Comparative Analysis of State and Civil Society Discourse on Human Rights Implementation and the Position of Roma in the Former Yugoslav Space

Abstract This study analyses the position of Roma people in the former Yugoslavia using state and civil society discourse on human rights implementation. It reveals that states are failing to give sufficient prioritisation to tackling longstanding discrimination and oppression. Instead of positive the effects predicted by complementarity theory, the findings reveal ‘frame dis-alignment’ between political elites and NGOs. This raises questions of legitimation and performativity in human rights practice. States’ failure to fully engage with NGOs’ policy claims effectively renders Roma people ‘semi-citizens’. Systemic reform by government is required in order to promote equal citizenship through civil society engagement in rights implementation.


Introduction
This study analyses the position of Roma people in the former Yugoslavia using state and civil society discourse on human rights implementation. Approximately, one million Roma are resident in the Former Republic of Yugoslavia (FYR) (CoE, 2015). The myriad identities and rich diversity of people now described by the umbrella term 'Roma' make it one of the largest minority groups in the EU (Tremlett, 2014). 1 European integration processes are transforming governance practices in the region. For some, the emergence of multi-level political structures and governing elites' espousal of rights may have had a positive impact on contemporary citizenship in the FYR (Castles, 2005). Yet for others, significant questions remain as to whether political rhetoric is matched by lived-reality: 'the predicament of Roma minorities as citizens of different European states is still one of the daunting challenges for scholars as well as policy makers' (Sardelić, 2013, p. 1). Historically, their position worsened with the fall of the socialist state system (Barany, 2002;Ringold, Orenstein, & Wilkens, 2005). Yet, of late they have borne the brunt of 'a new wave of "anti-Gypsism"', enduring 'severe poverty, racial hostility, the obliteration of their culture, and exclusion from public life' (Kostadinova, 2011, p. 163; see also Petr, 2004;Poucki & Bryan, 2014;Simeunović Bajić, 2011). This malaise has manifold causes (O'Higgins, 2010;Stauber & Vago, 2013) and has been compounded by the legacy of the Yugoslav Wars 1991(French & Nikolic-Novakovic, 2012Glenny, 1993;Latham, 1999). It has led to the Roma being described as 'the most discriminated ethnic minority' in the EU (Marin Thornton, 2014, p. 106), something that FYR countries have yet to address. Thus, as Sardelić (2015, p. 159) observes, the position of Roma minorities in the light of citizenship regime transformations after the disintegration of the former Socialist Yugoslavia . . . not only manifests in the rights dimension, but also in uneven access to citizenship with regard to the new polities.
The aim of this paper is to engage with these concerns and better understand how discriminatory pathologies endure despite a raft of human rights laws designed to offer protection. The research questions are: How can we characterise policy framing and issue-salience in state and civil society human rights discourse? What contrasts and commonalities exist? How does state-civil society input into human rights monitoring relate to policy theory on collaborative policy implementation? And, what do the findings tell us about civil society and Roma representation at the beginning of the twenty-first century?
The study findings reveal the discursive processes associated with human rights implementation contribute to the enduring failure to address Roma oppression. The following analysis provides evidence of conflict rather than consensus. Expressed in social theory terms, 'frame dissonance' is a feature of government and civil society discourse. As will be explained, in turn this raises questions over performativity and legitimation in human rights practice. The underlying problem is that state and civil society actors have contrasting priorities. Notably, civil society organisations emphasise the 'representation' frame. This is instructive and tells us much about the position of Roma and the nature of contemporary civil society in the FYR. In particular, it underlines how governments are not affording NGOs sufficient opportunities to input their views on human rights implementation, as well as public decision-making more generally. Instead, governing elites are placing rhetorical emphasis on 'soft' discursive rights and generally failing to detail effective implementation measures.
To explore the foregoing issues in further detail the remainder of this paper is structured thus: following an outline of the policy context and the methodology, attention then moves to the findings with comparative analysis of the discourse in state and civil society implementation reports. This first explores framing, before examining issue-salience. The conclusion offers theoretically informed reflection on the implications of the study findings for human rights implementation and the position of Roma people in the FYR.

