The Failure of Democracy in Turkey: A Comparative Analysis

Abstract Although Turkey took its initial steps toward establishing democracy in 1950, it has thus far failed to become a fully functioning democracy. Using the comparison cases of Spain and Greece, this article discusses two related variables that are likely to have thwarted the development of full democracy in Turkey: the country's experience with authoritarian rule, and the lack of elite settlement or convergence towards acceptance of the democratic rules of the game. The article ultimately contends that despite the EU's attempt to push Turkey towards full democracy in the modern day it is unlikely that it will become a fully functioning democracy unless it manages to achieve civilian elite agreement regarding the rules of the Turkish democratic game, and that Turkey's experience with authoritarian rule may, in turn, have hindered the development of such rules.

note regarding Turkish democracy is that since 1960 the military has been relatively actively involved in politics when compared to established democracies, with high-ranking military officials continuing to make periodic public pronouncements about major issues of the day, at times implying that force will be used to produce policy change. Moreover, the recent events associated with the Ergenekon organization, which is alleged to be an ultranationalist organization with ties to the military and to have been plotting the overthrow of the government, indicate at the very least that there is a large-scale belief within Turkey that a democratically elected government can indeed still be overthrown by the military. In short, the periodic party closures, restrictions on freedom of speech and military pronouncements make it clear that democracy is not fully established in Turkey. 5 In stark contrast, while some may question the quality of democracy in countries such as Spain and Greece, 6 it is no longer in doubt that they are fully functioning democracies. In the case of Spain, the adoption of the Law for Political Reform in 1976, elections to the Constituent Assembly in 1977 and the adoption of the new Spanish Constitution in 1978 were all clear steps forward in establishing democracy in Spain. The thwarting of a military coup in February 1981 and peaceful election of a socialist majority government in 1982, however, perhaps serve as better indicators that by the early 1980s Spain was a fully functioning democracy and in little real danger of reverting to authoritarian rule. Similarly, in Greece, the adoption of the 1975 Constitution after the collapse of a seven-year regime of colonels may be said to mark the beginning of the Greek transition to democracy, although -as discussed below -the legitimacy of the Greek Constitution was initially in greater doubt than was the case in Spain. As in Spain, the election of a centre-left government, the Panhellenic Socialist Movement (PASOK), in 1981 seemed to be a strong indicator that Greece had established a fully functioning democracy. 7 This article seeks to explain why Turkey has failed to become a fully functioning democracy by comparing it to the relatively more successful cases of Spain and Greece. The authors assume that historical processes that may be difficult to capture with large-N analyses must be explored in order to gain a better understanding of why Turkey has not yet become a fully functioning democracy. The argument in this article is thus based on a small-n analysis, using the general approach of Mill's method of difference, or Przeworski and Teune's most similar systems, 8 whereby the dependent variable varies across observations while the independent variables are held constant to the greatest extent possible. In this design, the potential independent variables that do vary across observations are among the likely causes of differences across the observations. Thus, the cases of Spain and Greece have been chosen because they are similar to Turkey on a number of key dimensions. These important similarities are outlined in the next section, and the subsequent sections highlight the key differences that we argue are likely to have produced differences in regime type across the three countries. These differences revolve around the interconnected phenomena of experiences with authoritarian rule and elite settlements or gradual elite convergence towards acceptance of the democratic rules of the game.

Economic Development
The academic literature on the relationship between economic development and democracy is vast and because of space limitations will not be reviewed here. A great deal of research indicates that 7 The election of socialist governments was significant in both countries because of the extreme hostility on the part of previous authoritarian governments to socialists and other leftist movements. The election of left-wing governments in both cases without violent opposition from the right thus indicated an acceptance that these groups had a right to compete freely in elections and to form a government if they won a majority of seats in parliament. 8  democracy and economic development are strongly connected, although the nature and direction of this relationship is somewhat ambiguous. 9 In terms of our research design, if a group of countries are relatively similar on this variable but still have differences on the dependent variable (being a full democracy), then it is unlikely that this particular independent variable -level of economic development -is the cause of the different outcomes on the dependent variable in these cases. Although Turkey's level of economic development in terms of indicators such as GDP per capita and employment in nonagricultural sectors of the economy have been lower than those of Spain and Greece at the time of their most recent transitions to democracy, all three countries have experienced relatively similar economic histories. Sapelli contends, for instance, that 'Southern Europe has specific characteristics that are rooted in a socioeconomic structure different to that of Continental Europe (including the British Isles), Central Europe and Eastern Europe.' 10 The main distinguishing characteristics of the south European economies include weak, initially foreign-dominated industrialization, rapid growth and rapid change from predominantly agrarian societies to serviceoriented economies. Compared to northern Europe, where the transition from an agrarian to a service economy was gradual via industrialization, the transformation from agrarian to service economy in southern Europe was rapid and 'has not been followed by a strong cultural change'. 11 This description of general economic development applies to all three of the case studies analysed here, although it must be acknowledged that parts of northern Spain did have the type of industrialization described above. 9  Generally, scholars writing on the Spanish and Greek transitions argue that economic modernization was unlikely to be the cause of transitions to democracy or the creation of fully functioning democracy in either of these countries; 12 for instance, Diamandouros, Puhle and Gunther note that the periods of political transition and democratic consolidation in Greece and Spain 'perfectly coincided with the severe economic crises of the second half of the 1970s and early 1980s', and yet they became fully functioning democracies. 13 Scholars of Turkish history and politics also indicate that, while the breakdowns of democracy in this country may be partly connected to economic difficulties, the primary causes of breakdowns have in fact been political. 14 In short, it seems that economic differences are unlikely to be the main cause of differences in regime type across the three countries.

