Abstract
Was Brazil’s President Dilma Rousseff victim of a coup or removed through a legal process of impeachment? The heated debate on the 2016 ousting of Brazil’s president testifies to the growing controversy around the definition of coups. Focusing on Latin America, we show that the use of coups with adjectives have become more frequent in public and scholarly debates. Occurring at a time when coups are becoming rarer, we argue that this development is linked to prevalence-induced concept change, meaning that when instances of a concept become less prevalent, the understanding of the concept expands. The meaning of coups has expanded through a proliferation of adjectives. Coups with adjectives are not new, but recent usage changes the concept from a classic to a family resemblance structure. Although this strategy can avoid stretching and increase differentiation, we urge caution and warn against harmful consequences, whether conceptual, theoretical, or practical.
“In Europe, with its more limited experiences of such matters, the expression coup d’état is not specifically distinguished from say, a coup de force, and we use the expressions, ‘military revolt’, ‘mutiny’, ‘rebellion’, ‘coup’, ‘revolution’ interchangeably without asking what precisely has happened. Latin Americans, with their closer acquaintanceship with the phenomenon, distinguish.”
Introduction
Coups d’état —however defined—have traditionally been regarded as the greatest threat to the survival of democracy. The sudden, often military-backed removal of the government customarily meant the end of a democratic regime. Today, however, despite a broad consensus that there is a global democratic recession and that liberal democratic values are threatened (Diamond, 2015a; Inglehart and Norris, 2016; Lührmann and Lindberg, 2019; Mechokva et al., 2017), there are fewer classical coups than ever, whether globally or in Latin America (Belkin and Schofer, 2005; Derpanopoulos et al., 2016; Djuve et al., 2019; McGowan, 2003; Marinov and Goemans, 2014; Powell and Thyne, 2011; Singh, 2014). In fact, consolidated democracies have grown almost immune to them (Svolik, 2015). Yet, we witness a puzzling increase in the use of the term (whether coup d’état or golpe de estado, as in Spanish and Portuguese), frequently combined with a qualifying adjective, in both academic and non-academic texts. 1 In this article, we analyze why coup with adjectives are on the rise at a time that coups occur less frequently, and what the analytical and conceptual consequences of this development are.
“Coups with adjectives” are not a new phenomenon. Terms like “military coup” or “self-coup” (autogolpe) have been in regular use within and outside academia for decades (Cameron, 1998; Fitch, 1977). However, their use has mushroomed in recent decades, particularly in Latin America. We found evidence of the recent invention or adaptation of qualifiers such as “soft” (Pitts et al., 2016), “parliamentary” (Santos and Guarneri, 2016), “presidential” or “democratic” (Varol, 2017), “constitutional” (Helmke, 2017; Yarwood, 2016), “market” or “neo-liberal” (Mauceri, 1995), “electoral” (Hellinger, 2005), “slow-motion” (Polga Hecimovich et al., 2017), “civil society” (Encarnación, 2002), and “judicial” (Yavuz and Koç, 2016). Many of these terms are piled together in Gentili’s (2016) volume on the impeachment of President Dilma Rousseff in Brazil. Although Bello (2016) and Keating (2012) register the use of coups with adjectives outside academia, the list above shows that their usage is not restricted to politicians or activists who have incentives to use real or invented coup-plots to implement emergency rule, arrest opposition leaders, clean out the bureaucracy, or brand the opposition as undemocratic.
We link the rise of “coup with adjectives” to the phenomenon of “prevalence-induced concept change” (Levari et al., 2018): when instances of a concept become less prevalent, the understanding of the concept expands to cover cases it previously excluded. Following Collier and Levitsky (1997), we argue that while earlier uses of coup with adjectives went down the ladder of generality or abstraction (increasing differentiation), new adjectives make up diminished or dismissive subtypes in which some constitutive elements of the concept are missing or only partially present. The examples of contemporary coups with adjectives mentioned earlier have in common that they do not satisfy all the criteria for the commonly accepted definition of a coup: the illegal overthrow of the government by other state actors. As a consequence, whereas a coup was formerly understood as a classical concept, the new usage approaches it as a family resemblance concept (Collier and Mahon, 1993; Goertz, 2006).
The choice of how to conceptualize a coup is not to be taken lightly since it carries normative, analytical, and political implications. In particular, in today’s democracies there is a conceptual peril of conflating a coup with legal tactics for government replacement. Identifying a phenomenon as an “impeachment” as opposed to a “coup” involves widely different interpretations, moral judgments, and consequences, as the latter could trigger international sanctions whereas the former should not. Therefore, academics as well as politicians and pundits should proceed with care when choosing concepts to define as consequential an event as the removal of a president.