Policy Context
Post-1991 Europeanisation has been accompanied by the development of a broad range of human rights instruments covering the FYR states. These include: the European Convention on Human Rights (Article 14); The Charter of Fundamental Rights of the European Union (with explicit reference to preventing discrimination linked to 'membership of a national minority'); the Employment Equality Directive (2000/78/EC); Racial Equality Directive (2000/43/EC); CoE Resolution 366 (2014) ('Empowering Roma youth through participation'); CoE Resolution 249 (1993) ('Gypsies in Europe: the role and responsibility of local and regional authorities') and the United Nations Convention on the Elimination of All Forms of Racial Discrimination.
Despite this, the latest survey data show: 'many Roma still face severe poverty, profound social exclusion, discrimination and barriers to exercising their fundamental rights' (European Union Agency for Fundamental Rights, 2014, p. 4). This paper therefore makes a timely contribution on rights implementation by providing critical discourse analysis of the reports submitted in 2013-2015 to the second cycle UN universal periodic review (UPR). This is the evaluation process that emerged in the wake of the 2006 UN General Assembly resolution (60/251) designed to assess progress in upholding human rights law. It is a procedure conducted under the auspices of the human rights council (HRC) (see Cochrane & McNeilly, 2013).
Crucially, both human rights law and the UPR are explicit in emphasizing the foundational role of civil society. This matters, for as Agarin (2014, p. 1972) outline, 'addressing the multiple and inter-connected issues facing Roma communities across Europe requires the full participation of Roma in social, economic and political life'. Thus, official EU guidance requires states to: 'invite non-governmental organisations and other civil society institutions . . . to participate in its Fundamental Rights Platform'. 2 Moreover, the Council of Europe's Fundamental Principles (CoE, 2002, p. 3) stress 'the importance and value of NGOs' contribution'. In turn, these principles have informed the UPR. It provides the opportunity for states and NGOs to submit formal written submissions to be considered by the HRC as part of the review process.

Method
Both the interpretive school of policy analysis (Hajer, 2003;Yanow, 1999) and the literature on social constructivism (Kukla, 2000) place emphasis on the political use of language in order to reveal policy actors' beliefs, values and understanding of social issues (Eden & Ackermann, 2004). These factors lie at the heart of policy analysis. Applied to the present research topic, as Tremlett and McGarry (2013, p. 20) observe, 'discourses surrounding Roma minorities are fundamental to understanding wider notions of social inclusion and cohesion' (cf. Imre, 2009;Nordberg, 2005). However, critical discourse analysis (CDA) has not previously been applied to the study of human rights implementation and Roma in the FYR. This is a key lacuna because CDA is concerned with 'the way social power, dominance, and inequality are enacted, reproduced, and resisted by text and talk in the social and political context' (Van Dijk, 2001, p. 352). Thus, making it is ideally suited to addressing the present study aims.
CDA was operationalised here by focusing on two aspects of language use, 'framing' and 'issue salience'. The former derives from the work of Goffman (1974, p. 21) and refers to a 'schemata of interpretation'. As Snow (1986, p. 464) notes, 'by rendering events or occurrences meaningful, frames function to organise experience and guide action, whether individual or collective'. In short, they 'shape and structure cooperative action by stressing specific values, facts and other considerations, endowing them with greater apparent relevance to the issue than they might appear to have under an alternative frame' (Nelson & Oxleya, 1999, p. 75). In this way they are pivotal to understanding rights implementation involving a range of actors spanning the public and civil spheres.
