News Media Access for Civil Society Actors During Democratisation Processes: The Case of the Transitional Justice Campaign in Uruguay

ABSTRACT Theoretical scholarship on media democratisation neglects the role of representing groups in civil society, which in new democracies emerging from authoritarian regimes are frequently marginalised. These groups may also contest the form of democracy that has been implemented, and the way in which it deals with key transitional issues such as past human rights violations. This study focuses on how such groups are represented in one post-authoritarian society: Uruguay, which returned to democracy in 1985. Through examining media access for civil society groups in the newspaper coverage of two critical plebiscites on transitional justice issues (1989 and 2009), combined with interviews with journalists and civil society representatives, it argues that there are significant barriers to media access for civil society actors in the return to democracy, which relate both to changes during the democratisation process and domestic journalistic norms. This signals the need for both greater attention to the normative media role of representation during processes of democratisation and a more critical approach to media democratisation theory.


Introduction
After a period of authoritarian rule, the restoration of a democratic media is a critical component of the political democratisation process. Mainstream theoretical literature on media democratisation, which largely emerged after the so-called "third wave" of democracies in Eastern Europe and Latin America, is broadly optimistic if pragmatic about this re-opening of broadcast and press outlets. Key formal changes include dismantling censorship and slander laws. In the early transition to democracy, normative media roles include pressure to support and provide information about the state, to enable trust in new institutions to be established, and changes in the political economy of the media are expected, including the emergence of many new publications in an increasingly competitive news market. Later in transition, "politics as usual" begins to dominate news coverage as democracy matures, which sees a shift from covering transitional issues to policy concerns. Yet in this scholarship, little attention is given to the democratic media role of representation during the process of democratisation. This is an essential component of democratic media in any society; however, it could play a particularly vital role in postauthoritarian contexts such as those in South America. Authoritarian states typically exert pressure on the media through censorship and shutdowns to stem and fragment dissenting groups. Thus, a key question is whether groups who have been excluded or censored from media outlets during the authoritarian period are able to gain access to publications in the new democracy, as well as how these groups are represented if they do succeed in gaining access. This could help provide evidence of substantive improvement in democratic performance of the media over time.
This paper directly engages with these questions by drawing on a content analysis of daily newspaper coverage of the two plebiscites on the Expiry Law (1989 and2009), along with interviews with journalists and human rights organisations. While these plebiscites may seem historical, current research on Latin American media and democracies increasingly recognises that processes of democratisation are not linear nor time-bound, whether this relates to media performance in persistent "insecure democracies" (Hughes et al. 2017), the rapid shifts between media models in some contexts (Hallin 2020), or the swing back to authoritarian tendencies with the "return of the right" (Cannon 2016).
The paper finds that there are significant obstacles to civil society voices challenging state-led narratives during the democratisation process, some related to domestic journalistic norms and others related to how the media shifts across democratisation. It argues that theoretical scholarship on media democratisation pays insufficient attention to the importance of media representation in post-authoritarian context due to its focus on the state, and that a more critical approach is necessary that refocuses attention on marginalised groups and more substantive indicators of a democratic media. The paper begins with a literature review of media democratisation, the media, democratisation and transitional justice in Uruguay, and source-media relations. It goes on to explain the methodology for the study. It then shares key findings regarding newsworthiness; the quantity of source access; and the quality of source access. Finally, it concludes.

Media Democratisation Theory
The dominant scholarship in the media democratisation field began in dialogue with transition literature in political science, engaging with the work of "transitologists" and utilising the "stages of transition" as a conceptual framework and theoretical approach where there was none from within the discipline (O'Neill 1998, 6). This rich literature developed out of the so-called "third wave" of democracy, which saw many states in Eastern Europe and Latin America return to democracy in the 1990s and early 2000s. Two of the main contributions towards theory building in media democratisation have used this approach to varying degrees. Rozumilowicz identifies four stages: pre-transition; primary; secondary; and late or mature (2002). Voltmer's earlier work also used more defined stages. However, later work acknowledges critiques from Carothers and others that the approach is teleological and insufficiently accounts for contextual differences. Thus, the broader categories of "before regime change", "transition" and "after regime change" are adopted (Voltmer 2013, 77). For Voltmer, like Rozumilowicz, the normative role of the media is contingent upon by the stage of transition: "depending on the particular stage of the transition process, the media affect the course of events in different ways" (ibid. 72-73).
