Diffusing the Abolitionist Norm in Japan: EU ‘ Death Penalty Diplomacy ’ and the Gap between Rhetoric and Reality in EU – Japan Relations

This article uses Börzel and Risse ’ s norm diffusion framework to conceptualize and evaluate the EU ’ s ‘ death penalty diplomacy ’ in Japan. Despite the ‘ exceptional ’ nature of Japanese politics with regard to the death penalty the EU has enjoyed numerous successes in its attempts to diffuse the abolitionist norm within Japan. These successes have occurred through both direct and indirect methods of norm diffusion, and through socialization, persuasion and functional emulation. Despite the dramatic increase in executions in Japan in 2018, in practice the EU and Japan have established an understanding that executions will in future stabilize at a low and symbolic level. The EU has also co-funded research that successfully challenged the Japanese government ’ s public opinion polling-based justi ﬁ cation for retaining the death penalty. This research also played an important role in socializing and persuading the Japanese Federation of Bar Associations to change its of ﬁ cial position and to support the abolition of the death penalty.


Introduction
On 17 July 2018, to great fanfare, the EU and Japan signed an economic partnership agreement (EPA) and a strategic partnership agreement (SPA). The first clause in the preamble of the SPA states that both parties reaffirm their 'commitment to the common values and principles, in particular democracy, the rule of law, human rights and fundamental freedoms, which constitute the basis for their deep and long-lasting cooperation as strategic partners' (Ministry of Foreign Affairs, 2019). However, this article questions the extent to which the 'values rhetoric' surrounding EU-Japan relations expressed here matches the reality. In particular, we address two clear, related problems in this discourse of shared values. The first is that there is no binding 'essential elements' human rights clause which links the EU-Japan SPA horizontally to the EPA, and provides for the possibility of the suspension or termination of the latter. The second is Japan's retention of the death penalty. There were 15 executions in Japan in 2018, 13 of which took place in July, the same month in which the EPA and SPA were signed. Taken at face value, this shows clear disregard for the persistent EU efforts to move Japan in an abolitionist direction. Johnson and Zimring (2009, p. 94) argue that it is now justifiable to claim that 'Japan's death penalty politics are exceptional' because Japan retains the death penalty, despite having experienced four 'precipitating circumstances' that should lead to abolition. These are: a transition from authoritarianism to democracy, the coming to power of a centre-left government; a high profile miscarriage of justice and an international commitment that requires a formal shift in policy (Johnson and Zimring, 2009, pp. 81, 90). Our core argument is that, despite the exceptional nature of Japanese death penalty politics and the unlikelihood of abolition in the near future, the EU has nevertheless enjoyed numerous successes in its attempts to diffuse the abolitionist norm to key actors within Japan. These successes have occurred through both direct and indirect methods of norm diffusion; namely, socialization, persuasion and functional emulation (Börzel and Risse, 2012).
The article is structured as follows. In the first section we argue that the EU is recognized in the expert literature as having played a substantial role in the creation and promotion of the global abolitionist norm, and also that the EU is recognized as the most important promoter of the abolitionist norm in Japan. The second section looks at our first case study; the connection between the EU-Japan SPA negotiations and the 13 Aum Shinrikyo executions in July 2018. We argue that although the number of executions seems high, in fact the EU has successfully communicated to the Japanese government that in the medium term it expects executions to stabilize at a low level. The third section looks at our second case study, where the EU funded Sato and Bacon (2015) to carry out opinion poll research, the findings of which successfully challenged the Japanese government's public communication strategy regarding the death penalty. The fourth section demonstrates that both direct and indirect mechanisms of norm diffusion were in operation, including functional emulation in the first case study and socialization and persuasion in the second. Based on these case study findings and analyses, we also propose and explain a new mechanism of norm diffusion, that of 'mediated local diffusion'. A brief conclusion summarizes the findings of the article.