This study's selection and use of coding frames builds on established mixed methods analysis. It uses deductive coding (Bowen, 2009;Joffe & Yardley, 2003;Sandana, 2012). This involves identifying incidences of key themes or 'frames' in the text of policy documents (Finfgeld-Connett, 2013). Thus, the initial coding frame (or list of themes) was grounded in the key literature on Roma people in the Yugoslav space (e.g. European Union Agency for Fundamental Rights, 2014;Jungić, Palić, Halitović, & Sivić, 2014;Kostadinova, 2011;Sardelić, 2013Sardelić, , 2015; for a full listing-see References). Examples of the initial frames include: 'equality', 'participation', 'rights', 'discrimination', 'inclusion', 'citizenship', 'representation' and 'participation'. Others, such as 'protection', were added during the coding process (see Table 1). This approach is both responsive and adaptable in the way it allows further analytical categories to emerge from the data. It is therefore capable of linking the analysis to wider debates whilst being sensitive enough to incorporate factors specific to a given context. In this way, it maximises the validity of associated findings (for a discussion see for e.g. Hannes & Macaitis, 2012;Wasserman, Michael-Clair & Wilson, 2009).
In addition to frame analysis, the texts were also coded a second time according to 'issue-salience'-or, the number of references to different policy issues (e.g. education, children, housing, employment, poverty, language, health, etc.). Both phases of the coding process were undertaken using electronic versions of the policy documents and appropriate software. The incidence of frames and number of references to different policy areas (issue salience) were logged into a database. Two statistical techniques were applied to the data: Chi squared and F-tests (two samples for variances). 3 Their purpose was to establish whether any differences in state and NGO framing in the human rights discourse were random, or statistically significant. Two key data sources were used in the study: 1. States' reports submitted to the United Nations 2013 -2014as part of the second cycle UPR and 2. Forty-nine reports on the implementation of human rights treaty obligations from a purposive sample of standalone civil society organisations and networks of non-governmental organisations across the region.

State and Civil Society Discourse on Human Rights Implementation
Analysis of the UPR data shows statistically significant differences in the framing of state and civil society discourse (P ≤ .001) ( Table 1). 4 This is a key finding and points to a major shortcoming in rights implementation for Roma people in the region. From a strategic perspective, it reveals that governments are prioritising contrasting issues and following different agendas to civil society organisations. Policy theory is instructive and sets out four ways in which this shapes human rights outcomes. First, issue-prioritisation is key to raising public awareness of discrimination and oppression. For example, it may boost the number of policy actors willing to address a given issue (increased 'buy-in') and lever-in additional resources, thereby increasing the likelihood of effective outcomes (Heath & Palenchar, 1997). Second, state-civil society divergence can lead to a number of policy pathologies (Hogwood & Gunn, 1984). These are wide-ranging and include the inefficient use of resources (for interventions are uncoordinated) and diminution of community trust and understanding (as policies lack apparent relevance and strategic direction). Third, in epistemological terms, state actors' failure to respond to civil society policy claims raises the prospect of implementation being an elite and technocratic exercise (Lowi, 1970). Again, this is problematic because it is not government but civil society organisations that possess necessary 'situated knowledge'-or first-hand experience of the inequalities and oppression facing Roma people in FYR communities. Such contextualised knowledge increases the potential for reflexive, policy learning that responds to local circumstances and addresses discrimination and inequality in an effective manner (Weible & Sabatier, 2009).
The fourth concern is democracy. Notwithstanding the political rhetoric on participation and engagement (OHCHR, 2011), failure to fully embrace and respond to civil society views means contemporary practice is lacking what the policy literature dubs 'input legitimacy' (Papadopoulos & Warin, 2007). With these issues in mind, attention now turns to comparison of state and civil society discourse. First in relation to framing and then policy areas (or 'issue salience').

(i) Framing
Discrimination is the (joint) first-ranked frame in the NGO discourse. In comparison, it is fifth ranked in the state discourse (Table 2). This disparity is emblematic of the longstanding oppression and inequality experienced by Roma people in the former Yugoslavia and beyond (Fekete, 2014). As existing studies attest, it is an issue that not only pervades social attitudes across the region, but has a negative impact on public service delivery. For example, 'practice in the field of social work with minority ethnic groups is often racist, especially when social service users are members of the Roma ethnic group' (Urh, 2008, p. 117). Whilst the current data reveal official acknowledgement of such shortcomings (e.g. 'the fight against social exclusion and discrimination of Roma . . . has been a priority area of concern', Government of Bosnia & Herzegovina, 2014, p. 8)-state discourse is less forthcoming on measures to address the issue.
