Talkin’ ‘bout a Negotiation: (Un)Transparent Rapporteurs’ Speeches in the European Parliament

For policies to be legitimate, both the policy process and the underlying reasons must be transparent to the public. In the EU, the lion’s share of legislation is nowadays negotiated in informal secluded meeting called trilogues. Therefore, presentation of the trilogues compromise by the rapporteur to the European Parliament (EP) plenary is, arguably, one of the few formal occasions for ‘transparency in process,’ i.e., public access to the details of actual interactions between policymakers. The aim of this article is thus to examine the extent to which rapporteurs are transparent about trilogue negotiations when presenting legislative compromises to the EP during plenary sessions, and to assess whether the extent of transparency is linked to the extent of conflict between legislative actors and to elements of the political context related to rapporteurs. To this purpose, we coded 176 rapporteur speeches and, on this basis, concluded that these speeches poorly discuss the trilogue negotiations. Interinstitutional negotiations are discussed in only 64% of cases, and even when they are, the extent of information about trilogues is generally small. While we do not find support for an effect of political conflicts, some characteristics linked with rapporteurs are significantly related to transparency in process of their speeches. This is the case for their political affiliation and their national culture of transparence.


Introduction
Transparency about the process leading to policy choices vis-à-vis those they concern is crucial for democratic representation and decision-making, including in the EU (Lord, 2013;Stie, 2013). Yet, in the EU, the lion's share of legislation is nowadays negotiated in informal secluded meetings called trilogues. During trilogues, representatives of the co-legislators (i.e., the Council and the European Parliament [EP]) and the Commission negotiate compromises that are then voted on by their institutions. Since trilogues are secluded, the public is de facto excluded from the negotiation process leading to EU legislation. It has thus been argued that decision-makers should at least provide retrospective information on the process that led to the legislative outcome. This information is necessary for citizens to control their representatives and for MEPs to vote on the compromises with sufficient information. Therefore, it is the cornerstone of the accountability of decision-makers to their constituencies, which is a foundation of the legitimacy of the EU legislative process. This is what Jane Mansbridge (2011) calls 'transparency in process,' i.e., public access to the details of the decision-making process, as opposed to 'transparency in rationale,' i.e., mere public access to the reasons for the decision.
In the ordinary legislative procedure of the EU (OLP), the presentation by rapporteurs (i.e., the main EP negotiators) of trilogue compromises to the EP plenary is, arguably, one of the few formal occasions for this 'transparency in process.' However, although the secrecy of the trilogues makes those rapporteurs' presentations particularly relevant to the accountability of the legislative process, few scholars have investigated the extent to which rapporteurs actually discuss the negotiation process in plenary. Hence, despite the crucial role of rapporteurs' speeches, we do not know yet the extent to which rapporteurs are 'transparent in process' when they present the outcomes of trilogues. In this context, this article contributes to filling this gap by (1) examining the transparency of rapporteurs' speeches during plenary meetings regarding trilogue negotiations, and (2) assessing whether the extent of transparency is linked to the extent of conflict-i.e., the extent to which actors disagree about the negotiated file-and rapporteurs characteristics. One can indeed argue that the process is more likely to be transparent in the case of 'hard negotiation,' since they have more concessions to justify. Arguably, the transparency of the process is also particularly important when the legislation is highly contested. Indeed, in this case it is more likely that citizens and other actors in society will want to hold those involved in EU policymaking processes accountable. Similarly, rapporteurs have different constraints and experiences according to their political groups and member states, and these characteristics are likely to influence the speeches they make.
Empirically, the analysis is based on an original dataset consisting of 176 rapporteur speeches. We manually coded each speech to construct a process transparency index assessing the extent to which rapporteurs discuss the negotiations leading to legislative compromise. We employed this index to evaluate the 'transparency in process' of the OLP and test our hypothesis about the effects of conflict. The remainder of the article is structured as follows: The next section addresses the role of rapporteurs' speeches for the transparency of trilogue negotiations, and highlights our contributions to the literature on the OLP. Section 3 develops our hypotheses regarding the factors that are likely to influence the degree of transparency of rapporteurs' speeches. Section 4 describes our data collection and the operationalization of our variables, while Section 5 presents our results. Eventually, Section 6 concludes.

