Recodifying the Law: A Metalinguistic Inquiry into the Recodification of Belgian Law Between 2014–2019

Legal scholars attribute a great deal of importance to the linguistic dimension behind recodification. According to them, language contributes greatly to the improvement of both the accessibility and clarity of the law. Nevertheless, little research on the linguistic aspects of (re)codification exists within both linguistics and legal theory. Consequently, it seems worthwhile to study this linguistic dimension more in depth. To this aim, the recent legislative proposals to recodify various economic, civil and criminal codes in Belgium serve as a useful case study to improve our understanding of the linguistic challenges of recodification. In this article we will look at the linguistic comments made by the Belgian Council of State, a state organ which offers technical advice to the legislator, on the relevant proposals and consider (i) the type of linguistic comments made, and (ii), their implementation by the legislator. On the one hand, such an analysis allows us to single out the linguistic problems accompanying the Belgian recodification and, on the other, to elaborate on them with the help of the insights provided by legal theory. In doing so, we hope for our analysis to be a useful point of reference both for research on the Belgian recodification and other legal/linguistic approaches to recodification in general.


Introduction
In recent years, a renewed scientific interest in the linguistic dimension behind the drafting of legislative texts has arisen [cf. e.g., 20; 53; 59; 61; 62; 63; 64]. In this regard, legal linguists, for instance, have considered questions like which language problems need particular attention when drafting legislation, how legal linguists themselves may help with drafting legislation and, with respect to legal accessibility, which measures may be taken to ensure a qualitative text. However, there is one specific form of legislative drafting of which the linguistic dimension has received less attention, i.e., the process of (re)codification.
The importance of this linguistic dimension of the recodification process, which aims to improve the accessibility and clarity of the law, was pointed out by legal scholars (Sect. 2). Accordingly, it is relevant to consider this linguistic dimension from an interdisciplinary perspective by combining the insights of these legal scholars with linguistic theories on legal drafting, domain-specific discourse and metalinguistic commentaries.
To this aim, the recodification proposals for various economic, civil and criminal codes during the Belgian 2014-2019 parliamentary term serve as a useful case study. More specifically, we will look at the linguistic comments made by the Council of State of the Kingdom of Belgium, a state organ which offers, among other, linguistic advice to the legislator (Sect. 3) on the recodification proposals. For our analysis, we will draw on the insights provided by research on legal drafting. We will consider (i) the type of linguistic comments made and (ii) their implementation by the legislator. In doing so, we hope to uncover more information about the linguistic challenges accompanying the recodification, in light of both (i) the original goals of the legislator and (ii) codification theory. The insights brought forth by this analysis can consequently serve as a point of departure for future legal linguistic research into the matter.
The structure of our contribution is as follows. In Sect. 2, we will discuss the recodification of the Belgian codes in the context of codification theory. Subsequently, in Sect. 3, we will explain the function of the legislative section of the Council of State, paying particular attention to the linguistic comments it provides. Section 4 details the way in which we analysed these comments. The results of this analysis will be discussed in more depth in Sect. 5. Finally, some concluding remarks regarding the recodification process are formulated in Sect. 6.

The Recodification
In this section we will discuss the recodification in Belgium and the importance that the proponents of the Belgian recodification and codification theory attribute to the linguistic quality of the text. For this purpose, we will first discuss the background of the Belgian recodification (Sect. 2.1), highlighting where the codes came from, what the aims of the recodification were and which criticisms 1 3 terminology; (v) restructure the code; (vi) improve internal coherence; and (vii) reform the law where reform was due [15]. 1 The recodification thus aimed to both reform the law and improve the linguistic quality of the codes. Experts from the Minister's staff, academia, the ministry, and practice helped (pro bono) with the preparation of these new codes. 2 Although nearly unanimous support existed in legal doctrine for the goals behind the recodification, the specific proposals received a mixed reception. Matthias Storme [54], for example, was openly and outspokenly critical of their legislative quality. As an expert in private law, his criticism was directed towards the proposal to enact a new book for the Civil Code regarding the law of obligations. Among others, he regretted the poor quality of the Dutch text which seemed to be a literal translation of the French text and included several language mistakes. Moreover, the proposal suffered from a lack of terminological coherence, was poorly structured and riddled with superfluous words [54: 17]. Willem Possemiers [38] raised similar concerns in regard to other books proposed for the new Civil Code [38: 219-220].
In the new text of the Matrimonial Property Law, he notes, for instance, the inconsistent use of terminology: both huwelijksovereenkomst/convention matrimoniale and huwelijkscontract/contrat de marriage are used to (presumably) denote "marriage contract" [38: 216].

Theory on Codification
Such arguments against the hasty attempts at codification and reform are not new. They have been developed in legal doctrine before, most famously by Friedrich Carl von Savigny, and have often pointed out the linguistic complexity of codification as well. Von Savigny recognised that codification can, in theory, improve both the accessibility (Sect. 2.2.1) and the content of the law (Sect. 2.2.2) [60 : 34]. However, in practice, these improvements require specific skills and knowledge for both the substantive (legal) and formal (linguistic) aspects of the codification (Sect. 2.2.3). Not every legislator is sufficiently endowed with these, which entails the risk of legislators making things worse than they were, for instance, by deteriorating both accessibility and content. Other authors have elaborated on this theory regarding the risks of overreliance on the legislator in the life of the law.