In contrast, the civil society discourse is highly critical of those in power: 'most persons reporting ethnic discrimination were Roma . . . which the Government should take steps to address' (ERRC & MRCCS, 2013, p. 1). In turn, the Civil Society Organizations' (CSO) reports detail how discrimination is manifest across policy areas. For example, 'Roma continue to face problems accessing health care due to registration restrictions, have lower education enrolment than other groups, and face discrimination in employment' (HRWBiH, 2014, p. 3). Notably, the discourse alludes to institutional discrimination: 'lack of prevention of discrimination against Roma remains a systemic problem owing to ineffective human rights framework' (AIS, 2014, p. 2). This is significant for it points to the way that discrimination is deep-rooted in state services across the region (McCrudden, 1982). In response CSOs are explicit in their call to political elites to implement 'targeted measures to reduce institutional discrimination against Roma' (IATRNE & RNCCP, 2014, p. 5). 'Rights' were first ranked in the state discourse and third ranked by NGOs. The key difference is that the state discourse focuses on the generic espousal of 'soft' rights-(e.g. 'the development of minority rights'-Montenegro, 2014, p. 6). As Trubek and Trubek (2005, p. 343) observe, these 'often have no legally binding force [instead they offer a . . . ] normative system employ[ing] non-binding objectives and guidelines to bring about change in social policy and other areas'. In contrast, the civil society reports look beyond these rhetorical devices to emphasise 'hard' rights. In other words, enforceable legal instruments. For example, it is difficult to argue that [. . . Macedonia] conducts systematic efforts towards harmonisation of the national legislation with international human rights instruments. . .
[government needs to] act upon the recommendation to accomplish progressively the human rights goals as set forth in Human Rights Council resolution 9/12,3. (CRPRC et al. 2014, p. 8) A further key contrast is evident in the 'representation/ participation' frame. Instructively, this was joint first ranked by civil society organisations, yet just fourth ranked in the state discourse. This underlines the 'disconnect' between state rhetoric on Roma engagement and contemporary rights implementation in the region. It also signals a failure to adhere to UN, EU and CoE policy protocols on civil society input. The CSO discourse repeatedly alludes to this and paints a bleak picture of exclusion. For example, 'there are currently no procedures in place to ensure consultation with the affected population on the forthcoming resettlement, as international standards provide' (AIM, 2014, p. 2), and 'the controversy of the consultation process on the draft-text was the exclusion of relevant CSOs from the process' (CRPRC, 2014, p. 2).
Further dis-alignment is evident in the 'protection' and 'inclusion' frames (the third and sixth ranked, and second and fourth in the state and CSO discourse, respectively). Textual analysis reveals that the government discourse exhibits complacency: 'visible progress has been made and the challenges involving their full social inclusion have been identified' (Government of Montenegro, 2013, p. 4). Whereas, the civil society discourse is forthright in condemning state failings. For example: 'apart from the formal establishment the National Roma Inclusion Strategy 2013 -2020. . . not much progress was made in the establishment and functioning of an integrated monitoring system' (IATRNE, 2014, p. 3). Here, governments' framing of 'protection' is noteworthy. It resonates with the literature on minority language rights (May, 2012, p. 18;Sasse, 2008). Specifically, the distinction between active and passive protection. Analysis of the state discourse shows when applied to Roma communities its connotation is invariably passive. Implicit is the idea that whilst governments may seek to offer protection, the status quo will be maintained. There will be no diminution in rights, yet crucially there will be no gains or transformation in the position of this national minority. In short, it conveys a sense of protection without empowerment. Expressed in terms of equalities theory (Fraser, 1997), there is an element of recognition yet thoroughgoing rights and redistribution (of power and resources) are overlooked.