Rapporteur Speeches and Transparency in Process
Broadly speaking, transparency relates to the availability of information (Meijer, 2013) and more precisely to "the extent to which an entity reveals relevant information about its own decision processes, procedures, functioning, and performance" (Grimmelikhuijsen, Porumbescu, Hong, & Im, 2013). In other words, the transparency of a decision-making process refers to the extent to which an actor makes information available about how and why decisions are produced to citizens and political representatives (Bovens, 2007;Naurin, 2017). A key dimension is that transparency must enable the external actors to evaluate the process (Warren & Mansbridge, 2013). From this perspective, Mansbridge distinguishes between whether the reasons for the outcomes are provided without detailing the process behind them (transparency in rationale) and whether the decision-making process is shared with the public (transparency in process). Importantly, such information on the process can either be provided in real time or in retrospect (i.e., after the process ended; Naurin, 2017). In the context of EU legislative decisionmaking, transparency in rationale would mean explaining why a particular piece of legislation is the best to solve a particular issue, while transparency in process would mean legislative institutions explaining how they arrived at the compromise they adopt. The extent to which legislative decision-making in the EU fulfils this second dimension is disputed, to say the least.
In the period since the early 2000s, legislative negotiations in the EU have undergone increasing informalization. The adoption of EU legislation requires that the EP and the Council agree on an identical text, which means that they have to reconcile their respective positions. Nowadays, most inter-institutional negotiations in the EU take place in informal meetings called trilogues (Laloux, 2020). In trilogues, representatives of the Council, the EP, and the European Commission negotiate informal compromises that can then be formally adopted by the two co-legislators. More precisely, the EP is represented by a negotiation team led by the rapporteur (i.e., the MEP in charge of the file) while the Council is represented by the rotating presidency (Roederer-Rynning & Greenwood, 2015). Trilogues are secluded, and the working documents are not made publicly available: This means that outsiders, including non-participating members of the legislative institutions (Brandsma, 2018;Leino, 2017), cannot observe the negotiations.
Hence, most of the substantive debates occur in nontransparent trilogue meetings. Outsiders must therefore rely on trilogue negotiators for information on the proceedings, that is to say, how the content of EU legislation has been designed and negotiated. Yet, few studies have investigated how negotiators report on trilogues even though they are the only source of information on the process leading to compromise. This means that the 'retrospect' transparency in process of the OLP in the EP plenary has not been assessed. Although several scholars have commented on the limited feedback from negotiators, only Brandsma (2018) has empirically studied it. Focusing on the public reports from EP negotiators to their committees during the negotiations, he found that they were generally limited, with negotiators not providing much information on what went on.
To our knowledge, no study has so far examined how trilogue negotiations are addressed in plenary meetings of the EP. This is surprising since plenaries are considered as the main institutional arena for public communication throughout the EU legislative decision-making process (Lord, 2018;Ripoll Servent, 2018). As stated by Christopher Lord (2013Lord ( , p. 1067: Of all the institutional settings through which [OLPs] meander, only plenary debates of the EP seem likely to meet what Anne Elizabeth Stie [2013, p. 75] defines as a requirement that there should be "at least one open setting where those decisions are tested and critically examined by popularly elected representatives in a manner that is publicly available and accessible.'' Whereas debates occur in other places during the EU legislative procedure, such as in the COREPER or the EP committee, the work in such a forum is often opaque and there is little public record of the debates therein, in contrast to the EP plenary (Lord, 2013;Naurin, 2010). In sum, the EP plenary is the most appropriate arena for public communication about legislative negotiations both externally, vis-à-vis citizens and national parliament and internally for MEPs that did not participate in trilogues.
A consequence of trilogues, however, is that plenary debates in the EP mainly concern issues that have already been negotiated and compromised on. As a result, the debates are not likely to have much impact on the legislation ultimately adopted, and rank-and-file MEPs-those who did not participate in the trilogues-cannot influence the legislation and therefore do not contribute to the process leading to it. The only opportunity the public has for transparency in the legislative process is thus in the hands of negotiators and more particularly rapporteurs. Rapporteurs do not negotiate alone with the Council. They are members of larger teams that also include other MEPs (Ripoll Servent & Panning, 2019). However, rapporteurs are those in charge of explaining and presenting the final compromise negotiated with the Council to the EP plenary, and thereby of justifying it in front of the public, including the course of negotiations (Garssen, 2016;Stie, 2013). Other members of negotiating teams can take the floor as well (usually the shadow rapporteurs) but their time for parole is significantly shorter (EP, 2019). Moreover, in contrast to the rapporteurs who present and justify the compromise, shadow rapporteurs do not speak as negotiators but express the opinions of their political groups.
All this means that rapporteurs are the ones in charge of opening the 'black box' of trilogues, not only to the public but also to their fellow MEPs. In other words, if we understand trilogues as an informal institution (Roederer-Rynning & Greenwood, 2015) rapporteurs' speeches are one of the only opportunities to ensure 'external transparency' therefrom, that is, transparency vis-à-vis those who are not involved in the negotiations and are therefore not members of the institution. However, there are no institutional constraints on what they can or cannot share (Garssen, 2016). Rapporteurs are not compelled to talk about negotiations, which raises the question of the extent to which they do so and thereby contribute to transparency in process of negotiation. If, following Mansbridge (2009), one accepts that transparency in process matters for the OLP, then rapporteurs' speeches are the key moment for transparency in such processes.