A key feature of the transitional stage is dismantling the architecture of authoritarian repression and constructing its democratic replacement: the lifting of censorship laws, desacato (insult) laws, and the introduction of laws protecting the freedom of information. The transition period is a time when publications will flourish in a newly competitive market and reporting may be particularly critical as a response to the period of suppression. Reliance on the media for political information is high, as citizens adapt to change and use information-dense newspapers to keep track of events (Loveless 2008). As consolidation is the period in which party politics becomes normalised, it is anticipated that newspapers will shed some of their initial over-zealous reporting and settle into following the political agenda. This is the point at which there may be a return to previous journalistic norms of, for instance, partisan journalism (O'Neill 1998;Voltmer 2013), though Randall notes this may be combined with practices internalised during the authoritarian era such as self-censorship (1993). Bennett (1998) observes that the media may pose a hindrance to consolidation through critical reporting that undermines nascent institutions. However, the role of the media is broadly considered to be positive: "[f]rom a normative perspective, one would expect the media to contribute to more transparent and responsive institutions and to a better-informed and more engaged citizenry" (Voltmer 2013, 96).
As this scholarship partly developed out of mainstream transition literature, it is open to some of the same critiques. First, there is a teleological expectation that the media will democratise once formal restrictions are lifted. There have been calls to test empirically what the media actually does during democratisation (Jebril, Stetka, and Loveless 2013), particularly in light of regular broad assessments of the media in Latin America as having persistent and critical democratic deficits Lugo-Ocando and Santamaria 2015). These are attributed to high concentrations of media ownership, vestigial laws from the authoritarian era, insufficiently radical reform in the postauthoritarian era, and a lack of journalistic professionalism that creates sufficient space between state and media to enable accountability or watchdog roles, often linked with a historically partisan press. As mentioned previously, studies have found that the democratic role of information is particularly valuable during democratisation (Schmitt-Beck and Voltmer 2007;Loveless 2008). In terms of the representation role, market diversity tends to be used as a proxy for diversity of representation, though the link is not grounded in empirical research. Thus, the predicted increase in publications in the transition period implies a concomitant increase in representation of groups in society. As previously highlighted, this development would be of particular value in post-transition societies in which certain groups have been marginalised.
Second, like the institutional focus of the transition literature, the state is identified as the key driver of media democratisation. Few would contest that the state is vital in creating a promising legal and commercial environment, however this includes a duty for the media to support the state in new democracies. Voltmer places particular emphasis on the role of the media in "making institutions … work", and the development of a culture of political communication is also stressed to facilitate democratic participation (2013,75). Again, this raises questions particularly pertinent to post-transition societies in which the state is the very site of key democratic obstacles created by pacted transitions, and these may be resisted by sectors in civil society as well as re-emergent political parties. In Latin American countries, the centrality of the state to media democratisation is further problematised by the role of the state in colluding with media corporations to the detriment of media pluralism (Segura and Waisbord 2016).
A key concern for the purposes of this study is that discussion of the representation role of the media is limited to reporting protests during the authoritarian period and, thus, is wholly located in scholarship on the pre-transition period. This is discussed in terms of the "witnessing" role of the media, which can bring legitimacy to the demands of protestors by covering their demonstrations and reporting their demands (Bennett 1998). However, it is not mentioned in scholarship in later stageseither in early transition or consolidation, even though it is now widely accepted that dictatorship-era issues are not neatly resolved at the point of return to electoral democracy. Key among these "pending issues" is transitional justice (Sikkink 2011). Overall, then, mainstream scholarship on media democratisation emphasises the normative role of the media in supporting nascent democracies; but says little about supporting those groups contesting the form of the state that emerges.

Media, Authoritarianism and Redemocratisation in Uruguay
Uruguay's reputation for democratic stability and quality press was initially eroded by political conflict in the late 1960s and decisively lost with the civic-military authoritarian regime of 1973-1984. Prosperity and democratic stability in early 1900s onwards sustained a long tradition of quality, partisan newspapers, which reflected the domination by parties of the political sphere. However, the late 1960s saw the breakdown of the country's exemplary tradition of press freedom as part of the government's response to the Tupamaro guerrillas. When a left-wing political magazine, Marcha, published an editorial in support of their activities, then-President Pacheco invoked the Security Measures Act for emergency powers to confiscate the print run of the edition in question. This began a series of suspensions, confiscations and selective censorships of left-wing and Communist newspapers which continued up to the military coup on the 27th of June 1973. This pre-emptive censorship weakened press freedom, both through direct measures and the beginnings of "docile obedience to government censorship, manipulation of information and outright distortion of the news" (Faraone and Fox 1988, 152).
When the military took power in 1973, the regime was distinguished by its high level of surveillance, unlawful detention and torture. Overall, approximately 160 detainees were disappeared, the majority of these in Argentina. A further 32 people died during torture and 26 were killed by extrajudicial execution (SERPAJ-Uruguay 1993). By December of 1973 "all Marxist and far left publications were banned" (Alisky 1981, 199) and the first wave of media restriction shifted to "absolute control" (Faraone 2003, 237). In attempting to report the human rights violations committed by the military, the country's journalists were largely spared the violence and intimidation seen in neighbouring Argentina (Knudson 1997). Rather, editors would receive phone calls from the regime with instructions not to report specific events, or to remove specific pages from their newspapers. 1 All but one of the country's long-standing newspapers were shut down during the dictatorshipthe surviving publication was El País, which remains the newspaper with the biggest circulation.