The European Union
Democratization researchers have also pointed to the potential importance of international factors in explaining democratic transitions and consolidations, and some contend that the European Economic Community/European Union (EEC/EU) may be relevant in guaranteeing a stable, fully functioning democracy. 15 In the cases of Greece, Spain and Turkey during their post-war transitions to democracy, though, all three countries had similar affiliations with the European Community (EC). Greece and Turkey signed similar association agreements with the EEC in the early 1960s, both of which envisioned full membership of the EEC. Spain also applied for associate membership in the 1960s but was eventually granted preferential trade status in 1970 instead. 16 Although Turkey's membership of the EU in the present day has been called into question because of the negative views of the German Chancellor Angela Merkel, French President Nicolas Sarkozy and European Council President Herman Van Rompuy, in the 1960s and 1970s Turkey was generally seen as being part of the 'the West', and there was no reason to suspect at the time that the country would not be eligible for full EC membership. 17 Moreover, although both Spain and Greece gained full entry into the European Community (Spain in 1986 and Greece in 1981), it seems unlikely that this was the main factor that made it possible for these countries to become fully functioning, stable democracies. Spain managed to cope effectively with an attempted military coup in February 1981, at a time when it was not at all certain that the country would be allowed into the EC because of refusal of the French government to accept another large predominantly agriculture-producing country into the EC. It was only in January 1983, when discussions were held between the French and Spanish foreign and economic ministries in parallel with negotiations for internal Community reforms and enlargement, that the French government announced that it would help Spain with entry to the EC. 18 By this time, as noted above, the Spanish 16 Shortly before Spain's application for associate membership was lodged, the Birkelbach Report was approved by the European Parliament in January 1962. The report stated that countries wishing to join the EEC were required to recognize the principles for membership outlined by the Council of Europe: democracy, respect for human rights and the rule of law. Associate status of the EEC was specified as a future possibility for countries that fulfilled these political conditions for membership but were not economically ready for full membership; Carlos  Similarly, although Greece's entry into the EC may have helped to turn a semi-loyal opposition into a loyal government, it is fairly clear that this 'semi-loyal' opposition was already becoming loyal long before Greece became a full EC member. 19 Much of the left, which was in opposition immediately after the colonels returned government to civilians, had already begun to moderate during the regime of the colonels, and the leader of the most significant leftist force, the Panhellenic Socialist Movement (Andreas Papandreou), began to rid his party of its radical militant activists in the lead-up to the 1981 general election and emphasized that the party would not act as a revolutionary movement. Contacts with West European socialist and social democratic parties were also increased, and 'socialism' was eventually dropped from the party's discourse. The legitimacy of the 1975 Constitution was never called into question during the 1981 election. 20 So Greece's democracy appears to have been fully functioning, with basic acceptance of the democratic rules, before the country gained entry into the EC.
Finally, in the case of Turkey, as noted above, Turkey's early association with the EEC appears to have failed to guarantee a fully functioning stable democracy; further, in the present day, although the European Commission has acknowledged that Turkey has progressed towards meeting the EU accession criteria and several packages of reforms and constitutional amendments have been adopted as a result of pressure from the EU, the Commission has also continued to note the significant failings of Turkish democracy, particularly in the realm of freedom of speech, but also in the area of the military's involvement in politics (see below). 21 That is, despite the lengthy association with the EEC/EC/EU and despite having been an official EU candidate country for more than a 19 Karakatsanis, The Politics of Elite Transformation. 20 Ibid., chapters 5 and 6. 21 See http://ec.europa.eu/enlargement/pdf/key_documents/2010/package/ tr_rapport_2010_en.pdf. decade, progress towards establishing a fully functioning democracy in Turkey has continued to falter, perhaps indicating the limits of the EU's influence. 22 In short, Spain and Greece managed to become fully functioning democracies before gaining full entry to the EC and without the pressure of EU conditionality. Turkey, despite similar associations with the EC and despite specific pressure from the EU regarding democratization, on the other hand, has failed to establish a fully functioning democracy, indicating that the impact of the EC/EU may ultimately be fairly limited in all three cases.

History of Military Interventions
All the three cases also share a history of military intervention in politics. As noted above, one of the key factors keeping Turkey out of the realm of being a fully functioning democracy is the continued involvement of the military in Turkish politics. The first such intervention in modern Turkish politics occurred in 1960, when a group of officers staged a coup and ran the government for a short period before returning power to civilian authorities; in 1980 a group of generals led a coup that resulted in a longer military regime, but still with a return to civilian rule in 1983. There were also significant interventions in 1971 and 1997, with the military forcing a change of government in both cases. Moreover, although power was officially returned to civilians in 1961 and 1983, it is clear that the military continued to play a significant role in determining 'national security policy', broadly interpreted to include most domestic issues. 23  In addition, the above-mentioned revelations of a possible plot for an overthrow of the Justice and Development Party government would also seem to indicate that significant numbers of officers and co-conspirators do not yet accept the notion of removing the military from political life in Turkey. Despite several reforms designed to reduce the role of the military in politics, 24 the latest European Commission report still notes that 'on some occasions, the Chief of General Staff made comments about ongoing court cases and investigations', 25 indicating that top-level military officials still perceive a public role for the military and that their comments are still newsworthy. This implies that in spite of the reforms, the military is still attempting to influence politics and is still believed to have such influence. The EU report also notes that 'No change has been made to the Turkish Armed Forces Internal Service Law, which defines the duties of the military and contains an article leaving the military wide room for manoeuvre to intervene into politics,' 26 and that 'No progress has been made concerning parliamentary oversight of the defence budget or on audit of the properties of the armed forces by the Court of Auditors.' 27 The implication is that the military still has considerable budgetary autonomy and is thus not entirely under the control of elected civilians.