Yet, how rapporteurs address trilogue negotiations in plenary meetings remains unknown. Generally speaking, the literature on plenary debates in the EP has mainly focused on identifying lines of conflict between MEPs, but has not addressed the transparency of the decisionmaking process itself (Laloux & Pennetreau, 2019). In particular, this literature largely has overlooked rapporteurs' speeches, even though they are a potential source of public information on the trilogues process. The only exception is the work of Garssen (2016), which aimed to identify the argumentation scheme at the disposal of rapporteurs when defending their work, and to assess the importance of those speeches. While not linked to trilogues, this work nevertheless showed the relevance of those speeches. As the following debate consists of other MEPs positioning themselves against the rapporteur's argument, the types of argumentation that can be used by proponents and opponents "is for the most part predetermined by the initial presentation made by the rapporteur" (Garssen, 2016, p. 26).
Knowing the extent of rapporteurs' communication about trilogues is normatively important for two reasons. First, transparency in process is crucial for public scrutiny of the legislative procedure. Scrutiny makes it possible to control EU legislators to hold them accountable (Curtin & Leino, 2017). However, in order to facilitate the negotiation process, negotiators are often required not to disseminate working documents, and the various actors involved prefer not to publicly disclose their positions (Reh, 2014). In such cases when visibility is lacking in the process leading to legislation, as in trilogues, accountability requires that decision-makers should at least provide a public account of the process leading to the outcome (Naurin, 2017;Warren & Mansbridge, 2013). In the OLP, trilogue negotiation implies that the 'burden of justification,' or 'narrative accountability' (Reh, 2014)-which arguably falls on the EP plenary-ultimately rests largely on the shoulders of the rapporteurs. A lack of transparency in the process would deprive citizens, national parliaments and other MEPs from 'their right to justification' (Stie, 2013), and thereby hinder them from monitoring the legislative decision-making process (Laloux, 2020;Leino, 2017). This could be problematic for the democratic legitimacy of EU policy-making, which relies inter alia on public scrutiny of the legislative process, and in the resulting accountability of EU legislators to their constituents and national parliaments (Lord, 2013). Similarly, this would also mean that MEPs lack information on the negotiations, and therefore their votes are not sufficiently informed. They vote on a take-it-or-leave-it basis on legislation that has already seen compromise, and they do so without knowledge of the process leading to the compromise. This lack of information has the potential to be problematic since it is the whole assemblyand not just the trilogue negotiators-that represents EU citizens and thus that brings collective legitimacy to the decision.
Second, since rapporteurs' presentation is important for the subsequent debates (Garssen, 2016), the extent to which they are transparent about trilogues might affect the negotiations process. In other words, if MEPs' interventions are linked to rapporteurs' initial speeches, the extent to which rapporteurs discuss the negotiation process might affect the extent to which this process is further debated by the whole plenary. Rapporteurs' speeches are in this way crucial to the transparency and legitimacy of the European legislative process; the information that they do or do not share influences the way the EP exercises its powers (Brandsma & Hoppe, 2020). In practice, this means that rapporteurs advise fellow MEPs on the need to uphold the EP's positions vis-à-vis the Commission and the Council and/or to accept the position of one or the other of these institutions (Lord, 2018, p. 7). Rapporteurs' communication about what takes place in trilogue negotiations not only constitutes transparency in process vis-à-vis their constituencies and the larger public, it also constitutes transparency among peers. The content of such negotiations is also important for accountability, which relies not only on transparency but also on the right of a forum to question the decisions of their representatives (Bovens, 2007). Therefore, one can argue that transparency about the trilogues process is necessary for substantive debate over legislation, and in turn for the plenary to fulfil its role: This makes rapporteurs' speeches a crucial component of both the transparency and legitimacy of the European legislative process.

The Expected Influence of Conflicts: Political Group and Nationality
We expect two kinds of variables to have an effect on the extent to which rapporteurs talk about the negotiation process, the level of conflict of the legislative file and the background of the rapporteurs. First, the degree of conflicts is likely to matter because this makes it more difficult to reach an agreement. By conflict we mean disagreement between legislative actors as to the content of the legislative act. The more divergent the positions of the EP and the Council, the more necessary it is for trilogue negotiators to make concessions to reach a compromise . Rapporteurs must therefore account for choices that do not necessarily correspond to the preferences of the EP as a whole, or of certain political groups in particular. Yet, compromises must be approved by their respective institutions before they can be formally adopted as a legislative act. In the EP, a majority in the plenary has to vote for the compromise, so if a negotiated compromise deviates too much from the positions defended by the EP, the rapporteur is confronted with a risk of defection, including within her or his own political group. Such a defection would not be without cost to the negotiators. Particularly, for rapporteurs the failure of an informal compromise would undermine their reputation, credibility and prestige in their committee as well as within their political group Mühlböck & Rittberger, 2015). Rapporteurs thus have an incentive to get their deal accepted, and the more concessions they make during negotiations, the more precarious their position. Moreover, deviating too far from the EP position is likely to entail a reputational cost for the rapporteurs. This could mean that rapporteurs represented poorly the EP during the negotiations because he or she was unable to defend the positions of its principal. As a result, this may diminish his or her reputation as well as the likelihood that she or he will be assigned other important tasks.