Upon the return to democracy in 1985, the approach of the transitional administration to dealing with past human rights violations was passivity and denial. A de facto amnesty for military personnel involved in human rights violations was agreed during transitional talks between outgoing military and incoming civilian regimes, known as the Naval Club Pact. After cases were brought on behalf of families of the disappeared in the early years of the return to democracy, this was formalised in 1986 with the Ley de Caducidad or Expiry Law, which constituted the single biggest legal and political obstacle to prosecutions for crimes committed during the dictatorship (Lessa 2013). The growing civil societyled anti-impunity campaign brought about two plebiscites in an attempt to have the law derogated. The plebiscite of 16th April 1989 saw the law upheld with 57 per cent support, while that of 25th October 2009 saw it upheld with 53 per cent of the vote. In the interim, the victory of the Frente Amplio (Broad Front) left-wing coalition in the 1994 election saw Uruguay join the "left turn" of the region. Slow but steady progress on impunity accompanied the Vazquez administrations (2005-2010 and 2015-2020), however Mujica (2010-2015) did not support the anti-impunity campaign. The Expiry Law was finally derogated in October 2011. This slow burn towards justice is why Uruguay is noted by scholars as constituting a "textbook" example of the "justice cascade" (Lutz & Sikkink 2000), which, emphasising the role of judicialisation, articulates the way in which some countries experience a delay in dealing with past human rights violations after the return to democracy. In Uruguay's case, the role of civil society actors was particularly key (Lessa 2013;Skaar 2013).
The restoration of the media was an altogether more straightforward process. The leftwing press was considerably depleted. However, many formal features of media freedom were immediately restored with the return to the guarantees of the 1967 constitution (others, such as access to state information would wait until well into the 2000s). A key new publication in the early return to democracy was La República, which launched in 1986 by Federico Fasano Mertens, an Argentinian who had worked as a journalist in Mexico as well as contributing to a UNESCO consultation on the media and social communication. Later, in 2006, La Diaria launched to cover human rights and social justice issues. Though frequently designated among the freest media in the region by Freedom House, Uruguayan media is blighted by similar concerns to its media democracy as others in the region, including high levels of concentration of media ownership. Recent reforms drawing praise from UNESCO have focused on broadcast media (Segura and Waisbord 2016). While these are positive and progressive policies, daily newspapers remain important for reporting social and human rights issues because they tend to set the news agenda for broadcast news and are more information rich (Bonner 2009).

Understanding Media Access: Theory and Concepts from Source-Media Relations
It is long-established that the mainstream media privileges official sources (Sigal 1973). This gives official sourcesthose in positions of authority or attached to powerful institutionsgreater and more reliable access to the media than non-official or less powerful sources such as members of the public, non-governmental organisations or social movements. This is as true of Western democracies as it is of Latin American ones (Waisbord 2000; Hughes and Mellado 2015). Beyond this, however, there is disagreement over the structure and causes of the dominance of official sources, and the extent to which and under what circumstances it can be challenged by other groups.
The first wave of source-media relations theory arrived in the 1970s-80s and is most closely associated with Stuart Hall and what became known as the Birmingham school of cultural studies, but also includes the work of the Glasgow Media Group (1976;1980) and Herman and Chomsky (1988). Though specific approaches differed, for each the outcome was a structured hierarchy of access to the mainstream media which had the effect of excluding marginal or unofficial sources. The dominance of elite sources in this model is fixed and cannot be challenged. For the Glasgow Media Group and Herman and Chomsky, this dominance is rooted in the political economy of the media, such as high concentration of media ownership, while Hall placed greater emphasis on the role of routinised journalistic practices including that sources be "objective" and that journalists use "authoritative" statements from "accredited" sources (Hall 1978, 58). Schlesinger (1990) questioned the assumption that there is always elite agreement over the meaning of social issues and that this is established before media coverage begins, rather than something formed as coverage unfolds. He also criticised the structural view for its lack of attention to different patterns of access across different media outlets and that its fixed nature "closes off any engagement with the dynamic processes of contestation in a given field of discourse" (ibid. 69). This view gained traction at a time of increasing use of public relations and media engagement by governments, businesses and NGOs (Davis 2000).
