The Meaning of ‘Limited Pluralism’ in Media Reporting under Authoritarian Rule

Research on mass media in authoritarian regimes focuses more on state mechanisms of control than on actual media reporting and on moments of crises much more than on times of stable functioning of the regime. In order to shed more light on the role of journalistic mass media in authoritarian regimes, this article deals with the actual limits of pluralism in media reporting regarding policy issues in ‘ordinary’ authoritarian politics. Looking at pluralism in sources (i.e., actors being quoted) and pluralism in opinion, the article also deals with the often assumed increasing degree of pluralism from TV over print media to the Internet. This study is based on a qualitative content analysis of media reporting on export pipelines in three post-Soviet authoritarian regimes (Azerbaijan, Kazakhstan, and Turkmenistan). The text corpus comprises 3,618 media reports from 38 different journalistic media outlets published between 1998 and 2011. Two major results of the study are, first, that concerning the degree of pluralism, the differences between types of media are country specific, and, second, that ‘limited pluralism’ seems to be a misnomer, as the political opposition—at least in our cases—regularly does not have a voice at all.


Introduction
Juan Linz has famously defined authoritarian regimes as 'political systems with limited, not responsible, political pluralism, without elaborate and guiding ideology, but with distinctive mentalities, without extensive nor intensive political mobilization, except at some points in their development, and in which a leader or occasionally a small group exercises power within formally illdefined limits but actually quite predictable ones' (Linz, 2000, p. 159). He went on to clarify that there is a 'fairly wide range [of pluralism] in which those regimes operate' (Linz, 2000, p. 161).
Since Linz published the first version of his definition in 1964, political scientists have extensively examined the meaning of limited pluralism and its empirical forms in the case of political forces, such as government institutions, political parties, and different kinds of elite factions. While Linz originally developed a typology based on several qualitative aspects, including the social origin of the ruling elites, their guiding mentality and the development stage of the regime, newer typologies of authoritarian political regimes focus more exclusively on the degrees of pluralism or political competition (cf. e.g., Diamond, 2002;Levitsky & Way, 2010;Roessler & Howard, 2009), where the central distinction is between hybrid regimes (which combine democratic and authoritarian features) and fully authoritarian regimes. Brownlee (2009) as well as Roessler and Howard (2009) use 'closed authoritarianism' to designate cases of extreme authoritarian control.
The role of mass media in authoritarian regimes, though, has gained little attention in political science, while the mainstream of media studies deals almost exclusively with democratic regimes. Existing research on mass media in authoritarian regimes focuses more on state mechanisms of control than on actual media reporting and much more on moments of crises for the regime (related to protests and potential democratization) than on times of its stable functioning.
In order to better understand the role of mass media in the functioning of authoritarian regimes, our analysis asks how pluralistic actual media reporting on an 'ordinary' policy issue is in fully authoritarian regimes.
Most literature on authoritarian regimes perceives mass media (often implicitly) as a mere transmitter of messages ('propaganda') produced by the ruling political elites (see Walker & Orttung, 2014, for a recent example with an explicit reference to mass media). Akhrarkhodjaeva (2017, summarized in Tables 2.5-2.7) has conducted a meta-analysis of political regime typologies and a dataset of electoral malpractice compiled by Sarah Birch, which-taken together-show that manipulation of mass media reporting is the second most common deviation from democratic standards in fully authoritarian regimes. In this view, there is no pluralism and those who are visible in media reporting represent the ruling elites. Accordingly, media appearances could be used to analyse the composition of the ruling elites.