Even before the founding of the Turkish Republic in 1923 the Ottomans experienced intervention by officers in the prestigious Janissary corps, with the latter often playing a role in succession decisions. Moreover, the restoration of a previously suspended constitution (suspended for 30 years by Ottoman Sultan Abdulhamid II) was forced by a group of officers in 1908. Like Spain and Greece, Turkey has a very long history of military coups. Turkish interventions in these periods also showed remarkable similarities 24 In particular, the number of military officials sitting on the National Security to those in Spain and Greece in the eighteenth and nineteenth centuries. 28 Historically, the Greek and Spanish militaries were also heavily involved in politics, with coups and counter-coups the norm at times in both countries. For instance, historian Paul Preston noted that between 1814 and 1981, Spain experienced more than 50 military interventions in politics, with the most recent coup attempt occurring in 1981, when a group of officers attempted to overthrow the new democratic institutions. 29 The most recent coup in Greece was in 1967, with the overthrow of a civilian government being presented as an attempt by a virulently anti-communist military to prevent a communist overthrow, although the real reasons for the coup appear to be far more complicated than this. 30 In both of these countries there was also serious concern about continued military interventions after their transitions to democracy in the mid-1970s, with reports and rumours of plots for coups coming to light through the early 1980s in Greece and the 1990s in Spain. 31 Despite the clear danger of military intervention in politics in these two countries, they managed to become fully functioning democracies. 32 Thus, it would appear that being prone to military interventions or having a military that was traditionally active in politics and/or regime change does not in itself hinder the establishment of a fully functioning democracy; regimes that have difficulties with military interventionism are not destined 28

Civil War
Both Spain and Greece experienced brutal civil wars involving street fighting between the left and the right and within left-wing groups (Spain between 1931 and 1936 and Greece between 1946 and 1949). Somewhat similar developments were witnessed in Turkey in the late 1960s. During the 1977-1980 period in particular, however, the country came to the verge of a civil war as the armed clashes between neo-fascist and leftist militants were accompanied by massacres and assassinations of public figures. 33 In the Spanish case, civil war was eventually ended by the establishment of the Franco regime, although violence against the left and Basques continued, particularly in the early days of the regime. In the case of Greece, the civil war ended with the establishment of a repressive regime that -as with Spain -targeted the left, with this regime eventually being overthrown by a military coup in the run-up to an election that was predicted to produce a victory for the Centre Union Party. In Turkey, instability relating to severe political tension between the government and opposition before 27 May 1960 and civil war conditions before 12 September 1980 were halted by military coups. In short, all three regimes experienced large-scale political violence between leftist and anti-leftist forces that any subsequent democratic regime would presumably eventually be required to come to terms with in establishing and stabilizing democratic institutions. 34 33 On 8 November 1978 Prime Minister Ecevit announced that 800 people were killed as a result of political violence, and that 1,052 rightist and 778 leftist militants were arrested during his term. The approximate death toll during Demirel's minority government (12 November 1979-12 September 1980) is thought to be around 1,500 (see Feroz Ahmad, The Making of Modern Turkey, London, Routledge, 1993, pp. 172 and 179). 34 We acknowledge that the Spanish Civil War was on a much larger scale than the left-wing and fascist/right-wing violence in Greece and Turkey, but merely wish to highlight the similarities in the nature of disputes and the existence of large-scale political violence across all three regimes. The impact of differences in civil war conditions across the three countries may deserve further exploration in future research, though.
Thus, we contend that the factors outlined above -economic development, relations with the EEC/EU, historical problems with military interventionism, and political violence -are unlikely to have caused the differences in regime type in the three countries. Instead, this article considers two potential alternative explanations for differences in the establishment of fully functioning democracy across the three countries: experience with authoritarian rule and elite settlement or convergence towards acceptance of the democratic rules of the game. We discuss each of these in turn, highlighting the nature of authoritarian rule in the twentieth century that is likely to have (or in the case of Turkey, have not) led to elite settlement or convergence in our three cases. We then discuss some of the key aspects of elite settlement and convergence that are likely to have produced differences in the functioning of democracy across the three countries.

DIFFERENCE 1: EXPERIENCE WITH AUTHORITARIAN RULE
Although periods of extremely repressive rule do not necessarily produce the 'right' configuration of attitudes to guarantee stable democracy, in countries where citizens and elites have started to engage in a democratic form of government and experience rising expectations regarding continued democratization, periods of authoritarian rule are likely to be anachronistic. Where there are civilian elites who have organized political parties and participated in democratic rule, repression seems especially likely to create a situation in which 'the democrats' realize that cooperation among them will produce mutual benefits. That is, opposition elites in such regimes are often able to unify in their deep-seated opposition to repression, with opposition to the authoritarian regime itself serving as a basis for cooperation. The memory of an authoritarian regime may also serve to continue to ensure conciliatory behaviour and support for the basic rules of the democratic regime. In the absence of an extremely repressive regime and the memory of such a regime, it may be more difficult for political elites to unify to try to guarantee the continued existence of their democracy.