Media theory emerging in Latin America in the 1970s-80s developed in the context of increasing influence of the United States on Latin American media content and corporations. In common with Hall's use of Gramsci, scholars argued that a veiled hegemony was being socially reproduced by the media (Somavia 1981). However, these scholars looked to alternative media to decentralise communications and demystify the process of journalism. There are other key features of Latin American journalism which affect the assumptions of Western-based source-media relations research. While Hall identified the journalistic value of objectivity as serving elite narratives on social issues, Latin American journalism is generally accepted to have no consistent tradition of this due to the enduring close relationship between the state and the media Waisbord 2000). Despite Hall's scepticism of objectivity and wider debates about its existence, this journalistic value can affect source use as journalists may seek more than one source to represent an alternative perspective to tell the "other side of the story". This can increase the range of views represented in an article and thereby act as leverage against the prestige of elite sources. In the context of Latin American journalism, elite prestige is not challenged in this way. Instead, as Waisbord describes, "the political clout of a few sources, rarely quoted or only cryptically mentioned, is often sufficient to print stories, making it unnecessary to comb other potentially knowledgeable parties or to search for alternative sources of information" (2000,103).
The field of source-media relations offers various concepts to enable the analysis of media access in journalism. A key journalistic practice shaping media access is, of course, source selection and the concept of source credibility informs this. Source credibility is "a person's believability as a source of information or as the degree to which information is perceived by a journalist as accurate, fair, unbiased and trustworthy" (Reich 2011, 51). Alongside credibility, other criteria for source selection can include (perceived) authority, reliability, accessibility, and the extent to which the source is perceived as media-friendly (i.e., their ability to give soundbites or concise quotes) (Thrall, Stecula and Sweet 2014). Unofficial voices or civil society groups have traditionally been hampered by their lack of resources (Goldenberg 1975;Gitlin 1980), however this has become more flexible since the advent of public relations (Davis 2000).
Research on the effect of elite conflict on source selection has found that disagreements at the government level can open the field for competing primary definitions of social issues (Miller 1993). A well-established model for assessing this is the indexing hypothesis, whereby the accessibility of the media to a range of sources depends upon, or is indexed to, the level of elite consensus around an issue (Bennett 1990). Where agreement is perceived, fewer sources will gain access; where conflict is observed, a greater range of views will be represented. In this way, the perception of elite debate can act as a valve on access for other sources. Finally, a useful concept for gauging the quality of media accessor the extent to which a source is able to give a full representation of the social issue they are commenting uponis Iyengar's distinction between episodic coverage, which tends to be event-driven, restricted to bare details and individual responses, and thematic coverage, which sets social issues in a broader social and political framework and thereby raises questions of causes and accountability (1991).

Methodology: Measuring Media Access Using the Case of the Expiry Law Plebiscites
This study focuses on the two plebiscites on the Expiry Law as a case study through which to measure media access for civil society actors at different points in the democratisation process. The plebiscites constitute a strong case study for media access as they were a key site of contestation between the hegemonic narrative of the state and competing narratives from groups in civil society, as described in a previous section. These two opposing narratives have been characterised as the "ethical" and "state" approaches (Barahona de Brito 1997). What's more, the plebiscites constitute a high-profile example of "formal political activity" (Hansen 1991) that are likely to attract news coverage.
To discern patterns of access as well as contextual and explanatory information, the study draws on a combination of content analysis and qualitative interviews. The newspapers selected for the study are El País and La República, which were the best-selling daily newspapers in Uruguay during the time of the plebiscites. As noted previously, El País survived shutdowns during the dictatorship by adhering to regime censorship; it is associated with the Blancos party and the centre-right. On the other hand, La República was established in the wake of the return to democracy, initially a campaigning newspaper that later became closer to the Frente Amplio, the centre-left coalition. As a new publication, La República is important as it represents a key feature associated with media democratisation in the literaturethe increase in publications, or media diversity, in the return to democracy, which is strongly associated with an improvement in media representation. The criteria for selection included articles (news articles, opinion columns and editorials) which focused primarily on the plebiscites and were published two weeks either side of their dates in order to collect a "reasonably representative" sample that accommodated news cycles and other influences on coverage patterns (Hansen and Machin 2013, 95). These were searched for and copied from the newspaper archive at the National Library of Uruguay in Montevideo during June-July 2015, as digital archives were incomplete. The total sample for this case study was 327 articles.
A two-level content analysis was developed to capture both features of the articles and the sources within the articles. The first level of analysis took the article as the unit of analysis (N = 327). This level analysed or noted the following: headline; valence (position on the plebiscite, expressed as "for", "against", "neutral" and "unclear"); and total number of sources, in order to discern journalistic practices relevant to explaining source access. The first level also included "theme". Thematic analysis has been described as the process of "establish [ing] what perspectives there are [in public debate], and then examin[ing] how they appear as themes in news accounts" (Philo, Briant, and Donald 2013, 29). Perspectives on approaches to transitional justice broadly represent the two competing narratives of the impunity questionthe logic of "ethics" and the logic of the "state" (Barahona de Brito 1997). The former is associated with civil society groups campaigning against impunity and the latter with the initial response of the state, which shifted over time. These were joined by a third category of "game", which was used to categorise articles that reported on the mudslinging between campaigns rather than dealing with substantive arguments for or against.