In such a situation, it is therefore necessary for the rapporteurs to be clear about the negotiation process to show that they have done their best. In other words, the plenary speech may be used by rapporteurs to justify his or her own actions in the process. One can argue, then, that it is in rapporteurs' interests to be clear about their reasons for deviating from the EP position in the negotiations process in the event that the compromise is deemed unsatisfactory if he or she wants it to be adopted. Indeed, MEPs may be more inclined to vote for a compromise they do not fully support if they know the concessions were necessary and the gains hard-won; that is, if they think the compromise is the best possible deal for the EP. This supposition is in line with Delreux and Laloux (2018), who showed that negotiators try to transmit the pressures from the inter-institutional forum to their institutions to find a deal. Moreover, MEPs arguably would be more convinced of the justice of a given outcome, and therefore to vote for it, if they were confident their side's position was considered seriously, and this requires transparency. Hence, our first hypothesis is: H1: The more conflict there is between institutions, the more transparent rapporteurs are about a trilogue negotiation.
We also expect a similar effect of intra-institutional conflicts, that is to say, when MEPs disagree as to the content of the file to be adopted. The rationale is similar: Rapporteurs want to see their compromise adopted, and will use the negotiations to push for that. Indeed, refusing a trilogue compromise is also costly for the institutions, entailing transactional costs and increasing uncertainty regarding the final output (Bressanelli, Koop, & Reh, 2016;Costa, Dehousse, & Trakalova, 2011). Therefore, in those cases, rapporteurs are likely to put more emphasis on the negotiations to show the cost of rejecting the deal for the EP. Moreover, discussing the negotiation might also be a means to put pressure on other MEPs. The EP might accept the positions of member states it does not fully agree with because it wants to appear 'responsible' by not blocking EU legislation, and because MEPs are sensitive to government pressure via the national parties (Bressanelli & Chelotti, 2016;Ripoll Servent, 2013). Therefore, rapporteurs might use negotiations to convince their colleagues that the compromise reflects the will of the Council, and that voting for it is the way to go. Accordingly, our second hypothesis reads: H2: The more internally divided the EP is, the more transparent rapporteurs are about a trilogue negotiation.
Second, besides conflicts, we also expect the elements of the political context related to rapporteurs to affect the degree of transparency in process of their speeches in plenary. Specifically, we expect that the extent to which rapporteurs are 'transparent in the process' will depend on (1) the size of their political group and (2) their national culture of transparency. As regards political affiliation, we expect rapporteurs from the two larger political groups-i.e., EPP and S&D-to be less transparent. The reason is that, because they have a larger share of the vote, these groups are more influential and can therefore be more confident that a majority will support their compromise, so their rapporteurs do not have to work as hard to persuade their fellow MEPs to support their position. This is especially true since those two groups also tend to form a grand coalition and vote together in the plenary, thereby reducing further the need to seek the support of other groups. In contrast, rapporteurs from smaller group do not have this advantage, and therefore might be more willing to justify the negotiation process in order to secure it. Hence our third hypothesis is: H3: Rapporteurs from the EPP and S&D are less transparent than rapporteurs from smaller groups.
Finally, we also expect one specific national characteristic of the rapporteurs to influence its culture of transparency and thereby the degree of transparency of rapporteurs' speeches: corruption. Indeed, the level of corruption of the member state might correlate with its culture of transparency. Transparency is widely regarded as a crucial tool in the fight against corruption (Tienhaara, 2020). For instance, Lindstedt and Naurin (2010) suggested that increased transparency of institutions helped to fight corruption under certain conditions. To our knowledge, the question of whether the opposite effect is true has not been a topic of focus in EU studies, yet, arguably more corrupt countries should tend to be less transparent as a matter of protection for corrupt officials. Along this line of reasoning, we can similarly assume that less corrupt entities develop more transparent cultures around conflict and negotiation, which would be reflected in the way their members account for the negotiation process. Our fourth hypothesis is thus: H4: The greater the level of corruption in the member state where the rapporteur comes from, the less transparent his or her speeches.

