obtain good data on colonial elite strength and interests, but we can use a variety of proxy variables. A new dataset on various indicators on colonies in Africa and Asia gathered by Ziltener et al. [Reference Ziltener, Künzler and Walter2017] is helpful here. While this dataset only includes colonies in Africa and Asia, it is a unique and rare dataset on the colonial period, and is the best option available at this point. Their research yields two relevant variables. The first is “Invest” which is a measure of investment in the colony for infrastructure, from roads to telecommunications, water and gas works (where “0” is no significant investment, “1” is moderate investment, and “2” is highest investment). This is one possible measure of urbanization and, following Coleman’s [Reference Coleman1954] theory, we predict that more urbanization led to the development of more middle-class colonized elites who in turn became anticolonial nationalists.

The Ziltener data also offers the variable EconTran (“economic transformation”). This is a composite index of the level of economic transformation the colony underwent under colonialism, based upon seven variables: trade policy (whether “open door” or preferential), trade concentration (measuring the colony’s trade with the metropole as proportion of total trade), investment concentration (dependence on direct investment from the metropole), investment in infrastructure, commercialization of plantations (relative amount of large plantations over small isolated plantations), and level of extraction of natural resources. The measure runs from “0”, which refers to no social transformation during colonialism, to “100”, which is the highest level of economic transformation. We take this as a measure of commercialization which we also expect to produce a colonized nationalist elite. We predict a positive relationship with anticolonial events.

Two other variables offer possible measures of colonized elite strength. The first is the presence of settlers in the colony. Breuilly [Reference Breuilly1982: 157] and others suggest that the presence of settlers negatively correlated with indigenous elite strength. A strong settler presence means a weaker local elite because settlers displaced them, while a weaker settler presence means a stronger local colonial elite because only locals can be deployed to occupy lower-level and mid-level positions in the colonial bureaucracy. Accordingly, we use Ziltener et al.’s [Reference Ziltener, Künzler and Walter2017] variable ForPre (for “foreign presence”) which is a categorical variable measuring the presence of settlers, where 0 indicates that less than 1% of the population were settlers and 4 is the highest possible score indicating that 10% or more of the population were settlers. We hypothesize a negative relationship between this score and anticolonial nationalism.

The second variable measuring elite strength is Ziltener’s variable for the level of social transformation during colonialism (SocTran). This is also a standardized aggregation of different variables meant to capture the overall level of social change in the colony during colonial rule. The variables compiled here are the number of immigrants from the colonizing country, the number of religious missions in the colony, the importation of foreign laborers to the colony, and the level of redrawing of territorial borders during colonialism. The measure runs from “0”, which refers to no social transformation during colonialism, to “100”, which is the highest level of social transformation. All of these variables suggest a heavy penetration of metropolitan power and control into the colony and thus a displacement of the colonized elite. We expect a negative relationship with anticolonial events: the lower the SocTran score, the less metropolitan control, and hence the stronger the indigenous elite.

As noted, global politics might also have played a role in developing colonial elites’ nationalist consciousness. During or in the wake of cataclysmic events such as wars, or during periods of intense inter-imperial rivalry, anticolonial nationalism is more likely to emerge, because these periods present new political opportunities. We test this using various variables. The first is war, which should theoretically increase the chances of the emergence of anticolonial nationalism. We created a binary variable set to one for the four main global wars in our period of interest: the Seven Years War (1756-1763), the Napoleonic Wars (1803-1815), World War I (1914-1918) and World War II (1939-1945). Because wars are not always global but may also be intra-imperial, transnational or regional, we also created a variable for these sorts of wars. We used the Correlates of War Dataset’s list of inter-state, extra-state, and intra-state wars to construct a list of total number of wars or conflicts that the metropole was engaged in per year [Singer Reference Singer1987]. The prediction is that these sorts of wars, whether global or not, would destabilize imperial rule and therefore create an opportunity for anticolonialism.

Another variable to test political opportunity is the metropole’s relative weakness in the international system. The weaker the metropole relative to other states, the more likely anticolonial nationalists will mobilize. This can be measured using the Correlates of War Dataset’s Composite Index of National Capability (CINC). Meant to measure a states’ relative power vis-à-vis other states, this is an index that takes six annual indicators of state power (total population, urban population, iron and steel production, energy consumption, military personnel, and military expenditure) and converts them into a share of those six measures on a world scale. For any given year, each state in the system has a score. The higher the score, the more power that state has relative to other states in the system. The lesser the score, the less power the state has vis-à-vis other states [Singer, Bremer and Stuckey Reference Singer, Bremer and Stuckey1972]. We hypothesize a negative relationship between this variable and anticolonial nationalism: the weaker the metropole (i.e. the lower the score), the greater the chance of anticolonial nationalism developing in that metropole’s colonies.