The qualitative material for this article is based on a review of the academic literature on coups in general and in Latin America in particular, and on close reading of social media and news related to Latin American politics, our area of research. To substantiate our claim of an increased use of coups with adjectives, we use quantitative evidence from Google Books’ English and Spanish corpora for the period 1800–2008 with the Ngram Viewer (2013) tool. 2 Furthermore, we mapped the academic use of coups using the Social Science Citation Index for English terms, and SciELO (Scientific Electronic Library Online) citation index for Spanish and Portuguese terms. 3 We compare the use of the coup concept with coup data from Powell and Thyne (2011) and Przeworski et al. (2013).
In the next section, we present empirical evidence to sustain our argument about the proliferation of the term accompanied by a variety of adjectives, and substantiate our puzzle: why does this occur during a period when actual coups in the world decline? In order to disentangle the puzzle, we then discuss the coup concept, its constitutive elements and structure, before we discuss strategies to avoid conceptual stretching, and show through examples that the current usage of coups with adjectives is shifting the concept structure from a classical to a family resemblance type. We argue that this shift can be understood as special form of prevalence-induced concept change. Finally, we discuss the implications and whether conceptual innovation has resulted in conceptual stretching or improved the utility of the term. Although we draw our examples mostly from Latin America, the discussion is expected to hold general validity.
The Fall of Coups and Rise of “Coups With Adjectives”
Globally, coup attempts have been declining since the latter half of the 1960s, and have been particularly rare after the Cold War. Based on data from Powell and Thyne (2011), Figure 1 displays the trend of coup attempts since 1950. The downward trend is uncontroversial and supported by several recent studies (Belkin and Schofer, 2003; Derpanopoulos et al., 2016; Djuve et al., 2019; Marinov and Goemans, 2014; Marshall and Marshall, 2018; Singh, 2014). In Latin America, with some exceptions such as in Honduras in 2009, coups have almost vanished.
Coup Attempts Worldwide and in Latin America.
At the same time, the use of the term coup has seen a puzzling revival with recent developments in Latin America. Reading the news, registering conversations on social media, and searching the literature, we observed that academics, politicians, and the media increasingly use the Spanish term golpe connected with an adjective to describe events that would not fall under the classical conceptualization. A non-exhaustive list of the most prominent and prevalent new adjectives includes “soft,” “parliamentary,” “constitutional,” “neo-liberal,” “market,” “electoral,” “slow,” “civil society,” and “judicial.” While often these terms are used by presidents to taint their opponents, they are also used in academic works, and thus should be analyzed rigorously. First, however, we need to establish whether our hunch about an increased use of coups with adjectives holds when confronted with evidence. We do this by studying the prevalence of the use of the term coup d’état and coup with adjectives in the English and Spanish corpora in Google Books using the Ngram Viewer tool. The results are displayed in Figures 2 and 3.
The Use of Coup d’état and Golpe de estado in Books.
The Use of Coup d’état and Golpe de estado in Academic Articles.
The use of the term “coup d’état” in English follows a pattern similar to that of the registered coup attempts (Figure 2), a steady increase until the 1960s, and after that a sharp decline until the final year registered (2008). The use of the term in books is highly correlated with actual coup attempts (Table 1). Notably, a similar decline is not present in the use of “golpe de estado,” which increases steadily from the 1950s and, with exception of a dip in the years prior to 2000, ends with the highest registered score in 2008. Although the Google books database covers academic literature, it also includes fiction and thus tells us little of the popularity and content of academic studies of coups. Looking only at academic articles in the Social Science Citation Index and SciELO (Figure 3), we find signs that studies of coups have become more frequent (solid and dotted lines), especially since the early 2000s. As academic production has increased overall, we control by mapping English language articles with “coup d’état” as topic as share of articles with “political science” as topic (dashed line). 4 The results show that the study of coups has also increased from the early 2000s as a share of all articles in the discipline, even though it peaked in the 1990s. Recent popularity may be connected to the growth of available datasets on coups (cited earlier), and the recent surge of studies of authoritarian regimes and leadership succession.
Correlations actual coups and attempts and mentions in Google Books.
Cells show correlation coefficients, significance level in parenthesis. “Coup d’état,” “Coup with adjectives,” “Golpe de Estado,” and “Golpe con adjetivos” from Google Books, observations with 1 year lag.
Our main interest, however, is with coups with adjectives. We searched Google Books for the above-mentioned coups with adjectives in English and Spanish, and found the following: in English there is no decline of their use as the number of coups declined, and in Spanish their use has expanded tremendously at the same time as the number of coups declined (Figure 4).
Coup With Adjectives in English and Spanish in Books.
Figure 4 displays the English and Spanish usage of coups with adjectives. Comparing Figures 2 and 4, we see that while the use of coup d’état has decreased with the numbers of coups, there is no dip in the use of coups with adjectives. 5 Furthermore, while the mention of coup d’état is positively and significantly correlated with actual coup attempts, the use of coup with adjectives is not after 1949 (Table 1). In Spanish, however, coups with adjectives were an extremely rare phenomenon until the 1960, but have consistently been mentioned in books since the 1970s confirming that it is indeed a more recent phenomenon. Figure 5 visualizes the association between the usage of coup d’état, coups with adjectives in English and Spanish, and actual coups since the early 1800s. With exception for the usage of the root term in English, the correlation between actual coups and usage of the term with or without adjectives seem to break down exactly when the number of coups decline (see also Table 1).