Improving Accessibility
The first goal of most codifications is to improve the accessibility of the law. Therefore, they collect rules found in different legal sources (for example customs, case law, legislation, and executive acts) and try to reformulate them in clear and 1 3 Recodifying the Law: A Metalinguistic Inquiry into the… accessible language. In other words, they aim to be a reliable and well-written guide to the law. It is, however, not self-evident that legislation should be more suitable for this function than legal doctrine.
Legal scholars have been able to write handbooks that systematise the rules of the law. Legal doctrine thus has been more than capable of producing a concise and accessible overview of the law of the land. Such handbooks will only gain general acceptance as a guide to the legal system if they are helpful to their users. To gain the preference of its readers a handbook must be a concise, frequently updated, comprehensible, authoritative text written in accessible, clear, and contemporary language, which must address current legal problems in a coherent terminology. Accordingly, the free competition between handbooks will in the long run deliver classics which are generally accepted as accessible and reliable authorities on the law [60: 35].
If, however, the legislator (i) enacts a piece of legislation, namely a code, that is intended as an overview of the law, (ii) makes it obligatory for judges and administrations (and by extension every legal practitioner) to refer to this code, and (iii) enforces this obligation through the establishment of a Court of Cassation and a Council of State which annul every decision of the judiciary and administration that fails to meet this obligation, this overview of the law will only gain general acceptance because it is backed up by political power. Consequently, there is little reason to believe that it will be the trustworthy or accessible guide the codifier was hoping for. After all, such a code, to which everyone would then be obliged to refer, is not selected on the basis of the formal or substantive quality of the text, but on the basis of the power and willingness of the legislator to enforce his own decisions [60: 36].
In theory, it is of course possible that the legislator chooses the most skilled jurists to write the code and match the qualities of a good handbook. This happened, according to some, for example, with the Digests of Emperor Justinian (sixth century, cf. [16: 10-13; 60: 44-45]) 3 and the Swiss Civil Code of Eugen Huber [66: 170-179]. The same strategy was envisioned by the drafters of the recodification when they appointed several expert commissions. Considering the remarks of the Council of State, summed up in Sect. 5 of this article, it seems that this attempt has not worked equally well for each recodification attempt.

Improving the Law
The second goal of most codifications is to improve the rules embedded in the law. In those cases, the norms are said to require clarification or adjustment. The former usually means finding a suitable criterion for delineation and/or proper written expression to lift uncertainty that arose from the previously used formulation.
It is not self-evident that changes in legislation are the best means to achieve this goal: (i) Legislation is not the only way to achieve clarification. If the law lacks clarity on a specific point of law, one can look for clarification in the formation of normative expectations and practice in society (that is opinio iuris sive necessitatis and usus that form customary law); the formation of a consensus in doctrine; or a court decision in the matter. There is no need for legislative intervention to clarify points of law, if one is willing to trust those processes [60: 29-30]. (ii) Legislation is also not the only way to adjust existing rules. If it is recognised that the rules endorsed by doctrine, customs and case law are ill-suited for social cooperation or difficult to reconcile with the moral beliefs of the times, change might appear endogenously. Scholars can abandon their doctrines, courts can adapt or even overrule their older case law and practitioners can change the opinio iuris and usus behind the customs. A legislative intervention is not required per se.
An intervention by the legislator might, however, still be the most appropriate tool if, for instance, the consensus forming process is hindered by a self-sustaining split in adherence to different authorities (legislative intervention then helps to break the stalemate of a stabilised disagreement [54: 17]), the chilling effects of legal uncertainty prevent a case from being brought in the first place (clarification of the law), political support is needed to legitimate the reform (legitimisation of reform of the law [54: 17]), the gradual case-by-case reform of the law would undermine legal certainty or be unbearably slow (declarative effect of legislation [17: 88-89]),or the coordination of several points of reform should take place simultaneously to work and/or prevent inconsistencies from arising (overhaul of the law [17: 88]).
However, one should keep in mind that when it comes to clarifications or adjustments, even the smallest deviation from the formulation in preceding case law or the smallest change to the text of the legislation might call into question centuries of clarifications accumulated by doctrine and case law. In short, clarifications and adjustments allow for an overall improvement of legal certainty only under very specific circumstances. Moreover, the clarification will only improve legal certainty overall if it is so precise and consistent with the rest of the law that the legislative provision (i) excludes all other competing interpretations (decisive for the existing conflict), (ii) will not call into question other aspects of the law (coherence with the rest of the law) and (iii) does not give rise to new discussions (clarity) [60: 36-37]. A reform should only be passed when there is enough knowledge to improve upon the status quo and by those who possess such knowledge [60: 37]. This sets a very high standard for the legal and linguistic qualities of each legislative proposal.

Application to Linguistic Aims
In order to improve the accessibility and clarity of the law by codification or recodification, the utmost care has to be taken of the legal and linguistic qualities of the text. Otherwise, the new or recodified code risks creating more confusion than it remedies. To this aim, in Belgium, the legislator can obtain support from the legislative section of the Council of State, a state organ which can be asked to (i) draft legislative texts or (ii) give advice on a legislative draft, and the legislative services of the parliamentary assemblies.

The Council of State
In what follows, we will consider the work of the Council of State (Fr. Conseil d'État) of the Kingdom of Belgium in more detail, paying particular attention to the linguistic comments it provides. Therefore, we will discuss the different sections of the Council (Sect. 3.1-3.2), the types of advice it provides (Sect. 3.3), the length and scope of the procedure (Sect. 3.4) and previous studies on the linguistic comments made by the Council (Sect. 3.5).

Sections of the Council of State
The establishment of the Council of State by the legislative act of December 23, 1946 [34] 4 was intended to fix two problems: the lack of effective and uniform judicial review of administrative acts and regulations, on the one hand [37: 22-30], and the poor drafting quality and lack of coordination between legislative acts, on the other [37: 21-23, 30; 58: 20].
Accordingly, it has two sections: an administrative and a legislative one [24: Art. 1]. The former is the highest administrative court of the country. It can, for instance, annul administrative decrees and individual decisions and hear appeals of cassation against rulings made by other administrative courts [24: Art. 14]. The latter drafts reasoned legal advice to both (a) federal and regional legislators when they enact legislation 5 and (b) federal and regional executive branches when they enact regulation [12: 4, 23; 24: Art. 2 and 3].
The advisory opinions discussed in this article are all drafted by the legislative section in regard to legislation. In what follows we only explain the rules governing this specific competence.

Tasks of the Legislative Section
On both the federal and regional level, the legislative competence is shared between government and parliament [13: Art. 74-75 and 132]. Both have the right to propose legislation or amendments to legislative proposals [13: 75]. Parliament then decides on all proposals and amendments [13: 76]. The government subsequently sanctions and promulgates the legislation [13: 109].
The rights and obligations of government and parliament towards the Council of State differ: (i) Parliamentary assemblies have the right to request advice on each legislative proposal or amendment. 6  The advisory opinions given on the recodification were all either mandatory requests by the government or voluntary requests made by parliament regarding legislative proposals (i.e. types i and ii, not type iv). The rest of our discussion-on the scope of the opinions (Sect. 3.3) and procedures (Sect. 3.4)-will therefore be limited to opinions of those two categories.