Focusing on What the EU Does, Not What it Is
The EU, the Abolitionist Norm and Japan In his seminal article on normative power Europe (NPE), Manners demonstrated that there have been a number of cases where the EU has played an 'important, if not crucial' role in bringing about abolition in acceding and neighbourhood states (Manners, 2002, p. 248). Manners also suggests that his study of what he refers to as the EU's 'most visible' global human rights policy demonstrates the EU's normative power, because the EU has helped to 'accelerate' the abolitionist movement (Manners, 2002, p. 251-2). Johnson and Zimring have also noted that Europe has been the epicentre for reform of capital punishment, and that 'Europeans' success on their home turf has left death penalty activists with the energy and resources for other geographic zones', resulting in 'aggressive advocacy' for abolition throughout the world (pp. 8, 333, 96). Hood and Hoyle (2015) explain in their landmark study of the death penalty how 'mounting political pressure' from the Council of Europe and the EU was a key element of 'what generated the new wave of abolition' from 1989 onwards (p. 29). It was important to Hood and Hoyle that there was a formal requirement of abolition for membership of these two bodies, and also that both offered a principled opposition to the death penalty as a violation of fundamental human rights. Hood and Hoyle also recognize that the EU carried out an important 'diplomatic offensive' on the death penalty on a number of different fronts: the adoption of the EU guidelines on the death penalty; recognition of the charter of fundamental rights; regular EU human rights dialogues and seminars with a range of third countries; spending more than €23 million on death penalty reform and abolition research projects since 2000, and the periodic sponsoring by the European Parliament of UN General Assembly resolutions calling for a universal moratorium on the death penalty (2015, pp. 29-32). Finally, Maiko Tagusari, a prominent Japanese capital defence lawyer and human rights activist, has argued that the EU has been really important to the creation and promotion of the abolitionist norm. Although it was the Council of Europe who first tried to reach out to Japan, they do not have enough power or budget, so they could not effectively follow up on their own activities. But the EU is an actual power, and has also been the most important actor in promoting the abolitionist norm in Japan. The EU is special, different (Interview 1 1 ).

Case Studies on Human Rights Norm Diffusion in Third Countries
Somewhat curiously, there has been very little work on EU human rights norm diffusion in third countries beyond the European neighbourhood. To understand why this is so, it is necessary to return to Manners' article on NPE. For Manners, not only is the EU constructed on a normative basis, but this 'predisposes it to act in a normative way in world politics … the most important factor shaping the role of the EU is not what it does or what it says, but what it is' (p. 252). The EU possesses the 'ability to define what passes for "normal" in world politics', which is 'ultimately, the greatest power of all ' (pp. 236, 253). There are four stages in Manners' analysis. He first explains the EU's normative difference (its historical context, hybrid polity and political-legal constitution), and its normative basis (the five core norms of peace, liberty, democracy, the rule of law and human rights). He then offers an explanation of how norms can be diffused, identifying six methods of norm diffusion (contagion, informational, procedural, overt and transference diffusion, and cultural filter). Finally, Manners uses his norm diffusion framework to analyse four clusters of death penalty cases in acceding and neighbourhood states, and identifies which methods of diffusion were deployed in each of these clusters (Manners, 2002, pp. 244-52).
Identifying the EU's normative difference and normative basis in this way was groundbreaking. But in this article we focus on developing the third and fourth elements in Manners' analysis by using a norm diffusion framework to analyse a concrete country case study in which the EU has attempted to diffuse the abolitionist norm. Following the publication of Manners' seminal article, most subsequent analyses of NPE that have framed the mainstream debate have focused on developing the first element of Manners' analysis, looking at normative difference, and the issue of what type of power the EU is, or should, be. In this article, however, we park the issue of EU actor identity. This is partly because the specific analysis of norm diffusion is often marginalized in this highly politicized meta-debate and partly because the concepts of normative, civilian, postmodern and ethical power are not very clearly defined. Pace (2007), for example, has suggested that the concept of NPE is a 'semantically empty notion'. Lastly, there is considerable overlap between these various concepts and each approach asserts that the EU does or should seek to diffuse human rights norms. If our focus is on EU norm diffusion, it is therefore not strictly necessary to take a position on what type of actor the EU is.
According to Forsberg (2011Forsberg ( , p. 1191, in the debate over NPE at least five kinds of criteria for identifying a normative power have been used: having a normative identity, having normative interests, behaving in a normative way, using normative means of power and being able to achieve normative ends. We focus on the fourth and fifth of these criteria, with an emphasis on whether the EU is able to achieve normative ends and assessing the extent to which EU contributes concretely to the diffusion of human rights norms. We suggest that whether the EU has the capacity to define what passes for normal is not the main question. The main question is whether, how and to what extent the EU is able to diffuse the human rights norms that constitute the basis of NPE. Only then can we build a meaningful picture of whether the EU actually makes a normative difference, rather than simply being normatively different. We should do as Manners does in his 2002 article, where he actually discusses case studies of norm diffusion, but not as he says when he argues that we should focus on what type of actor the EU is. Manners' own norm diffusion criteria, and his demonstration of how such criteria can be applied to case studies, have not been picked up on and developed in mainstream EU studies. This is partly because subsequent scholars have followed Manners' headline advice to focus on what the EU is, as a result of which meta-arguments about the nature of EU identity have dominated mainstream scholarship on NPE. But it is also partly because Manners own criteria were 'briefly outlined', 'unclear' and 'unspecified', and not convincingly developed and systematically applied throughout the course of his article (Forsberg, 2011(Forsberg, , p. 1196. We seek to address these shortcomings by adapting and applying a different set of criteria developed by Börzel and Risse (2012), which have been widely used in the Europeanization literature. Some of the criteria Börzel and Risse use overlap with and build on those used in the 'spiral model' approach (Risse and Ropp, 2013), which itself builds on pioneering work on the 'boomerang effect' by Keck and Sikkink (1998). Börzel and Risse have historically confined the scope of application of their diffusion framework to EU member states, acceding states, states in the EU neighbourhood and other regional organizations (Börzel and Risse, 2012, p. 6). However, we seek to demonstrate how their approach can be expanded to analyse EU relations with third countries beyond the European neighbourhood.