Actual Coups vs Mentions of Coups with or without Adjectives in Books.
Table 1 shows correlations between mentions of coups with or without adjectives in English and Spanish in Google Books and actual coups and coup attempts as measured by Powell and Thyne (2011) and Przeworski et al. (2013). When looking at the full period from 1804, all variants of coups in Google Books are positively correlated with actual coups, and mentions of “coup d’état” is strongly and significantly so. Since 1949, when Powell and Thyne’s data start, however, we observe that coups with adjectives in Spanish show a negative and significant correlation with coups and attempts, while for coups with adjectives in English or coups in Spanish, the correlation is either negative or zero and non-significant. Table 1 also confirms that the decoupling or negative correlation between coups with adjectives in English and Spanish and real coups and attempts, are recent phenomena. In sum, we find support for our initial observation that there is an increase in the use of coups with adjectives, especially in Spanish, and that several of the adjectives are of new date.
What accounts for the increase in the use of coups with adjectives at a time when coups are declining? First, for Latin America we believe it is related to two key developments. Second, and most importantly, we argue that it is related to a change in the concept structure of the term coup, which constitutes a special type of prevalence-induced concept change (Levari et al., 2018).
Two developments in Latin America help explain the rise of coups with adjectives in particular in the Spanish language. One is the new political instability, which refers to the many cases of early presidential termination by impeachment (Pérez-Liñán, 2007), or through resignation after massive protests (Hochstetler, 2006). Not only was the use of impeachment, or impeachment-like procedures, new to the region, it was also new that democracy tended to survive the fall of presidents. Second, the revival seems linked, but is not exclusive, to the wave of leftist regimes in the region and the fall of presidents of the left. While the phenomenon Pérez-Liñán and Hochstetler analyzed mainly occurred with neoliberal presidents, a recent development is that left-leaning presidents have begun to fall (Presidents Rousseff in Brazil 2016, Lugo in Paraguay 2012, and Zelaya in Honduras 2009) or to be questioned for their democratic merits (Presidents Maduro in Venezuela and Ortega in Nicaragua). In a region where United States meddling and coups directed against the left were trademark, the fall of presidents of the left has revived bad memories. 6 Yet, this explanation is incomplete, unless we can explain how events formerly not recognized as coups are now labeled as coups with adjectives.
Our key hypothesis is linked to prevalence-induced concept change, or concept creeping, which is a theory from psychology that holds that “when instances of a concept becomes less prevalent, the concept may expand to include instances it previously excluded” (Levari et al., 2018: 1465). Over a series of experiments, Levari et al. showed how the reduced prevalence of a phenomenon made the understanding of the phenomenon expand so that the participants over-reported its occurrence. This explains why even though modern societies have made progress in fighting poverty, hunger, and infectious diseases, people still believe the world is going in the wrong direction (Levari et al., 2018: 1465). We contend that the same takes place with coups; as they become less frequent and less dangerous to democracy, observers expand their understanding of what a coup is. To be sure, our findings are not strictly consistent with this theory, since it is the use of coups with adjectives, and not the use of the term coup alone, which has increased. This finding is even more puzzling since adding adjectives to a classical concept, such as a coup, reduces its extension (coverage) by increasing its intension (meaning) (Collier and Mahon, 1993; Sartori, 1970). To make sense of why the use of coups with adjectives increases in a period when actual coups are at historical lows, we argue in the remainder that the prevalence-induced concept change has not altered the constitutive elements of a coup, or what a coup is, but rather changed the concept structure of coups from classical to a family resemblance type, thereby reducing the threshold for an instance to qualify as a coup.
What is a Coup?
A coup is not an essentially contested concept like democracy or populism (Gallie, 1956; Mudde, 2004). Even though disagreement exists, there is a relatively broad academic consensus on what constitutes a coup. The term, however, is much used outside academia, and controversial because it is politically loaded and holds normative and practical implications. The United States, for instance, is bound to suspend foreign aid if a recipient experiences a coup, and the European Union upholds similar clauses (Marinov and Goemans, 2014: 805). In Latin America, the Inter-American Democratic Charter of the Organization of American States calls for immediate action in case of a coup (“unconstitutional interruption”), and foresees the potential suspension of the member state in question (see, OAS, 2001: arts. 19–21). In addition, a distressed president has incentives to brand his belligerent opposition as coup-mongers, while the opposition would like to cloak themselves in innocence as fighters for freedom and democracy against an authoritarian-minded president. Likewise, coup-makers often attempt to legitimize their own coups by calling their actions “revolutions,” and themselves a “movement.” Just as “one person’s terrorist is another person’s freedom fighter” (Wardlaw, 1989), one person’s coup-maker can be another person’s democratizer.