Qualitative Analysis of Rapporteurs' Speeches
The rapporteurs' speeches were collected in an automated way using R on the legislative database of the EP. We selected specifically rapporteurs' speeches in plenary sessions, and debating trilogue compromises; due to the nature of these criteria, all the files in our sample had been completed at the time of collection. Moreover, we only selected speeches that were made after the end of negotiations and therefore preceded final plenary vote; this explains the total of 176 coded speeches. Since the speeches were made in different languages, we translated into English using Google Translate to carry out the analysis, following on previous work that has shown this method to provide valid results (de Vries, Schoonvelde, & Schumacher, 2018).
To assess the transparency of rapporteurs' speeches about the trilogue process, we developed a 'process transparency index' based on manually coding speeches.
Since theory is scarce about the transparency of rapporteur speeches, we opted for an inductive approach, which is common in cases when mismatches are observed between theory and empirical observations (Timmermans & Tavory, 2012), or when theory is lacking (Jebb, Parrigon, & Woo, 2017). As is usually the case with inductive coding, this process took place in three phases (Charmaz, 2014). During a first investigative phase, both researchers coded 15 speeches in order to identify the elements present in rapporteurs' speeches.
Once the codebook had been inductively established and its categories stabilized, the same 15 speeches were coded a second time by the two coders. Through this process, 12 different categories were included in the codebook.  Position of the EP When rapporteurs refer to EP preferences in relation to legislation and/or the specific positions resulting therefrom in the trilogue negotiations.
"Parliament also wanted to have better control on the establishment of the criteria and the procedure for the designation of the registry by using delegated acts.''

Position of others actors
When rapporteurs refer to the Council's or the Commission's preferences in relation to legislation and/or the specific positions resulting therefrom in the trilogue negotiations.
"The Council supported the Commission's proposal at 30% in a non-binding format.'' Integrated claims: EP When the rapporteurs refer to the EP's gains during the trilogue negotiations, i.e., whether Parliament's preferences that were actually incorporated in the negotiated compromise.
"In parliament, we accepted the structure as proposed by the commission but wanted to provide for additional safeguards, such as for the respect of users' privacy and security, consumer protection and human rights. I am especially glad that we could strengthen the text on safeguarding human rights and the rule of law.'' Dismissed claims: EP When the rapporteurs refer to the EP's concessions during the trilogue negotiations, i.e., whether Parliament's preferences that were not included in the negotiated compromise.
"We wanted quantified targets, but this has not been achieved at EU level.'' Integrated claims: others actors When the rapporteurs refer to the Council's or Commission's gains during the trilogue negotiations, i.e., whether Council's or Commission's preferences that were actually incorporated in the negotiated compromise.
"Just as the council came to meet us with the wetlands, we had to compromise on the so-called compensation mechanisms and the reference value for forests.'' Dismissed claims: others actors When the rapporteurs refer to the Council's or Commission's concessions during the trilogue negotiations, i.e., whether Council's or Commission's preferences that were not included in the negotiated compromise.
"The Council agreed to withdraw the amendments concerning derogations from the Regulation on the protection of personal data (GDPR Regulation) aimed at creating specific derogations for statistics from this Regulation.'' of the coding categories. During the second phase, another 30 speeches were coded in two rounds of 15. After each round, a comparison of the coding was conducted, making it possible to refine the coding criteria and thus ensure better inter-coder reliability. Knowing that our coding had reached a sufficient level of equivalence that the results would not be due to chance, we then entered the third coding phase. Each researcher coded the remaining rapporteur speeches of the sample (and removed those that did not correspond to the selection criteria). Of course, induction does not mean a complete lack of theoretical background (Wacquant, 2002); our operationalization of transparency in process is based on existing work on transparency. More precisely, we build our categories by adapting the work of Brandsma, Curtin, and Meijer (2008;Brandsma & Schillemans, 2012), who identify three broad dimensions of the decision-making process about which decision-makers must be transparent when accounting for their decision-process: input, process and output.
First, Brandsma argues that, to hold decisionmakers accountable, principals must be able to compare between outcomes and preferences. This requires agents to account for the inputs and outputs of negotiations. Information about the initial preferences and positions of the actors involved is necessary to understand the basis for the negotiation, and therefore to assess them. This information corresponds to our 'positions' categories, for the EP and other actors. It is indeed necessary to distinguish between the EP and the other institutions because the rapporteur is first and foremost the negotiator of the EP; she or he is therefore primarily responsible for defending its position against the Council and the Commission, and wild be judged accordingly by the MEPs. Arguably, this makes it more important for rapporteurs to be transparent about the EP's positions and outcomes in front of the EP plenary.