There have been some previous attempts to look at the death penalty and international norm diffusion in Japan (see for example Bae, 2011;Obara, 2011), but these have not been cumulative and the authors have also often come from different research fields such as comparative politics and legal studies. More recently, Weyembergh and Wieczorek (2020) identify problems with the implementation of the death penalty provisions of the EU-Japan mutual legal assistance agreement. Andreescu and Hughes (2020) argue that Japanese public support for the death penalty is not as high as the Japanese government claims, but do not consider the role of external abolitionist actors. There are abolitionist studies published in Japanese which adopt a philosophical or criminal justice perspective and a smaller specialist literature on public opinion and the death penalty. However, these studies have a heavily domestic focus, and there are no significant studies in Japanese that address the role of the EU in diffusing the abolitionist norm. At present there is also very little systematic case study research in English on EU human rights norm diffusion to third countries beyond the neighbourhood, and almost no work of this Paul Bacon and Hidetoshi Nakamura kind on the Asia-Pacific. There is potentially a major and profoundly under-investigated research programme here. Lastly, it is surprising that there has been no significant attempt to challenge, support or extend Manners' evaluation of the diffusion of the abolitionist norm in his discipline-shaping 2002 article, which is one of our objectives in this article. It is time to recognize and address these various oversights and pursue this particular normative turn not taken.

Conceptualizing EU Human Rights Norm Diffusion in Third Countries
Börzel and Risse define norms as 'expectations of appropriate behaviour based on a given collective identity', and diffusion as 'a process through which ideas, normative standards, or … policies and institutions travel across time and space' (Börzel and Risse, 2012, pp. 4, 5). Following their example, the dependent variable we analyse here is 'domestic or regional institutional change which can be traced to the EU's direct or indirect influence' (p. 14). In its attempt to diffuse human rights norms the EU often relies on external incentives (conditionality), on the one hand, and technical and financial assistance (capacitybuilding), on the other. Conditionality 'tries to manipulate the cost-benefit calculations of target actors through creating positive and negative incentives', while capacitybuilding, 'by contrast, provides target actors with additional resources enabling them to make choices to begin with' (Börzel and Risse, 2012, p. 7).
While Börzel and Risse identify conditionality and capacity-building as one mechanism, we follow Risse and Ropp, (2013, pp. 14-16) in dividing them into two separate mechanisms. We argue that the funding for the Sato and Bacon research (discussed below) can be considered as a form of successful capacity-building that provided a platform for local norm entrepreneurs and policy advocates to engage in the direct socialization and persuasion of domestic target actors in a manner that is congruent with EU policy preferences. When lesson-drawing is in play, one should expect a selective adoption of institutional solutions, as they need to be tailored to the problems at hand. Lesson-drawing is based on instrumental rationality, since it follows a functional logic (Börzel and Risse,p.10).
Finally, Börzel and Risse (2012, p.5) identify three types of rationalities or logics which underlie norm diffusion: instrumental rationality, or the logic of consequences; normative rationality, or the logic of appropriateness and communicative rationality, or the logic of arguing. Normative rationality is a 'thin' logic of appropriateness, where actors can be socialized into following rules, and recognize that the dominant script is regarded as the 'right thing to do'. This rationality or logic can be distinguished from communicative rationality, where actors can be persuaded at a thicker or deeper level than is the case with normative rationality; they internalize a particular norm as the right thing to do not simply because they are capable of identifying with a dominant script which recognizes this to be the case, as with normative rationality, but because they now themselves believe that the norm itself is substantively true or correct (Börzel and Risse, 2012, pp.6-8). Connecting this discussion of rationalities to mechanisms of diffusion, instrumental rationality corresponds to the possibilities of conditionality and functional emulation, normative rationality corresponds to the possibility of socialization and communicative rationality corresponds to the possibility of persuasion. The various mechanisms of diffusion are set out and divided into direct and indirect mechanisms in Table 1.