In the seventeenth century, a coup was the exceptional and illegal act of a prince against other parts of society in order to deal with an extraordinary situation. Its goal was to restore order, and the prince’s illegal act was considered legitimate since its aim was to benefit the general interests (Bartelson, 1997; Martínez, 2014). 7 With the end of absolutism and the establishment of the principle of popular sovereignty, which separated the ruler(s) from the State, the coup concept changed into how we know it today (Bartelson, 1997). One classic definition comes from Luttwak who defined a coup as “the infiltration of a small but critical segment of the state’s apparatus which is used to displace the government from the control of the remainder” (Luttwak, 1969: 12). Coup-makers use the state machinery to take control over the political levers. According to Luttwak, a coup shares characteristics with revolution, putsch, rebellion, pronunciamiento, civil war, and national liberation. Occupied with differentiating the coup from these phenomena, Luttwak focused on the state turned against the state, and distinguished a coup from other take-over strategies in its being performed by a small group. Another classic definition that seeks to distinguish a coup from a barrack revolt (or cuartelazo) is Finer’s (1962: 154), 8 “the seizure and elimination of the person of the head of state.” Finer (1962: 155) also pointed out that a coup is “speedy and immediate,” but did not include these characteristics as part of the definition. In fact, the term coup in French or golpe in Spanish means a quick movement, a stroke, a blow, or a sudden attack. We stick, however, to Finer’s parsimonious strategy, and exclude speediness and other common traits such as secrecy and small group size of coup-makers from our definition. Had we added these criteria the distinction between the old and new conceptualizations of coups would become even starker.
Based on the triple criteria of target, perpetrator, and tactics, Powell and Thyne summarized the literature and arrived at the following precise definition: “[a coup attempt includes] illegal and overt attempts by the military or other elites within the state apparatus to unseat the sitting executive” (Powell and Thyne, 2011: 252). Neglected in classic definitions, but emphasized by Powell and Thyne and McGowan (2003), is that the tactics used must be illegal. Powell and Thyne (2011: 251) write, “this is important because [it] distinguishes coups from political pressure, which is common whenever people have freedom to organize.” This illegal act also normally involves a suspension of the constitutional order, whether democratic or autocratic, thus connecting this criterion to the original root of the concept in the seventeenth century. The earlier omission of illegality was due to three factors which are as follows: (1) the definitions aimed to differentiate between illegal power grabs—thus the criterion was superfluous; (2) the authors were occupied with military coups—again, the criterion was superfluous; (3) many coups occurred in regimes that failed to meet the criteria for rule of law and democracy, which made the legality/illegality distinction less important.
The legal-illegal distinction is especially relevant in a democratic context, and therefore in today’s Latin America from where we draw many of our examples. In contrast to previous periods, most countries in Latin America since the Third Wave are democracies with at least a veneer of rule of law. Therefore, it is imperative to distinguish coups from legal removals of the heads of state such as impeachments or popular recalls, which have become increasingly common (Helmke, 2017). Before the Third Wave, the pitfall of coup definitions was conflating coups with other illegal attacks on the government, for example, military rebellions. 9 In today’s democracies, the conceptual peril is conflating a coup with legal tactics for government replacement.
Coup d’état —Constitutive Elements
Looking further at the constitutive elements—the defining criteria—of the concept, we proceed with Powell and Thyne’s (2011) definition because it summarizes well what a coup is.
Actor/Perpetrator
A coup is performed by actors within or belonging to the state. Although the military is often the focus, and a major historical concern in Latin America, it is not the only actor that can perpetrate a coup. Coup-makers may also include the judiciary, congress, and the civil bureaucracy. This criterion excludes external agents as principal coup-makers, thus distinguishing a coup from an invasion or a revolution (David, 1987: 8), but it does not mean that external actors cannot influence coup-makers, as has been the experience with the United States in several cases (see for example, Thyne, 2010).
