The Arab uprisings and the return of repression

ABSTRACT The Arab uprisings of 2011 led to a reassessment of comparative politics research on authoritarianism in the Middle East and North Africa. The Arab region made its way from area studies into mainstream comparative politics, and research foci have shifted towards civil-military relations and repression. Ten years later, we observe higher levels of repression across the region, reflecting a diversity of repressive trends. Advocating comprehensive research on this variation, we review recent literature that tackles various dimensions of repression in Arab autocracies. In addition to disaggregating forms and targets of repression, we call for its justifications, agents and transnational dimensions to be considered next to the implications of digital technologies of coercion. We also reflect on how repression affects the possibility of doing research and how we can investigate the proposed dimensions of repression.


Introduction
The Arab uprisings of 2011 caught researchers studying authoritarianism in the Middle East and North Africa (MENA) off guard. Throughout the region, citizens contested long-standing authoritarian regimes. This astonished both observers who had propagated an 'Arab exceptionalism', painting Arab states as different from other world regions and resistant to democracy, and researchers studying the durability of authoritarianism. The mass protests challenged arguments about the strength of incumbent governments vis-àvis citizens, revealing severe crises of legitimacy. The region-wide contention in 2010/11 was an exceptional moment that has become the starting angle for current analyses of Arab politics. Like the political liberalizations in the early 1990s, the Arab uprisings initially triggered a wave of optimistic expectations for democratization among researchers. As a meaningful phenomenon of global relevance, the Arab uprisings have provoked the interest of mainstream political science. Over the past decade, the Arab region has gained importance for theory-building in comparative politics (CP), thus becoming attractive beyond area studies (see the discussion of the Area Studies Controversy in the introduction by Bank and Busse). However, this has also been criticized by scholars from the region as 'Academic Tourists Sight-Seeing the Arab Spring' (Abaza, 2011). The Arab uprisings are now cited as one epochal wave of contention comparable to the Third Wave of Democratization or the Colour Revolutions, elevating them to a critical juncture in the sense of this Special Issue's introductory article.
One year after the beginning of mass protests, a special issue in this journal proclaimed, 'it is clear that the end product of the MENA peoples' demands will be a more accountable political system' (Pace & Cavatorta, 2012, p. 135). A few years later, the region offers a more sobering picture, as counterrevolution did not wait long to arrive. Although rulers fell in Tunisia, Egypt, Libya and Yemen, the only country where democratization has remained on track towards consolidation is Tunisia. The trajectories of other states look less favourable, as Egypt is under the grip of a military dictatorship more repressive than ever under Mubarak. The Bahraini monarchy has also repressed a strong protest movement with regional assistance. In the violent, internationalized conflicts in Libya, Syria and Yemen, statehood is failing, and besides official security forces, militias play a significant role in killing and displacing large parts of the population. The bulk of Arab countries somehow muddled through under the pretence of reform, mixing some political, but mainly material concessions with a strong crackdown on activism. Except for Tunisia, all states display a higher degree of repression today than they did in 2010.
In this article, we review CP research that has emerged in the context of the Arab uprisings, with special attention to repressive variation among Arab autocracies. We call for a comprehensive, disaggregated conceptualization of repression, arguing that such a research focus helps understand many political developments in Arab states since 2011. The regional dynamics corroborated previous findings of research on the repression-dissent nexus. Nonetheless, empirical developments shed new light on theoretical and conceptual assumptions (Davenport & Moore, 2012). The topic of repressive variation among non-democratic regimes has been under-researched. Given the pervasiveness of authoritarian rule in the MENA, the region seems particularly well-suited to advance this research agenda, which we aim to do.
The remainder of the article is structured as follows: after a brief overview of authoritarianism research before and after the Arab uprisings, we show the heightened level of repression across MENA states. We then outline various repressive trends that we observe empirically in different states and review innovative research that pushes the boundaries of previous scholarship on repression. We distil various dimensions of repression that have emerged as important since the Arab uprisings and that we believe to be relevant for studying repression more generally. For this purpose, we look at established categories of repression, such as its levels, forms, agents and targets. Going beyond these common disaggregation measures, we propose a perspective incorporating justifications, multiple levels of analysis and new digital tools that authoritarian incumbents use to control their citizens. We contend that considering these aspects can illuminate current political developments and future dynamics across the region and beyond. We then tackle questions of fieldwork and research ethics, as well as sources and methods for studying repression before concluding with the lessons learned.

