The Transposition of International Criminal Law Concepts Into National Jurisdictions: The Case of Genocide

Abstract This article discusses preliminary findings of a study on the transposition of the legal concept of genocide into 131 national jurisdictions. The specificities of this transposition into national criminal systems, as well as those related to the international legal definition of genocide, are described in the first part. The communicative situations in which the concept of genocide has been transposed are then examined in order to show their scope and breadth, and to which extent they contribute to the transformation of the concept of genocide. Trends related to the object of transformation in the definition and their effect on meaning are subsequently outlined. The findings point to a situation where, despite having been the object of multiple consensus at the international level, the concept of genocide has been transformed by the vast array of domestic legal languages and legal systems into which it has been transposed and thereby reinforce the relation between the configuration of the language and law, and the difficulty of translation.


Introduction: From the International Legal Definition of Genocide to Its Domestic Versions 1
The legal concept of genocide 2 has evolved in a process that involves three major phases: creation of the legal concept by the international community, transposition of the international concept into national legal languages and systems by means of translation, and interpretation of the concept by national courts. In this article, the focus will be on the first two phases. The concept as created by the international community will be presented as the source text. It is followed by an overview of domestic definitions of genocide (target texts) in 131 national jurisdictions, along with key contextual aspects underlying their domestic transposition and potentially affecting the global concept. Relations between contextual aspects and shifts will be drawn in order to substantiate if and to which extent context affects the transposition and translation process and its results. Foremost, the paradigm of transposition and translation of the concept of genocide is briefly contextualised.
The production of multilingual international legal texts and their transposition into domestic systems fall under the scope of legal translation (Prieto Ramos 2011: 204). Even if its role is often downplayed if not squarely ignored (Öner and Banu Karadağ 2016: 334), legal translation is a key component of the negotiation between 1 The author wishes to thank Fernando Prieto Ramos and Jean-Claude Gémar for their significant suggestions on earlier drafts of this paper. 2 For the purpose of this study on the large-scale and self-reliant transposition of the concept of genocide and its subsequent effects on the meaning of genocide, the concept of genocide is confined to the field of law. the international language and law, on the one hand, and national languages and laws on the other hand (Šarčević 1997: 64).
Legal translation is a 'special and specialised area of translation activity' (Cao 2007: 7) and it is generally admitted that it differs from 'general' translation in three main aspects (Cao 2007;Gémar 1995a;Harvey 2002;Šarčević 1997): 1) specificities of texttype and genre (Prieto Ramos 2014a), 2) functions of legal texts (Dullion 2007;Nord 1991); and 3) legal meaning being conveyed (Cornu 2005). All these aspects are closely related and "what matters most for legal translation is the characterization of groups of texts corresponding to specific varieties or styles of legal language, and this is generally a question of text producers and purposes in communicative situations" (Prieto Ramos 2014a: 263).
In the case of the transposition of the legal concept of genocide into 131 national jurisdictions, text-type and function are stable. At the national level, the concept of genocide is transposed in legislative texts (i.e. in a criminal code or an implementation law) whose function is instrumental (Nord 1991). Target texts are of the same type and share the same function as the source text. Given their stability in the context at hand, text-type and function have been discarded from our study as they do not provide insight into how and why the concept of genocide evolves. Instead, the focus is placed on legal meaning conveyed and contexts of implementation.
Šarčević points that for many, translation mainly consists in 'transcoding a message from one language into another, whereby the primary goal was to preserve the meaning of the message ' (1997: 55). In legal translation, the message is made up of the text, but also, of equal importance, of the intent (1997: 55) and because 'a text derives its meaning from one or more legal systems, legal translation is essentially a process of translating legal systems ' (1997: 229). If many strategies are readily available to transcode the text and the intent from one legal system to another, that historically span from a strictly literal approach to a co-drafting approach (1997: 24), '[n]o translation technique is a priori more adequate than another' (Prieto Ramos 2014b: 124). In a given context and for a given task, the adequate strategy will be determined through a thorough analysis of legal pragmatic considerations, including the communicative situation and macro-context (2014b: 122-23).
This study covers 131 domestic definitions of genocide. It is thus not realistic to perform a thorough analysis of all legal aspects involved in each context of implementation. Yet, in order to contribute to our understanding of the scope and breadth of contexts in which the definition of genocide has been transposed, two aspects have been identified in each domestic context: legal language and legal system.
It is generally accepted that the configuration of language and law has direct consequences on the degree of difficulty of translation (e.g. Gémar 2002b: 168;de Groot 1987: 798-800;Tabory 1980: 146;Wagner and Gémar 2015: 2). For instance, de Groot identifies five degrees of difficulty of legal translation based primarily on the extent of affinity between legal systems and secondarily on the extent of affinity between legal languages: 1) legal systems and legal languages closely related (easy); 2) legal systems closely related and few similarities between legal languages (somewhat easy); 3) within a uniform system (somewhat easy); 4) different legal systems and hardly related legal languages (difficult); 5) different legal systems and linguistically related legal languages (very difficult) (1987: 798-800). Another example is Gémar's classification of legal translation based on cultural, linguistic and legal aspects: 1) unilingual States with one legal system (e.g. Brazil, France, Mexico, Netherlands); 2) two or more legal languages with one legal system (e.g. Switzerland); 3) two or more legal languages and legal systems (e.g. Belgium, Canada, India) (2002: 168). All in all, the configuration of the law and the language has to be taken into account as it may condition the transformation of the legal meaning from the source text to target texts, particularly when the concept is the subject of universal transposition through multiple system-bound channels, rather than as part of a single institutional translation process of a unique instrument.