Scope of the Advisory Opinions
Due to historical sensitivities, the Belgian Council of State-in contrast with its French, Dutch and Luxembourgish counterparts-is only competent to provide legal advice (on the wording of the text, and on coherence and compatibility with other legal provisions), which often includes linguistic remarks as well, and not policy advice (on the desirability, appropriateness and relevance) [57: 180-183].
All advice should be reasoned [24: Art. 2, § 1]. This implies that the Council explains why it arrived at the findings set out in the opinions [12: 33; 57: 183-184]. The scope of the advisory opinions varies as follows: In the legislation on the Council of State, we find five types of advice that are mandatory. First of all, the Council of State will always provide advice on: (i) the competence of the federal or regional legislator to enact the proposed legislation [24: Art. 3, § 2]; (ii) the legal basis, that is the compatibility with higher norms [12: 10-12; 24: Art. 84, § 3]; and (iii) the fulfilment of formal requirements (e.g., obligation to seek advice from certain institutions) [12: 10-15; 24: Art. 84bis-ter].
The federal and Brussels legislator are subject to additional mandatory rules: (iv) the Council will consider the correspondence between the French and Dutch 6 The chair of an assembly can be required to request such advice by one third of its members, half of the members of one language community (on the federal level or in Brussels) and twelve members of the parliamentary committee which settles conflicts of competence between the House and Senate (on the federal level) [24: Art. 2, § 1-4; 10: 10]. 7 In this text "legislative proposal" refers both to a bill introduced to parliament by the government (projet de loi), and to a bill introduced to parliament by a member of parliament (proposition de loi). 8 There is no obligation to request advice for amendments [12: 7, 9].

3
Recodifying the Law: A Metalinguistic Inquiry into the… versions of the text [24: Art. 48]; and (v) in case of a proposal for federal legislation, it will also determine whether the legislation is within the exclusive competence of the House of Representatives, or a shared competence of House and Senate [24: Art. 2, § 1 and 3].
If the time limit (see 3.4) allows it, the Council of State prefers to discuss: (vi) the internal coherence of the text (e.g., structure, layout and correspondence between the text and purpose of the proposal); (vii) the external coherence (in relation to other already existing texts at the same level in the hierarchy of norms); and (viii) the drafting of the text, in particular whether it is in accordance with the customary rules (legal and linguistic) on legislative techniques [12: 11; 57: 223-231; 58: 20]. Sometimes the Council of State will propose an alternative text [12: 11].
Many of the comments provided by the Council are therefore linguistic in nature. The types of advice that are per se linguistic, are thus partially mandatory (correspondence between Dutch and French text) and partially optional (the drafting). The other types of advice will, from time to time, include linguistic comments. Previous statistical research by Hendrickx [18; 19], who studied a corpus of 2000 advisory opinions on both proposals of legislation and regulation from 1980 to 2000, found that linguistic comments could be found in 75% of the opinions [19: 36]. Those included, among others, comments on correctness (27%), vagueness (14.5%), ambiguity (5%), cumbersomeness (5%) and redundancy (7%). Due to some changes in the legislation on the Council of State [25], the author of the research warned that the findings might no longer be representative of current practice [19: 43].
Based on its experience, the Council also published a guide on legal drafting which the legislator can use when drafting legislation and which contains many tips on language use [11].

The Procedure
The authority that requests the advice can steer-according to the rules discussed below-(i) the number of members of the Council that will look at the legislative proposal and (ii) the length and scope of the procedure and can also (iii) provide further explanation. Consequently, the legislator can (indirectly) influence the depth and scope of the opinions [21: 771].
An auditor of the Council prepares the advisory opinion with the help of a documentalist [12: 20-22]. The default rule is that the opinion is given by either a French or Flemish-speaking Chamber consisting of three members of the Council of State and two assessors [12: 21-22; 24: Art. 81]. However, if the proposal raises questions in relation to the competence of the federal or regional legislator (about 5% of the requests) [10: 41], it is discussed in a United Chamber consisting of three Flemish and three French-speaking members of the Council of State and four assessors [24: Art. 85bis; 57 : 24]. If the authority that requests the advice so wishes (less than 1% of the requests [10: 41]), the General Assembly of all twelve State Counsellors (and the assessors) of the legislative section will convene to give the advice [24: Art. 79 and 85; 57 : 29].
In principle, advisory opinion is handed out in the same order as it was requested [24: Art. 84, § 1]. However, in over 99% of the cases, the requesting authority sets a time limit [10: 39]. The requesting authority has the right to request that the advice is delivered within 60 or 30 days [24: Art. 84, § 1, 1° and 2°]. The time limit can be extended with the consent of the requesting authority [12: 12-13]. If the requesting authority requests advice within 30 or 60 days the Council has the right to limit the advice to three points, namely (a) the competence of the legislator (federal or regional competence; and the competence of the Senate); (b) the legal basis (compatibility with higher norms); and (c) the fulfilment of formal requirements [12: 10-12; 24: Art. 2, § 1, 3° and Art. 84, § 3; 57: 191-215]. Taking into account that in the judicial year 2017-2018 the Council received 2,115 requests for advice and has delivered opinions about 1939 proposals, containing 48,439 provisions, the Council is-due to its scarce resources-factually obliged to exercise this right for most requests that are to be delivered within 30 days [10: 32, 33, 39; 12]. If the requesting authority invokes urgency, which happens in about 5% of the cases [10: 39], the advice will be delivered within 5 days [24: Art. 84, § 1, 3°). In that case the Council is obliged to limit the advice to those three points [24: Art. 3, § 2 and Art. 84, § 3]. If the urgency is not satisfactorily motivated the request is inadmissible [12: 24; 14: 41]. In those cases where (i) an authority requested the advice of the General Assembly or (ii) the Council of State is required to sit in a United Chamber the time limits are raised to from 60 to 75, from 30 to 45 and from 5 to 8 days [12: 13, 15-16; 24: Art. 84, § 1].
The procedures allow for a dialogue between the Council of State and the institution that proposed the legislation. The Council has the right to hear a person authorised by the chair of the parliamentary assembly [21: 771] or the minister [24: Art. 82]. The authorised person himself can also seek contact with the auditor [12: 20]. The requesting authority can specify legal questions on which it would prefer to receive advice [12: 23].
In Table 1 we summarise the choices made by the requesting authorities (the Minister of Justice and Chair of the House) that steered the scope and detail of each advisory opinion. The proposals for recodification discussed in this article state that the requesting authority aims for an improvement of the quality of the text. However, the choices that were made indicate that the requesting authority had a limited interest in the legal and linguistic advice provided by the Council of State on those proposals. This was also the case for those recodification proposals which were set to replace two century-old codes and could, according to codification theory, call into question the cumulated case law and doctrine based on these codes. The requesting authority never used the right to ask that the advice would be given by the General Assembly to the Council of State. This limited the number of people that went through the text. The Minister or the Chair of the House of Representatives requested that the advice would be given within 30 days for 6 proposals and within 60 days for 3 proposals. This meant that the Council of State was-for practical reasons-required to limit the advice to the three mandatory matters, namely competence, legal basis and fulfilment of formal requirements. This is not an isolated case. Both members of the Council of State and legal doctrine have argued that the legislator shows little interest in the advice of the Council Recodifying the Law: A Metalinguistic Inquiry into the… Table 1 Scope and procedure of each advisory opinion  and, in particular, in the quality of the legislation. Accordingly, the legislator makes choices with respect to the time limit, the scope of the advice and the implementation of the advisory opinions [58: 26, 924]. From a political perspective, this might be a reasonable strategy, because a lot of time could be lost drafting a good text without this resulting in any meaningful gain of political capital useful to survive the next election.