Research on authoritarianism and repression before and during the Arab uprisings
By the end of the Cold War, hopes for democratization in the Arab world imbued scholarly works, before this wishful thinking gave way to research on actual developments in Arab politics. In the 2000s, comparative political scientists focusing on MENA states studied the recipes for regime durability in the region, mostly focusing on formal political institutions or elite-level politics. Repression was assumed to be a relatively constant factor (Albrecht & Schlumberger, 2004, p. 375). Studying political phenomena 'beyond coercion' was prioritized, probably because blatant political violence in the public sphere was rare. Hence, despite the presence of systematic repressive policies, research on repression was underrepresented, with some notable exceptions (Bellin, 2004;Hibou, 2011).
Since 2011, CP research has tackled such puzzles as the uprisings' motivations and triggers, the spread of protests and divergent trajectories (Yom, 2015). Mirroring empirical developments and their consequences for scholarship, research foci have diversified and reconfigured (see Valbjørn, 2015;Kao & Lust, 2017;Bank, 2018; and the introduction by Bank and Busse). Repression made a strong return to research regarding two central topics: the protestrepression nexus and civil-military relations. The relevance of repression as a research topic was evident in the varying state reactions to the mass protests in terms of policing. Military behaviour decided on the survival of political regimes in 'dictators' endgame' scenarios (Bellin, 2012) or 'endgame coups' (Koehler & Albrecht, 2021), inspiring studies that brought civil-military relations back into research. Counterrevolutionary activities by persistent autocrats made repression even more worthwhile studying. Some scholars date the beginning of counterrevolutions -understood as 'collective and reactive efforts to defend the status quo and its varied range of dominant elites against a credible threat to overturn them from below' (Slater & Smith, 2016, p. 1475) -to mid-March 2011, the time of the Saudi-led GCC intervention in Bahrain and other instances of increased regime violence (Lynch, 2012, p. 131). At the latest in May 2011, governments upscaled their repressive strategies to avoid the outcomes of less fortunate autocrats (Heydemann & Leenders, 2014, p. 87). In 2013, Heydemann predicted that Arab authoritarianism would become 'darker, more repressive, more sectarian, and even more deeply resistant to democratization than in the past' (Heydemann, 2013, p. 72). Indeed, in the years that followed, repression rose markedly throughout the region, as the following section explains.

Repressive trends after the Arab uprisings
The overall rise in levels of repression throughout the MENA region is empirically evident in quantitative data from the Political Terror Scale (PTS), which has measured the violation of physical integrity rights since 1976 (Gibney et al., 2019). The data combining reports from Amnesty International and the US State Department in Figure 1 show that the levels of repression in the MENA were already rising in the late 2000s, but they jumped significantly in 2013 and peaked in 2015. Between 2013 and 2018, they remained at an unprecedented regional high: the Political Terror Scale had never reached similar regional scores for several years in a row.
These data strongly suggest that repression has indeed been rising in the Arab world over the last decade. 1 In that sense, the critical juncture of the Arab uprisings was a game-changer that has lifted the level of repression in the region to heights unknown in the last half-century. Nevertheless, this trend mirrors aggregate data. Arab countries today exhibit a large variance in repressive practices that is not reflected adequately in average levels. To capture and structure this variance, we suggest distinguishing four repressive trends in various Arab states: (1) reduced repression, (2) counterrevolutionary upscaling of repression, (3) repression during violent internationalized conflicts, and (4) readjusted repression.
These empirical trajectories yield different consequences for the relevance of concepts and the usefulness of theoretical arguments. In contrast to conventional approaches that categorize cases according to how protests unfolded in 2011 and where leaders fell (e.g., Hinnebusch, 2015), our trajectories describe the situation in early 2021 and thus present research themes that have emerged as important for analysing the Arab region. In this sense, we move beyond the Arab uprisings angle. The protests in Algeria, Sudan, Iraq and Lebanon in 2019 showed that although the wave of diffusing protests in 2011 was unique, unexpected mass demonstrations occur outside of this contingent window. Repression is complementary to societal mobilization and thus pivotal for evaluating the potential for and processes of (de)mobilization and (de)radicalization. Mirroring Goldstein's and Davenport's well-known definitions, we understand repression as state-or government-controlled action that discriminates against persons or organizations by the actual or threatened use of negative sanctions in the form of direct violence and coercion or indirect and psychological forms of coercion and intimidation in an attempt to deter specific activities and/or beliefs perceived to challenge existing power relationships or key governmental policies. (Edel, 2018, p. 20) Repression played out differently across Arab states. Below we present four dominant trajectories and how they relate to existing theoretical assumptions about repression. All these trajectories point to dimensions of repression that we discuss in more detail in the following section.
(1) In the singular case of reduced repression, the democratization process in Tunisia brought the harsh repression of the Ben Ali era to an end. However, the nascent democracy did not fully support the truth commission that investigated past state violence, as old elites in parliament ended its mandate. Changes in the security apparatus were minor and state archives were not opened to investigators. Nonetheless, the decrease in the level of state repression has been the most dramatic worldwide over the past decade. 2 This corroborates the claim by dominant research that democracies employ less repression than autocracies. At the same time, the case suggests that legacies of the prior state structure and security sector dampen the magnitude of the effect of regime change on repression.
(2) In Bahrain and Egypt, a counterrevolutionary upscaling of repression has taken place. In both countries, mass protests of unprecedented scale had seriously challenged centres of power. Governments responded with massive crackdowns and repression has remained on a high level, even after it crushed oppositional mobilization. In Bahrain, about one in four citizens took to the streets -arguably the largest per capita protests in human history (Yom, 2014, p. 52). The confrontation with monarchical minority rule led to a protracted uprising with harsh repression. In Egypt, mass mobilization first prompted Mubarak's resignation and the military establishment taking over. With mass protests instigated against the Mursi presidency, the military regained direct power through its 2013 coup. In both Bahrain and Egypt, spaces for opposition and civil society have declined significantly. The ratio of political prisoners per inhabitants has skyrocketed. Nevertheless, the two states employ different repressive practices. For example, hundreds of Bahrainis have been stripped of their citizenship (Babar, 2017), while Egypt has become infamous for many forced disappearances. Overall, the upscaling can be traced back to the magnitude of the perceived and real threat to the regime, and the mutual dependence of powerful security apparatuses and the regime centres.
(3) In Libya, Syria and Yemen, repression has occurred in violent internationalized conflicts that have led to state failure and profound transformations of political structures. The large-scale and lethal violence employed by the state apparatuses was not just a consequence of the escalating conflict dynamics. Especially in Libya and Syria, state repression arguably was the most important trigger of the violent conflicts. The repressive environments before 2011 had made the organization of civil society in Libya and Syria virtually impossible. As citizens revolted in the context of the region-wide uprising, state violence triggered escalating cycles of protest and repression. Orders to violently repress the protest movements led to splits in the security apparatus, provoking some military officers to defect from the regime and fight against it. In Yemen, the defection of General Ali Mohsen al-Ahmar prompted a split of the armed forces. Research building on the civilmilitary relations literature has shown that these splits and defections were largely due to a specific composition and structure of the security sector, as well as the ruling elites' relations to (certain) military units, police and militias. Close ties to the regime and privileging some parts of the security sector over others are key explanatory factors for defections. Consequently, state structures broke down, violent non-state actors assumed a large role and the conflicts were sectarianized (see also the contribution by Valbjørn in this Special Issue), not least by international actors.
(4) In the Arab countries where rulers or ruling families remained in power, repression was readjusted. This was the case in Jordan, Morocco, Saudi Arabia, Oman, Kuwait, UAE, Qatar, and Algeria and Iraq until 2019. Whether protests in 2011 were large, small or almost absent, they never posed an immediate threat to regime survival. Observing the transformations in other countries, incumbents appeased protest movements through material and some political concessions, in addition to repressing them. Over time, they introduced more restrictive laws, legally backing repression. Almost all Arab monarchies fall into this group (Yom, 2014), jointly with other oil-rich states that distribute spoils along clientelist networks. The higher legitimacy enjoyed by some monarchies made protesters more likely to demand reform instead of the revolutionary fall of a regime.
This ties in with the general question of how governments combine different strategies of political rule: whether they rely more on legitimation or repression, co-optation or divide-and-rule strategies. However, an analysis of these regime strategies should be complemented with a realistic assessment of their success. The uprisings in Libya and Bahrain show that a mechanistic ascription of stabilizing power to either oil or monarchism does not hold. Likewise, one could regard some readjustment strategies as authoritarian upgrading (Heydemann, 2007). However, in view of lessons from 2011 as well as recurrent protests, we should not assume that all upgrading is sophisticated and ensures regime durability; rather, the consequences of regime strategies need to be assessed empirically in terms of implementation and possible side effects.
As noted, the breakdown of state structures is one central outcome of the political developments in 2011. Thus, it seems important to bring the state back into research, complementing the previous focus on regime structures. Crucially, regime and state breakdown did not always coincide: while the regime in Tunisia broke down but left statehood intact, Syria saw a continuity of regime agents and practices in the face of state breakdown (Heydemann, 2018;Schlumberger, 2016). It is important to 'distinguish more carefully between "regime" on the one hand and "state" on the other' (Schlumberger, 2016, p. 37) in order to grasp the political realities ten years after the uprisings and to analyse the prevailing repressive trends, as characteristics of the state, regime and challenge jointly shape the features of repression (Josua & Edel, 2015).