Second, information about the outputs of negotiations is also necessary i.e., what happened to the initial positions. Hence, the question of which institutions' claims were and were not integrated into the final compromise was included among our categories. We make a second distinction here, which is added to the first one: It is important to distinguish between integrated and dismissed claims. Indeed, it is likely easier for rapporteurs to talk only about negotiation successes, while transparency requires talking about the negotiations losses as well.
Eventually, as noted by Brandsma and Schillemans (2012), information about inputs alone are not in themselves sufficient to properly report on negotiations. Procedural issues (the context and sequence of negotiations) are also crucial in determining whether alternative outcomes could have been achieved (Behn, 2001). This means including practical information about the process-trilogues and their dynamics in our casewhich is to say, whether the rapporteur could have achieved a different outcome. Table 1 displays the coding strategy and our resulting category.
This coding enabled us to build an index of the transparency of rapporteurs' speeches. The index is composed of the following seven categories: (1) negotiations process (which combines hard and smooth processes and a specific trilogue category), (2) position of the EP, (3) position of others actors, (4) integrated claims-EP, (5) dismissed claims-EP, (6) integrated claims-others, and (7) dismissed claims-others. These categories cover the three dimensions necessary for citizens and MEPs to evaluate the results of the negotiators: Firstly, the positions of the actors and what happened in the compromise, but also practical information on the conduct of the negotiations, so that the EP can assess whether other results could have been achieved. All the categories are presented in detail, including examples in Table 2. On this basis, the process transparency score of each speech is the sum of the categories that compose it: one point for each category. This process transparency score is the dependent variable of our analyses.
Regarding the independent variables, following Cross and Hermansson (2018), we measured inter-institutional conflicts by the length of the negotiations. Our assumption is that, in more conflictual cases, more time is needed to reconcile the positions of the actors. We used the result of the vote in the EP to measure the extent of intra-institutional conflicts. For our third hypothesis, we used a dummy variable taking the value of '1' for EPP and S&D rapporteurs, and '0' for the rapporteurs of the other groups. We measure corruption using the latest report of the Corruption Perceptions Index provided by Transparency International. This index measures the perceived levels of corruption in the public sector of worldwide countries. Scores ranged from 0 (highly corrupt) to 100 (very clean).
Two control variables were included in the model. First, we controlled for the number of words in the speeches, since we assume that the more rapporteurs speak, the more likely they are to address the negotiations. We also control for the scope of a file, measured by the number of Eurovoc descriptors (Van Ballaert, 2017), as it is likely that rapporteurs have to spend more time describing files dealing with many subjects, thus leaving less time to discuss the negotiations in their speeches.

Modestly Transparent Speeches, Regardless of Conflicts
In this analytical section, we firstly present a description of our coding results. Then, we explore the data descriptively by examining variation according to the characteristics of the rapporteurs and of the committee. Eventually we test our hypotheses using negative binomial regressions. Figure 1 displays the distribution of the process transparency index, as well as the occurrence of each category. First, looking at the distribution of the process transparency index of the speeches, a first lesson is that it is left skewed. More than 35% of the rapporteur speeches do not mention any of our categories relating to the trilogue negotiations (64 out of the 176 speeches), while fewer than 10% of the rapporteurs in our sample mentioned five or six of the categories (respectively nine and six speeches). Another lesson from this observation is that no single speech within the sample discusses all seven categories. Hence, we can conclude that rapporteurs' speeches presenting compromises reached in trilogues are modestly transparent regarding the process that led to them. Of course, some of the information may flow informally from the rapporteur to the parliamentary committee and, via the latter, is disseminated within the political groups. Nevertheless, rapporteurs' speeches during plenary sessions, as provided for in the formalized institutional procedure, do not guarantee access to information on these negotiations. In other words, rapporteurs' speeches are not sufficient to ensure that MEPs are informed about the negotiations process when they vote. Figure 1 also shows that, when talking about the negotiations, rapporteurs focus more often on the EP (successes), and on describing how negotiations as such generally went. That is to say, they mainly refer to the EP's position during trilogue negotiations with the Council and the Commission (a bit over 30% of speeches), as well as the course of the negotiation process: hard or smooth (a bit over 30% of speeches). In a quarter of the speeches in the sample, the bargaining successes of the EP (integrated claims) is mentioned. Trilogue negotiations are discussed in 20% of the speeches. The only category relating to the actors involved in the negotiations that is regularly mentioned is the bargaining position of other actors (32 speeches, i.e., 18% of cases). In con-trast, rapporteurs rarely mention the concessions made by other actors or their success in the negotiations. The demands of other actors that ended up being dismissed in the final compromise are addressed in under 4% of the speeches (six speeches). Similarly, the demands of other actors that were integrated into the compromise are addressed in slightly under 8% of cases (14 speeches). Noteworthy is that rapporteurs do not often talk about the EP's failures during negotiations either-or at least much less than about its successes. The EP's dismissed claims are mentioned in 19 discourses (a bit under 11% of cases).