However, there is also an alternative view in the literature which stresses the need of the ruling elites to get reliable information regarding public sentiment concerning important policy issues. It has also been argued that media reporting can be used to give criticism a controlled channel of expression in order to avoid unexpected and harder to control eruptions of public anger in the form of protests. Based on a large-scale analysis of online censorship by the Chinese government King, Pan and Roberts (2013, p. 339) argue that the Chinese leaders 'seem to recognize, looking bad does not threaten their hold on power so long as they manage to eliminate discussions associated with events that have collective action potential'. The authors claim that 'this 'loosening' up on the constraints on public expression may, at the same time, be an effective governmental tool in learning how to satisfy, and ultimately mollify, the masses'. Heydemann (2007, p. 21) claims that Arab authoritarian rulers 'recognize the value of these technologies [i.e. new media] as steam valves: outlets that mitigate social pressures that might otherwise become politicized'.
In this context, it is also often argued that the Internet offers a new opportunity for dissenting voices in authoritarian regimes (for an exemplary critical discussion of this argument related to our case countries see Imamova, 2015;Pearce, 2014). Looking at pluralism in mass media reporting, our analysis also tests the hypothesis of an increasing degree of pluralism from TV over print media to the Internet (i.e., news websites in the case of journalistic mass media).

Operationalizing Media Pluralism
At its core, media pluralism is a normative concept, related to the democratic idea of free debates, implying the ability to challenge existing power relations and to engage in a debate based on the merits of the better argument (cf. e.g., Hrvatin & Petković, 2015;Karppinen, 2013). At the same time, it is not possible to determine an ideal score of perfect pluralism. As pluralism is not a value in itself ('the more, the merrier'), it is restricted to (what is deemed to be) legitimate ideas presented in an accepted manner with reasonable arguments.
In this context, Valcke, Picard and Sükösd (2015) differentiate between external pluralism, the plurality of media outlets and media ownership, and internal pluralism, the plurality of opinions in actual reporting. As the authors highlight, although the conditions of external pluralism 'increase the possibility of achieving the objectives of pluralism, they do not guarantee it because they are not necessary and sufficient conditions for its existence' (Valcke et al., 2015, p. 2). Nevertheless, attempts to measure media pluralism focus on external pluralism, not only because it is easier to measure, but also because policy measures to improve-or in the case of authoritarian regimes, restrict-media pluralism are foremost directed at external pluralism (cf. e.g., Aslama et al., 2007;Picard, 2000).
However, in order to assess 'limited pluralism' as a core feature of authoritarian regimes and, thereby, to understand the visibility of alternative opinions in authoritarian regimes, internal pluralism is the vital indicator. In this sense, 'political pluralism in the media refers to fair and diverse representation of, and expression by (i.e., passive and active access) various political and ideological groups, including minority viewpoints and interests, in the media' (Hrvatin & Petković, 2015, p. 113).
In order to assess the 'limited pluralism' in media reporting, this analysis will focus on actual reporting (internal pluralism) in the case of 'ordinary politics', i.e., debates about a policy issue. For an assessment of pluralism in everyday politics, the policy issue under study should not directly challenge regime legitimacy; however, it should be related to political decisions and be of great relevance for the respective country, so that the political leadership cannot simply ignore the issue. Moreover, the selected issue should allow for more than one policy decision as an outcome, so that there is-in principle-room for a pluralistic debate. To allow for a comparison between countries, the respective policy issue should fulfil these criteria in all case countries over a longer period of time.
In order to measure internal pluralism, this analysis refers to 'sources' (i.e., various political and ideological groups) as well as 'opinions' (i.e., various viewpoints and interests). The first aspect indicates the variety of people or institutions being quoted by journalists. The key question concerning 'limited pluralism' here is to what extent voices not belonging to the ruling elites are being quoted and which voices in particular. The second aspect indicates a variety of opinions (independently of the author).
The key question concerning 'limited pluralism' here is to what extent diverging and conflicting opinions can be voiced in the media.
For the actual content analysis, the most popular media as well as the media outlets of the major political camps should be included. The focus of this analysis is exclusively on journalistic mass media, as social media require a different form of analysis.