What to study: Dimensions of repression since the Arab uprisings
Across the Arab region, incumbents increased and/or modified their repressive practices, showing that one of the few law-like relationships in political science also applies in the Middle Eastern context: the higher the threat perception by incumbents, the more repression they employ (Davenport, 2007). The Arab uprisings not only provoked immediate crackdowns but also evoked a sustained feeling of insecurity among powerholders. The latter imposed new restrictions trying to compensate for the severe crises of legitimacy, of which the uprisings were symptomatic (Josua, 2017). Rather than striving for new social contracts in order to overcome these crises and regain legitimacy, most incumbents have intensified repression in order to stay in power. This partially lies in the natures of regimes that display neo-patrimonial features and thus make pacted transitions unlikely (Stepan & Linz, 2013). Besides their 'will to repress,' strong security apparatuses make for a high 'capacity to repress' (Bellin, 2004(Bellin, , 2012. Where this is not the case, external actors are often willing to help out and support incumbents, financially or militarily. The overall rise in levels of repression in Arab states is an insight in itself. Still, the quantitative level alone provides only limited information on political dynamics. An exclusive focus on aggregate numbers obscures the variegated forms that acts of repression can assume. Acknowledging repressive variation is important because of its diverging consequences. Researchers sometimes study the consequences of repression with a perspective of an over-simplified dichotomy of deterring vs. radicalizing effects. However, the conditions under which repression occurs are decisive (Opp & Roehl, 1990, p. 523), as the effects of repression are highly contextual. Often the main effect on opponents is tactical shifts (e.g., Grimm & Harders, 2018;Taylor & Dyke, 2004). Furthermore, short-and long-term consequences can differ considerably (De Jaegher & Hoyer, 2019). Thus, while it may seem tempting to hypothesize one dominant effect of repression, past research has proven that such an approach does not lead us very far. Since repression has more complex effects than previously imagined (Sullivan & Davenport, 2017), our conceptual tools and theoretical approaches need to resonate with this empirical reality. Thus, currently emerging trends acknowledging the importance of microlevel dynamics in dissident movements and disaggregating characteristics of repressionwhich are at the centre of this article -represent a promising way forward.
Below, we present various issues related to repression that have emerged as important since the Arab uprisings and that have received attention by scholars in both North America and Europe alike. 3 To grasp the variance of repression in the region, we propose taking several dimensions into account. After discussing consequences of repression in the current regional context, we delineate not only the traditional categories of forms and targets of repression, but also its justifications, visibility, agents and its extension beyond borders and into the digital sphere. Manifold links between these aspects invite a rethinking of research foci, concepts and categories, and a cross-fertilization of research approaches.