The discourse under the 'citizenship' frame underlines ongoing failure to extend full and equal rights to Roma across the region. The emerging narrative lends empirical support to Gëzim's (2015, p. 197) warning that: 'despite the legally enshrined promise of equality [. . . the FYR is witness to] the emergence of hierarchical citizenship, in which some groups [. . . are] more equal than the others'. Thus, the state discourse tends to be generalised and declaratory in nature, as well as short on immediate measures, instead promising future action (e.g. 'intensive activities are expected in terms of addressing the issues that predominantly involve legal status', Government of Montenegro, 2013, p. 6). In contrast, civil society discourse concentrates on the immediacies of statelessness and needs of internally displaced people. Many of the latter are those forced to move as a result of the post-1991 conflict. CSOs repeatedly highlight key policy failures and the barriers facing such people. For example: 'there are obstacles to obtaining such [legal citizenship] status for the Roma and Ashkali displaced from Kosovo. Few possess personal documentation, such as passports, required to obtain residency' (AIM, 2013, p. 2), and 'since Roma often lack personal identity documents, those who are stateless have found it difficult to meet the requirements to obtain Croatian citizenship by naturalization' (HRWC, 2013, p. 5).
Amongst the remaining frames, compared to the FYR governments, NGOs place greater emphasis on 'integration' (fifth ranked compared to eighth). A key difference is the fact that the state discourse is principally concerned with normative statements. In other words, describing a desired situation or ideal yet to be achieved. For example, 'working towards Roma inclusion [and] enhanced possibilities for integration in the labour market' (Government of Macedonia, 2014, p. 3), and 'the objective of this spatial planning policy is comprehensive integration of Roma into the Slovenian society' (Government of Slovenia, 2014, p. 5). In contrast, civil society language is both critical and practical in orientation. Examples include: 'institutions at the same or different levels of authority were often not familiar with each other's implementation of certain Roma integration measures. They usually acted in an uncoordinated manner' (IATRNE, 2014, p. 5); and 'the situation of Roma in Montenegro. . . there still are complex problems related to their integration, combating poverty and addressing their social, economic and cultural rights' (MIC, 2014, p. 7).
In the UPR data both states and civil society organisations make limited direct reference to Roma people in terms of 'equality' (seventh-and ninth-ranked frames, respectively). However, a key difference is that, in contrast to NGOs, states make repeated reference to affirmative action (and to a lesser degree equal treatment). As an extensive literature attests (cf. Bagilhole, 2009), this is problematic. For whilst such approaches may secure short-term gains, they fail to address the underlying structural and cultural causes of inequality. For example: 'RE [Roma and Egyptian] children have been enrolled in city schools in order to eliminate segregation in education. . . mainly following the affirmative action principle' (Government of Montenegro, 2013, p. 4).

(ii) Issue-salience
Analysis reveals that, when state and civil society organisations are compared, there are also statistically significant differences in 'issue salience'. In other words, the level of attention afforded to different policy areas (P ≤ .001) ( Table 3). 5 This matters because it shows that, compared to those representing Roma communities, governing elites in the FYR hold a differing view of the issues that need to be addressed (Table 4). For example, the lead frame in the state discourse was education. Whereas for NGOs it was employment. This is significant for labour market discrimination contributes to widespread poverty amongst Roma families (see Fischer, 2011;Jungić et al., 2014;Vladisavljev & Kiers, 2004). As one NGO notes, 'despite measures, funds and efforts invested by Croatia in improving the economic status of the Roma to increase their employment, satisfactory results have not been achieved' (CCC, 2014, p. 7). Thus, governments' lower prioritisation of the employment reflects a failure to adequately address the deep-set inequalities faced by Roma people in a context where jobless rates are as high as 80%.