How do we know when attempts at human rights norm diffusion have been successful? Risse and Ropp identify several benchmarks. These include the following: 1. Non-state actors accept a particular human rights standard as binding for themselves and the non-state actor in question offers some sort of statement that it intends to accept at least voluntary codes of conduct as obligatory, in relation to the particular human rights norm under consideration. 2. Behavioural change of state actors. 3. State ratification of relevant international human rights conventions and protocols. 4. Transposition of these international commitments into domestic law. 5. Sustained compliance with international human rights standards in the case specified (Risse and Ropp, 2013, pp. 6-10).
We argue below that the research by Sato and Bacon (2015) played a significant role in the socialization and persuasion of the Japanese Federation of Bar Associations (JFBA) to change its position, thereby meeting the first criterion identified above, demonstrating that the EU has successfully engaged in norm diffusion with regard to the abolition of the death penalty in Japan.

The SPA and the Aum Shinrikyo Executions
Conditionality is ultimately not meaningfully in play with regard to the EU's 'death penalty diplomacy' with Japan. This can be seen in the absence of a horizontal clause in the EU-Japan SPA linking it to the EPA. This was a surprise to informed analysts of the relationship. The SPA negotiations between the EU and Japan were initially based on the EU-Canada template, where there is a horizontal clause linking the SPA and EPA documents, with regard to two 'essential elements' of the agreement, which are respect for human rights and nuclear non-proliferation. The essential elements clauses are present in the EU-Japan SPA but there is no clause connecting them to the EPA, and providing for the suspension or termination of that agreement. Such connecting clauses do exist in agreements the EU has concluded or is close to concluding with other countries in the Indo-Pacific region, such as Canada, Singapore, South Korea and Vietnam. This absence can therefore be regarded as a diplomatic victory for Japan. Japan was initially reluctant to  Börzel and Risse, 2012, p. 14 Mechanisms that feature in our Japan case study are in bold. Mediated local diffusion is a mechanism we identified through our analysis of this case.
negotiate an SPA at all, despite the constant rhetorical appeal by both parties to shared values. Ultimately, a compromise was reached in the SPA that reflected these concerns. The essential elements clauses are present in the political agreement and are legally binding, providing for the suspension or termination of the political agreement. However, they are not directly linked to the economic agreement. According to a senior Japanese official at the Ministry of Foreign Affairs the horizontal clause was removed during rounds 9-13 of the SPA negotiations (Interview 2). Japanese negotiators were concerned, not just by the political symbolism and the implication that the treaty was unequal, but also by the possibility of being subject to legal proceedings if they failed to respect the essential elements clause as it related to human rights, and in particular, the death penalty. Could the 'mere' fact of having the death penalty itself constitute an abuse of the essential elements clause? Might a significant increase in the number of executions create a risk that the agreement could be suspended? And what might constitute a significant increase? The EU's official preference, as expressed in the official guidelines on the death penalty is for an official moratorium leading to abolition. However, according to a senior member of the European External Action Service with detailed knowledge of the SPA negotiations, the EU indicated that it was prepared to tolerate the continuation of executions in Japan, as long as they remained at a low and symbolic level, as they had between 2014 and 2017. The EU communicated informally during the SPA negotiations that continued use of the death penalty would be tolerable if executions did not exceed five per year (Interview 3).

Contextualizing and Explaining the Aum Shinrikyo Executions
The EU-Japan SPA and EPA were signed on 17 July 2018. This date is sandwiched between two sets of executions of members of the Aum Shinrikyo (hereafter Aum) cult, which took place on 6 July (seven executions) and 26 July (six executions), respectively. On the face of it, Japan tested and dramatically exceeded the informal limits to executions that were communicated to it during the political negotiations, in the very weeks surrounding the signing of the EPA and SPA.
But how should the Aum executions be interpreted? Do they herald a return to consistently higher annual numbers of executions? We suggest that this is unlikely, for a number of reasons. First, the sentences for all Aum members facing charges had only recently been finalized. There is a strong norm in Japan that capital punishments are not administered while death row prisoners may possess valuable information or may be required to testify at the trials of other defendants. Also, there is a further norm that once all convictions have been secured, those who are convicted of the same crime are executed on the same day. To put it bluntly, from the government's perspective the Aum executions were addressing a backlog of prisoners who had been waiting on death row for some time, until the final Aum convictions could be achieved.