Consequences of repression
Studying the effects of repression is paramount for understanding the dynamics of political contention. The consequences of repression also include medium-and long-term effects. However, only ten years after the Arab uprisings, long-term effects of heightened repression are yet to be observed. While some recent research has pointed to medium-and long-term effects of repression such as polarization (El Kurd, 2019;Nugent, 2020), most of the relevant literature deals with short-term effects. 4 The forms, levels, and degree of indiscriminateness evoke different reactions by protesters that, in turn, shape regime responses. For instance, brutal repression has at times led dissenters to shift their tactics towards more violence, as in Syria in 2011/12. Similarly, in Bahrain, some low-scale insurgency activities were reported between 2012 and 2017. In other cases, the opposite was true, and higher levels of repression crushed the opposition. For example, Egyptians have remained largely non-violent (except for contention on the Sinai) since 2013, downscaling protests to more flexible and decentralized forms (Ketchley, 2017, p. 141).
In the counterrevolutionary and persistent autocracies, repression had some short-term 'success' in containing dissent. However, possibilities for future protests remain. Though facing more obstacles, some demonstrations have occurred under harsh repression. Examples include protests in Egypt against the sale of islands to Saudi Arabia; and even in war-torn Syria, unfaltering citizens in Idlib protested against the regime in 2019. Massive teachers' strikes in Jordan in the fall of 2019 and frequent socioeconomic unrest related to oil or food prices epitomize the tendency to mobilize for single-issue topics (see the article by Weipert-Fenner in this Special Issue). Thus, even after protests seem to be pacified, authoritarian incumbents can never be sure of their grip on power.
Repression may result in higher mobilization of peaceful protesters and even trigger violent responses. Such a backlash originates from dynamics on the microlevel, namely bystanders and activists feeling anger or outrage over use of violence that they perceive as disproportionate or unjust. These emotions spur mobilization even in high-risk settings, as shown for Egypt under El-Sisi (Ayanian & Tausch, 2016). For example, the lethal violence that Algerian security forces employed against mass protests in 1988 led to a backlash resulting in political liberalization, free elections that were then aborted, and finally civil war.
In order to understand such varying consequences of repression, it is crucial to pay attention to its contextual characteristics. We support the growing scholarly consensus that the targeting and forms of repression are important in this regard. Our own contribution is a call for also considering justifications, visibility and agents of repression. These factors determine how repression is perceived, which, in turn, shapes reactions to it.

Forms and targets of repression
A common way of disaggregating repression is to look at its forms and targets. As mentioned, in protest policing it makes a great difference whether security forces employ lethal repression or not. We distinguish between three different forms of repression: constraining, incapacitating and eliminating repression, which differ in their effect on the direct targets. Measures to constrain challengers range from subtle to grave restrictions of civil liberties as well as various forms of physical harassment that leave some room for manoeuvre. By contrast, incapacitating and eliminating repression completely forestall the targets' capacities to continue their activism. Incapacitating repression mostly takes the form of imprisonment, house arrest, exiling and temporary disappearances, whereas eliminating repression results in the target's death. For example, in 2019, the governmental reaction to protests in Algeria largely took the form of constraining repression through preventive tactics such as cutting off access to protest sites and moderate policing. Security routines in Iraq differed markedly, as security forces employed lethal force, killing hundreds of protesters. Despite a similar challenge in these two republics, the patterns of repression displayed remarkable variance. Besides the type and level of the challenge, state resources and preferences, regime characteristics and past experiences thus influence the choice of certain forms of repression (Josua & Edel, 2015). Although monarchies used less highintensity repression (Yom, 2014) in line with established findings, they display wide variation (Lawson & Legrenzi, 2017). To explain the impact of regime types, the proposed distinction between incapacitating and eliminating repression could be helpful.
Constraining repression is reconfigured across the region, as legal frameworks have become much stricter. 5 For instance, Gulf states increasingly strip dissidents of their citizenship (Babar, 2017). Restrictive laws with broad definitions of offences pave the way for the arbitrary repression of regime opponents that is now enhanced through 'legality.' Thus, repression is practiced more openly, while international actors such as the US government have toned down their criticism of human rights violations.
The targets of repression vary according to whom incumbents perceive as a threat. They may be dissidents or protesters, but also members of certain ethnic, cultural or sectarian communities, ideologically defined groups such as secularists, communists or Islamists, and intellectuals or academics. 6 With the electoral victories of Muslim Brotherhood parties in numerous Arab states, their opponents tried to push back by outlawing the organization as such (Darwich, 2017). This move has been successful in some states but not in those where the groups have been historically more embedded in societies and integrated into the political systems. In virtually all states, journalists and bloggers have increasingly been restricted by legal frameworks.
Whether repression is targeted or indiscriminate is an important distinction that shapes the consequences of repression. Lawson and Legrenzi (2017) showed that to explain different repressive outcomes in Gulf states we need to analytically distinguish between the level of repression and its degree of indiscriminateness. As Blaydes (2018) found for Iraq under Saddam Hussein, targeted repression is often used when surveillance capacity is higher. Nugent (2020) demonstrated that targeted repression led to the polarization between groups and thus hindered co-operation among the opposition in Egypt, while indiscriminate repression in Tunisia laid common ground that later enabled co-operation among political forces during the transition. The extent to which repression is targeted is also important in relation to digital techniques of repression, which we discuss below.