In order to identify sources of information (as opposed to mere references to actions by the same people or institutions) and, even more so, to identify opinions, the qualitative content analysis has to be done manually with the support of specialised coding software. In the text corpus, we have coded all sources being quoted (including interviews and guest authors). In order to assess plurality of opinion, frames 1 related to the policy issue have been coded, as well as whether the respective frame is thought to be adequate or not (i.e., whether the frame 'explains' the respective policy issue). Finally, positive and negative references to specific policy options have also been coded.

Case Selection
For our analysis, we have opted for authoritarian states in the post-Soviet region, as they qualify as most similar cases in terms of historical and geopolitical context. Moreover, with Azerbaijan, Kazakhstan, and Turkmenistan, three authoritarian states can be selected for which the construction of oil and gas export pipelines is a relevant and similar policy issue over a long-term period. This policy issue is controversial, as pipelines have been proposed in all cardinal directions: north to/via Russia, east to China, south to/via Iran or Afghanistan and west to Turkey and/or the European Union. Plans for the construction of oil and gas pipelines from the Caspian region have been drafted since in the late 1990s. All major projects starting in the three countries were completed by 2011 (for an overview of the post-Soviet pipeline infrastructure see Heinrich, 2014; for related official discourses see Heinrich & Pleines, 2015). Accordingly, our analysis looks at mass media reporting on export pipelines from 1998 to 2011.
Comparative political science literature (namely Diamond, 2002;Levitsky & Way, 2010;Roessler & Howard, 2009) largely agrees that Azerbaijan and Kazakhstan belong to the category of full or hegemonic authoritarian regimes, while Turkmenistan is often described as closed authoritarian. This assessment is also confirmed by political regime indices (namely Freedom House, Polity IV, Bertelsmann Transformation Index and Economist Democracy Index); Azerbaijan and Kazakhstan are clearly in the group of authoritarian countries, while Turkmenistan receives the most extreme values (for an overview see Akhrarkhodjaeva, 2017, Tables 1.6 and 1.7).
Country rankings of press freedom match the overall assessments of the political regimes. In the 'Freedom of the Press Index' (Freedom House) both Azerbaijan and Kazakhstan score between 65 and 85 on a scale from 0 to 100, where a score above 60 indicates 'not free'. For both countries, the long-term trend is a slight worsening of the situation. Turkmenistan scores above 85 for the full period under study. In the 'Press Freedom Index' compiled by Reporters without Borders all three countries are continuously ranked among the 80 worst countries, Turkmenistan is often among the 10 worst (an overview of the rankings is provided by Pleines, 2014).
Media consumption in the case countries is dominated by largely government-controlled TV broadcasting, which is the primary source of information for the vast majority of the populace. In Azerbaijan and Kazakhstan, the circulation of independent newspapers and journals has, since the 1990s by and large, been restricted to the major cities because many people cannot afford to buy print media and there are logistical, commercial, and political restrictions to country-wide distribution (Caucasus Analytical Digest, 2011;Junisbai, Junisbai, & Ying Fry, 2015). The importance of the Internet has increased significantly over the period under study. According to 'Internet World Stats', Internet penetration (in percentage of the population) rose from 0.1% in 2000 to 44% in 2011 in Azerbaijan and from 0.5% to 35% in Kazakhstan. 2 Turkmenistan is in a separate league, mainly due to direct state control and censorship of all mass media in the country (Anceschi, 2011). In order to ensure control, access to the Internet has been heavily restricted in the country. Internet penetration still stood at a mere 2% in 2011. 3 Our analysis covers the reporting by national newspapers, journals, TV stations, and professional (journalistic) Internet sites in Azerbaijan, Kazakhstan, and Turkmenistan. Only mass media with nation-wide (or capitalbased) coverage that addresses a national audience in the respective country were included. Media that contained, on average, less than one report on our topic per year was not included. News agencies were not in-1 Frames can be defined as the basic cognitive structures that guide the perception and representation of reality (Gitlin, 1980, p. 6). 'To frame is to select some aspects of a perceived reality and make them more salient in a communicating text, in such a way as to promote a particular problem definition, causal interpretation, moral evaluation, and/or treatment recommendation for the item described' (Entman, 1993, p. 52, emphasis in the original). 2 Internet World Stats, available at http://www.internetworldstats.com 3 Internet World Stats, available at http://www.internetworldstats.com cluded because they do not directly participate in national debates.