Source Text: International Legal Definition of Genocide
The legal concept of genocide has been chosen as the source text for this study for three main reasons: it is stable, it is relatively new and it is defined.
The term and the concept of genocide were coined by the Polish-Jewish lawyer Raphael Lemkin. In Axis Rule in Occupied Europe: Laws of Occupation, Analysis of Government, Proposals for Redress (1944), he proposed the following definition of genocide: "A co-ordinated plan of different actions aiming at the destruction of essential foundations of the life of national groups, with the aim of annihilating the groups themselves. The objective of such a plan would be disintegration of the political and social institutions, of culture, language, national feelings, religion, and the economic existence of national groups and the destruction of the personal security, liberty, health, dignity and even the lives of the individuals belonging to such groups. Genocide is directed against the national group as an entity, and the actions involved are directed against individuals, not in their individual capacity, but as members of the national group. " (1944: 79) Lemkin's definition served as one the foundations of the Convention on the Prevention and Punishment of the Crime of Genocide adopted in 1948, 3,4 which encompasses the first legal definition of genocide:

Article II
In the present Convention, genocide means any of the following acts committed with intent to destroy, in whole or in part, a national, ethnical, racial or religious group, as such:  Despite frequent calls for amendments (Schabas 2010: 536 -40), the international community has agreed upon the definition of genocide numerous times and over time. Indeed, the international community has reused, verbatim or with limited changes, the Genocide Convention definition in numerous treaties 5 , including 50 years later in the Rome Statute of the International Criminal Court 6 . Given its stability at the international level, it can be expected that the concept of genocide will suffer limited shifts when translated into national legal languages and legal systems.
Ever since the adoption of the Genocide Convention, 131 States have integrated the definition of genocide into their criminal system in order to exercise jurisdiction over such crime, and it has been applied in more or less 40 cases around the world (Rikhof 2009: 26-38). Hence, given its relative novelty and limited application thus far, it is still feasible to explore the meaning of genocide in all its national forms and in its entire scope, from its genesis to its finality.
The fact that the concept of genocide has been defined is another important factor. Besides providing textual boundaries in which to observe shifts (Chesterman 2005: 26) legal definitions 'promote clarity by reducing indeterminacy and help achieve consistency' (Šarčević 1997: 153). As far as their translation is concerned, legal definitions leave little room for creativity (1997: 158). As for legal definition of international crimes, including genocide, they are regarded as 'hard rule' 7 that are universal in application (Van Sliedregt 2012:854) and scholars in the field of law recommend to 'align to, or even incorporate, the (exact) definition of international crimes ' (2012: 849-50).
The assumption is that if this stable, new and well-defined concept is no longer universal in its domestic forms, a much greater transformation of other international concepts transposed at the national level could be inferred by extrapolation. In sum, the Genocide Convention contains the original meaning of genocide as agreed upon by the international community. It is the source text against which domestic definitions of genocide are being compared.