The regime established by the victors of the Spanish Civil War was an extremely repressive one, particularly at the start. Estimates vary, but clearly hundreds of thousands were executed. Anyone who had been a member of a trade union, a Masonic lodge, any of the 497 THE FAILURE OF DEMOCRACY IN TURKEY: A COMPARATIVE ANALYSIS republican parties or a left-wing political party, a supporter of Basque or Catalan nationalism, or who had in anyway 'helped to undermine political order or . . . impeded the Movimiento Nacional' was likely to be imprisoned or executed. 35 Military tribunals were often used for trials for political crimes, and in 1963 the Public Order Court was established to deal with those suspected of 'undermining the foundations of the state, altering political order, or creating anxiety for the national conscience'. 36 Although some of the repression was relaxed and some liberals and Christian democrats were eventually able to publish mild criticism of the regime, the repression of communists and Basque separatists continued during the Franco regime. 37 In addition, Franco's regime was lengthy, lasting for more than 35 years. It can thus be argued that the period of war and subsequent state violence against the population and segments of the political elite had a substantial impact on the Spanish population and its post-Franco leadership, including their attitudes and behaviour during the post-Franco settlement period to be discussed below. 38 The post-civil war regime in Greece, although less brutal than the war itself, was nonetheless fairly repressive. The Greek public was officially divided into those deemed to be 'nationally minded' (non-communists who were judged to be loyal to the regime) or 'not nationally minded' (communists, family members of suspected or actual communists, and some centrists or centre-leftists who were deemed to be 'fellow travellers'). The latter were ineligible for professional licences, employment in the civil service, driver's licences, passports or university entrance. Martial law courts could prosecute communists and other leftists, including those from the centre-left, as well as those of more centrist views, and sentence them to death, and there was systematic surveillance of hundreds of thousands of 35  citizens. Elections were regularly held, but the regime clearly had illiberal elements. 39 In 1967 a military regime was then established that lasted until 1974. In the colonels' regime, civil servants and school and university teachers whose allegiance was in doubt were dismissed. Lawyers and judges who showed too much independence were harassed and dismissed. The press was censored and students were subjected to brutality. 40 Leaders of resistance groups were sentenced to lengthy prison terms, and rather than targeting one part of the political spectrum, the colonels repressed individuals from most of the political spectrum. 41 The worst and most visible act of brutality came in November 1973, when students occupied Athens Polytechnic and university buildings in Salonika and Patras. Coup leader and post-coup Prime Minister (and eventually President) Georgios Papadopoulos sent in troops and tanks to crush the students; this was apparently carried out with severe brutality and at least 34 students and others were killed, hundreds of others were wounded and almost a thousand arrested. 42 The treatment of the students in Athens and other locations was met with extreme revulsion, and although not on the scale of the inhumanities carried out in the early Franco regime, this, along with the lengthy period of repression after the civil war -Greece was ruled by repressive regimes for more than 25 years -may have been enough to convince citizens and civilian opposition from all political leanings that an authoritarian regime was to be avoided at all costs in the future. Turkey was governed by the single-party regime of the Republican People's Party (RPP) between 1923, the year when the republic was proclaimed, and 1946 and had no real experience with democratic rule prior to this. 43 Completely free elections were not held until 39 43 The Ottoman constitution of 1876 did actually foresee a parliamentary monarchy. However, the only period in the twentieth century that this constitution could be said to be in effect was between 1908 and 1913. The 1908 elections were considerably freer, and the following year the constitution would be amended so as to limit some of the sultan's powers (including his right to dissolve the parliament). Many parties 1950, but because of the extremely disproportionate multi-member district plurality system, the new opposition party -the Democrat Party (DP) -suddenly won an overwhelming majority of seats in parliament and eventually became a repressive force in itself and was toppled by the military in 1960. The military regime was short-lived, however, lasting approximately one year. During this period, the ousted DP leader, Adnan Menderes, and two of his cabinet ministers were tried and executed. On the positive side, a new liberal constitution was designed and ratified. As noted above, however, civil unrest re-emerged and the military stepped in again via a memorandum in 1971, forcing a government resignation and warning civilian elites to solve the economic and political crises that were bringing the country to the brink of civil war. Some heed was paid to the memo; but, following the 1973 elections that brought back full civilian rule, street fighting restarted amidst bickering on the part of the political elite, and the military intervened with force in 1980. The generals who engineered the 1980 coup were determined to restructure all the essential institutions so as to make Turkey a fairly illiberal 'democracy' where freedoms of expression and association would be limited. In Ahmad's words, 'all political life came to a standstill as the political parties were closed down and former politicians banned from participating in politics. Before some semblance of political life was restored, Turkey's institutions -the constitution, the electoral law, the universities emerged during this period, but the political arena would soon be dominated by violence between rival parties. A number of political homicides took place and these parties relied on different factions within the military. The 1912 elections were rigged by the Committee of Union and Progress (CUP), which would eventually establish its single-party rule by a coup d'état in January 1913. The regime would further transform into a CUP dictatorship as a response to the assassination of Prime Minister Mahmud Sevket Pasha in June 1913, who was appointed to this post by the CUP after the January coup.
In the Republican era, two brief multiparty periods were experienced: in 1924-25 and 1930. The first one was marked by the formation of an opposition party against the RPP, founder of the Republic, by a group of dissident political elites of both civilian and military origins. Mustafa Kemal Pasha, however, did not tolerate this move and the opposition party was disbanded in 1925. In 1930, though, it was Mustafa Kemal who encouraged the formation of an opposition party. Having attracted a large amount of support from discontented masses, this new party would soon be regarded as a threat to the new regime and be forced to dissolve itself.
-were radically amended to depoliticize the country'. 44 The coup had severe consequences: 650,000 people were taken into custody, 517 people were given the death penalty and 50 people were executed (18 leftist, 8 rightist and an Armenian militant were hanged; the others were executed for non-political violence); 299 people died in prison; 30,000 people fled Turkey and became political refugees; 14,000 people already in exile were stripped of their citizenship. 45 Compared to the very long experiences of repressive regimes in Spain and Greece, this was relatively short, lasting 38 months. Also in comparison to Spain and Greece, the repression of elites in particular was far less widespread in Turkey: after the 1960 coup, three Democrat Party officials were executed, but there was no substantial repression of the neo-Democrat parties. After the 1980 coup, mainstream politicians from the centre-left Republican People's Party and centreright Justice Party were sent to the military's holiday camps for a brief period and then banned from politics for 10 years, although this ban was removed after only five years by a referendum. One politician, Bülent Ecevit, was imprisoned twice after giving interviews with the foreign press after the 1980 coup, and the political magazine he published was also closed down. Some officials of the Islamist National Salvation Party also spent a few years in prison after this coup, whereas those of the neo-fascist Nationalist Movement Party served lengthier prison terms. Overall, though, the level of repression of elites after the 1980 coup was relatively minimal.