Justifications and visibility of repression
Highly visible forms of repression attract the attention of the broader population and sometimes even alert the international community. Anti-regime mobilization on the domestic level, or shaming and sanctions on the international level, are likely consequences. To avoid such backfire scenarios, incumbents often try to either justify repressive acts or decrease the latters' visibility: They hide their repressive measures from the public or put forward rhetorical or procedural justifications to influence how the former are perceived (Edel, 2018). This can take the shape of securitization (Pratt & Rezk, 2019). Scartozzi (2015) demonstrated how the changing narrative in Syria influenced popular support for the regime in the early years of the civil war. Autocrats often frame dissenters as enemies of the state, as was the case in Egypt when Muslim Brotherhood supporters were massacred after the 2013 coup (Edel & Josua, 2018).
Such rhetorical framings can be backed by arguments about legality. Linking back to legislation shrinking spaces for societal organization and free expression, governments use their discretion to claim illegality and forestall activism (Josua, 2020). Declaring that dissidents or groups operate illegally and should be shut down on legal grounds is a powerful tool in the hands of autocrats, not least towards external observers. Incumbents often claim that protesters are criminals to hinder larger mobilization or sympathy among the larger population. Another popular frame is to accuse opponents of being terrorists (Josua, 2020). The nascent research areas of judicial repression and justifications widen the common focus of repression research to go beyond protest policing.

Agents of repression: Bringing civil-military relations back in
Kao and Lust called for considering human agency more in the study of Middle East politics (2017). While we deem structures as essential in fostering (and, importantly, foreclosing) political dynamics, we concur that agency is an essential factor, also with regard to repression.
From the 1950s to the 1970s, the heydays of military coups in Arab republics, the relevance of the military was obvious to analysts of the region. More recently, only a few scholars have studied coup-proofing strategies (Kamrava, 2000) or military elites. Since 2011, however, militaries have reemerged as critical actors. Stacher argued that '[t]he Arab uprisings' most visible product to date has been the militarization of politics and societies' (Stacher, 2015, p. 260). The crucial point in every uprising was how the endgame questions were answered, namely 'would the military defect?' and 'would the military shoot the protesters or not?' (Bellin, 2012, p. 130). Thus, Albrecht (2015) suggested reconsidering the effectiveness of coupproofing.
When studying political violence, it is essential to consider the whole security sector rather than the military only (Santini & Moro, 2019). A nuanced analysis situates the military within the structure of the security apparatus, including militias, police and gendarmerie forces, as well as intelligence services. Forces other than the regular military often shape outcomes, mirroring conflicts within the security apparatus. In the case of Tunisia, for example, the interplay between police forces, presidential security and military was decisive for regime breakdown (Holmes & Koehler, 2020). The growing political power of security forces, such as the police (Abdelrahman, 2017), and the balancing of different security institutions are consequential (Albrecht et al., 2016;Springborg, 2016). Disenfranchised branches are more likely to defect from the ruler by refusing to engage in repression, depending on their corporate interests and composition in terms of recruitment patterns (Grewal, 2019).
Beyond state agents, repression by militias and thugs merits attention. In some contexts, such as Iraq, no state actor has a full monopoly on the use of violence. The brutal repression of protesters by security forces and militias in 2019 and early 2020 corroborates claims in the literature that extrajudicial killings become more likely when militias are involved (Mitchell et al., 2014). In other instances, the deployment of non-state actors is part of the elites' calculus to decrease the visibility of violent repression and to deny responsibility.
Business interests also influence the armies' behaviour, as they are often powerful economic actors. Conversely, business elites are involved in violence, financing repression not least by bankrolling non-state actors. While the role of Gulf businessmen funding jihadist groups in Syria and elsewhere is well-known, private entrepreneurs paid thugs to attack demonstrations in Cairo, notably in the infamous Battle of the Camel in February 2011 (Ketchley, 2017, p. 66). When studying elite shifts (and continuities) throughout the region, the intertwinement of armies and business thus deserves special attention (Grawert & Abul-Magd, 2016).

Transnational and subnational repression
While leaders on the national level often order repression to preserve their rule, it is vital to acknowledge the roles both transnational and subnational dimensions of repression play. In violent conflicts with foreign interference, international actors are important in terms of aiding repression, boosting the capacities of militaries and militias. Beyond the context of violent internationalized conflicts, security services have extended their reach to the transnational environment to target dissenters (Tsourapas, 2020). Moss (2016) described how Libyan and Syrian diaspora populations in the US and UK experience transnational repression. Yom (2016) highlighted cross-policing, with security apparatuses prosecuting dissenters in other Gulf monarchies. Further repressive practices also diffuse across the region (Bank & Edel, 2015;Darwich, 2017;Josua, 2020). On the individual level, tracing a single ruler's learning from negative examples in other countries and subsequent strategic decisions can offer interesting insights, as Leenders' (2013) study of Syrian decision-making demonstrates. Thus, it is worthwhile studying repression within a region from the perspective of recent research on international learning, diffusion and cooperation of authoritarian regimes.
Besides the transnational dimension of repression, its sub-national level is increasingly being considered. Numerous scholars have pointed out the disparities within states, with protests emanating from disenchanted peripheries. Thus, the provinces are important sites of mobilization and offer reality checks of policies made in the capital (Schwedler, 2012). Likewise, the forms of repression that governments choose, as well as the visibility and impact of coercion, also depend on the location within a given country. This effect can go both ways, increasing lethal repression in the countryside where visibility is lower, or in the capital where the symbolic value of protests and sensitivity are higher. In Cairo, for example, the securitization of public spaces depends on their strategic importance, and sites of protest shape interactions between dissidents and security forces (Ketchley, 2017: 145ff.).