However, states' prioritisation of education does at least indicate official awareness of its foundational role in tackling discrimination. Notably, this takes place by breaking the inter-generational transfer of prejudice through citizenship education that emphasises tolerance and cultural respect. This is of key importance for, as Posavec and Hrvatic's (2000, p. 97) work in Croatia reveals, there is 'limited acceptance of the Roma culture (language, art, history, customs) [. . . and a strong] need for intercultural education for all pupils'. In the FYR, educational inequalities have been exacerbated by systemic or institutional practices that largely go unaddressed in the state discourse. Thus, as a broad literature attests, Roma children's lower educational attainment, poor attendance and early school leaving have been attributed to institutional prejudice. As De Beco, Hyll-Larsen and Balsera (2009, p. 10) conclude 'their education is of inferior quality and that the assessments on the basis of which they are transferred to "special schools" are in direct violation of the prohibition of non-discrimination' (see also , Jugović & Doolan, 2013, p. 364;Macura-Milovanović, Munda, & Peček, 2013;Sunčica & Mojca, 2013, p. 83). Such issues are given prominent attention in the NGO discourse. For example: 'an ongoing problem is the overrepresentation of Roma pupils in special education. . . where children are isolated from the rest of the school population and offered a substandard curriculum that does not prepare them for educational success or employment' (ERRA & NRCCM, 2014, p. 3).
The current analysis of the UPR discourse shows that there is broad state-NGO concordance in the level of attention to children's policy (second and third ranked, respectively). This is welcome, for existing studies provide an extensive litany of problems facing Roma children in the region. These include: failings in relation to adoption services (Ahern, & Rosenthal, 2008, p. 102), meeting the health and care needs of Roma children exposed to war (Barenbaum, Ruchkin, & Schwab-Stone, 2004), vulnerability to human trafficking (Poucki & Bryan, 2014), malnutrition (Machel, 2001) and poverty (UNICEF, 2012, p. 219). Yet despite these profound challenges the state discourse remains predominantly descriptive in nature. Often it outlines broad policy goals rather than specifying the means by which they will be achieved. For example, 'we will extend support for children on the streets, [. . .leading to the] improvement of the situation of Roma children' (Government of Bosnia Herzegovina, 2014, p. 12). In contrast, the NGO discourse is more critical, often offering an indictment of governments' failure to secure progress. For example, 'the lack of awareness of specific risks and their consequences, and needs of children from vulnerable groups is evident. . . It was especially evident in the case of Roma children' (CMCRS, 2014, p. 6).
As Stambolieva's (2015, p. 379) account underlines, variable welfare provision in the region is an inheritance from the past and a function of the transitional context presented by the former Yugoslav republics. It can also best be understood in relation to  democratisation and the challenges posed by internationalisation. These factors combine to create a context in which the Roma are often subject to inferior or subordinate modes of welfare citizenship. As Milcher's work (2006, p. 127) notes, 'besides socioeconomic determinants of welfare, the Roma identity significantly influences welfare [uptake] levels in Southeast Europe'. Health policy has long been an area of concern (Antic, 2005). It was second ranked in the NGO discourse, yet only sixth in the state discourse. Civil society organisations identify a range of related issues and challenges. For example: one of the greatest areas of concern is the exercise of the right to health for Roma people. The number of Roma medical practitioners is still very low, as is the number of primary healthcare services close to Roma settlements. (ROMA et al., 2014, p. 4) Housing is another aspect of welfare where 'members of the Roma population stand out as the most vulnerable population' (UNICEF, 2012, p. 71). Again, the state discourse is overly descriptive. Often it downplays or ignores implementation details. For example: 'the Law on Social Housing recognises the Roma as a particularly vulnerable social group with the right of priority to the resolution of the housing situation' (Government of Serbia, 2013, p. 16) and 'the Government Commission approved the concept of modernising Roma settlements' (Government of Slovenia, 2014, p. 6).
The current analysis also reveals that 'access to justice' has a higher priority in the civil society discourse (fifth ranked) compared the state discourse (seventh). Under this frame a prominent, reoccurring issue is justice for Roma women who have been victims of domestic violence. Here again, there are concerns about systemic inequalities and oppression of national minorities. For example, one NGO alluded to 'institutional discrimination of marginalized women who are victims of domestic violence. [In particular,] the rejection of women under the age of 18 to be admitted at shelters, most prominent [are] young Roma women in underage marriages' (AHC 2014, p. 6).