Target Texts: Domestic Definitions of Genocide
In 2012, Amnesty International (AI) reported to the Sixth (Legal) Committee of the United Nations (UN) General Assembly that 118 UN Member States (out of the 193) had implemented the definition of genocide into their national criminal law (2012: 13). In order to have an accurate picture of the current state of implementation of the definition of genocide, we reviewed the criminal law of the 75 UN Member States that had not transposed the definition in 2012. As of November 2018, a total of 131 States have criminalised genocide. Below is a list of those States. 7 As opposed to soft rules, forum-specific norms and national rules: "The first tier of ICL [international criminal law] contains 'hard' rules that are universal in application, e.g., crime definitions. The second tier relates to so-called forumspecific norms, e.g., distinct rules of procedure and evidence. The third category of ICL concerns the general part of domestic law where ICL, rather than imposing a single uniform approach, allows and constrains a 'margin of state discretion to apply local law to the prosecution of ICL offences'. The fourth tier consists of 'default ICL' in case there is no appropriate or available domestic law to apply. […] Universality is found in the core of universally binding law." (Van Sliedregt 2012: 854) of a language criterion. It is justified by practical reason, but also concealed strategic motives.
The method calls for the investigation of the concept in its natural languages 10 . It is not realistic to expect sufficient knowledge of the 76 languages in which the definition of genocide has potentially been translated into. Therefore, the corpus of national definitions comprises only legal definitions translated into official languages in which we are proficient: English, French and Spanish.
Those 3 languages are prevalent in the overall state of implementation of the definition of genocide. Amongst the 176 potential definitions of genocide in force in the world today, 11 almost half of them (42.6%) are written in English, in French or in Spanish. As shown in the figure below, they are the only 3 languages with a high occurrence level in the overall state of implementation of the definition of genocide. [https://www.admin.ch/gov/en/start/federal-law/classifiedcompilation.html, accessed 13 February 2019]) has been discarded given that they have no legal force and can not be interpreted by national courts. 11 Assuming that all unilingual States have one legal definition of genocide, that all bilingual States have two legal definitions of genocide, and so on. However, as seen for Mauritius and Seychelles (infra. 6), it is possible that States do not use all of their official languages in their legislative activities, hence the expression 'potential definitions of genocide in force in the world today'. Not only are those languages salient, but they are also highly visible. They rank on the top 3 of the global Calvet Barometer and amongst the top 6 on both scales of source language and target language for translation 12 . Regarded as lingua franca in legal activities, they also have an historical value (Mattila 2012: 38). English and French are the official working languages of numerous international organisations (2012: 40), and are sometimes considered to be 'more equal' than other languages (2012: 44). As for Spanish, it is an official language of most intergovernmental institutions and it accounts for an extended linguistic community (2012: 52).
These 3 languages provide a basis for comparative analysis within the 'same' language. Indeed, the source text has been adopted in all UN languages, including English, French and Spanish. The transposition process qualifies as intralingual (Jakobson 1959: 233), which corresponds to the highest degree of difficulty on de Groot's scale (different legal systems and linguistically related legal languages [very difficult]) (1987: 800).
Despite their similarities or common origin (Mattila 12  2012: 177), legal languages are rarely identical. Legal French of France is not identical to that of Canada or Burkina Faso. The same scenario applies to legal systems. If it is true that former or actual territorial dependencies have adopted, to some extent, the legal system of settlers, they have also retained part of their customary law and intertwined them in such ways that the result is inextricable (Scassa 1997: 251-54). It also holds for any movement of the law across boundaries, for instance the German Civil Code adopted by China, Greece and Japan; the literal application of the Swiss Civil Code in Turkey; the adoption of the Code de Napoléon in all conquered territories, etc. (Öner and Banu Karadağ 2016: 336, note 4). As law is a social phenomenon (Šarčević 1997: 13), it is bound to grow apart from its origin (Gémar 1995b: 31) and to acquire its 'own life' (Sirois 2000: 538). As a result, legal languages and systems are unique and scarcely sustain no harm when crossing linguistic and legal boundaries (Gémar 2008: 327).
In sum, the paradigm at hand potentially accounts for 131 unique configurations of language and law, which is ultimately more important in the determination of the representativeness of the corpus than any other factor. A lack of diversity would only provide insights into how the concept of genocide has shifted when transposed into a few contexts (e.g. group of civil law or unilingual States), whereas highly diverse contexts of implementation provides a better picture of the extent to which it might evolve when transposed worldwide. In determining the representativeness of the corpus, the emphasis has therefore not been so much on the need to represent each and all linguistic and legal systems accurately, but rather on the need to include the broadest range of communicative situations.
All in all, the language criterion is strategic in the sense that it focuses the analysis on high occurrence languages with high visibility whilst providing a basis for intralingual comparison in a very diverse set of contexts. And ultimately, that is why the limitation of three languages is not so much an obstacle after all.