This very brief overview of periods of repression in Spain, Greece and Turkey has highlighted the variable levels of brutality experienced in the three countries. Though the Turkish authoritarian regimes clearly involved tremendous brutality, the length of these was relatively short and the extent of repression comparatively limited. The number of executions and imprisonments compared to the Franco regime was low, and the range of elites who were executed or 44  fully functioning, stable democratic regimes. 49 The early discussions of agency distinguished between the transition phase and the smooth functioning of democracy phase of this process, indicating that the factors that might bring democracy into being may be very different from those that help to maintain it. Elites are crucial in both phases, though, and it is possible that the manner in which elites handle the transition phase may have some bearing on whether the regime becomes a fully functioning, stable democracy or not. In a series of publications, Field, Burton and Higley 50 attempted to outline an agency-based theory of elite settlements and convergence, which are extremely relevant to the establishment of a fully functioning democracy. In their analysis, elites can be conceptualized as: people who are able, through their positions in powerful organizations, to affect national political outcomes individually, regularly, and seriously . . . a nation's top leadership in all sectors -political, governmental, business, trade union, military, media, religious, and intellectual -including both 'establishment' and 'counterelite' factions. A national elite can be said to encompass 'all those persons capable if they wish, of making substantial political trouble for high officials (i.e., other elite persons who happen to be incumbents of authoritative positions) without being promptly repressed '. 51 According to this body of research, an important distinction between types of elites can be drawn between those who are not unified and those who are consensually unified. In the former, elite factions are deeply distrustful of one another, interpersonal relations among elites do not cross factional lines, and factions do not 49  generally cooperate to contain societal divisions or avoid political crises. Consensually unified elites (in pluralist political systems, that is) may take opposing ideological or policy stances but they consistently refrain from pushing their disagreements to the point of violent conflict; elites share a tacit commitment to abide by common codes of political conduct centring on a norm of restrained partisanship, and there is an extensive web of interpersonal relationships across factional lines. 52 The former of these elite structures is generally believed to be associated with political instability, including frequent or expected forcible seizures of government power; the latter is believed to be associated with peaceful democratic rule, and thus it is clear that without a transition to this type of elite -according to the elite settlement and convergence arguments -establishing a fully functioning stable democracy is not possible. 53 Elites can become unified fairly suddenly via an elite settlement, usually at the end of a period of costly but inconclusive conflict such as a civil war in which no party emerges as the clear winner, or as a result of a major crisis of the state, often revolving around the failures of the head of state (e.g. monarch). 54 Alternatively, elites may gradually converge towards acceptance of the democratic rules, often after some of the warring groups of elites enter sustained, peaceful collaboration in electoral politics and manage to dominate government executive power by winning elections; other groups of elites who may have initially been hostile to this new set-up eventually realize that the only way to gain power is to abandon the ideological and policy stances that keep them from winning elections and adopt those of the winning coalition. 55 In the process, previously anti-system, or semi-loyal, elites are converted into system loyalists, willing to play the game according to the rules set by their opponents. 56  with each other regarding the democratic rules of the game. On the other hand, the two-step process of elite convergence eventually leads to the creation of a fully functioning democracy, despite the absence of private face-to-face negotiations. The degree of elite unity varies substantially across our three cases. Spain represents a now-classic sudden elite settlement model, Greece represents a case of elite convergence, while Turkey represents a case of elite disunity, particularly a lack of settlement and lack of gradual convergence. Given the potential importance of this variable in explaining the differences in regime type across the three cases, it is worth exploring further how and why the three countries differ on this variable.
The events surrounding the elite settlement in Spain have been outlined by other scholars, 57 but a few of the key points deserve to be revisited here, given their stark contrast with events in Greece and Turkey. In Spain, previously warring factions from the 1931-6 republican period and subsequent civil war were initially stopped by the Franco regime, and his death opened up the possibility of a more permanent settlement across these groups. Such a settlement was far from inevitable, and several alternative approaches after Franco's death were possible. King Juan Carlos and the member of the elite he put in charge of the transition, Adolfo Suárez, however, appeared to realize that the best hope for permanent peace and stability was a settlement, with the alternative perceived as likely to produce significant elite and civil unrest and perhaps a return to the levels of violence experienced during the civil war. Suárez also seemed to realize that the best hope for settlement would be via private discussions with elites who had the potential to disrupt the transition and to cause serious problems of governance if not included. This approach began before any official discussion of a constitutional settlement, and focused on discussions with the major opposition, who had already begun to make demands for the legalization of all parties and had begun to form a provisional government (and had also organized public demonstrations and strikes). 58 Early on, Suárez began negotiating with these elites and also with representatives of the right prior to drafting the Law for Political Reform, which paved the way for the creation of an elected constituent assembly to draft a new constitution.

THE FAILURE OF DEMOCRACY IN TURKEY: A COMPARATIVE ANALYSIS
After the approval of the Political Reform Law, contacts between Suárez and the opposition were formalized with the creation of a nine-member negotiating committee that aimed to represent the main groups in the opposition. 59 The drafting of the crucial electoral law that would determine the composition of the constituent assembly reflected the discussions Suárez was conducting with the committee of nine on the one hand and the Franquist right on the other. The law incorporated the wishes of the Socialists and Communists for proportional representation but also included 'correctives' to over-represent the two largest parties in most districts; correctives that had been promised to the right, which was concerned with potential fragmentation of Spanish politics. This system proved successful in preventing fragmentation while simultaneously not excluding important groups. 60 The system of informal negotiation continued with the development of the Pacts of Moncloa, which aimed to stabilize labour relations during an increasingly precarious economic situation while the parties continued to develop the constitution.
The formal constitutional-drafting phase of the process began with the appointment of a seven-man ponencia (drafting subcommittee). The party newly formed by Suárez, the Union of the Democratic Centre (UCD), had won the plurality of seats in the constituent assembly and had three representatives in the ponencia; the Socialists, Communists, the Catalan nationalist Convergencia democratica de Catalunya, and the party formed to represent the Franquist right, the Alianza Popular (AP), each had one representative (importantly, the subcommittee contained no representative of the Basques). Deliberations were entirely confidential. However, once a draft of the constitution became public and interest groups and other political parties were invited to participate by offering amendments, a more conflictual pattern of decision-making emerged, with the Socialist representative walking out of talks when an amendment on church-state relations was introduced by the right. Initial discussions of the draft of the constitution in the Congress of Deputies took place publicly, and decisions were taken using relatively close votes and majoritarian principles. The UCD and AP between them had enough votes to push 59  provisions through and could easily have imposed a new constitutional order on the remaining parties; however, it soon became clear that enacting a constitution that excluded the support of the regionalist or leftist parties would limit the legitimacy of the new regime severely. Thus, a new decision-making style was introduced in which four deputies from the two largest (and opposing) parties -the UCD and Socialists -met at a restaurant to attempt to resolve outstanding differences. By the following morning they had reached compromise agreements on all the major contentious religious issues (disestablishment, education, divorce and abortion), labour relations and other contentious issues. Subsequently, differences across groups were dealt with in a similar manner. Parliamentary debates that had previously been lengthy and rancorous suddenly became brief. 61 Gunther highlights the importance of face-to-face contacts between Prime Minister (or, officially, President of the Government of Spain) Suárez and opposition leaders and particularly the cordial relations that developed between Suárez and Communist Party leader Santiago Carrillo and Socialist leader Felipe González, both of whom had been considered to be enemies of the regime only a few years before this. These friendly relations continued across party lines in negotiations over the constitution and the regional autonomy statutes too. 62 The only issue that could not be dealt with completely in this way was the issue of the Basques, although agreements were eventually developed and most of the Basque political factions have pledged loyalty to the Spanish constitution and renounced armed conflict.