Initially, the intention was for all 13 Aum members to be executed together. However, Ministry of Justice (MOJ) officials ultimately decided to split them into two groups, and interestingly, their decision anticipated 'possible international outcry' over capital punishment (Asahi Shimbun, 2018). The first seven to be executed were the most senior and were chosen based on the number of their convictions, their degree of involvement in the crimes, and especially their rank in the cult. They included Chizuo Matsumoto, the founder of Aum Shinrikyo. These were the largest number of executions on a single day for the preceding two decades, nearly doubling the previous record of four. The Ministry of Justice had been bracing for 'fierce criticism on the 'mass execution'', but there were no widespread protests, and the ministry therefore felt that there was no reason to postpone the executions of the second group of prisoners' (Asahi . The Japanese government's sensitivity to potential criticism demonstrates the strength of the global abolitionist norm. There were also a number of other reasons why timing was a factor with these executions. The government wanted to bring closure on the Aum cases before the ascension of the new emperor in 2019; the transition period between emperors is meant to be a period of peace and tranquillity (with no executions). A further timing issue is the fact that the Aum killings are indelibly linked in the Japanese psyche to the Heisei period (1989-2019) and there was a widespread sentiment in Japan that it was best to conclude 'Heisei business' within the Heisei period (see, for example, Asahi Mainichi Shimbun, 2018b).
The Aum members who were executed were also securitized and othered as 'cult' members guilty of 'terrorist' acts, both by the government, and right across the mainstream media spectrum, including the more liberal newspapers (see, for example, Asahi Shimbun (2015) and Mainichi Shimbun (2018a). The 13 Aum executions are framed in Japan as a set of exceptional cases and should not therefore be considered as 'routine' executions. If we accept the governmental and societal framing that the Aum cases were exceptional, then Japan can be interpreted as not having exceeded the informal limits which were communicated to it by the EU. From an abolitionist perspective it is of course deeply regrettable that so many executions happened in such a short space of time. But there is a strong likelihood that Japan will now return to the pattern established in the 4 years preceding 2018; symbolic executions of three or four people each year (which would continue to meet the requirements informally communicated by the EU). At the time of writing (February 2021) and justifying this interpretation of the dramatic spike in executions in Japan in 2018, there were 'only' three executions in 2019 and none in 2020.

Challenging the Government Narrative on the Death Penalty -The Sato and Bacon Opinion Poll
The global abolitionist norm, which the EU is recognized as having played a significant role in creating and promoting, has had an impact on debates within Japan. For example, in late 2016 the JFBA issued the landmark Fukui declaration indicating that they would for the first time commit to the abolition of the death penalty. This declaration made several references to the development of a strong abolitionist norm around the world (JFBA, 2016).
EU strategy has also evolved in Japan; in 2014-15 the EU, along with three European foreign ministries, funded a project that engaged concretely with local actors and local debates (Sato and Bacon, 2015). This was achieved by funding an alternative opinion poll challenging the validity of the 5-yearly poll carried out by the Japanese government, which purported to identify a consistently high level of public support for the death penalty. The findings from this EU-funded capacity-building project received media attention in Japan (Japan Times, 2015) and had an impact on debate; two major official declarations by the JFBA, including the Fukui declaration, also refer approvingly to the findings of alternative opinion poll research as a key justification for abolition (JFBA, 2015(JFBA, , p. 4, 2016. One important aspect of the EU's human rights strategy in Japan has therefore been to focus on the issue of public opinion. A Cabinet Office survey conducted in late 2009 showed that a record 85.6 per cent of Japanese favoured maintaining the death penalty. Both the United Nations Human Rights Council and the United Nations Human Rights Committee have recommended that Japan abolish the death penalty regardless of public opinion. However, the Japanese government instead has chosen to maintain the punishment, citing and foregrounding the justification of public support. This position is stated clearly in Paragraph 104 of Japan's Sixth Periodic Report to the UN Human Rights Committee, where the death penalty is identified as 'a critical issue constituting the backbone of Japan's criminal justice system, and therefore needs to be carefully examined in all respects … with the fullest attention given to the people's opinion (Government of Japan, 2012, p. 20). The tone of the Japanese government's argument is dramatic, asserting that the death penalty constitutes the backbone of Japan's criminal justice system suggests that this system might collapse if the death penalty were to be abolished. Sato and Bacon (2015) tested the validity of the Japanese government's arguments on its own terms by questioning whether Japanese public opinion really does support the death penalty to the extent that the Japanese government claims. Are the Japanese public deeply persuaded of the need for the death penalty, or are they 'merely' socialized into it, and if the latter is the case, is there scope for their re-socialization towards an acceptance of abolition? If the degree of support that is claimed for the death penalty can be challenged, the Japanese government's communications strategy regarding the death penalty would be undermined. The Japanese Cabinet Office conducts a public opinion poll on the death penalty roughly every 5 years. The comparative reference point for the Sato  With funding from the EU and other European foreign ministries, 2 Sato and Bacon conducted a survey on a similar scale to that