To test whether global economics exacerbated or hampered colonial elite interests, we code years for periods of global economic contraction and expansion. We use Goldstein’s specification of Kondratieff cycles [Goldstein Reference Goldstein1988]. This is typically used in various other time-series analyses of the world system, such as Boswell’s [Reference Boswell1989] examination of colonial expansion in the world-system [Boswell Reference Boswell1989]. Coding “0” for economic contraction and “1” for expansion, we therefore predict a negative relationship between this variable and anticolonial nationalism.

Table 1 presents the main theories and concepts, their measurement, and the predicted relationship.

Table 1 Concepts & Variables

Regression Results

Because of missing data for some of the variables, we run separate models to maximize cases and test each set of theories before proceeding step-wise to find the most powerful predictors. Table 2 shows the different models and their results. Model 1 tests the cognitive-cultural approach associated with Anderson’s “imagined communities” thesis. Model 2 tests the World Society variant of the cognitive-cultural approach. Model 3 then tests emulation and contagion effects. In these first three models, there are only two variables of significance: (a) the number of IGOs in the world system (β = -1.83 at the <0.01 sig. level) and (b) the number of previous anticolonial events in the same geographic region (β =.41 at the <0.05 sig. level). However, regarding IOs, the relationship is the opposite of what is predicted. Accordingly to World Society theory, more IGOs in the system should increase the probability of anticolonial nationalism, but the results show that anticolonial nationalism is more likely when there are less IGOs. Alternatively, the number of prior anticolonial events in the region is in the predicted direction: anticolonial nationalism is more likely in one colony when other regionally proximate colonies had also seen anticolonial nationalism. This suggests a positive emulation effect. In short, of all the cognitive-cultural approaches, only the theory of emulation is relevant. Literacy or metropolitan universities, contra the Anderson thesis, do not predict anticolonial nationalism; neither do the presence of IGOs.

Table 2 Regression Coefficients Examining Anticolonial Nationalist Events between 1700-1990

Robust standard errors in parentheses. Cubic splines and region dummies are included in all models.

The coefficients of the splines, region dummies, and constant are omitted from the table.

Sig: * = < 0.05 (two-tailed test).

** = < 0.01 (two-tailed test).

Models 4, 5 and 6 test different variants of the colonial economics and politics explanations. Given that the variables from the Ziltener, et al. [Reference Ziltener, Künzler and Walter2017] dataset are restricted to colonies in Asia and Africa, we separate them first. Model 4 tests those variables, revealing that both the economic and social transformation indicators are significant and in the direction predicted. This model suggests that high levels of economic development in the colony and lesser metropolitan penetration increased anticolonial nationalism, which confirms the hypothesis that anticolonial nationalism emerged from a comparably strong colonized elite seeking to realize their interests through national independence.

Keeping in mind the regional limitations to the data in Model 4 (i.e. restricted to Asia and Africa), Model 5 separately tests the possible impact upon anticolonial nationalism of the extent of railways in the colony. Meant to measure the sociospatial practices of the colonial state and hence the creation of an integrated and cohesive economy, it covers more years and regions than the variables in Model 4. This analysis confirms that anticolonial nationalism was more likely in those territories where there was a more integrated economy (β = 0.068 at the <0.01 sig. level). Model 6 examines the other variants of this approach, testing whether global political opportunities as measured by wars and relative metropolitan strength, or whether economic grievances related to global economic cycles increased the likelihood of anticolonial nationalism. Of these, only the variable for the metropoles’ relative strength, CINC, is significant and in the predicted direction: anticolonial nationalist events were more likely in the wake of a decrease in the metropole’s relative strength in the international system (β = -7.677 at the <0.01 level).