States' UPR discourse is notable for the dearth of attention given to intersectionality. In other words, the multiple and simultaneous identities of Roma people crosscut by age, sex, (dis)ability and so on (see, for example, Crenshaw, 2000). As Woodcock (2010) such essentialising of the Roma often leads to reductive and negative stereotyping. For example, one NGO alluded to how, 'Roma women continue to be exposed to multiple discrimination both from mainstream society and from members of their own communities, particularly in marginalized and non-urbanized communities' (IATRNE, 2014, p. 7). Another referred to how 'the position of multiply discriminated groups of women (Roma, elderly, rural women, women with disabilities, lesbians, single mothers, etc.) is particularly difficult. The mechanisms for independent monitoring of implementation of state policies are undeveloped' (AWC, 2014, p. 2).

Discussion
The foregoing analysis shows that states' implementation of human rights law in the Former Yugoslavia is centred on three main frames: 'rights', 'inclusion' and 'protection'. 6 In contrast, civil society organisations give principal attention to 'discrimination' and 'representation'. 7 This is significant because it not only reveals a disjuncture between NGOs and governments; it also underlines that FYR states are giving insufficient priority to tackling the discrimination and oppression faced by Roma in the region. The overall effect is 'institutional de-coupling'-or, the situation in public administration whereby political elites pronounce one thing but practice another (see Dahlström, 2004;Meyer & Rowan, 1991). In the words of one NGO report, 'lack of prevention of discrimination against Roma remains a systemic problem owing to ineffective human rights framework' (AIS, 2014, p. 6). The disjuncture between state and civil society discourse is evident across frames. For example, in the case of 'rights', the foregoing analysis shows that whilst state reports generally espouse 'soft' rights, civil society organisations give greater accent to 'hard' rights. The former are symbolic, rhetorical devices, the latter are enforceable legal instruments. The state-civil society 'disconnect' is also evident in references to welfare; with states failing to match civil society organisations' prioritisation of issues such as employment and health. Moreover, the analysis shows governments across the region giving insufficient attention to labour market discrimination, with concomitant negative effects on Roma well-being and economic status. All of this supports Kostadinova's observation (2011, p. 163) that, whilst anti-discrimination laws may 'address Roma recognition issues via provisions related to minority physical existence. . . [. . .They] lack redistribution provisions per se'. Thus, there is a need to strengthen human rights implementation in order to deliver equitable welfare for Roma people in the FYR.
On political citizenship the findings also resonate with Sardelić's analysis (2013, p. 5) that, 'even when Roma minorities are included in the body of citizenry, such inclusion is often marked by discrimination and inequality in accessing certain rights and participating in political communities'. The current study findings supports this. 'Representation' is joint first-ranked frame in the civil society discourse, yet fourth ranked in the state reports. This tells us much about the nature of civil society and Roma representation at the beginning of the twenty-first century. As Devic (2006, p. 257) notes, 'the tendency in transnationally exported definitions of the practices of a "good" civil society is to neglect the limits imposed by the local context'. Accordingly, the key human rights issues revealed in this study need to be seen in the context of an extended history of marginalisation and its impact on Roma associative life and mobilisation. Prior to 1991 civil society in the Socialist Federal Republic of Yugoslavia was suppressed by authoritarian rule (Križan, 1989;Trajkovski, 1999). In the wake of the Socialist Republic, a transitional or hybrid civil society has emerged. Whilst it exhibits features familiar elsewhere, it also has singular aspects. Notably, it lacks elements of the mobilising structures and criticality seen in more established liberal democracies (Gordon & Durst, 2004). Crucially, for Roma people, as the present findings suggest, there is a need for 'strengthening advocacy and campaigning skills [amongst CSOs]. . . and providing inputs into public policy, all of which will assist them to impact positively on nationally-led development policies' (Sterland, 2006, p. 2; see also Bosanac & Zdravković, 2012).