conducted by the Japanese government. The strategy that these researchers developed, in consultation with the EU, was as follows: if it could be shown that support for the death penalty was significantly lower than that identified by the Japanese government, in an alternative survey conducted on a similar scale, then this would be a highly significant and policy-relevant finding. The 2014 survey commissioned by the Japanese government shows that 80 per cent of respondents consider the death penalty to be 'unavoidable'. The EU rationale for funding the Sato and Bacon research project was that if the evidence produced in the Cabinet Office survey could be challenged, the retentionist stance and justificatory strategy of the Japanese government could then be brought into serious question. And indeed, it turns out that the 80 per cent retentionist figure masks a significantly more complex reality, and that many Japanese voters are not, in fact, strongly committed to the death penalty (Sato and Bacon, 2015, pp. 24-8). This EU strategy ultimately proved to be successful; the findings from the project had an important impact on the JFBA's decision to change their official position on the death penalty, as we demonstrate below.

A Closer Look at the Key Sato and Bacon Project Findings
The Sato and Bacon survey initially asked the same question about abolition and retention of the death penalty as the government survey had done. The question was 'Which of the following opinions concerning the death penalty do you agree with'?
The three response options were: • The death penalty should be abolished.
• The death penalty is unavoidable.
• Do not know/difficult to say. (Sato and Bacon, 2015, p. 24) In the Sato and Bacon study, support for the death penalty was slightly higher, at 83 per cent retentionist, compared with the government survey figure of 80 per cent (Sato and Bacon, 2015, p. 24). After this first stage, the results of the two surveys were comparable and within the margin of error for a sample of this size. Because of this success, it was therefore legitimate to isolate, and to look in more detail at the views of these 83 per cent retentionist respondents in the Sato and Bacon survey, and probe for more information about their commitment to the death penalty, with more specific questions. The four key findings from the research can be summarized as follows: First, the Sato and Bacon survey also offered an additional general question on the death penalty, this time with five response options, offering additional degrees of conviction in support of or opposition to the death penalty. The question was: People have various opinions about the death penalty. Do you think that it should be kept as a form of criminal penalty or do you think it should be abolished?
The five response options were: • The death penalty should definitely be kept.
• The death penalty should probably be kept.
• Cannot say.
• The death penalty should probably be abolished.
• The death penalty should definitely be abolished. (Sato and Bacon, 2015, p.2 5) Finding 1. From responses to this question, it emerges that in fact only 27 per cent of respondents could be considered to be committed retentionists (Sato and Bacon, 2015, pp. 24-5). This is where the language used in the questions takes on additional significance. 'The death penalty is unavoidable' is a strong formulation, which could be taken to suggest that more than 80 per cent believe that the death penalty should definitely be kept. However, even when one makes the simple move of introducing a five-point scale, only 27 per cent of respondents argue that the death penalty should definitely be kept. By a trick of phrasing, the government survey invites us to imagine that more than 80 per cent of the public are committed retentionists, when the real number of people who are genuinely committed to the death penalty is roughly a third of that figure. On closer inspection, therefore, the depth of public commitment to the death penalty is not strong. As Sato and Bacon note, 'the current government-survey question does not adequately capture the degree of support for the death penalty. Behind the supposed majority support lies a minority of respondents who are really committed to keeping the death penalty. This calls into question what the government is actually measuring, and on what basis the execution of prisoners is being justified' (Sato and Bacon, 2015, p. 25). Using the language of the norm diffusion paradigm, we suggest that only 27 per cent of retentionist respondents are persuaded, but the other 56 per cent who make up the 83 per cent in our survey are 'merely' socialized into accepting the death penalty. Finding 2. Of the 83 per cent of respondents who initially answered that the death penalty was unavoidable in the Sato and Bacon survey, 71 per cent said that they would accept abolition as government policy, if the Japanese government were to abolish the death penalty (Sato and Bacon, 2015, p.27). In other words, even those who initially identify themselves as abolitionist are not so seriously committed to the death penalty. Again using the language of the norm diffusion paradigm, we can say that there is considerable scope for the re-socialization of this 71 per cent. Although they profess to support the death penalty, only a smaller number are genuinely persuaded, and two-thirds could comfortably be socialized to accept a new reality where the death penalty was abolished. Finding 3. Further, 72 per cent of respondents argued that abolition of the death penalty would affect their daily lives 'not at all' (31 per cent), or that they 'do not know' (41 per cent) how abolition would impact on them (Sato and Bacon, 2015, p. 26). This hardly suggests that the public is firmly committed to retention and deeply concerned about the consequences of abolition. Finding 4. When asked who should decide the future of the death penalty, fewer than half (40 per cent) of the respondents thought the decision should be based on the results of public opinion surveys conducted by the government. (Sato and Bacon, 2015, pp. 25). To put it another way, the Japanese general public does not share the Japanese government's view that public opinion should be the most important determinant of Japanese government policy on the death penalty.