Implementation
Legal doctrine believes that (i) the advice is-as a rule-followed [57: 189] and (ii) the legislator is more likely to follow advice of the Council if the advice is technical [57: 188], which is, for instance, the case for many language-related recommendations.
The statistical research mentioned in Sect. 3.3 studied the implementation of a sample of 57 and 19 advisory opinions from the 1980s and 1990s respectively with a relatively high number of remarks. For the 1980s, it found that 76% of the linguistic comments were fully implemented by the legislator, 10% partially implemented and 10% not implemented. In the 1990s results were slightly lower, with full implementation accounting for 67%, partial implementation for 10% and no implementation for (20%) [19: 41].
Compliance was particularly high for comments relating to cumbersomeness (81%), redundancy (79%) and correctness (76%), but lower for ambiguity (70%) and vagueness (67%), where Hendrickx thought political motives played a role, since a lack of clarity can be a manner to achieve political compromise [19: 42]. At the time, the Ministry of Justice-which was also the requesting authority for most recodification proposals-had the lowest compliance rate, namely 59% (ranging between 0 and 89%), but Hendrickx clarifies that the small sample sizes for each ministry might make this result unreliable [18: 405-409].

Aim
Considering the linguistic dimension attributed to the process of (re)codification and the mixed reception of the recodified Belgian codes (Sect. 2), it seems promising to look at the linguistic comments made by the Council of State. By means of a qualitative assessment of these comments (cf. Sect. 4.2), we aim to (i) uncover the (recurring) linguistic problems attributed to the Belgian recodification proposals in the Council's advisory opinions and (ii) discuss these in light of both the recodification-goals of the proposals and codification theory itself (cf. Sect. 2.2).

Methodology
To this aim, we will conduct a twofold analysis: (i) The first part of our analysis (Sect. 5.1) entails in a qualitative analysis of the linguistic comments made by the Council of State, which consists in two steps. First, we will label certain comments made by the Council of State as "linguistic comments", distinguishing them from other non-language related comments by the Council of State. Under linguistic comments, we understand all comments which explicitly mention either language, writing or style and/or regard the orthographic, grammatical, terminological, syntactic or pragmatic aspects of language use. Secondly, we established an inductive, qualitative classification of these linguistic comments based on the aspects of language and the language phenomena they regard. It is the goal of this classification to pinpoint any recurring problems (possibly) specific to the recodification proposals, as identified by the Council. To aptly describe these problems, we will also distinguish between different subcategories of the comments made by the Council, provide more detailed information if necessary and offers examples from the French versions of the drafts and opinions. Quantitative data will also be provided, both in absolute and relative frequencies, to highlight the frequency of the categories in and across the proposals. However, it should be noted that these quantitative data are entirely dependent upon our own qualitative operations. At the end of the section, there will be a quantitative (regarding the number of comments across the different opinions) and qualitative overview (which discusses the results in light of our research aim (cf. Sect. 4.1). (ii) For the second part of our analysis (Sect. 5.2), we will consider the implementation of linguistic comments by the legislator, for which we will look at the 7 preliminary drafts which have eventually been turned into legislation and published in the Belgian Official Gazette. More specifically, following Hendrickx [18], we will make a distinction between comments that are (a) fully, (b) partially or (c) not implemented. Once again, we will also consider whether quantitative and qualitative differences occur between the different codes. We will, however, limit ourselves to the implementation of the specific comments, one of the three main sections within the advisory opinion. This specific comment-section entails an article-by-article assessment of the proposal and includes most of the linguistic comments singled out by our analysis (cf. Sect. 5.2). Due to this article-by-article discussion, the implementation of these comments is relatively easy to verify in light of our current study. This in contrast to the implementation of: • The introductory remarks. These deal with problems reoccurring at various places in the proposals, making their implementation more difficult to account for with respect to the specific comments. • The concluding remarks. These also treat problems that reoccur at various places in the proposals. Additionally, three out of the seven law proposals did not include a concluding remarks-section, which is why this section will be left out of the analysis as well.
Since the scope of the introductory remarks may apply to the complete proposal, and therefore weigh heavier on the linguistic quality of the text, the results of our 1 3 following analysis cannot be regarded as being representative of the entire proposals, but only with regard to the specific comments made by the Council of State.
For the discussion of our results, we will mainly depart from the insights provided by linguists and jurists on legal drafting. Additionally, we will draw on discourse linguistics (with regard to legislative texts as a domain-specific discourse tradition-cf. Sect. 5.1.2) and metalinguistic research, since all comments considered here are, in fact, metalinguistic in nature. All examples are translated with the help of De Valks Juridisch Woordenboek [56], which contains English translations for most Belgian legal terminology, and the EUR-Lex-corpora on SketchEngine [52].