The second level of analysis took each source as the unit of analysis (N = 496). Each source was categorised by type (politician, military, trade union, NGO, academic, legal, judiciary, church representative, family of the disappeared, member of the public, anonymous). For the purposes of comparison, these are grouped into official and civil society sources in the analysis that follows, with the civil society group comprising NGOs, families of the disappeared, members of the public, lawyers, academics, and church representatives, and the official source group comprising politicians, members of the judiciary, and the police. To gauge quality of access, sources were also coded for the way in which they were quoted or otherwise referenced, these categories were: direct individual; first person (i.e., the author of an article); press release; publication; paraphrased; and reference only.
Qualitative interviews were carried out with 14 interviews with journalists, editors and representatives from key anti-impunity groups in Uruguay in two stages: first in October-November 2014 and then August-September 2016, during which pertinent results of the content analysis were shared with interviewees. Representatives for the principal antiimpunity groups were approached directly and interviews were carried out with key organisers from Madres y Familiares de Uruguayos Detenidos Desaparecidos (Mothers and Families of Uruguayan Detained and Disappeared); Servicio Paz y Justicia (SERPAJ); Observatorio Luz at the PIT-CNT union; and Fundacion Mario Benedetti. Journalists and editors were initially approached on the basis of either (1) their names being noted in bylines in articles collected during the content analysis or (2) their former or ongoing role at either of the two newspapers selected for the content analysis (in a few cases, journalists had moved on from their previous posts). These were then "snowballed" on the basis of suggestions by interviewees; consequently, a limitation of this study is that the sample of interviewees could have been more representative of the timeline across democratisation and across publications. Ten interviews were carried out with two editors (El País and La Diaria) and eight journalists. The journalists were either working for a newspaper or news outlet (Sudestada and El Observador) or were freelance with former roles at either La República or El País.

Newsworthiness and Media Access: The Return of "Politics as Usual" as an Obstacle
For civil society actors to gain access to the media, the issues they are campaigning for must first be considered newsworthy. Overall coverage for the 1989 plebiscite was far higher than that of 2009, with 273 articles across the two newspapers in 1989 and only 54 in 2009. This represents an 80 per cent decline in overall coverage. Journalists and editors offered three explanations for this. First, one interviewee identified the outcome of the first plebiscite as essential in understanding the decline in coverage. 2 This suggests that the chilling effect of the 1989 plebiscite identified in scholarship extended to the media; in turn, that the media followed the political and public agenda. Second, interviewees described the increasing low news value of "old" human rights as time went on. One recalled an editor of quality dailies lamenting that he was "tired" of dictatorship-era human rights. 3 Lastly, the second plebiscite on the Expiry Law took place at the same time as a presidential election and another plebiscite on the voting age. 4 This is important for two reasons. First, that this was the presidential election in which Mujica campaigned and won, without stating any position on the plebiscite. Second, the two plebiscites being held at once had the effect of conflating and thus equivocating the two issues. The newspapers would feature both in single articles as "los dos plebiscitos" (the two plebiscites).
The way in which this decline was manifested across newspapers indicates that campaigning newspapers are no more resistant to change. Figure 1, below, shows that La República featured 2.5 times more articles than El País in 1989. However, in 2009, there was significantly less coverage in both publications. For El País this represented a decrease of 76 per cent and for La República a decrease of 82 per cent. The analysis also makes concrete the previously explained different editorial positions of La República and El País on the human rightsat least in 1989. Overall, the 1989 coverage showed greater polarisation in positions on the issue. In the 1989 plebiscite, El País supported the yellow vote, or vote in favour of the Expiry Law, and La República supported the green vote, or the vote against. La República ran extensive campaigning, with slogans reading "I vote for happiness. Vote green." across its masthead, and full-page adverts promoting the green vote alongside the newspaper's logo. The campaign was attributed to the newspaper's founder and editor of the time, Federico Fasano Mertens, described by an interviewee as "the great communicator, impeller and agitator for the 'no'". 5 By way of contrast, El País ran quarter-page replicas of a yellow voting ballot over several days; a comparatively muted intervention. In 1989, 65 per cent of the coverage in El País was identified as negative towards the cause of the anti-impunity campaign. By comparison, 58 per cent of the coverage in La República was supportive of the "no" vote against the Expiry Law; 9 per cent against/negative; and 12 per cent balanced (N = 196). Although the latter is more balanced, overall, these numbers clearly indicate that the editorial positions of each newspaper were reflected in the way the plebiscite was covered.