For Azerbaijan and Kazakhstan, the database includes a large part of the most popular (print, TV, and Internet) and important media for the major political camps as well as national specialised business journals from 1998 to 2011, if these exist. The most popular TV stations and print media were identified based on viewer statistics, print circulation figures, public surveys on media consumption, and expert assessments. For Internet sites in Azerbaijan and Kazakhstan, lists of the most frequented news websites were used. 4 For Turkmenistan, which has strong state control over all media, only the state TV channels have been included in the analysis.
The resulting text corpus comprises 3,618 media reports on export pipelines published between 1998 and 2011 by 38 different journalistic media outlets. As explained in section 2, our analysis considers pluralism in sources (i.e., actors being quoted) and pluralism in opinion based on manual software-aided coding. 5

Results
As TV dominates media consumption in all three case countries, we begin with an analysis of pluralism of sources, i.e. actors being quoted, in TV reporting. For TV channels, the text corpus comprises a total of 925 'quotes', i.e., direct or indirect quotes plus interviews. Of these, 565 are from Azerbaijan (AZ), 124 from Kazakhstan (KAZ), and 236 from Turkmenistan (TKM). If we consider all pro-government actors, i.e., politicians and state officials who are part of the ruling elites, their share in the total number of TV quotes stands at 52% for Azerbaijan, 59% for Kazakhstan, and 79% for Turkmenistan, as shown in Table 1. At the same time, a real opposition, i.e. politi-cians openly opposing the government, is only verifiable in Azerbaijan, where it accounts for 1% of all quotes.
It is telling for the special position of Turkmenistan that the president personally accounts for nearly two thirds of all pipeline-related quotes, combined with officials of the regime increasing to a total share of 79%. All the remaining quotes come from foreign politicians and business people, mainly in the form of selected quotes from official (i.e., diplomatically phrased) press conferences after meetings. Compared to Turkmenistan, there is more diversity in Azerbaijan and Kazakhstan. First, the quotes from representatives of the regime are much less dominated by the presidents personally; instead, a larger number of people speak for the regime. Although the outspoken political opposition is in fact banned from TV appearances, independent domestic actors, namely experts and representatives of (foreign) private business, are clearly visible.
The interesting question, therefore, is whether this broader variety of people being quoted is reflected in a broader variety of opinions in media reporting. In a first step, we look at different groups of arguments-'frames'-used to justify or explain pipeline decisions. In the case of export pipelines, the 'classical' frames are 'geopolitics', whereby pipelines are considered as a way to foster alliances in foreign policy, and 'profitability', i.e. pipelines are a means of generating financial income for the country (see Heinrich & Pleines, 2015, for details on these frames).
Our results show that some 'apolitical' frames, namely 'diversification' and 'technical feasibility', are quite popular, while controversial issues like the 'environment' or the 'resource curse', i.e. the negative consequences of a resource boom, are by and large neglected. However, in each country, the five most popular frames are men- Table 1. Share of different groups in total number of quotes in TV reporting. Source: Authors' own analysis and calculation. tioned in at least 10% of all media reports. That means that there is variety when it comes to arguments about specific policy decisions (i.e., pipeline routes in our case). However, this variety is largely consensual. In the case of Turkmenistan, the 369 reports included in the analysis comprise only 13 remarks questioning the appropriateness of a specific frame; nine of them relate to 'political feasibility', thus in fact supporting the official project of a pipeline through Afghanistan despite questions about its political feasibility. Kazakhstan has the highest number of critical comments about frames, a total of 118 accounting for 8% of all reports included in the analysis. Here, two-thirds refer to 'geopolitics' and 'profitability'. In both cases, the largest share of critical comments can be attributed to business journals and news websites. In Azerbaijan, critical comments about the adequateness of specific frames are more evenly spread. Throughout the full text corpus, there is no recognisable pattern, neither concerning media type nor political orientation. However, in total only 4% of media reports in Azerbaijan include a critical reflection about frames.