Limits of the Analysis
Regarding the scope of our research, we would like to single out three principal research limits.
First, it should be noted that we only considered the linguistic comments made by the Council of State 9 for our analysis. We did not check the draft proposals on language mistakes or terminological inconsistencies that might not have been mentioned or have been overlooked by the Council of State. Furthermore, we only considered the published versions of the legislation for the implementation of the comments. We did not check the accompanying parliamentary documents, which entail the parliamentary discussion of the Council's advice, for possible reasons why a certain comment would not have been implemented.
Secondly, as pointed out before, our analysis implied a qualitative operation. Our discussion of the results, including also descriptive statistics (see below) is therefore totally dependent on this qualitative operation. Furthermore, other researchers might have employed other qualitative classifications or assigned different properties to linguistic comments-cf. in this case the different classification used by Hendrickx [18] (cf. Sect. 3.3)-, which would possibly have led them to different insights.
Lastly, we will only report the absolute and relative numbers of comments as descriptive statistics. We wish to stress that the reader should not try to compare legislative proposals or categories of comments based on either absolute or relative numbers. We give a short list of non-exhaustive reasons why this would not be possible for these data: Legacies and Gifts) might have more descriptive word groups (e.g., "replace article … with …") that will not end up in the law than those proposals intro-ducing new books in a code. This can affect our wordcount, which is why the relative frequencies allow no real comparison. (iii) It is to be expected that if an equal time limit applies, the longer a legislative proposal is, the less time the Council will have to consider linguistic features. Moreover, the time limit for the Council to provide advice differs for several proposals (cf. Table 2). The workload of the Council varies too. (iv) The number of State Counsellors involved in the advisory opinion on the Elimination of the Commercial Code was three times higher than those involved in the other proposals. (v) Those responsible for the drafts were aware that the Council of State was not the end of the legislative process. Consequently, they anticipated that they would still have the opportunity to continue working on further improvements regarding both language and content of the law during the procedures of the Council and the parliamentary discussions. Several improvements-which were not part of the comments made by the Council-were included in later versions of the drafts. (vi) The number of comments for some categories (e.g., comments on non-substantive reformulations, Sect. 5.1.4) is so low, that it is impossible to attribute any statistical value to their relative frequencies (cf. Table 2).

Results
Our analysis revealed a total of 1024 linguistic comments. In what follows, these comments will be further analysed by means of an inductive qualitative classification, which will specify the different types of comments (Sect. 5.1) and their implementation in the published legislation (Sect. 5.2)-cf. our methodology in Sect. 4.

Types of Linguistic Comments
All linguistic comments were divided into four main categories: language comments (LC), discourse-specific comments (DSC), translation comments (TC) and comments on non-substantive reformulations (CNSR). The first category, the language comments (Sect. 5.1.1), regards all comments on mistakes against the Dutch or French standard language. The discourse-specific comments (Sect. 5.1.2), on the other hand, concern comments on features which are problematic or wrong only within the specific context of the law because they are not in accordance with the legal (discourse) tradition or cause problems for the specific text function. The third category, the translation comments (Sect. 5.1.3), include all comments which explicitly address formulation-related differences between the French and Dutch versions of the law (= inequivalencies). The final category, which regards comments on non-substantive reformulations (Sect. 5.1.4), must be understood specifically in light of the context of recodification. Part of the recodification process included the linguistic modernisation of the codes, without substantively changing the legal rules in question. In such cases, the Council recommended the legislator to mention this in the explanatory memorandum.
Looking at our inductive classification from a quantitative perspective (Figs. 1 and 2), we find notable differences. The most frequent category are the discoursespecific comments, which are accounted for 927 times (equal to 90.5% of all linguistic comments). They are followed by the translation comments (53 occurrences or 5%) and the language comments (41 occurrences or 4%). The final category, the comments on non-substantive reformulations, are only marginally accounted for: 3 times, which makes up less than 0.5% of all linguistic comments.
The following subsubsections will discuss the aforementioned categories in more depth and provide some examples, as well as distinguish subtypes for the first two categories (cf. Sects. 5.1.1 and 5.1.2). Section 5.1.5 provides an overview of the distribution of comments across the different opinions.

Language Comments
Language comments were earlier defined as comments on mistakes against the two standard languages. Within this category, a further distinction can be made between grammatical (1), idiomatic (2) and orthographic (3) comments: (1) À l'article 1:31, alinéa 2, en projet, il y a lieu, dans la version française, de remplacer le mot « visé» par le mot « visées» [3: 106] 'In article 1:31, subparagraph 2, of the legislative proposal, in the French version, it is necessary to replace the word "visé" by the word "visées".'  The grammatical comments, exemplified by (1), regard all comments on conjugations and accords. This is the most frequent subtype within the language commentcategory, with a total of 26 occurrences. The idiomatic comments (2), which are accounted for 5 times, have to do with mistakes against more or less fixed expressions. Finally, the orthographic comments (3), accounted for 10 times, include all comments on misspelled words. A long list containing no less than 28 such orthographic errors, 10 like compétencs (instead of compétences) and sociétets (instead of sociétés) can be found in the concluding remarks 11

of the advisory opinion on the Code on Companies and Associations [3: 111-112].
When we look at the distribution of language comments across the different opinions (cf. Table 2), we observe that these comments can be found in six of the three opinions. More specifically, the language comments are absent in the advisory opinions on the Civil Code, Legacies and Gifts, the Matrimonial Property Law and Book 5 (Obligations).
Considering their textual distribution from a qualitative perspective, we must observe that, within the specific comments, these comments are rarely formulated in the main text of the advisory opinions, but instead in footnotes or between parentheses (cf. e.g., [4: 28; 8: 121]). Accordingly, it seems that the Council of State did not-or could not, due to the great workload of the Council and the time limits set by the legislator (cf. Sects. 3 and 4.3)-specifically focus on language correctness, treating them almost exclusively as sidenotes. However, within the concluding remarks of some opinions-for the new Code on Companies and Associations [3: 105-112], the Elimination of the Commercial Code [4: 56-58] and Book 8 (Evidence) of the Civil Code [7: 20-21]-, the Council of State formulated a limitative list of linguistic problems occurring within the laws, which could also contain language comments. Language comments were especially frequent in the list made for the Code on Companies and Associations. In this case, they generated a total of 31 comments. This is the normal approach of the Council: if language mistakes in a proposal are plentiful, the Council will, due to the time limit and its workload, not be able to deal with them one by one. It will then take a sample of the text and discuss the mistakes that were made in the general comments, which the legislator can then use to check for other language mistakes during the revision of the texts.