We turn now to examine the variation in the process transparency index according to the characteristics of the speaker. First, the functioning of the EP is organized around groups. Therefore, one of the questions that arises is whether the degree of transparency of the rapporteurs' speeches varies according to their political affiliation. As can be seen in Figure 2, the average transparency of rapporteurs' speeches according to the different political groups is relatively low but varies by twice. The least transparent group is the EFDD, with an average of 1 on the process transparency index. It is noteworthy that the two main groups in the EP-the Christian Democrats and the Social Democrats-are respectively in the penultimate and antepenultimate positions, with average scores below 1.5. This is in line with our expectation that rapporteurs from those groups have less need to be transparent. The Greens have an average transparency score above 1.5, just behind the ECR-the third most transparent group, with a score close to 2. Finally, with a score between 2 and 2.5, the Liberals have the second-best average transparency score, just behind the left-wing group, which is close to 2.5 on average on the process transparency index. Second, beyond the political groups, another crucial dimension of the EP, like other parliaments, is the territorial dimension, i.e., the member states as far as MEPs are concerned. Figure 2 also displays the mean process transparency score per country. With an average transparency score of 3.5, Danish MEPs are the most transparent, and are an outlier compared to their peers. They are followed, at a good distance, by the British MEPs-2.5 on average on the transparency index. Next, come 17 Member States whose rapporteurs' speeches range from an average of 2 to 1. Italian rapporteurs are, on average, just below 1, followed by Portuguese, Maltese and Latvian rapporteurs. Finally, the only Slovakian rapporteur of our sample did not provide any information on the trilogues in his speech, and therefore scores 0 on the process transparency index. Hence, there seems to be a slight tendency for the rapporteurs from the countries that joined the EU earlier to produce more transparent speeches. Nevertheless, this tendency is mitigated by the position of Italy as well as, to a lesser extent, the Netherlands (and also by the rapporteurs from Spain and Portugal, if only the CEECs are considered as new Member States).
We also examined variation in the process transparency index according to the public policy areas of the file, as reflected in the structure of parliamentary committees (see Figure 3). Since the EP committees have different working habits and are concerned with policy domains at varying levels of politicization (Laloux & Pennetreau, 2019), it is possible that their internal functioning affects the transparency of their rapporteurs' speeches. Again, the average transparency per parliamentary committee varies but remains generally low. Speeches by rapporteurs from the committee on agriculture are the most transparent, with a score around 2.5. Next comes a group of six committees with a score between 2 and 1.5. This first group shows that the importance of EU competencies does not seem to influence the degree of transparency of rapporteurs according to their parliamentary committees. Indeed, the committee dealing with industrial research and energy issues-an area in which the EU has only limited competence-falls between those relating to agriculture and the internal market, areas in which the EU has extensive competence. A second group of nine committees has speeches with an average score between 1.5 and 1, confirming our observation. Finally, the few speeches made by rapporteurs from the committees on Constitutional Affairs and on Budgetary Control are not transparent at all, with a score below 0.5.
To test our hypotheses, we conducted regression analyses using the score of our process transparency index as the dependent variable. Specifically, since this score is basically a count variable, we conducted event count models, here negative binomial regressions. Table 3 shows that none of our hypotheses regarding conflicts can be supported. Neither of our measures of conflict has a statistically significant effect on transparency. Therefore, we find no support for our assumptions that rapporteurs are more transparent in the most conflictual cases, for neither intra nor interinstitutional conflicts. This result may be explained by the fact that the countervailing dynamic is also going on in some cases. One could as well argue that the greater the conflict, the less likely a  rapporteur would be to go into it in public. Indeed, showing that they have made many concessions could isolate them and put them in a difficult situation in the EP, which would perhaps even cost them votes. If the two hypotheses work concurrently, this would explain the results obtained, i.e., the absence of a meaningful relationship.
In contrast, the hypotheses regarding rapporteur characteristics tell a different story. As we expected, rapporteurs from large member groups tend to be significantly less transparent than those from smaller ones. This supports our hypothesis that they have less need to be transparent since they also have less need to partner with other groups. Similarly, there is a significant negative relationship between corruption and transparency. In other words, rapporteurs from more corrupt countries tend to be less transparent about trilogues. This result could imply that, as we expected, the (dys)functioning of politico-administrative systems influences the culture of transparency and negotiation at the EU level. However, just making information available will not prevent corruption if such conditions for publicity and accountability as education, media circulation, and free and fair elections are weak. Many studies have shown that multiple factors interact (Camaj, 2013;Lindstedt & Naurin, 2010), moreover, policy instruments deployed to promote transparency or fight corruption may have little impact on the perception of these phenomena (Dunlop, Kamkhaji, Radaelli, Taffoni, & Wagemann, 2020). A focus on this specific issue of the relationship between member state corruption and the degree of transparency of the incoming rapporteurs' discourse is therefore necessary to better understand what is going on.