Digital tools and repression
Many initially hailed the Arab uprisings as examples of social mediasupported civic self-empowerment, and research investigated how Internet usage facilitated protests. Over time, however, governments have learned to exploit the manipulative potential of digital tools. Digital repression connects to and cuts across the aforementioned dimensions of repression. Digital tools are used for repressive ends in two ways. First, digital repression is a form of repression in and of the digital sphere, via the censoring of web contents or the denial of access to Internet services (to individuals or collectively). In that narrow sense, it is part of constraining repression. Second, the use of digital tools can enhance offline forms of repression; that is, it serves as a way of economizing other forms of repression by employing them in a more targeted way. Beyond state actors who are involved in digital repression, private transnational actors play an enabling, if not critical, role.
While some Arab countries have always fundamentally constrained activities in the online sphere, their regional neighbours have followed suit since the Arab uprisings. The Freedom of the Net index substantiates this claim when comparing data from the 2011 and the 2018 reports. On a scale from 0 (full freedom of the net) to 100 (worst restrictions on the net), Saudi-Arabia moved from 70 to 73, while severe drops were seen in Egypt (54 to 72) and Bahrain (62 to 71). Widely used tactics encompass the full range of digital repression, from censoring specific contents to complete Internet shutdowns and surveillance practices supported by artificial intelligence. During the 2011 protests, for example, Bahraini authorities actively used social media to monitor the population, censor activists and spread propaganda. Tactics ranged from information gathering to trolling the 'disrupted social media space by assimilating it as part of the regime's surveillance apparatus' (Jones, 2013, p. 82). In the context of the 2017 Qatar crisis, allegedly false news implanted onto a website was the pretextual trigger of the boycott by the UAE, Saudi Arabia and others. The ensuing propagandistic media onslaught, led by the latter states, was intensified with Twitter bots (Jones, 2019). During the 2019/20 protests in Iraq, Internet access was frequently shut down, and 'nightly internet curfews' were introduced. Most Internet providers blocked prominent social media channels like Facebook, WhatsApp, Twitter and Instagram for almost two months. 7 Similar practices prevail throughout the region.
Tech companies and governments outside the MENA region often facilitate digital repression. Already before 2011, regimes bought digital surveillance software from Western firms. 8 Private tech companies with headquarters located in democratic countries struck lucrative deals with authoritarian governments that enabled thorough repression through pervasive surveillance and control. An example is the UK-based Gamma Group, which has distributed surveillance software to multiple countries in the region since 2010. 9 While some countries have outlawed the sale of programs that obviously serve domestic repression, more innocuous Western firms such as Google and Twitter are also complicit in cooperating with autocratic governments, providing user data to them and censoring content upon request. Thus, the digital dimension of repression is closely intertwined with questions of actor constellations and transnational assistance.
Digital tools not only offer an additional layer of repression that pervades spheres perceived as private. They also impact on offline repression, enhancing the repressive capabilities of authoritarian governments. Emerging research asks whether digital repression influences the balance of repressive tactics. It has been suggested that digital repression complements rather than substitutes 'traditional' repression (Frantz et al., 2020). There is evidence that digital surveillance can change the forms of repression. In Syrian regions with higher Internet penetration, regime violence is employed in a more targeted way, while repression is indiscriminate elsewhere (Gohdes, 2020). Digital repression also enables transnational repression, as states control and harass their critical citizens living abroad through online means (Moss, 2018).
Thus, digital tools provide new channels to collect information and to disseminate and manipulate narratives, and new areas where constraining repression is used. However, repression of the online sphere does not substitute for incapacitating and eliminating forms of repression. In the reassertion of authoritarian self-confidence, repression sometimes escalates, as the assassination of the journalist Jamal Khashoggi shows. Saudi authorities had accessed conversations between Khashoggi and a fellow dissident, whose phone (besides others) they had infected with spyware. 10 This use of digital tools preceding a killing epitomizes the close connections between digital and other forms of repression.

How to study repression
The heightened repression since the Arab uprisings has repercussions on the practical level of doing research. While repression as a phenomenon needs to be studied, gathering data on such a sensitive topic is never easy (Clark & Cavatorta, 2018). At the same time, repression is the prime reason why conducting political science research in the MENA in general has become riskier in recent years. Repression has inter alia targeted academics, deterring scholars from tackling certain topics and impeding the possibility of doing fieldwork. This has consequences for the ways in which researchers can conduct their work in the first place. In line with the third key issue Bank and Busse identify in their introduction to this Special Issue, below we describe the ethical and practical difficulties that MENA researchers encounter in general and when working on repression in particular, before laying out some remedying research strategies.