AZ (TV) KAZ (TV) TKM (TV)
When it comes to concrete policy decisions, i.e., an opinion for or against a specific pipeline project, the large majority of reports avoid any clear position. In all three countries, about 60% of assessments made neither support nor criticize the pipeline projects they are reporting about. Quite often this makes for very dull reading with long lists of technical information about pipeline routes, partners, and through-put volumes.
In order to understand pluralism of media reporting in a political regime, it is also important to assess differences by media type. This question relates to the idea that-although mass media reaching the whole population, namely TV, is strongly controlled by the state-there are niches of pluralism in authoritarian regimes which are, in principle, accessible for large parts of the population. Although most people never bother to get access, in times of growing discontent with the regime, these media outlets-and the journalists working there-may grow into a more important role. Traditionally, small newspapers with an intellectual image were the major representatives of this pluralism in the media landscape. Increasingly, the Internet has taken over this role. As this study focuses on journalistic mass media reporting on policy issues, the relevant part of the Internet are news websites.
In order to allow for a more fine-tuned differentiation, we have divided print media into pro-government, independent, and opposition. In the case of Kazakhstan, there also is a critical mass of business newspapers and journals which-similar to the 'Financial Times' and 'The Economist' in Great Britain, for instance-address first of all a business audience, but offer comprehensive reporting about political events.
The data for Azerbaijan and Kazakhstan, reported in Tables 2 and 3, clearly support the assessment that TV reporting is closest to the government and thus the least pluralistic. The share of quotes from official representative of the regime is above 50% and the share of formally non-aligned experts does not exceed 5%. At the same time, reports with a neutral stance about the policy issue dominate with about 70%.
The picture for the other media types, though, is much less clear cut. Azerbaijani print media fit the expectation of more pluralism. Concerning the share of quotes from representatives of the regime, the difference between pro-government and oppositional print media is much less distinct than the difference between print media in total and TV. News websites in general are similar to oppositional print media in the share of quotes from official representatives of the regime. Formally nonaligned experts, though, are most visible online. At the same time, the share of neutral reports is highest for news websites.
In Kazakhstan, however, business-oriented print media offers the highest degree of pluralism as far as quotes Note: As one newspaper has been coded as independent, it counted neither as pro-government nor as opposition. As a result, the sums of print-pro and print-opp are smaller than the total for print. Note: The three business print publications included in the analysis are not coded as pro-government or opposition, so there is no overlap between the three sub-categories for print publications. As three non-business newspapers have been coded as independent, they counted neither as pro-government nor as opposition. As a result, the sums of print-pro, print-opp and print-bus are smaller than the total for print.
are concerned. It has the lowest figure for the share of quotes from regime representatives and the highest share of quotes from experts. Kazakhstan's news websites have an even higher share of quotes from business people, but the share of experts is lower only in TV reporting. At the same time, news websites have the highest share of non-neutral assessments, being the only media type in Kazakhstan where less than half of all assessments are neutral.

Conclusions
Our results for Azerbaijan and Kazakhstan as typical fully authoritarian regimes illustrate that internal media pluralism is so limited that it is completely toothless in political terms. The political opposition is not visible in mass media reporting at all-outside its own small print outlet in Azerbaijan. As a result, pluralism is restricted to experts and foreigners. As the policy option preferred by the regime is often not clear at the time of reporting, most media outlets opt for a neutral stance in order to err on the side of caution. Consequently, even in 'ordinary' politics-which pose no threat at all to regime survivalcontroversial debates about different policy options do not take place in the mass media. That is why the situation in Turkmenistan-at least in relation to internal media pluralism-seems to differ more in degree than in kind. Though there is a difference between types of media, with the exception of TV it is not clear-cut. In both Azerbaijan and Kazakhstan, TV is clearly the least pluralistic medium in terms of quotes and the most neutral in terms of opinion. Between the other media types the differences are minor when it comes to pluralism of sources, with the business press in Kazakhstan being the only outlier offering more pluralism. As far as ex-pression of opinion is concerned, news websites in Kazakhstan offer slightly more non-neutral assessments than the other media types, while those in Azerbaijan offer markedly less.