The Sato and Bacon survey showed that, when one digs deeper and goes beyond the headline 80 per cent plus figure, in fact just over one-quarter of respondents were committed retentionists. The parallel survey also highlighted the fact that more than two-thirds of retentionists would accept abolition if the government decided to abolish the death penalty. Were the government to change its stance on the death penalty, Sato and Bacon have provided reliable evidence to suggest that Japan's citizens would follow suit. The majority of the public are in favour of the death penalty if asked in general, but how strongly or how unconditionally they want to retain it is a different matter (Sato and Bacon, 2015, p. 40). Correspondingly, there is no basis to the Japanese government's somewhat dramatic claim in the State report that the death penalty is a 'critical issue constituting the backbone' of Japan's criminal justice system. It is therefore questionable to base the justification for the maintenance of the death penalty on public opinion, because the depth and intensity of support for the death penalty is not as strong as is claimed. The Japanese public is socialized but not persuaded to support the death penalty. Furthermore, there is evidence that they can be re-socialized to accept abolition given leadership by the Japanese government, if it is prepared to move forward.

Direct, Indirect, and Locally-mediated Diffusion
In this section we draw together the various findings from the two cases, within the adapted Börzel and Risse framework. This framework makes a distinction between two types of diffusion mechanisms: First, ideas, policies, and institutions might diffuse through direct influence mechanisms. An agent of diffusion actively promotes certain policies or institutional models in her interactions with a receiving actor or group of actors. Second, diffusion also occurs through indirect mechanisms, namely emulation. Here, the action starts at the receiving end. Actors need to solve a problem or to overcome a crisis and look around for 'best practices' and institutional solutions that serve their needs (logic of consequences). (Börzel and Risse, 2012, p. 5) With regard to the Aum case, we need to analyse two actions by the MOJ: • The specific decision to have two separate sets of executions for the Aum death row prisoners in July 2018 (a logic of consequences). • The general decision to keep executions at a low and symbolic level since then (also a logic of consequences).
In the first of these cases, the EU is part of a coalition of norm senders, as one of the principal promoters of the general abolitionist norm. In the second, the EU is a polity with whom Japan recently signed a binding political agreement and agreed informally to low levels of executions. In both cases the Japanese MOJ engaged in functional emulation: it had to 'solve the problem' of being significantly out of step with the global abolitionist norm and was mindful of the limits to which it had informally agreed. The MOJ responded reluctantly and strategically, using an instrumental rationality, or logic of consequences, to make some concessions to EU pressure and to the powerful international logic of appropriateness surrounding executions. In the specific case of the Aum executions the Japanese government avoided a crisis by splitting the executions. In the more general case, the Japanese state has reverted to a low level of executions. The MOJ was of course not persuaded or socialized in either case, but recognized that it was prudent to moderate its behaviour in the face of influential abolitionist actors, such as the EU. As such, the naming and shaming that can go along with persuasion and socialization (Risse and Ropp, 2013, p.14) were not in play.
With regard to the second case, the Sato and Bacon research and its impact on the JFBA, there are five main interactions, each of which constitute successful examples of diffusion: • the direct influence of the EU on the JFBA via the global abolitionist norm through socialization (a logic of appropriateness) • the direct influence of the EU on Sato and Bacon, through capacity-building • the direct influence of the Sato and Bacon research on the JFBA Death Penalty Abolition Committee, through capacity-building • the direct influence of the JFBA Death Penalty Abolition Committee on the JFBA Executive Board through socialization (a logic of appropriateness) • The direct influence of the JFBA Death Penalty Abolition Committee on the general JFBA membership through a combination of socialization and persuasion (logics of appropriateness and arguing). and file. Tagusari claims that it is most accurate to characterize the JFBA Executive Board as having been socialized: To have the Fukui declaration adopted, the draft declaration had to be approved by the JFBA board. This consists of the representatives of all of the local bar associations, and among them there were aggressive supporters of retention of the death penalty, or people who were working for victim's rights. So the most challenging part of the process of the adoption of the Fukui Declaration was convincing the board members of the JFBA. To get to the point where they agreed to have a vote was a major achievement (Interview 1).