Discourse-Specific Comments
The discourse-specific comments appeared most frequently in our corpus, making up 90.5% of all comments. These comments have to do with the discourse-specific features and needs of law texts and can be further divided into four categories: general discourse comments (Sect. 5 Fig. 3).
In our classification (cf. Table 2, Sect. 5.1.5), these comments constitute the most frequent category in all texts, with the pragmatic comments being the most frequent subtype in the first six opinions and the terminological comments being the most frequent subtype in the last three. This high frequency, in comparison to the frequency of the other categories, is in part due to the complexity of this category: whereas the language comments-category (cf. Sect. 5.1.1) exclusively regards grammatical, idiomatic, or orthographic comments, the discourse-specific category entails four subcategories, with the terminological and pragmatic comments each entailing two ulterior subcategories, cf. Sects. 5.1.2.2-5.1.2.3.

General Discourse Comments
The first subcategory of discourse-specific comments is entailed by the general discourse comments, which we named as such since they do not single out specific problems at specific passages in the proposals, but rather include general statements about the quality of the proposals' linguistic organisation (4). 12 (4) L'avant-projet constitue de toute évidence un texte qui n'a pas fait l'objet d'une relecture globale après l'assemblage de ses différentes composantes [3: 5] 'The draft proposal is clearly a text which has not been subject to a thorough proofreading after the different parts [of the draft proposal] were merged together.' In doing so, the Council of State can also call out the lack of effort put into the linguistic redaction by the legislator-which happened for instance for the Code of Companies and Associations-even though within legal doctrine said redaction is considered a crucial part of the recodification process (cf. Sect. 1.2). During our analysis, we accounted for a total of 7 comments pertaining to the general discourse comment subcategory. Notwithstanding their relatively low quantitative frequency, the weight of these comments on the linguistic quality of the proposal is very big, since their qualitative scope regards the proposal in its entirety.

Terminological Comments
The second subcategory concerns terminological comments, which were accounted for 391 times. Within this subcategory, an ulterior distinction can be made between comments on used terminology (243 comments) and comments on definitions (148 comments).

Comments on Used Terminology
The comments on used terminology regard whether terminology was used correctly, consistently throughout the entire law proposal, in accordance with the legal (discourse) tradition and clear enough to secure legal certainty. Some examples of these comments are given in (5) Particularly interesting with regard to the recodification process is (5b): the use of the collocation père et mère ("father and mother") creates problems about other forms of parenthood like same-sex parents, which were already recognised by older legislation. Such terminology, therefore, hinders the (linguistic) modernisation aspect of the recodification. Similar problems arose with other outdated terms as well, like porteur des titres ([3: 32], "security holder"), which no longer exists in Belgian law, and photocopies ([3: 58], "photocopies"), which cannot be used to refer to electronic documents.
The uniformity of the terminology, on the other hand, appears to have been a particular problem in the proposals for a new Code on Companies and Associations [3: 20-21] and Book 2 of the new Criminal Code [9: 91]. Here, the Council of State explicitly pointed out the terminological variation in the introductory remarks and recommended avoiding terminological inconsistencies as much as possible. These inconsistencies, in turn, form a major problem when it comes to the legislator's goal of "cleaning up terminology" (cf. Sect. 2.1). An additional problem concerning terminological uniformity was that terminology not only had to be used consistently within each legislative proposal, but also across the different recodification proposals, which was not always the case. This has, for instance, been pointed out in the advisory opinion on the Elimination of the Commercial Code [4: 34].

Comments on Definitions
Since the recodification introduced a lot of new legal terminology, the Council dedicated particular attention towards the definitions provided by the legislator. These comments deal with the correct and clear formulation of definitions, whether they are applicable throughout the entire law proposal and whether they are in accordance with the legal tradition. Requests made by the Council of State to provide definitions for certain terms also belong to this category. These comments are exemplified in (6) Both comments in (6) regard circular definitions. Another interesting example, particularly in the light of the recodification, regards a comment on the definition of vol ("burglary") in the proposal for a new Criminal Code [9: 84]. In this case, the Council noted that the legislator proposed a new definition of burglary, but that the previous definition was already very clear and therefore guaranteed a uniform jurisprudence. This led the Council to question whether jurisdiction would profit from a new definition-or recodification-of the term burglary. Consequently, the new definition proposed by the legislator neither formed an improvement compared to the previous one, nor reflected case law on the matter.
The comments on definitions also concerned the key notions of each legislative proposal, like entreprise ([1: 4-5; 4: 33-39], "corporation") in the two proposals regarding the Economic Code, and société cotée ([3: 17-18], "listed company") and société à responsabilité limitée ([3: 9-11], "company with limited responsibility") in the Code on Companies and Associations. In the case of the definitions for entreprise ("corporation"), for instance, the Council pointed out that this notion was not exclusive to Book XX of the Economic Code in Belgian law [1: 4] and that, regarding the compatibility with higher norms, specific rules and definitions also existed on the European level [4: 37]. In such cases, the comments on definitions not only regard the legal scope of a singular term but also the legal scope and legal certainty provided by the new (proposed) code in its entirety.

Pragmatic Comments
The third group of discourse-specific comments, labeled pragmatic comments, regards all comments on formulation-related and interpreting problems above the terminological level. In our classification, this is the most frequent subcategory of the discourse-specific comments, amounting to 519 comments. Once again, a further distinction can be made, with comments on discourse efficiency, on the one hand, (379 comments) and comments on the thematic and formal coherence of the law, on the other (140 comments).

b. On ne comprend pas qui est visé, dans le texte français, par le pronom « elles» à l'alinéa 4. Il convient d'utiliser le mot « ils» ou préférablement un terme plus précis
'We do not understand who is intended, in the French text, with the pronoun "elles" [they/ feminine] in subparagraph 4. It would be more correct to use the word "ils" [they/masculine] or preferably a more precise term.' c. Au paragraphe 2, les mots « que l'interruption de mandat ait ou non été convenue de commun accord» sont dépourvus de portée normative dès lors qu'on se trouve nécessairement dans l'un ou dans l'autre cas. Ces mots seront donc omis [3: 47] 'In paragraph 2, the words "whether or not the interruption of the mandate was established by mutual agreement" have no normative consequence since one of the two situations is always applicable. These words, therefore, need to be left out.' d. À l'alinéa 3, l'expression « causés par …» est impropre. Il convient d'écrire: le « dol, la violence ou l'abus de circonstances, imputables à un tiers complice d'une partie contractante ou à …» [6: 12] 'In subparagraph 3, the expression "caused by …" is insufficient. It would be better to write: the "fraud, violence or abuse of circumstances, attributable to a third-party accomplice of a contracting party or a …"' In (7a), for instance, the Council of State remarked that the overuse of adverbs affected the clarity of the text. In (7b), a problem of coreference was highlighted, since it was not clear to which objects the pronoun elles ("they/feminine") referred, leading to an unintelligible text. The word group singled out in (7c), in turn, has no normative consequence, and so the Council advised the legislator to remove it. Finally, in (7d), the Council proposed its own formulation to substitute another formulation, which the Council deemed insufficient.