Conducting research in repressive environments
Political scientists conducting research under conditions of authoritarianism face distinct challenges. Most facets of academic freedom, such as the freedom of individual scholars, independence of research institutions and campus integrity, are precarious in autocracies (Spannagel et al., 2020). This becomes even more relevant when repression itself is the object of study. The high sensitivity of the topic and challenging access to data complicates the study of issues related to repression. Consequently, scholars often selfcensor their choice of topics and cases, as conducting research would put interlocutors and themselves at high risk.
The Academic Freedom Monitoring Report by the Scholars at Risk network reports 255 incidents of academic freedom violations in the Arab MENA states since early 2011, including 88 cases of killings, violence or disappearances and 80 cases of imprisonment. 11 In Egypt, for example, '21 extrajudicial killings and 1,181 student arrests' have occurred, and universities are treated 'as military facilities that fall under military jurisdiction' (Saliba, 2018, p. 315). Everywhere, the red lines of what is permitted have become less clear against the backdrop of nervous security apparatuses. Thus, field research has become nearly impossible in many parts of the MENA, particularly in counterrevolutionary states and those mired in violent conflict. Interlocutors are afraid of negative repercussions or suspicious of foreign interference. When researchers take ethical considerations seriously, the current situation forestalls many avenues of research to preserve the safety of researchers and their collaborators and respondents.
High levels of repression especially affect scholars living and conducting research in the respective countries. This is most concerning for citizens of MENA states, who are threatened, jailed, or convicted, sometimes in absentia. Regarding research on sensitive topics such as repression, scholars from outside the region have appeared less vulnerable to reprisals by the states they study. 12 However, even international researchers have come under attack during their field research, notably two PhD candidates based in the UK. Most unsettling to the research community was the torture and killing of Giulio Regeni in Egypt, who had studied independent unions, on the fifth anniversary of the January 25 Revolution. Furthermore, the UAE convicted Matthew Hedges, who was researching the Emirates' national security strategy, of spying and jailed him, with several months spent in solitary confinement in Abu Dhabi. Such outrageous treatment of scholars sent strong deterring signals to researchers all over the world to either abandon their research topics or to avoid travelling to certain countries. The old perception that Western passports somehow protect scholars has been shaken up. As the red lines are less clear and governments feel more threatened, fieldwork is an endeavour that many scholars may find too risky. At the same time, local researchers are in even greater danger, as repression might adversely affect their physical wellbeing, reputation, livelihoods, and families (Ryzova, 2017, p. 511).
In the current repressive environment, conducting comparative research is particularly precarious. As different levels of sensitivity prevail in various Arab states, what can be studied in one country is impossible in the next. When research on the ground is possible, scholars who are new to the field in particular must rely on advice by experienced colleagues on how to minimize risks, how to frame research projects towards interlocutors and the authorities, and how to act when interacting with security apparatus (Clark & Schwedler, 2018).

Sources and methods for studying dimensions of repression
The obstacles outlined above notwithstanding, there are multiple possible sources and research strategies at hand for studying various dimensions of repression. For investigating levels, forms, and targets of repression, reports from human rights organizations are a good starting point. Conducting interviews with dissidents and other victims of repression is mainly possible in former autocracies (Nugent, 2020). In other cases, exiled citizens can narrate their experiences, offering vital insights into the consequences of repression (Pearlman, 2017). Shedding light on the actors, Grewal (2019) conducted the first survey of retired military officers in Tunisia. Apart from conventional written and oral sources, videos documenting repressive incidents have become a resource of transparency ever since the killing of Khaled Said, which was the trigger for mobilization in Egypt in 2011. Videos are also important for investigating state crimes, as the Syrian Archive epitomizes (Deutch & Habal, 2018). Open-source documents that are freely accessible offer novel possibilities for research. For studying historical instances of repression, it is insightful to consult archives, memoirs, and biographies, as scholars did for the cases of Iraq or Syria (Blaydes, 2018;Bou Nassif, 2020). Through recent advancements in data collection, event data have become more sophisticated and can sometimes be helpful, especially when multiple datasets are integrated (Donnay et al., 2019). However, Arabic news outlets are under-represented in cross-country event datasets, and the further development of quantitative text analysis for Arabic media is overdue (see the contribution by Weipert-Fenner in this Special Issue). Justifications of repression can be studied with document and speech analysis (Edel & Josua, 2018). To demonstrate the microfoundations that underpin the after-effects of repression, El Kurd (2019) and Nugent (2020) conducted lab experiments in the West Bank and Tunisia, respectively.
Digital tools can only partially remedy decreased access to the field, as the online sphere underlies the restrictions sketched out in the section before, and both ethical and cybersecurity questions need to be considered. Selfcensorship and social desirability affect the collection of data online when online surveillance is high. Nevertheless, social media (meta-)data and Twitter posts by both government and opposition are valuable, if not representative, sources (Jones, 2013). A significant advantage of social media data is that it provides detailed, unobtrusively collected information. Within certain limitations, online surveys offer unprecedented and time-sensitive access to certain groups. For example, Ayanian and Tausch (2016) conducted an online survey of Egyptian activists promoted via Facebook and Twitter during a phase of intense contention.
Triangulating sources and employing multi-method research designs to corroborate results should be the gold standard, especially when sources include data gathered online (Monier, 2018). While collecting data on repression remains difficult, recent research has shown that there are innovative ways to make sense of what is happening. Should the trend towards higher repression recede in some countries, previously untapped sources might emerge so that we can, at some point, scrutinize the dimensions and impact of repression more thoroughly.