Victim/Target
The victims of a coup d’état are the government, head of government and/or state. The goal of coup-makers is to implement a change in government—either by taking power themselves (e.g. military in government) or by putting someone else in charge (e.g. military as o poder moderador). Some scholars, however, insist that a coup is an attack on the regime rather than the government (Belkin and Schofer, 2005; Lunde, 1991; Varol, 2017), in other words that the coup is also directed at the set of rules that regulate how the state is governed. By narrowing the victim to the government, a coup needs not by definition change the nature of the regime, for instance a coup in an autocracy may foster a new autocracy (Aksoy et al., 2015), but regardless of who is defined as target, a coup will hold consequences for the regime. 10
How/Tactics
A defining characteristic of a coup is that the government removal is illegal or also unconstitutional since it normally involves at least the temporary suspension of the constitutional order. This criterion distinguishes a coup from legal procedures of removal allowed in democracies (impeachment, popular recall, vote of no confidence). Although the legal/illegal distinction seems clear-cut, this is not always easy to determine. As scholars, we may read and interpret the laws and constitution to assess the legality of the removal of a head of state. Our interpretation, however, may not matter to the actors on the ground. Two issues of controversy arise: (1) How should we interpret cases in which a state institution is given the right to adjudicate conflicts between other state institutions and take measures to end them? If a Supreme Court or the military is entitled with such powers and they order the arrest of the president, is that a coup? (2) Who is entitled to determine the legality of the actions that removed a president? In the literature, the issue is not discussed. As for large-N datasets, should we base our operationalization of legality on press reports, on the local constitutional court (the final interpreter of the constitution), or on an international institution such as the European Union or the Organization of American States, which are entitled to take diplomatic action? Or should we as researchers (in political science—not law) make the call based on our own interpretation of local laws or general rules of democracy? Any of these strategies can be chosen. None is unproblematic. A Supreme Court in a less than fully democratic regime may not be trusted, as its evaluation of legality may be politicized. A recent example is the coup in Honduras in 2009 when the Supreme Court took side with the opposition to President Zelaya and declared his removal by the military as legal. Evaluations by international institutions may be equally tainted by political considerations; and the press and scholars may have access to imperfect information sometimes riddled with personal bias.
The Conceptual Structure(s) of a Coup d’état
Following Goertz (2006), the constitutive elements of a phenomenon make out the ontology of a concept, what a phenomenon really is. Concept structure is linked to how the constitutive elements are connected in the construction of a concept, and is important for measuring and quantifying a concept. There are two main schools of constructing a concept. First, one can treat all the constitutive elements of the concept as necessary and jointly sufficient for the phenomenon of interest. This constitutes the classical or Aristotelean way of constructing concepts (Sartori, 1970, 1984). The other is the family resemblance approach, which does not identify any necessary criteria, all the constitutive elements need not be satisfied in order for a particular case to be regarded as an instance of the phenomenon of interest (Goertz, 2006). Therefore, concept structure is important for identifying which cases or events qualify as instances of the phenomenon of interest, and for how to deal with conceptual stretching (Collier and Levitsky, 1997; Collier and Mahon, 1993; Sartori, 1970).
As regards concept structure, coups have traditionally been understood as a classical concept consisting of a set of necessary and jointly sufficient criteria. Powell and Thyne (2011) see the three constitutive elements as necessary and jointly sufficient in order to classify an event as a coup. Coups are also typically dichotomous: either it (or an attempt) has occurred or it has not. Although it does not follow by definition, a dichotomous measure of a concept lends itself to a classical concept structure.
The Coups They Are a-changin’: Conceptual Stretching and Coup with Adjectives
With concepts, we aim to capture identical or similar events under the same concept, and use that concept to distinguish a class of events from others. In other words, as empirically oriented scholars, we are interested in validity—the idea that concepts accurately measure something in the real world—and differentiation (Goertz, 2006; Sartori, 1970). Comparing cases through complex concepts is not an easy task, and we can apply several strategies to avoid problems of conceptualization. Sartori was preoccupied with conceptual stretching; a problem fed by conceptual traveling. He argued that when scholars expanded the comparative perspective (i.e. increasing the N) they tended to broaden the meaning of the concepts in order to be able to incorporate under its expanded rubric the larger realm of observations. This way, social scientists ended up saying less and doing it less precisely (Sartori, 1970: 1033–1035). Falling into the trap of conceptual stretching means identifying two different phenomena by the same name, or making “pseudo-equivalences” that do not differentiate between phenomena. Sartori identified the problem and offered the remedy, a ladder of abstraction. In order to increase the extension of a concept, one needs to reduce its intension—the number of defining characteristics. This way, a more universal concept would not lose precision. The other strategy is to increase intension—climbing down the ladder of abstraction—by adding secondary characteristics to the root concept in order to differentiate. The concept would thereby cover fewer cases, losing extension while gaining in precision.
Collier and Levitsky (1997) point out that climbing the ladder of abstraction, or the ladder of generality as they called it, avoids conceptual stretching, while descending the ladder increases differentiation. However, one cannot achieve both goals simultaneously. Using democracy as an example, they identify three strategies to avoid stretching while enabling differentiation. The first is to use diminished subtypes whenever one of the defining attributes is missing. This strategy is akin to converting the concept structure from a classical to a family resemblance type, where the diminished subtype is distinguished from the root concept by adding an adjective. The second strategy is to increase the precision of the definition whenever faced with new empirical developments. The third strategy is to shift the overarching concept itself. Collier and Levitsky’s example is to change democratic regime with state. We now use these insights to analyze the concept of coup (with adjectives).
Distinguishing Between Coups—Classical Strategies
It is not only Latin Americans that love to distinguish, as Finer’s initial quote suggests, academics do as well. Therefore, despite the commonly accepted definition of an illegal overthrow of the government by actors within the state, social scientists have used different strategies to differentiate between types of coups. The first strategy has been to go down the ladder of abstraction by adding an adjective that regards a criterion non-essential to the definition. The second strategy has been to walk up the ladder of abstraction, turning the term into a subcategory of a broader concept.