A tentative explanation for these differences might be that those media outlets which are most clearly associated with the opposition, like oppositional newspapers and also news websites in Azerbaijan, have to be careful not to overdo their criticism as they are under special surveillance. Media outlets which are considered to be closer to the regime, like business print media and many news websites in Kazakhstan, may find it easier to voice some different opinions as their loyalty to the regime is less likely to be questioned.
In summary, using the terminology of discourse theory, one can state that fully authoritarian regimes-as long as they remain stable-enjoy discursive hegemony in mass media not only in the discourse about regime legitimacy but also in 'ordinary' policy discourses.
Thus, if we take TV reporting as a 'mirror' of the ruling elites, we obtain some insights into the elite structure in the countries under study. First, in Turkmenistan, authoritarian rule is clearly more personalised. Even on ordinary policy matters, it is first of all the president who personally speaks for the regime. In Azerbaijan and Kazakhstan, however, it is a broad group of political elites and state officials who represent a more collective leadership to TV audiences. Business representatives form a distinctive and highly visible group in TV reporting in Azerbaijan and Kazakhstan-where they collectively get more quotes than the respective president. This highlights the role of business actors (including foreign ones) in patronage schemes, or 'pyramids of power' (Hale, 2015), in fully authoritarian regimes. In Turkmenistan, however, business as a relevant actor is marginalized. This again points to more centralised rule in Turkmenistan.
However, unlike the Chinese leadership, the example which we quoted in the introduction, the ruling elites in fully authoritarian regimes of the post-Soviet region do not seem to use any mass media as a way to obtain a second opinion on policy issues or to manage public dissent. This is in contrast to 'authoritarian upgrading', which uses mass media to check public sentiment, to build pre-emptive consensus and to channel dissent (cf. e.g., Cavatorta, 2010;Heilmann, 2010;Heydemann, 2007). Nevertheless, all three post-Soviet regimes have achieved a remarkable degree of stability.
In this context, our results point in two directions reaching beyond our own research. First, as mass media do not give a voice to the political opposition in fully authoritarian regimes, a focus on social media is justified. Although, authoritarian regimes-including in Azerbaijan and Kazakhstan-have increasingly used social media to promote their own agenda, harass its critics or dissuade Internet users from political activism (Anceschi, 2015;Freedman & Shafer, 2014;Lewis, 2016;Pearce, 2014Pearce, , 2015Pearce & Kendzior, 2012), social media are still the only communication channel through which oppositional activists can reach a broader audience. The decline of journalistic mass media vis-à-vis social media does not necessarily favour the opposition, but it offers a new arenaone which is also used by journalists who have been ostracised by the regime. In this respect, Turkmenistan as a closed authoritarian regime presents an extreme case because here not even social media can fulfil that function.
Second, while the difference between fully authoritarian regimes like Azerbaijan and Kazakhstan on the one hand, and closed authoritarian regimes like Turkmenistan on the other, is clearly visible in the degree of internal pluralism, its impact on domestic politics should be limited, as opposing voices are not represented in mass media in either regime type. If there is a distinction between full and closed authoritarianism, it most likely lies in the interaction between political elites, not in mass media communication with the broader public. Thus, concerning media pluralism, the distinction between fully authoritarian regimes and hybrid regimes might be more relevant. Though a systematic comparison is still lacking, a cursory look at media reporting in hybrid regimes demonstrates that the opposition-though discriminated against-is clearly visible in mass media reporting, e.g., in the post-Soviet region in the cases of Kyrgyzstan, Ukraine, or Russia (especially before 2008).