During the 2009 plebiscite, El País did not run any campaigning material, while La República featured a more stylised "YES" (this time, for the nullification of the Expiry Law) logo and a series of "memorials" for the disappeared. La República intensified its position on the issue in the 2009 sample with almost 90% of articles (N = 37) in support of the "yes" vote, whereas El País ran more articles that expressed a positive perspective on the nullification -44 percent for or positive about the nullification of the Expiry Law and 17 per cent against (N = 18). In effect, the coverage in El País became more balanced. An interviewee with an editor at El País linked this to professionalisation at the newspaper, which constituted a shift away from partisan reporting (interview with editor, 2016). On the other hand, while the editorial position of La República remained favourable to the nullification of the Expiry Law, the reduction in coverage of the 2009 plebiscite lends support to the view of interviewees that it became less of a campaigning newspaper and more closely aligned with the Frente Amplio and its agenda once the party took power in 2004. As one journalist put it: "La República had famous front pages about human rights questions … but a change of government, a change of [editorial] profile. It's natural." 6 Overall, the changing patterns of newsworthiness and waning support for the antiimpunity campaign described here confirm the prediction of media democratisation literature that there is a clear return to "politics as usual" after transition (Voltmer 2013). While this is considered a positive indication that democracy is becoming normalised, this study highlights that this shift in what is considered newsworthy is not necessarily a "natural" or linear progression of the democratisation process. As discussed earlier, in Uruguay a state-led policy of impunity included narratives that the transition was over and that the country must look forward. In "supporting" the state agenda on political issues, then, either overtly as in the case of El País or by coming to see human rights violations and subsequent campaigns as "old" news, the valve on mainstream news media access began to be closed. We also see the way in which the persistent tradition of partisan journalism played a contributing role in this.

Who Gets to Speak? Civil Society Exclusion from Influential Press
The analysis of overall source access shows that while official sources dominated coverage in the 1989 plebiscite, civil society actors gained more access in 2009. In 1989, official actors gained access to 57 per cent and civil society groups 24 per cent (N = 372). In 2009, the access of civil society sources increased to 50 per cent and official actors reduced to 36 per cent (N = 124). As the previous section indicated, 2009 was also marked by decline in newsworthiness and lack of support for the "yes" vote by the Frente Amplio. Thus, civil society sources were able to gain more access to coverage when the issue was less newsworthy.
Returning to the question of whether the publication matters, Figure 2(a) and (b), below, break down the proportional (%) access gained by sources to the 1989 plebiscite coverage. It shows that civil society actors gained more access to La Repúblicaalmost 25 per cent more than in El País. Therefore, it is clear than La República played a significant part in the overall access of civil society actors to overall media coverage of the plebiscite in 1989. Though official voices still dominate, with 47 per cent of all access to its coverage, civil society sources gained 30 per cent. Compared to the 6 per cent access gained by civil society actors to El País in the same year, this provides further evidence that civil society sources have a better chance of gaining access to new publications in new democraciesat least at an early stage. By 2009, Figure 2(b) indicates that this difference is less pronounced, with an increase in access for civil society actors in both publicationswith 50 per cent access to coverage in El País and 64 per cent in La República. In turn, the proportion of access for political actors is reduced in both publications. Again, the decline in newsworthiness of the 2009 plebiscite and disinterest of politicians in the issue created an opportunity for civil society voicesand the less contentious nature of the issue by then meant it was less controversial for even El País to report.
El País featured very few civil society sources. The newspaper quoted only five NGO sources in its coverage of both plebiscites. Similarly, the newspaper quoted only two trade union sources (the first, in 1989, was quoted in an article criticising the use of trade union funding of the campaign to annul the Expiry Law). In addition to this, two relatives of disappeared persons were used as sources, no churches in 1989, and only one member of the public. Combined, this constitutes a clear pattern of routinely neglecting to use civil society sources. This may not be considered a problem when the campaign was comparatively well-represented in La República; however, interviews with NGOs clarified that when they assessed their access to the media, what they were looking for was access to El País. Access to La República was considered as given: "But they are notoriously identified with the left, yes? They are newspapers defined as the left, so in those cases it is logical [to gain access] because they maintain an identification with that sector that is interested in solving issues of the recent past.
[But] the others? No." 7 This preference for media access was not related to editorial position, then, but to reach, influence and the opportunity this gave for speaking to the political elite as well as wider sectors of the general public. As the above interviewee went on to describe it, left-wing publications "are not media that impact on the majority of people". 8 A representative from Famidesa confirmed that media access to El País was effectively a way to communicate with politicians.