Walking Down the Ladder of Abstraction
A paramount example is the military coup. Adding the adjective military is uncontroversial, since it narrows the focus to which actor performs the coup: the extension is thus reduced by excluding purely civilian coups. Such a restriction of scope makes sense if the focus is on military tactics, strategies, planning, and organization, or on government strategies toward the military to prevent coups (Belkin and Schofer, 2005). Since the military forms part of the state, distinguishing military coups from others does not challenge the root concept.
Other coups with adjectives within this strategy are the neoliberal coup (Mauceri, 1995), adding a political direction, and the “promissory coup” where coup leaders promise return to democracy once order is reestablished (Bermeo, 2016). One can also find references to right-wing coups and leftist coups (Hamilton and Hamildon, 1983; Tovias, 1984), as well as democratic coups—which end an autocratic regime and install a democracy—and non-democratic coups—which end democracy and install autocracy (Varol, 2017). Similarly, Aksoy et al. (2015) distinguish between reshuffling and regime-change coups in autocracies depending on coup-makers intent and the outcome of the coup.
Examples of walking down the ladder of abstraction with coups are presented in Table 2.
Examples of walking down the ladder of abstraction with coups.
A: Actor/Perpetrator = state; B: Victim/target = government; C: How/tactics = illegal; D: State actor/perpetrator = Military; E: Outcome = democracy; F: Outcome = autocracy; G: Outcome = neoliberal politics.
The Self-Coup—Walking Up the Ladder of Abstraction
Reminiscent of Naudé’s definition from the seventeenth century, the modern self-coup or autogolpe is a more troublesome concept. Although illegal and supported by force or the threat of force, and also perpetrated by state actors, the autogolpe—which has also been called a constitutional or a presidential coup (Helmke, 2017; Roberts, 1995; Varol, 2017: 30)—changes the target from the head of government to other state institutions such as congress or the judiciary. The term was popularized with President Fujimori’s actions against congress and the constitution in Peru in 1992. Cameron (1998: 125) defines it as “when a president closes the courts and the legislature, suspends the constitution, and rules by decree until a referendum and a new legislative elections are held to approve broader executive power.” By changing the target, we argue that scholars have stepped up the ladder of abstraction, and identified the self-coup as a subtype of a coup understood as directed at the regime. Regime is a more abstract term than government and involves all key state institutions that share or divide power.
Whereas there are dangers with this strategy, as one risks conflating coups with other phenomena, it avoids stretching. A further critique points to the term itself: self-coup leads one to think of shooting oneself or slashing out at oneself, which is inappropriate given that the one who gains from performing the self-coup is the one leading it. The autogolpe does not refer to self-harm but to the harm to others. To avoid conceptual confusion, a more appropriate term may be Svolik’s incumbent takeover. However, the autogolpe or self-coup has gained academic ground and is referred to regularly in the press. A Google Scholar search of self-coup (in July 2019) resulted in over 1700 hits, and autogolpe returned over 5500 hits. 11 Thus, the term is here to stay.
The Coup as Family Resemblance: New Strategies for Differentiation
Distinguishing between similar phenomena by adding adjectives to classical concepts reduces the extension of the concept by increasing the intension. Thus, classical strategies of using coups with adjectives cannot account for its increased use at the same time as the number of coups is reduced. In fact, much of the current usage and rise of coup with adjectives falls outside the classic conceptualization, and changes the concept structure of coups into a family resemblance types where adjectives create diminished subtypes. In effect, diminished subtypes expand the concept to include instances it previously excluded, and since their increased usage occurs simultaneously with the reduced prevalence of the phenomenon, our findings are congruent with the theory of prevalence-induced concept change (Levari et al., 2018).
To make sense of the proliferation of new adjectives that do not satisfy all of the three commonly accepted criteria for a coup, we argue that many coups with adjectives follow a conceptualization of coups or mandate interruptions based on a family resemblance structure (Figure 6). This entails that there are several subtypes of coups, but no necessary or sufficient condition is common to them all. By keeping the classical constitutive elements, that is, perpetrator, target, and tactics, four combinations arise:
(a) If the perpetrator is a state agency, the target is the chief executive, and the removal is illegal or unconstitutional, the act is the classic
(b) If the chief executive is removed through illegal means but the perpetrator is not a state agency, the act is a
(c) If the perpetrator is a state agency and the tactics are illegal but the target is not the chief executive, the act may be called a self-coup or
(d) If the perpetrator is a state agency and the target is the executive but the process is conducted through legal tactics, the act is diversely called an
Coups as a Family Resemblance Concept.