Although some members of the board were not personally persuaded of the case for abolition, they came to believe that allowing a vote was the right thing to do. With regard to persuading the rank and file membership of the JFBA the Sato and Bacon research boosted the educational capacity of the JFBA to make the case for abolition. For this cohort, according to Tagusari, there was a mixture of socialization and persuasion: Before the Fukui declaration, when we tried to persuade local bar associations, I used the research myself. The research was also of course used by other JFBA representatives we dispatched, who were trying to persuade normal JFBA members to support the Fukui declaration. Actually, the research was useful to persuade retentionists or people who did not know much about the system or its administration, or did not have any specific idea about whether to abolish the death penalty or not. The research helped the members of the JFBA Death Penalty Abolition Committee to persuade the ordinary members of the JFBA to agree. This is because the research showed that the Japanese people are not as committed to the death penalty, and that helped us to make the argument that the Fukui declaration should be supported (Interview 1).
It is clear that the EU-funded Sato and Bacon research was an important capacity-building effort that increased the socialization and persuasion capabilities of the JFBA Death Penalty Abolition Committee. Risse and Ropp (2013, p. 16) characterize capacity-building as 'creating the preconditions so that logics of appropriateness can apply' and that is exactly what happened in this case. What this also therefore suggests is that, building on Börzel and Risse's framework, we need to specify a third category of diffusion: one where the EU successfully builds the capacity of local non-state actors (through financial assistance or knowledge transfer, among others) enabling them to act as direct diffusers of the human rights norm in question, in the local domestic context of the target state. In this case, the local academic norm entrepreneurs were Sato and Bacon and the JFBA. Sato and Bacon were epistemic entrepreneurs and the local policy advocates were the members of the Death Penalty Abolition Committee of the JFBA. From an EU perspective, the diffusion has become indirect, because this is now being carried out by local intermediaries. These local non-state actors are themselves engaging in direct diffusion, but the EU is now working through intermediaries and proxies, local norm entrepreneurs or norm carriers in the target state. We tentatively refer to this process as 'mediated local diffusion'.

Conclusion
Although Japan's death penalty politics are exceptional, the EU has enjoyed some success in diffusing the abolitionist norm in Japan. We argued that the EU has played a substantial role in the creation and promotion of the global abolitionist norm and is recognized as its most important promoter in Japan. Furthermore, we showed how Japanese MOJ officials have engaged in functional emulation of the EU, both in their timing of the executions of Aum Shinrikyo members in July 2018 and in their general commitment to keep execution levels at a low and symbolic level since 2018, as informally agreed with the EU in negotiations over the SPA. Through our case study research we identified a mechanism that we have labelled 'mediated local diffusion'. This occurs when local norm entrepreneurs and policy advocates attempt to diffuse human rights norms on behalf of the EU in the domestic environment of the target state, having been funded, educated or otherwise empowered by the EU. We showed how the EU engaged in capacity-building in Japan by funding research by Sato and Bacon. This in turn gave abolitionist policy advocates on the Death Penalty Abolition Committee of the JFBA greater argumentative and educational capacity with which to persuade and socialize other members of the JFBA to sign up to the 2016 Fukui declaration.
When considering whether the EU is a normative power, we believe that there should be a greater focus on concrete case study work on EU human rights norm diffusion in third countries, focusing on what the EU does rather than what type of actor it is. There is a vast universe of unexplored EU human rights norm diffusion case studies in the wider world beyond the EU neighbourhood. We hope that we have provided a persuasive illustration of how original case study work can be conducted using the adapted Börzel and Risse framework as a template. The framework could be used to analyse other EU attempts to improve the criminal justice system in Japan; for example, by promoting better treatment of prisoners on death row. There is also considerable potential for comparative case study research on the EU's attempts to diffuse the abolitionist norm in other Asian states. Lastly, the Börzel and Risse framework adapted could also be used to analyse EU attempts to diffuse various other human rights norms in the region; for example, by addressing human trafficking and forced labour practices in South-East Asia.