Comments on Thematic and Formal Coherence
The second group of pragmatic comments includes all comments on internal coherence, both thematic and formal. Consider in this regard the examples in (8) Both comments (8a) and (8b) regard the thematic coherence of the law. In the case of (8a), paragraph 3 states the same thing as the preceding paragraph. As such, it is redundant and should be left out. The comment in (8b), in turn, expresses that the title of a designed article does not correspond to its content, which is why the Council of State suggested another title. In (8c), on the other hand, formal problems with the legal text occur: by not numbering an enumeration the legislator might give rise to the idea that the different parts of this enumeration should be understood as different subparagraphs, instead of subordinated thoughts.
This internal coherence appears to have been a particular problem in the proposals for a new Criminal Code and a new Book 5 (Obligations) of the Civil Code. In the introductory remarks of its opinion on the Criminal Code, the Council stated that (i) the legislator would do best to establish as much internal coherence within the proposal as possible [9: 87-91] and (ii), the proposal displayed an atypical layout for a law text [9: 92]. For Book 5 of the Civil Code, on the other hand, the Council noted within its concluding remarks that many of the internal references (that is, to other parts of the draft) were wrong. As such, the Council provided a limitative list of places where this problem occurred [6: 27].

Lexico-Grammar Comments
The final subcategory of discourse-specific comments regards comments on the lexico-grammar choices made by the legislator, which were accounted for 10 times. Legal discourse does not only have its own terminological and formal tradition but also its own lexico-grammar tradition. This lexico-grammar tradition refers to the fact that legal texts prefer to consistently use the same lexico-grammar patterns. Some examples include the use of the definite article de ("the") instead of pronouns such as hun ("their") (9a) and the use of verbs in the present indicative if the legislator wishes to express an obligation (9b). 'To express an obligation in a law it is more appropriate to use the present indicative instead of the future tense.' In both cases, the legislator does not make a mistake against the standard language but opts for a lexico-grammar element which is considered less prototypical for legislative texts. Therefore, the Council advises on using the more typical lexicogrammar elements.

Translation Comments
The third main category of comments concerns the translation comments, with 53 occurrences. These comments are mandatory (cf. Sect. 3.3), since all federal legislation is bilingual and the Council of State needs to single out inequivalencies between the Dutch and French versions of the law, which would otherwise create legal uncertainty. These translation comments are exemplified in (10) The translation comments appear in 8 of the 9 opinions and are only absent from the advisory opinion on the proposed Matrimonial Property Law (cf. Table 2

Comments on Non-substantive Reformulations
The final category of comments-the comments on non-substantive reformulations-were only accounted for 3 times (cf. Jurists are inclined to interpret reformulations of legislative texts as substantive changes to the laws themselves. To avoid such a "wrong interpretation" [6: 5], the legislator needs to make explicit in the explanatory memorandum that the reformulation merely had a formal purpose. The fact that the Council of State only formulated 3 such comments does not mean that this problem did not occur more frequently: 2 of the 3 comments were formulated within the introductory remarks-in the opinion on the proposed Code on Companies and Associations [3: 5] and the proposed Book 5 of the Civil Code [6: 4-5]-and therefore addressed problems occurring at multiple places within those drafts (cf. Sect. 5.2.2). Furthermore, this specific problem with the recodifications was highlighted by various legal scientists, like Storme [54] (cf. Sect. 2.1), who noted this to be a particular problem in the proposed Book 5 of the Civil Code. Table 2 gives an overview of the comments' distribution across the different advisory opinions:

Textual Distribution
The numbers outside of the parentheses (the numbers above) regard the absolute frequencies of comments. The numbers between the parentheses (the numbers below) concern the relative frequencies of comments, which regard their occurrence per million words. As mentioned in Sect. 4.3, is it impossible to use these data for a contrastive analysis between the different texts due to a variety of reasons (cf. the reasons listed in Sect. 4.3). In this regard, we specifically want to draw attention to the fact that the length of certain proposals differs greatly (cf.   it comes to the definition of vol ("burglary") in the proposal for a new Criminal Code [9: 84], we also mentioned that in this case, the legislator was neither able to improve the content of the law, nor reflect the case law on the matter (cf. Sect. 5.1.2.2). Furthermore, collocations like père et mère and archaic terminology like porteur de titres and photocopies do not contribute to a more modern law book.
Similar observations can be made for the benefits that legal doctrine assigns to the recodification (cf. Sect. 2.2). However, these observations are only valid for the first versions of the legislative proposals, since the legislator can rework and improve the draft afterward. Furthermore, our observations are limited to the language of the drafts, and more specifically to the opinions of the Council of State. The legislator can have valid reasons to not implement the comments of the Council. Consequently, it seems worthwhile to look at the implementation of these linguistic comments and consider why some of them were not implemented by the legislator (cf. Sect. 5.2).