An electoral coup can be understood as the use of elections as a tool to illegally keep or change power, and as such also resembles the concept of electoral fraud. The concept has mainly been used not only by political actors, among them is Venezuela’s President Maduro in connection with the parliamentary elections in 2015, but also by academics and political observers to describe electorally dubious processes in Jamaica, Nicaragua, Venezuela, and Honduras (e.g. Hellinger, 2005; Kitson, 1982). Although, when presidents use the term one may doubt the allegations of illegality, the term indicates that actors from outside the state act in an allegedly illegal manner to influence the electoral outcome in a way that is not congruent with the preferences of the majority, thus satisfying one or two of the three criteria of a coup. Furthermore, for coups of this type large masses must be involved to muster votes or organize a fraud, which breaks with the understanding that a coup is performed by a small group. Finally, an interesting case is the market coup (“golpe de mercado”). The term was popularized in Argentina in 1989 when market actors bought hundreds of millions of subsidized dollars depleting the country’s international reserves and starting a chain of negative economic consequences that led to President Alfonsin’s early resignation (Smith, 1990). Thus, no state actors were involved, but in contrast to the above-mentioned coups of this type, the actions were legal, and the term only satisfies one of the three criteria.
The slow-motion coup has also appeared to describe Venezuela’s transition from democracy to authoritarianism, and is defined as a process “whereby a consolidated democracy slowly descends into a hybrid regime and from there into authoritarianism, through the creation and implementation of conscious decisions made by its ruling elite” (Polga Hecimovich et al., 2017: 37). As such, the process involves state actors, illegal actions, and/or use of force, but the victim is not the president. The process thus resembles the judicial coup, or an incumbent takeover, but is less specific on the perpetrator. Finally, the qualifier “slow” breaks with the understanding of a coup as something swift and quick. In the end, coups of this type can be understood to satisfy two of the three criteria.
The constitutional coup describes the use of constitutional measures to remove a president, or prolong his stay in power. The latter case is used particularly in the African context (Camara, 2016; Kotze, 2017), 13 and is similar to the judicial coup mentioned earlier. In Latin America, the term has mostly been used by political actors and the press, but appears frequently in Spanish in Google Books since the 1980s. The term constitutional coup does not indicate an actor, and it admits to the (controversial) constitutionality of the action; hence, the term only satisfies the criterion that the chief executive is the target, and where it is used to describe a president’s expansion of term limits, the term actually does not satisfy any of the criteria for a coup even though the actions are identified by that term.
Adding Adjectives, Subtracting Meaning: Discussing Coup With Adjectives
The cases presented earlier lack either one or more of what have been considered the necessary and jointly sufficient attributes of the root concept. Therefore, they constitute diminished subtypes. The increased use of coups with adjectives has changed the concept structure of coups from a classical to a family resemblance category and driven its prevalence upward in a time when coups are diminishing globally, consistent with prevalence-induced concept change. In some cases, the terms describe actions by masses instead of a small group (e.g. electoral coup), slow instead of quick processes (e.g. slow-motion coup), or actions that are not shredded in secrecy (e.g. constitutional or parliamentary coup). Therefore, several of these examples also fail to satisfy logical but omitted criteria from modern definitions (secrecy, speediness, and small group size).
Diminished subtypes may increase differentiation and avoid stretching (Collier and Levitsky, 1997). The root concept is left intact, but it is applied with qualifications to cases that are not considered “full and complete.” This strategy allows for flexibility, so that new forms and shapes of mandate interruptions can be captured by a well-known concept. For example, the coup concept itself has changed considerably since Naudé’s definition (Martínez, 2014), and the Third Wave democratization and the new political instability in Latin America contributed to conceptual changes that reflected empirical developments, for instance by insisting that coups should consist of illegal actions (Powell and Thyne, 2011). For terms such as democracy (Collier and Levitsky, 1997) or populism (Weyland, 2001), changing the concept structure from classical to family resemblance has been a successful strategy to capture empirical developments. And although Sartori (1984) was skeptical toward family resemblance concepts, these are now well established in the social sciences and their use does not inhibit positivist, causal analysis (Goertz, 2006).
Nevertheless, we identify three problems with the use of diminished subtypes and the coup concept with a family resemblance structure. The first is that it inhibits the recognition of political evolution. Second, it creates the danger of causal and conceptual confusion. Third, getting a coup wrong may entail serious political consequences.
First, contrary to other concepts such as populism—which expanded to account for the increasing presence of populists (Rooduijn, 2018)—we link the changing structure of the coup concept to its decreasing empirical presence. The relative scarcity of coups, we argue, has led to prevalence-induced concept change. By adding adjectives, observers have increased the extension while reducing the intension of the root concept, therefore overreporting coups at the same time as their occurrence ebbed. For Latin America in particular, changing the concept structure inhibits observing the real progress made in reducing coups as threats to democracy. Even though scholars and the media mention coups (with adjectives) more often than before, the phenomenon is minimal in the region.