Overall, this analysis indicates that, as expected, official sources dominate coverage of political events. Yet it also demonstrates a variation on the indexing hypothesis (Bennett 1990). While this predicts that elite disagreement on an issue can create an opportunity for non-official sources, elite passivity on an issue also appears to be a situation in which non-official sources can gain access. The clear exclusion of anti-impunity civil society voices from El País well into what would be termed "mature" democracy presents a problem for the assumption that news media become more democratic the further it moves away from the authoritarian period. Rather than developing a competition for access that might enable campaigners to "speak to" politicians, as per Schlesinger's more dynamic pattern of access (1990), access is structured in a manner that still supports the state narrative on impunity established at the point of transition. While this constitutes a troubling authoritarian enclave in the democratic press, the ongoing tradition of partisan news media is also relevant in explaining how this was enabled to persist.

Which Civil Society Sources? NGOs and Source Credibility
Turning to look at which civil society actors appeared as sources, the analysis indicated two significant patterns in access. Firstly, media access remained stable for NGOs and trade unions. Secondly, it increased for families of the disappeared and lawyers. These are represented in Figure 3, below, and will be discussed in turn.
Firstly, the access for NGO and trade union sources is stable in both plebiscites. NGOs gained 37 per cent of civil society source access to the 1989 coverage and 34 per cent in 2009, respectively. This is the highest share of access of any other civil society group. Meanwhile, trade unions gained 11 per cent of civil society source access in both years. That both groups maintained access across the 20-year period is perhaps not surprising given that they are the instigators and protagonists of the anti-impunity movement in Uruguay.
As discussed in a previous section, the ability of sources to gain access to the media partly depends upon the extent to which they are perceived as "credible".  broad remit and the PIT-CNT has been deeply involved in the anti-impunity campaign since its inception. 9 Moving on to those types of civil society source that gained more access to coverage over time, the most significant change is the increased access of lawyersfrom 3 per cent in 1989-27 per cent in 2009. This is related to the announcement by the Supreme Court of Justice (SCJ) on 19th October 2009just six days before the plebiscitethat its judges had found the Expiry Law unconstitutional. This announcement was made in concluding its deliberations on the Sabalgasaray case and newspapers were informed in advance. Thus, the 2009 plebiscite constituted an opportunity for legal sources to gain access. Substantial legal progress began to be made with the victory of the Frente Amplio in 2005. From 2006, trials in Uruguay began under the shadow of the Expiry Law. Lawyers thus gained an increasingly loud voice in the anti-impunity campaign, which their increased access reflects. Here we see patterns of source access that appear to more closely resemble the view that it is a competition between variably resourced groups in society (Davis 2000), however the intervention of the SCJ also represents the growing judicialisation of the anti-impunity campaign in Uruguay and thus the opportunity for sources with greater perceived credibility to enter the fray (Reich 2011). Gaining Quality Media Access: "Impunity Isn't Just the Expiry Law … It's a Culture" The previous sections have focused on the quantity of access gained by sources. This section moves on to discuss findings regarding the quality of access, using measures discussed in the methodology section.
In terms of how sources were cited, the key finding is that of the total sources in coverage of both plebiscites (N = 496), 66 per cent were direct quotes by individually named sources. In many cases, particularly with politicians, these direct quotes ran for paragraphs without much intervention from the journalist and were also used in headlines. For example, an article in El País before the 1989 plebiscite was headlined "Medina: 'Peace and the dignity of the army are at stake'" (2nd April 1989). This is not to say, however, that all actors had the same quality of media access. When this overall figure is broken down by source type across the total coverage, 75 per cent of official sources (N = 258) were quoted individually and directly, compared to 56 per cent of civil society sources (N = 151). However, this improved over time. In 1989, official sources were quoted directly more than civil society sources in 1989 by a difference of nearly 20 per cent. In 2009, civil society sources had caught up with official sources, with 66 per cent (N = 62) being quoted directly compared to 64 per cent (N = 45) of official sources. This indicates a better quality of access for civil society actors later in democratisation, with the caveat that the issue was considerably less newsworthy.
The tendency towards a declaratory style of journalism in the daily newspapers under analysis means the possibility for sources to gain access through direct quotations is high; however, this practice does not necessarily produce the depth of representation associated with it unless the journalist also includes deeper contextual information. As indicated above, the tendency to construct news stories from lengthy quotes generally precluded this. Therefore, key to media representation for anti-impunity campaigns is not only gaining access to newspapers, but having the issue represented in a deeply contextualised, "thematic" way that is able to challenge more simplistic "episodic" narratives of "turning the page".
The second method of gauging quality of access is assessing which themes of an issue are foregrounded in coverage and which are marginalised. In the 1989 coverage (N = 273), 38 per cent of articles reflected the "state" theme and 21 per cent the "ethics" theme. In 2009 (N = 54), this had shifted to seven per cent reflecting the "state" theme and 72 per cent the "ethics" theme. Thus, the "state" theme declined over time, which dovetails with the increase of civil society access during the 2009 coverage. In the 1989 coverage, the "state" theme was articulated in a number of ways. It was most frequently associated with notions of the transition being over (often expressed as an instruction to "dar la pagina" or "turn the page") and national reconciliation being the prioritywith the implication that addressing human rights violations was incompatible with this. El País featured this theme almost exclusively. A less frequently iteration of this theme was the invocation of fear of returning to military regime. A recurrent phrase in both newspapers was whether each side would "adhere to the ballot box" and each side both sought and gave assurances that it wouldsimultaneously signaling democratic integrity of the process while creating an atmosphere of uncertainty.