The process by which elites were transformed from disunity to consensual unity in Greece is a contrast to the Spanish process. Given the violence of the Greek civil war, the subsequent distinction drawn between 'nationally minded' versus 'suspect' citizens (anticommunist versus communist), and the severe repression of the left carried out by the right-wing government after the civil war via 'loyalty boards' and martial law courts, it would be expected that some sort of reconciliation would be required before Greece could establish a fully functioning, stable democracy. During the rule of the colonels several attempts at developing an anti-dictatorship unity were, in fact, attempted but these generally failed due to personal animosities and suspicion between the left and right, 63 thus making the possibility of a negotiated transition like Spain's unlikely. 64 The 1974 transition to democracy in Greece was extremely abrupt, dominated by a single partisan individual -Constantine Karamanlisand, in contrast to Suárez's approach in Spain, this transition leader did not attempt any formal or informal negotiations between elites. Even the decision regarding who would lead the transition to democracy -initially reached via discussions amongst elites -was ultimately made in a less than negotiated manner that had the potential to provoke suspicion and hostility. That is, once the military decided to hand power back to civilians (after being embarrassed by failed mobilization efforts against Turkey), the civilian leadership from the right and centre-right and several military officers decided that the leader of the National Radical Union (Kanellopoulos) should lead the transitional coalition government. The new government was due to be sworn in three hours after this decision was made but during this short period the agreement was broken by a smaller group of the negotiators who decided, without informing the others, that Karamanlis should be invited to lead the transition government. 65 Once he was sworn in, Karamanlis single-handedly chose the members of his 1974 national unity government, which included none of the main left-wing parties or politicians, although a handful of prominent politicians who had been persecuted during the military dictatorship was included. Karamanlis also personally decided the timing of the first post-junta elections and a referendum on the monarchy, unilaterally legalized the Communist Party of Greece, lifted all the restrictive post-civil war measures, and commuted the junta leaders' death sentences to life imprisonment, all without negotiation with elites from the left, right, or from the military. In the elections held prior to the drafting of a new constitution, the electoral law used for these included a strongly majoritarian electoral formula of 'reinforced' proportional representation, and Karamanlis's New 63 Karakatsanis, The Politics of Elite Transformation, pp. 47-51. 64 It is contended, however, that although these talks failed to produce a unified solution to rule of the colonels, the face-to-face contacts between previously warring elites served to create a sense of unity and mutual civility between them, and these helped to build mutual trust and respect between civilian politicians (Karakatsanis, The Politics of Elite Transformation, pp. 7-8). 65  Democracy Party won 73.3 per cent of the parliamentary seats with only 54 per cent of the popular vote. 66 Thus, the 'constituent assembly' itself clearly failed to represent the socio-political divisions in Greek society accurately. In fact, the so-called constituent assembly did not perform the function that would normally be expected of such a body. Again, in contrast to the ponencia created to draft the Spanish constitution, in Greece Karamanlis and a handful of close advisers drafted the constitution themselves and submitted it to parliament for debate. The constitution was considered by two special parliamentary committees whose composition represented the relative strength of various parties in parliament, thereby ensuring a majority for Karamanlis's New Democracy Party on both. The opposition fervently objected that most of its amendments had been rejected and walked out of the debate, refusing to attend most of the subsequent sessions. They also boycotted the signing ceremony on 9 June 1975. Moreover, the members of the constituent assembly who remained for the entirety of the debate did not have time to finish discussing all of the articles of the constitution because Karamanlis had stipulated that the debate would be limited to three months and refused to negotiate this stipulation. 67 The difference between this process and the careful cross-party negotiations in Spain could hardly be starker. Not surprisingly, the main opposition party leader in Greece, Andreas Papandreou (PASOK), denounced the constitution, arguing that many of its provisions -particularly those regarding executive power -were 'totalitarian'. He also initially claimed that if his party were to win enough seats to form a government in future elections, he would dissolve parliament and hold elections for a new constituent assembly to write a new constitution. Prominent politician George Mavros, leader of the Centre Union, also initially denounced the constitution, as did the Communists. 68  appeared to be a semi-loyal party that threatened the new Greek democratic regime, while several other parties were also questioning the legitimacy of the new regime.
Karakatsanis contends that Karamanlis's domination of the transition may, in fact, have helped contribute to establishing democratic stability in Greece because Karamanlis was acceptable to virtually all Greeks by 1974 and he attempted to act as a national rather than partisan leader. 69 Although Karamanlis's National Radical Union party (precursor to New Democracy) had been responsible for many of the repressive measures taken in the postcivil war era, his approach had clearly changed by the time he returned from his self-imposed exile in Paris to lead the transition, and he made vast reforms to the Greek right. His legalization of the Greek Communist Party of the Interior and his adoption of civil and human rights provisions, as well as social welfare and gender equality provisions are all evidence of his moderation. Furthermore, after the fall of the colonels, there was widespread diffuse support for democracy and virtually any democratic alternative was perceived as acceptable. It was recognized that Karamanlis had the authority and prestige to lead this process. Even prominent leftists spoke of him approvingly. 70 Also important in the Greek transition is that the apparent semiloyal rhetoric of PASOK was ultimately not met with action. Once PASOK took control of government in 1981 it carried out very few of the actions it had promised and Papandreou had, in fact, already begun to distance himself from the more extreme elements in the party in the run-up to the election, as noted above. Thus, the main party of the left had begun to moderate. In addition, although the Communists -also a relatively popular movement in Greeceopposed the electoral system imposed on them and contested many aspects of the new constitution, the party still pledged in 1974 to adhere to democratic, parliamentary government. 71 Thus, by 1981, there was widespread acceptance of the democratic rules established by Karamanlis, both on the part of the public and on the part of elites.