Regarding the control variables, only the number of words has a statistically significant impact on the degree of transparency of rapporteurs' speeches. Logically, the more a rapporteur speaks, the more likely it is that he or she will mention something related to the negotiations. This also means that we found no evidence that among rapporteurs the scope of a dossier reduces the extent of transparency. As a robustness check, we also conducted OLS and negative binomial regressions, with similar results. To explore our data, we also conducted a probit regression for all the categories of our index. Whereas the majority did not differ from the main model, interestingly, the length of the procedure was significantly linked to how rapporteurs describe the negotiations. Specifically, the longer a procedure is, the more rapporteurs use the 'hard process' category and, conversely, rapporteurs use more the 'smooth process' category for shorter procedures. This suggests that, at least, rapporteurs report properly to the plenary about the difficulty of the negotiations.

Conclusion
The aim of this article has been to examine and explain the extent to which EP rapporteurs are transparent about trilogue negotiations when presenting legislative compromise to the EP plenary. To do so, we assessed the effect of legislative conflicts and rapporteurs' characteristics on this transparency in process. We coded 176 legislative speeches and, on this basis, came to the conclusion that 'transparency in process' is rather poor during EP plenaries. The inter-institutional negotiations are discussed in only 64% of the cases and, when so, the extent of information about them is generally little. In other words, rapporteurs' speeches do not guarantee that MEPs have sufficient information about the negotiation process when they vote on the compromise. This result may be due to organizational constraints on the conduct of the plenary sessions, leaving little time for the rapporteurs. Rapporteurs have to find a balance between information on the legislative provisions, and information on the negotiation process itself. Our results indicate that this balance leans towards the former. Another result is that rapporteurs tend to focus more on the EP's positions and outcomes than on those of the other institutions. This suggest that they knowingly act as agents of the EP when discussing negotiations.
Regarding the factors explaining transparency in process in the plenary, contrary to our expectation, we do not find a significant link between legislative conflicts and transparency in process. However, the hypothesis that transparency in process is linked to the individual characteristics of the rapporteurs is supported. We observe that EPP and S&D rapporteurs are significantly less transparent, which may explain the general findings about low transparency. It might also be a problem since, as these two groups are the largest in the EP and often come together in a grand coalition, many (important) files are dealt with by their members.
Another important finding is that corruption reduces transparency. The higher the level of corruption in a member state, the lower the transparency of speeches made by rapporteurs from that country.
From a normative point of view, these are not good results. The legitimacy of EU policymaking is partly based on public scrutiny of the process, which is why trilogues have been criticized for their lack of transparency in process. While this concern could be partially allayed by rapporteurs' public reporting on the negotiations, we did not find evidence that this was the case. Rapporteurs' speeches during EP plenaries do not seem to provide enough information on trilogue negotiations for outsiders to be able to assess how negotiations went. What is more, we did not find evidence that more conflicts result in more transparency, whereas one could argue that the trilogue concessions that conflicts induce require more justification. Hence, these results cast further doubt on the extent to which plenary debates in the EP are able to fulfil their role in providing justification for EU legislation and therefore to legitimize the EU legislative process.
We conclude by stressing that these results beg further research on the transparency of trilogues. So far, existing studies have mainly focused on the availability of working documents, and less on the actual reporting of negotiations by their participants. Potential avenues for future research include consideration of the longitudinal perspective. Going back to the early days of trilogue negotiations would make it possible to observe whether their gradual institutionalization and the importance they have garnered with respect to OLPs have influenced the way rapporteurs report on them. Similarly, a deepening of the analysis through political affiliation could yield other results. Indeed, the political groups in the EP are not always coherent or homogeneous. Analyzing the transparency of speeches based on national political party affiliation might lead to different results. It could also be interesting to test some individual variables, such as the political seniority, former professions, level of education, and even the age or gender of rapporteurs.
Further, the fact that rapporteurs do not speak much about trilogue negotiations in the plenary raises the question of what they talk about when they justify legislative compromise. Further studies could investigate rapporteurs' speeches more generally, not only through the lens of transparency in process, which would enable assessment of how legislations are justified in the EU. More particularly, examining the transparency in rationale of rapporteurs' speeches would make it possible to assess whether this compensates for the lack of transparency in process, and thereby to assess the transparency of rapporteurs' speeches as a whole. Finally, scholars could also investigate the quality of account given in the other institutions participating in trilogues. Such research would improve our understanding of how policy choices are justified in the EU, and thereby make it possible to better assess the democratic legitimacy of such policies.