The "ethics" theme was in 1989 expressed chiefly through the expression "truth, justice and never again" or "verdad, justicia y nunca más", which is synonymous with anti-impunity campaigns across Latin America. Naturally, this theme was raised repeatedly by civil society groups, trade unions, sympathetic politicians from the Frente Amplio and in the editorials and opinion columns of La República in 1989. The unanimous declaration by the Supreme Court of Justice of Uruguay just days before the 2009 plebiscite that the Expiry Law was unconstitutional brought about a significant shift in patterns of source access. Following this, the "ethics" theme in 2009 was more frequently expressed through the argument that the Expiry Law, and so impunity, was unconstitutional. Though this is part and parcel of the spectrum of transitional justice, it is qualitatively different to the message of the civil society-led campaign, which continued to be focused on truth and justice.
This analysis of the quality of media access clarifies the way in which the reliance on direct quotes to construct a news article, associated with the prestigious status of official sources (Waisbord 2000), presents a problem for "thematic" coverage (Iyengar 1991). In the absence of a professional norm to be "objective" or "impartial", there is no pressure for journalists to present additional information or viewpoints that might necessitate the use of secondary sources or a greater number of sources in an article. When questioned about this, one journalist cited the pressures of deadlines for this practice, while another argued that in Uruguay journalists cannot add "interpretation" as this would be considered biased in a way that presenting only direct quotes was not (interviews with journalists, 2016). These explanations notwithstanding, it is clear that the use of direct quotes as the basis for news articles in daily newspapers is a persistent feature of news reporting in Uruguay that is most likely linked to the partisan nature of publications as well as the prestige of official sources. This finding helps to further illustrate the ways in which domestic factors interact with processes of media democratisation in ways that make the performance of normative roles of the media in transitional context highly contingent. Rather than being anchored to the stage of transition, as emphasised by mainstream media democratisation literature, domestic journalistic norms that shape the quality of representation in daily news media appear to be more fixed over time.

Conclusion
This study shows that a range of significant obstacles confronted the civil society-based anti-impunity campaign in attempting to gain access to the Uruguayan news media during democratisation. These are not easily explained by clear categories of "democratic" or "non-democratic" news media performance; instead, they call for greater attention to the fundamental role of domestic contexts in processes of media democratisation. A critical issue was the way in which El País systematically excluded civil society voices, and so the anti-impunity narrative, from its pages. In a post-authoritarian context this looks very much like vestigial censorship (Randall 1993) and points toward the potential for authoritarian enclaves to persist in the media as well as the state. However, in the context of media democracy in Uruguay this structured lack of source access is explainedand to some extent excused by journalistsas the persistence of partisan journalism across daily newspapers and the tendency to use single sources within news articles rather than introduce opposing views for balance.
As such, these findings present a challenge to existing media democratisation literature. While this literature accurately describes some features observed in the newspaper coverage in this study, it neglects to pay attention both to the repercussions of these for representation of groups in civil society and to the representation role as a vital component of media democratisation in itself. In a post-authoritarian context, this latter omission is key: it is the difference between an approach based on procedural democratic models and a more substantive, critical approach to democratisation. Thus, the study is an important departure from existing literature both in its focus on the media and civil society in the context of processes of democratisation and its more critical approach to how media democratisation ought to be defined and measured.
While further research is required, a sketch of a critical model of media democratisation might include: beginning from the assumption that democratic processes and transitionera issues are not linear nor time-bound and nor is democratic media performance; bringing the role of the media in the representation of marginalised groups to the centre of the analysis, on the basis that the potential for reconfiguration of social relations in a postauthoritarian society is a more substantive democratic indicator than the day-to-day of party politics; better articulating how domestic journalistic norms and press models relate toand resistexpectations of media democratisation, over and above vital changes to legal frameworks; and a more nuanced examination of what supporting new democratic institutions might mean in contexts where the media and the state have historically had a very close relationship.
Finally, the case indicates that more empirical and theoretical research is needed into the media and processes of transitional justice in a range of contexts. Scholars in the field of transitional justice have outlined the ways in which news media could theoretically play a role in enabling the public to understand different transitional justice mechanisms and building support for them (Price and Stremlau 2012). However, this study indicates that these may be difficult to achieve in different media environments and highlights the need for transitional justice campaigns and media projects to strategise around domestic norms and practices.