In particular, as Marin Thornton (2014, p. 117) explains, in order to secure effective representation and 'be less discriminated against, the Roma need to amass more power by acceding to key political and social positions'. Yet, as the locus for such action, civil society in the FYR presents formidable challenges. Not least, because, as a series of analyses underline, there's a need to ensure that the most vocal civil society organisations on Roma matters actually represent the communities they purport to speak for. For example, Trehan (2001, p. 134) underlines that, 'for a number of reasons, some of the most high profile [NGOs] whose work primarily focuses on Roma communities in the region lack grass-roots constituencies'. In a similar vein, Kóczé (2012, p. 19) refers to 'civil society development as a project of the elite' and refers to 'formal' and 'informal' sides of the Roma movement. With the former being 'more procedural driven. Particularly, concerning the funding and supportive framework. . . [whereas, the latter] is closer to the local community' (p. 20; see also Kóczé & Rövid, 2012). Gheorghe (2013, p. 28) puts this in stark terms, 'there is a dramatically widening gulf between the "clubs" of Roma political elites-both at national and transnational levels-and the communities they are supposed to represent'. It is in this context that the current analysis makes a timely contribution. It provides empirical data showing the nature and extent of the disconnect between political elites and Roma CSOs when the views of a broad cross section of small standalone NGOs and alliances and networks of civil society organisations are analysed. In short, it underlines Trehan's (2010, p. 53) observation that Roma representation needs to be seen in the context of: 'subaltern counterpublics' stemming from 'the multiplicities of public spheres [. . . for] civil society is not a monolithic entity, but represents a site of contestation of multiple, criss-crossing communities and publics'.
The present analysis of state-civil society perspectives on human rights implementation can also be usefully related to theory on collaborative policy implementation. In conceptual terms, instead of the synergies predicted by complementarity theory (Klijn & Skelcher, 2008), as noted, the disjuncture in state and civil society discourse associated with the second phase UPR shows that governments are not affording NGOs sufficient opportunities to input their views on rights implementation. This is consonant with Bachrach and Baratz's (1962) classic work on how political elites may exclude, ignore or downplay exogenous policy claims. Thus, the present findings provide a further example of how states can undermine NGO advocacy for minorities (see also Schnellbach, 2012, p. 510). In the words of Sardelić (2015, p. 159), this may effectively render Roma people as 'semi-citizens'.
In social theory terms, the study findings also engage with the distinction between legitimacy, legitimation and performativity. In the former case, as Rawls (2003, p. 185) notes, 'political legitimacy aims for a public basis of justification and appeals to free public reason, and hence to all citizens viewed as reasonable and rational'. It is the approach sought by the United Nations whose normative vision of 'legitimate' human rights implementation is one shaped by civil society policy claims. In contrast, 'legitimation involves communicative actions aimed at managing the public's perception that government actions are effective in promoting their desired ends, whether that is in fact true' (Moore, 2001, p. 707). It is a scenario is also captured by the term 'performativity'. In other words, the situation when government practices seek to give the impression of responsiveness to external policy demands (Price & Shildrick, 1999, p. 241). As the present analysis shows, contemporary practice in the FYR tends to reflect the latter approach.
The current study suggests an agenda for future research. In particular, it will be useful to explore organisational practices, contestation and the internal power dynamics within both state and civil society organisations, as managers and leaders seek to set priorities for implementing human rights in the FYR. Furthermore, future research needs to examine the interplay of intersectionality, institutionalism and human rights for Roma people. In particular, the way that both government and NGOs mediate claims-making in ways that are simultaneously concerned with ethnicity and a combination of other 'protected characteristics' (such as sex, sexual orientation, faith (and non-belief), age and (dis-)ability).
In summary, this study highlights a number of key issues and challenges in human rights implementation for Roma people in the Former Yugoslavia. Whilst there is a need for strengthening a range of civil society capacities, on the part of government, there is an urgent need for systemic reform in order to better promote engagement and greater responsiveness to claims-making by civil society organisations acting for Roma communities. Such action is imperative in order to address differentiated citizenship in the region: an enduring pathology stemming from the deep-set, continuing patterns and processes of discrimination facing Roma people in the former Yugoslav space.

Acknowledgements
The author gratefully acknowledges the helpful and constructive feedback of three anonymous referees when revising an earlier draft of this paper.