Karakatsanis points out that the Greek case indicates that radical anti-constitution rhetoric does not necessarily translate into semi-or 69 Karakatsanis, The Politics of Elite Transformation pp. 72-3. 70 Ibid., pp. 75-6. 71 Ibid., pp. 9-10. disloyal behaviour. 72 Such rhetoric can be used mostly to attempt to win votes and/or because it is perceived that this is simply the way that the opposition should behave; Andreas Papandreou's failure to change the constitution and his gradual dropping of language signifying that he would do so indicates that gradual convergence may indeed occur. Thus, Greece appears to represent a two-step model of elite convergence: Karamanlis and New Democracy managed to dominate the transition and early period of Greek democracy, and, while the main opposition, PASOK, initially appeared to oppose the institutions of this new democracy, it seemed to realize that it would have to moderate in order to avoid the position of being a permanent loser in the new regime. This convergence, in turn, provided a degree of stability to Greek institutions that would be likely to ward off attempts to overthrow it -although, as noted above, there were still rumours of such plots. The fact that these did not come to fruition may thus be a strong indicator of the widespread acceptance and legitimacy of Greek democratic institutions, which would make an overthrow extremely difficult indeed. Karakatsanis's analysis also indicates that for this particular method of achieving elite consensual unity via gradual convergence to succeed, the person or group leading the transition must have widespread legitimacy and democracy must be seen by elites and the masses as the only option available (whether due to international circumstances, memory of brutality or from experience with a poorly functioning authoritarian regime, and so on).
The case of Turkey represents neither one of elite settlement prior to the launch of a new democratic regime nor gradual convergence. Instead, the Turkish case appears to be marked by elite disunity, as defined above. After the 1960 military coup in Turkey, a new constitution that was more liberal than the previous one was drafted by a Constitutive Assembly and ratified by public vote. The way that the assembly came into existence and the debates surrounding the new constitution were democratic to some extent. The constitutionmaking process in 1961 did, however, have a crucial deficiency: members and supporters of the Democrat Party (DP) -the former governing party that was overthrown by the junta that carried out the coup -were excluded from the entire process. 73 Apart from raising questions of legitimacy for the new constitution (and even for the new political order) of post-1960 Turkey, that exclusion would also result in a reluctance in neo-Democrat parties to defend the constitutional order in times of difficult political crises entailing the military's involvement in politics. 74 It is important to note also that in contrast to the Greek 1975 constitution, the 1961 constitution in Turkey was not designed by an individual or group that had widespread popularity or legitimacy. The constitution-writing process in Greece was far from consensual but the fact that it was designed by an elite who was elected with widespread popular support in an election in which other major (and minor) parties were also allowed to compete freely would seem to lend far more legitimacy to the rules and procedures designed by Karamanlis and his advisers than was the case with the 1961 Turkish constitution.
The making of the post-coup 1982 Turkish constitution was even less democratic than the 1961 constitution's design process. In the latter case, most of the members of the Constitutive Assembly were elected, but not with direct elections or universal suffrage. 75 In addition, in cases of discord between the civilian (Representatives' Assembly) and military (Committee of National Union) branches of the Constitutive Assembly, a mixed commission consisting of members from both branches would meet to resolve the disagreements. If this commission could not resolve a disagreement, the civilian branch was ultimately prescribed to prevail. 76 During the 1982 constitutionmaking process, however, the military branch (Council of National Security) of the Constitutive Assembly had an undisputed supremacy over the civilian branch (Consultative Assembly). All members of the Consultative Assembly were appointed by the military, and none of them was a member of any party. Indeed, all political parties and party members from the pre-coup period were excluded from participating in this process and banned from politics for several years after the coup. The final draft of the 1982 constitution would be introduced to popular vote for ratification, but criticizing it was forbidden. 77 Tanör and Yüzbaşıoglu contend that democratic constitutionmaking principles comprise free public opinion, freedom for political parties, elected assemblies and democratic representation, and free elections, and these authors posit that the Spanish and Greek democratic transitions in the 1970s conformed to democratic principles. In their view, though, the methods practised in Turkey after the 1980 military coup fell short of these principles and are therefore closer to the constitution-making models experienced in Pakistan, Albania, and some Latin American countries. In this respect, they stress that the high level of 'yes' votes at the 1982 referendum on the Turkish constitution does not necessarily imply the presence of a strong, voluntary and conscious social approval behind the Turkish constitution. 78 Thus, the Turkish case follows neither the Spanish nor Greek models. Elites have neither had the opportunity to develop a lasting cross-party settlement nor have they gradually converged towards the acceptance of any particular regime rules designed by a popular, elected government. In contrast, in the Greek case of gradual convergence, the constitution was designed by a very popular leader who had been elected by the Greek public, with no limitations imposed on opposition participation in those elections. While the constitution was not designed in a consensual manner, it was at least designed by a leader whose party had clear public support. In Turkey, both the 1961 and 1982 constitutions were designed by groups of elites who did not necessarily have public support, as the public was not directly consulted except to ratify the documents produced by the assemblies. Thus it has been possible for almost all elites to question the settlements, and in the instance of the 1982 settlement, the document produced was far from democratic in any case. 79 The lack of settlement or elite convergence towards a set of democratic rules has implications for political stability and particularly the role of the military in guaranteeing stability. Even in the face of political and economic turbulence in Spain during the transition to democracy,  79 Elite settlement may, of course, not guarantee democracy and instead could produce an authoritarian settlement. Given the configuration of interests in Turkey, this sort of settlement seems extremely unlikely. Thus it appears that any movement towards elite unity in acceptance of the basic rules of the regime will revolve around democratic rather than authoritarian rules. elite settlement regarding the democratic rules -made possible via consensual negotiations -also allowed elites to converge to fight off the military threat in the early 1980s and beyond. The gradual convergence experienced in Greece appears to have had a similar effect.