The last part of our analysis helps us adjudicate among all of the different theories by putting the significant variables from the prior tests into the same models. But because of the spatial limitations with the Ziltener et al. [Reference Ziltener, Künzler and Walter2017] data, we run two different models. The first test, Model 7, keeps the Ziltener variables and thereby tests economic development, metropolitan penetration, IGOs in the world, regional emulation, and relative metropolitan strength. Because the data is from Ziltener, this only includes territories in Asia and Africa. We find, first, that the cultural-cognitive variables (measuring diffusion and emulation) reduce in explanatory power and lose their significance, thereby suggesting that these variables were actually capturing other processes in the previous models. We therefore find little support for cognitive-cultural theories of anticolonial nationalism. The results also show that the CINC variable, measuring the metropole’s relative power, loses its significance, but the variables for economic development and the extent of railways retain their significance (while the latter increases its explanatory power from β= 0.068 to 0.217 at the <0.01 sig. level). This lends powerful support to the argument that the creation of a developed, spatially-extensive and bounded economy through the sociospatial practices of the colonial state and the attendant development of a modernized colonial elite made anticolonial nationalism more likely.

Model 8 removes the Ziltener et al. variables to expand the cases temporally and spatially. The result is the same for the cultural-cognitive variables: emulation through regional anticolonial nationalism is not significant; while diffusion through IGOs is negatively correlated, again going against the expectations of World Society theory. Relatedly, the extent of railways retains its power and significance, supporting the argument that the sociospatial practices of the colonial state were important (β =0.079 at <0.01 sig. level). But one difference is that the variable for the metropole’s relative power is significant again, although at the lower 0.1 level (this is not reflected in Table 2 as we only indicate results with a significance level below the 0.05 level). This lends itself to two possible interpretations.

The first is that the political-economic explanation of anticolonial nationalism is strongly supported, with the possibility that imperial weakness matters in some cases but not all. The second is that the results regarding the latter issue of imperial weakness are partly driven by the cases in the model, due to data limitations. The cases included in Model 7 with the Ziltener et al. variables are restricted to certain colonies in Africa and Asia and are part of the later wave in the longer history of anti-colonialism.

This latter interpretation is strongly supported when we look at the actual cases in the different models, as listed in Table 3. The Table shows that most anticolonial nationalist events occur in the latter historical period (the second wave, from 1879-1990) and in Africa. Missing from the models using the Ziltener data are cases in other regions, most importantly the Americas, where the earliest instances of anticolonial nationalism erupted.

Table 3 Anticolonial Declarations by Region and Historic Wave

*Null cases are excluded (N = 8)

(a) North, Central, South & the Caribbean

To deal with this problem, we ran the models as before but on different time periods and hence different cases. Following the graph of the different waves, we ran the models to cover the years 1700 to 1878, and then the same models for the years 1879 to 1990. The results are given in Tables 4 and 5. They show that, for the first wave of anticolonial nationalism, the cultural variables are important: specifically, literacy rates and the number of universities in the world-system best explain the emergence of anticolonial nationalism. They also show support for the role of emulation through anticolonial nationalism within imperial fields. These findings correspond with Benedict Anderson’s [Reference Anderson1983] theory of the birth of anticolonial nationalism among creole elites in the early South American colonies and the Philippines. However, for the second wave, the cultural variables are not significant. Instead, the more important predictor is the extent of railways. This supports the political-economic explanation combined partly with the imperial weakness explanation (the variable for relative state power, CINC, is significant in this second wave).

Table 4 Regression Coefficients on Models for Historical Wave 1.

Note: Robust standard errors in parentheses. Cubic splines are included in all models.

The coefficients of the splines and the constant are omitted from the table.

Sig: * = <0.05 (two-tailed test).

** = < 0.01 (two-tailed test).

Table 5 Regression Coefficients on Models forHistorical Wave 2

Note: Robust standard errors in parentheses. Cubic splines are included in all models.

The coefficients of the splines and the constant are omitted from the table.

Sig: * = <0.05 (two-tailed test).

** = < 0.01 (two-tailed test).

These findings suggest the need to consider the historical specificity of different theoretical explanations. Cultural theories that explain how the nation came to be “imagined” fare best for explaining the historical origins of anticolonial nationalism. Our findings suggest that the nationalisms in the first wave covering 1700 to 1878, i.e. the earliest nationalisms in the Americas, can indeed be explained by reference to literacy and education which enabled Creoles and other colonized peoples to first invent and see themselves as part of a national community. Therefore, print capitalism and education predict when anticolonial nationalist events erupted in this first historical wave. But things are different when we come to the later wave of anticolonial nationalism which began in the 1870s. In regards to this historical movement, our findings suggest that cultural factors serve, at best, as necessary rather than sufficient conditions. Instead, for this wave, the political-economic dynamics of elite conflict best explain the timing of anticolonial nationalist events. In this latter period, literacy and education do not matter as much as integrated colonial economies, attendant elite political-economic interests, and the political opportunities offered when the metropolitan-imperial state decreased its relative power.