Implementation of Linguistic Comments
For the second part of our analysis, the implementation of the linguistic comments, we will look at the laws as they were eventually published in the Belgian Official Journal. We are, however, only able to look at the implementation of seven of the nine proposals, since Book 5 of the Civil Code and Books 1 and 2 of the Criminal Code have not yet been turned into laws.
Furthermore, as already mentioned in our methodology (cf. Sect. 4.2) we will limit ourselves to the implementation of the specific comments. This is done because (i) both the introductory and concluding remarks contain problems that reoccur at various places in the proposals, making their implementation more difficult to account for, (ii) the introductory remarks also include descriptive comments made by the Council, which require no further action from the legislator, and (iii) because the concluding remarks were absent in three of the seven proposals.
The specific comments of the seven accepted laws (cf. Table 3) amount to a total of 650 comments. However, 31 of the comments lost their relevance, since during the redrafting process the paragraphs or articles to which these comments belonged were deleted and their linguistic problems with them. Consequently, we will look at the 619 comments (equal to 60.5% of the linguistic comments singled out in our analysis) whose implementation we can check in the laws themselves. In doing so (Fig. 5), we find that 325 of the linguistic comments of the Council of State were fully implemented by the legislator, 93 comments were partially implemented and 201 comments were not implemented at all. In relative data (Fig. 6), this means that the legislator acted upon more than 2/3 (67.5%) of the linguistic comments, with 52.5% being fully and 15% of the comments being partially implemented.
It should be noted that the non-implementation of comments does not necessarily imply a lack of care from the legislator. The legislator can have other opinions than the Council of State on what kind of language formulae are desirable or not (see, for instance, our discussion regarding the implementation of definitions in Book 3 of the Civil Code below). Furthermore, we would like to stress that this implementation only regards comments related to language. Consequently, optimisations regarding the content of the law, which may have been included in the final version of the drafts, are not included in our analysis.
When we only consider the implementation of specific comments across the different texts (cf.  In what follows, we will consider this last code in a more detailed manner. We will therefore look at (i) the implementation of the different types of comments and (ii) compare these data with the explanation of Book 3 provided by Sagaert [51], who was one of the main legal architects of the code in question.
The specific comment-section for Book 3 contained 1 language comment, which addressed an orthographic mistake, 140 discourse-specific comments, including 73 terminological comments, 66 pragmatic comments and 1 lexico-grammar comment, and 8 translation comments. Their implementation is illustrated in Fig. 7: The terminological comments made by the Council are discussed by Sagaert [51]. Here, Sagaert makes explicit that (a) in Book 3 of the new Civil Code few definitions were provided, (b) the Council commented at various points on this fact, but (c) the comments on missing or vague definitions were mostly not implemented in the law, for which Sagaert provides two reasons: (i) The authors of Book 3 of the new Civil Code intended the code to be an "instrument for society" [51: 406], starting therefore from a functional approach towards legislation [51: 407]. This implies that, rather than providing the exact definition of legal rules within the law, the legislator intends to give the judicial power the freedom to consider the implications of legal rules and terminology from context to context. (ii) Imposing too fixed definitions might cause the law to be outdated rather quickly [51: 407]. In this regard, the use of "open categories" contributes to the durability of the law book [51: 410].
Concerning (i), the use of "open categories" can here be seen as what has been called "purposeful vagueness" within legal linguistics [22: 18; 23: 170; 61: 13-14] and is best exemplified by the open categories de bonne foi (in good faith) and de mauvaise foi (in bad faith) [8: 18, 38], which the legislator left without a definition in the final version of the law: what good or bad faith is, can be left for a judge to decide.
Looking at the unimplemented terminological comments in the opinion relating to the relevant law book, we notice that 32 of the 53 of these comments were comments on definitions. However, this particular argument does not explain why 4 pragmatic comments (cf. Fig. 7) were not implemented by the legislator, in particular those which entail more technical matters, like the comments in (12a) on discourse efficiency, and (12b) on the internal coherence of the text. The same applies to the terminological comments not covered by this explanation, like comments on unnecessary (13a) or wrongly used terminology (13b).
Why such comments were not implemented by the legislator is a question that our current analysis cannot answer. The same is true for similar unimplemented comments in other codes. More qualitative research might enable us to better understand the other motives for their non-implementation. Moreover, it would be interesting for future studies to also consider the parliamentary documents of the House of Representatives, which, due to reasons of space, were not considered here. These documents entail the discussion of the advice in parliament and can therefore shed light on the (ideological) stances taken concerning these comments.

Conclusion
Our analysis set out to uncover the difficulties of 9 recodification proposals within Belgian law to learn more about their linguistic dimension and, more specifically, linguistic challenges. Legal scholars have often emphasised this linguistic dimension of recodification processes (cf. Sect. 2.2), making the Belgian recodification proposals an apt experimental ground for an evidence-based study on the matter. We specifically looked at the linguistic comments made by the Council of State. We departed from a twofold qualitative analysis, which considered (i) the types of comments made by the Council (Sect. 5.1), and (ii), their implementation by the legislator (Sect. 5.2).
Our analysis illustrated that the Council of State spent particular attention on the linguistic redaction of the texts, at times even calling out the insufficient linguistic quality of the law and the lack of effort put into them by the legislator (Sect. 5.1.2.1). We accounted for a total of 1024 linguistic comments, many of which-like the language comments (Sect. 5.1.1), most of the pragmatic comments (Sect. 5.1.2.3) and the lexico-grammar comments (Sect. 5.1.2.4)-the Council is not obliged to make (cf. Sect. 3.4). These comments ranged from formal and language-related problems, like the orthographic comments in (3), the circular definitions in (6) and the translation comments in (10), to more substantial matters, like the new definition for vol ("burglary") discussed in Sect. 5.1.2.2.
These problems can be regarded as problematic when it comes to both the goals set out by the legislator (cf. Sect. 2.1) and the benefits assigned to recodification by legal doctrine (cf. Sect. 2.2). We refer to our discussion in Sect. 5.1.5.1, where we discussed how these comments were problematic when it came to the legislator's goal of promoting legal certainty (which is also one of the benefits legal doctrine assigns to recodification), enhancing readability, cleaning up terminology and improving internal coherence.
When it comes to the implementation of these comments, which we could check with regard to the specific comments of the seven proposals which have been turned into law, we noted that the legislator acted upon 418 (67.5%) of the 619 comments, with 325 (52.5%) being fully and 93 (15%) being partially implemented. 201 comments were not implemented by the legislator.
Due to the different scopes of comments and circumstances of the advisory opinions (cf. Sect. 4.1.3), on the one hand, and reasons of space, on the other, we were unable to compare both the distribution and implementation of comments in a more detailed manner. Regarding the implementation of the specific comments, however, we observed a particularly low implementation in Book 3 (Property) of the Civil Code, in particular concerning the terminological and pragmatic comments. For the former, we could-concerning the comments on definitions-find some explanation in an explanatory article provided by one of the authors of the original proposal [51]. This was, however, not the case for terminological comments not related to definitions (12) or the pragmatic comments (13), which our current analysis was unable to account for.
Concluding our contribution, we can state that the advisory opinions of the Council of State pointed out the linguistic complexity behind a recodification process. However, further research on the matter is necessary, for which our current contribution hopes to be a valuable starting point. This could concern both contrastive analyses between some of the legal proposals and a more detailed, particularly more qualitative look at the non-implementation of comments, e.g., based on the parliamentary documents (cf. Sect. 5.2). When it comes to a comparative approach, it might also be insightful for future research to compare the Belgian recodification process to other recent recodification processes, like the Norwegian Klarspråk-project (2010-2016).