In Turkey, on the other hand, the failure to develop an elite settlement on key issues -including the functioning of Turkish democratic institutions, the very controversial issue of the role of religion and rights of religious groups and the treatment of ethnic minority groups -has meant elite disagreement on the basic rules of the regime and unwillingness to defend democratic rules (particularly since they played no role in creating them in the first place). This, in turn, has meant that civilian elites have been less willing or able to amass the power required to counter the military threat. They have also had more difficulty in converging towards a general acceptance of the meaning of -and acceptable limitations on -free speech, which has significant implications for respect for the latter in Turkey. CONCLUSION This article has argued that experience with authoritarian rule and elite settlement or convergence are likely to be key explanations for the differences in regime type between Spain and Greece, on the one hand, and Turkey, on the other. The former two countries experienced violent civil wars followed by years of repression; experiences that are likely to have had a significant impact on the historical memory of elites and masses, making them more prone to try to avoid repeating these circumstances. In this respect, it should be acknowledged that the Civil War in Spain may be just as important as the Franco dictatorship in determining the behaviour of Spanish elites. Turkey has also experienced conditions approaching civil war but these experiences were generally halted by military interventions involving relatively limited brutality and repression. The elites and 'ordinary' cadres of the mainstream political parties in Turkey were rarely harmed in times of military intervention, whereas elites of all political persuasions were repressed in Greece's most recent experience with authoritarian rule (the regime of the colonels). Thus, in Spain and Greece, when democracy failed, the consequences were severe (especially in Spain); in Turkey, when democracy has failed, the military has served as a welcome stabilizer. This outcome is positive in the sense that the level of human suffering is likely to have been reduced in Turkey, but it may be one of the factors that has made it unnecessary for elites to attempt to settle their differences to try to make democracy work.
As noted above, one of the key failures in terms of elite settlement or convergence in Turkey is the failure to come to some agreement regarding the place of religion in Turkish politics and society. This potentially makes Turkey's situation more difficult than was the case with Spain, where by 1975 the Church had distanced itself from the Franco regime and was fairly accommodating to the democratizers. In Turkey, circumstances were and are quite different. First of all, there is no 'the mosque' in Turkey equivalent to 'the Church' in Spain since secularism in Turkey came into being as the state's establishment of control and domination over religion. When it comes to political Islam, the Islamist party was quite strong even in 1970s, and Islamists gradually accumulated power during 1980s, but it was not until the mid-1990s that the Islamist-secularist cleavage would become important. Since that juncture, the scale of tension between Islamists (or religious conservatives) and secularists has probably been far greater than was the tension between liberals and conservative Catholics in Spain during its transition (but perhaps similar to the situation in 1930s Spain). This cleavage may reduce the likelihood of a consensual democracy in Turkey. As noted above, however, one reason for this modern-day tension in Turkey is likely to be the failure of elites to even attempt to settle among themselves consensually what role religion would play in the Turkish polity. This failure dates back to the establishment of the 1924 constitution and has continued to be repeated in later constitutional arrangements, as well as other junctures along the way, and continues to provide one of the justifications for limitations of freedom and acceptance of the military presence in Turkish politics. It is fairly clear that gradual convergence on the role of religion in a fully functioning Turkish democracy is unlikely, and so it appears that the only alternative to the current situation is elite settlement on this issue.
Similarly, the Kurdish question continues to stand out as a factor that has made full democracy difficult to achieve in Turkey. Spain had to deal with a comparable problem, the Basque question, both before and after its transition to democracy (and particularly increased Basque violence around the time of the transition). This may be roughly comparable to Turkey's Kurdish problem, but two crucial differences between the two cases remain: the Basque country is more developed than much of the rest of Spain whereas the Kurdishpopulated parts of Turkey are the poorest regions of the country. Second, the insurgency of ETA -the Basque separatist movementcaused the death of fewer than 1,000 people in Spain while it is estimated that more than 30,000 people lost their lives in Turkey due to the PKK's rebellion. The impacts of these differences on democracy are clearly worth exploration in future research.
Nevertheless, we contend that our findings still have clear implications for ongoing attempts to establish stable democracy in Turkey. Although the EU is attempting to use carrots and sticks to pull and push Turkey toward democracy, its success has thus far has been fairly limited. Our findings indicate that the EU's approach may not be that helpful because of the failure of any real elite settlement or convergence towards acceptance of any particular set of democratic rules. Until such settlement or convergence occurs, it seems unlikely that Turkey will move significantly forward in meeting the EU's political requirements for full membership.
Since the election of the Justice and Development Party (AKP) -a party with religious background -in 2002, the party leadership has periodically raised the issue of developing a new, civilian-designed constitution. Based on the advice of constitutional scholar Ergun Özbudun, the government initially appeared to hope to adopt a Spanish-style consensual approach to constitutional design. This issue was dropped from the agenda as a result of the main opposition Republican People's Party's firm resistance to any attempt at constitutional reform led by the AKP, highlighting the increased importance of the above-mentioned Islamist-secularist cleavage. Attempts to conduct a broad constitutional reform to democratize the Turkish polity are likely to crop up again, however. As argued in this article, the lack of such reform may create a serious stumbling block on Turkey's road to creating a fully functioning democracy. Thus, it seems that, as argued by Grigoriadis, 80 some consensus between the conservative Islamists and the secularists must be achieved to resolve social divisions in Turkey and ultimately to develop a fully functioning democracy in Turkey.