Lessons learned
With the Arab uprisings of 2011, the sustained crises of legitimacy in many states in the region developed from latent to open, manifest conflicts between political elites and opposition forces. Instead of renewing social contracts and regaining legitimacy, governments have resorted to higher levels of repression and reconfigured it in all Arab countries except Tunisia. We have reviewed and discussed recent research that has analysed repression in the MENA region and call for investigation not only of the levels, but rather the specific characteristics of repressive measures. Oversimplified analyses of causes and consequences of repression often blur our understanding of political dynamics more than they illuminate them.
Time will tell whether the current high level of repression in Arab states is a temporal or a permanent feature. Many regimes had some ephemeral 'success' in containing dissent, but compensating low levels of legitimacy with repression may only lead to temporal regime durability rather than actual 'stability.' As conditions and motivations for large-scale mobilization linger, dissidents adapt to repressive environments with tactical shifts. The 2019 dynamics in Sudan, Algeria, Iraq, and Lebanon demonstrate that the Arab world is anything but immune to large-scale protests. Without some form of limited pluralism as a safety valve and responsiveness to political and economic demands, the narrowing of boundaries could lead to new revolutionary situations. Unexpected factors might become influential as 'there is now a widespread feeling that everything is potentially political' (Valbjørn, 2015, p. 229).
The increased repressiveness in the Arab world mirrors global developments. The features and mechanisms we have described are not unique to the MENA region, but are observable elsewhere as well. This global trend is an argument against Arab exceptionalism and for studying MENA developments using generalizable theoretical approaches. There has been a surge of repression in many other states, especially in the context of protests, as in Belarus, Chile, Hong Kong and Venezuela. These cases also highlight that opportunities for achieving real and lasting democratic change often remain limited even in the face of large mobilization, as in Egypt and Bahrain in 2011, or in Algeria and Iraq in 2019/20. Around the globe, a climate of fear coincides with a toning down of human rights discourses and a sense of impunity for powerholders. Against the backdrop of digitalization and citizens' empowerment through social media, states have employed new digital tools for repression. This poses serious challenges for oppositional activities but also for researching repression.
Therefore, tracing repressive dynamics within Arab states is increasingly vital to draw lessons for other world regions. Our MENA insights recommend cross-fertilizing protest-repression research with literature on non-violent resistance, civil-military relations and authoritarian rule, as well as engaging more systematically in cross-regional comparisons (Bank, 2018). Insights into psychological micro-level mechanisms could also be a promising way forward. Tapping into innovative strategies of data collection and analysis may partly compensate for the difficult research situation on repression.
Our discussion shows that it is important to differentiate forms, targets, agents and justifications of repression in order to better understand varying dynamics and consequences. We stress multi-level analyses of transnational and subnational aspects of authoritarian strategies and the integration of digital tools, legal foundations and framing of repression. Many of the most pressing and interesting research puzzles lie at the intersection of more than one dimension. It is only through the processes of disaggregating repression and rethinking concepts and categories that we can get to the bottom of causal relations and mechanisms.
We propose that future research on repression, both in the MENA region and globally, should focus on three interplays. First, underlining the 'nonmonolithic nature of regimes' (Chenoweth et al., 2017) and the importance of regime type, we recommend studying the interplay of mobilization and regime constellations, taking (regime) agency more seriously. Second, we call for bringing the state (back) in once again by differentiating between state and regime. Considering the whole security sector and militias and factoring in state capacities and path dependencies of state institutions helps to make better sense of repressive outcomes. Third, repression as a strategy of power maintenance interacts with legitimation strategies, but they do not linearly substitute one another. Rather, repression itself is latently in need of legitimation, making rhetorical justifications and legal foundations of repression a significant future research area. Notes 1. To some extent, this mirrors the violent conflicts in the region. When we exclude data for the war-torn countries Syria, Libya and Yemen, the level of repression peaks in 2013-2014. 2. According to the V-Dem Indicator, which measures how strongly civil society organizations are repressed ('CSO Repression'), the Tunisian score improved from 1.14 to 3.47 points between 2010 and 2018 on a scale from 0-4, 4 signifying absence of CSO repression (https://www.v-dem.net/en/analysis/VariableGraph/). 3. Researchers from the region figure less prominently in pertinent literature, not least because they are directly affected by such policies; see also the section on conducting research in repressive environments below. 4. Other consequences of repression include international reactions such as sanctions and economic effects, which are beyond the scope of this article. 5. For example, the score of all Arab countries measured in the Rule of Law Index of the World Justice Index dropped between 2015 and 2019 (https://worldjus ticeproject.org/our-work/research-and-data/wjp-rule-law-index-2019#&gid= 1&pid=2). 6. See also the section on conducting research in repressive environments below. 7. https://netblocks.org/reports/social-media-partially-unblocked-in-iraq-after-50days-18lJJrBa 8. https://www.zdnet.com/article/wikileaks-microsoft-aided-former-tunisianregime/ 9. https://citizenlab.org/2015/10/mapping-finfishers-continuing-proliferation/ 10. https://marsad-egypt.info/en/2019/08/27/digital-authoritarianism-is-on-therise-in-the-middle-east/ 11. https://www.scholarsatrisk.org/academic-freedom-monitoring-project-index/ 12. We are aware that a dichotomous perspective of a researcher's positionality does not do justice to the multifaceted personal life conditions in a globalized world nor to intersectional approaches to privilege. Nevertheless, we focus on the legal aspect of citizenship, which is most important for reprisals from state actors following research activities.