Second, applying a family resemblance structure has created causal and conceptual confusion. By conflating coups with other concepts, the plethora of event that coups with adjectives refer to are also assumed to share the same or similar causes and consequences. Figure 6 shows that for every subtype of coup with adjectives under the family resemblance structure, there exists established concepts such as revolution and impeachment. The strategy of applying diminished subtypes allows for differentiating the key concept while traveling, but it also creates confusion between, for example, a civil society coup and a revolution, and between a parliamentary coup and an impeachment. It further creates pseudo-equivalences between phenomena as different as an impeachment and a revolution by placing both under the conceptual umbrella of a coup, which inhibits rather than improves our understanding and leads to faulty conclusions.
We argue therefore that applying a family resemblance structure and diminished subtypes obfuscate rather than clarify the coup concept. Even though adding adjectives can enhance the precision in order to capture new developments, such as the new political instability or new ways of democratic deterioration, one problem remains, the recent usage of diminished subtypes has not led to an agreement over a more precise concept. The new subtypes point toward different processes which cannot be captured by a single new criterion and several adjectives such as “slow” or “constitutional” contradict the original meaning, constituting dismissive subtypes. In some versions, state actors are not involved, in other the actions are legal, in some the president is the target, in other all institutions but the presidency are the target. Unlike similar developments with the concept of democracy, which generated a more precise definition, coups with adjectives that constitute diminished subtypes obfuscate rather than precise the root concept.
The conflation of coups with adjectives with other established concepts and phenomena lead to causal confusion and flawed analyses. Scholars explaining democratic backsliding and recession often highlight the differences between today’s dangers to democracy and classical coups. Causal processes leading to democratic backsliding are slow and incremental rather than quick, led from the government rather than against the government, and weakening institutions rather than aborting them (see for example, Bermeo, 2016). As Svolik (2015) points out, democracies grew immune to coups, but not to incumbent takeovers. Conflating these processes is likely to mislead causal analysis and policy prescription since when the diagnosis change, so should the therapy.
Finally, classifying a change in government as a coup may hold real-life consequences. During the Cold War, the military in Argentina, Brazil, and Paraguay classified their coups as revolutions and their organizations as movements to legitimize their actions. Today scholars, observers, and presidents on the losing end of an impeachment classify such events as coups (with adjectives) to delegitimize their opponents. A crisper definition is in order since labeling an event as a coup may generate political actions of grave consequences such as the withholding of aid, the suspension from international organizations, the triggering of international sanctions, and even foreign military intervention. To reduce the dangers of wrongful categorization of coups, and to distinguish politically motivated accusations or whitewashing, a clear definition with a set of jointly necessary and sufficient conditions is more useful than a family resemblance structure.
Conclusion
Whereas the threat of coups is lower than ever before, the dangers of concept misformation of coups is rising. Even though we can identify several recent interesting and positive conceptual developments in the literature that distinguish between different types of coups (Aksoy et al., 2015) or between coups and mutinies (Dwyer, 2015; Johnson, 2018), our main concern has been with the proliferation of diminished subtypes of coups. In today’s democracies, we particularly warn against conflating coups with legal tactics for government replacement. We have shown how the use of coups with adjectives, in particular in the Spanish language, has increased considerably in both the scholarly and non-scholarly literature while the prevalence of coups has decreased both globally and in Latin America. We argue that the reason for this puzzling development lies in prevalence-induced concept change, which has come in the form of changing the concept structure of coups from classic to family resemblance.
Originally suggested as a way of preventing conceptual stretching and creating analytical differentiation, we showed that when applied to coups, this strategy prevents concept stretching only partially and at a high cost. The risk is that proliferating adjectives end up diluting the noun: a coup becomes tantamount to just any mandate interruption and, as such, is virtually meaningless without adjectives. This further creates conceptual and causal confusion, and inhibits observers from identifying real world improvements. Politically, the spreading of coups with adjectives can also have damaging consequences. Classifying as coups phenomena that belong to another class, such as impeachments, raises a set of normative and, more consequentially, policy-related questions, as powerful actors could set measures in motion according to the label that best fits their interests.
We conclude with a humble reminder for scholars and a warning for practitioners. The reminder, however obvious, is that not every removal is a coup, as a Latin American political activist ironically quipped, “golpes eran los de antes” (“real coups were those of yesteryears”). The warning should alert political leaders and activists that stretching the concept by accusing today’s adversaries of coup-mongering may hit them back tomorrow.
Footnotes
Acknowledgements
The authors thank Carlos Gervasoni, Mariana Llanos, Miguel De Luca, Michael Alvarez, and two anonymous reviewers for comments, and Octavio Amorim Neto and Michaël Tatham for conceptual and technical input.
Authors’ Note
A previous version of this article was presented at the 9th Congress of Asociación Latinoamericana de Ciencia Política (ALACIP), Montevideo, Uruguay, 26–29 July 2017, and at the American Political Science Association Annual Meeting, Washington DC, 29 August–1 September 2019.
Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Andrés Malamud acknowledges the financial support of FCT project UID/SOC/50013/2013.