Diversity of Language and Law Configurations
The diversity of the corpus in terms of legal and linguistic configurations covered is illustrated in the figure below. 14 Percentages have been calculated per level (per circle) and rounded so that there are slight discrepancies from one level to the next (e.g. in the third circle, 9 States CIV/CUST+1L equals 12.9% and in the fourth circle, 9 States of CIV/CUST+1L/FR = 12.7%). In the middle circle, the corpus of 71 States is divided into 2 groups: monosystemic States and plurisystemic States. There are slightly more monosystemic States (57.7%) than plurisystemic States (42.3%).
The second circle further divides the primary groups by type of legal system. On the right side, monosystemic States are divided in two groups: the CIV tradition with 23 States (32.4%) and the CL tradition with 18 States (25.4%). On the plurisystemic side, a first section clusters States that have a mixed legal system influenced in part by CIV. They primarily belong to the CIV/CUST group (15.5%). A second section accounts for States with a mixed legal system influenced in part by CL, including 5 States with a CL/CUST system (7%). The last plurisystemic section accounts for States with a mixed legal system influenced by both CL and CIV traditions: 8 out of 10 States of that group have a CL/CIV system (11.3%). In sum, there are 3 major trends related to legal systems: 1) CIV (32.4%); 2) CL (25.4%); 3) CIV/CUST (15.5%).
The figure shows that, despite being limited to 3 languages, the corpus provides 3 clusters of States with similar linguistic and legal configurations, as well as 18 less frequent or 'marginal' configurations, including 14 configurations limited to 1 or 2 States.

Identification of Shifts in Target Texts
To explore the transposition and translation process and its effects, shifts have been identified in target texts, along with their effect on the meaning (broader, constraint or no effect on meaning). Two examples of domestic definitions are provided below. Objects of shifts are underlined, and are followed by a label in-between brackets identifying its effect and classification as follows: + (broader meaning),  (constraint meaning) and  (no effect on meaning) 15 , LEG (conceptual shift), LEX (lexical shift), SYN (syntactic shift) and STY (stylistic shift). Numbers following each label refer to 15 Effects are determined in the context of the domestic text and legislation. For instance, in the Lesothan example (1), the third shift (genocide  offence of genocide) has no effect on the global meaning of genocide. On the contrary, the seventh shift (act  act or omission) expands the concept because in the original definition (or other dispositions of the text in which it is defined) does not provide explicitly for the possibility of genocide to be committed by omission.

Trends: Object and Effect of Shifts in Translation
A total of 1,021 shifts and their effect have been identified in the corpus of 75 definitions. Such data provide insight into diverse phenomena, including translation trends, object of shifts and movement of the meaning. In other words, they show how the concept has been modified, what has been modified (the form or the substance) and the effect of those modifications on the meaning (broader, constraint or stable meaning).
For example, Figure 3 shows the distribution of shifts by object and effect. Amongst the 1,021 shifts, 21% (214) are LEG shifts and 79% are LING shifts, including LEX, SYN and STY shifts. For every LEG shift, there are more than 3 LING shifts. LING shifts are further divided by components: 20% of the 1,021 shifts are related to LEX, 32.5% to SYN and 26.5% to STY. The proportion of LEG shifts (21%) is similar to that of LEX shifts (20%) and overall there are slightly more SYN (32.5%) and STY (26.5%) shifts.
In the outskirt circle, we distributed shifts by effect. Almost half of all shifts have no effect (all 'Null' sections) on the meaning (44.3%). Shifts with constraining effect (all '-' sections) cover 33.4% of all shifts and the remaining 22.3% broaden the meaning (all '+' sections).
Most LEG shifts have a broadening effect (12.4%) and very few of them have no effect (3.2%). On the opposite side, most SYN and STY shifts have no effect (16.5% and 15.9% respectively) and fewer have an expanding effect (2.4% and 2.2% respectively). 16 The following abbreviations are used in this figure and the followings: LEG (legal shift), LING (linguistic shift), LEX (lexical shift), SYN (syntactic shift) and STY (stylistic shift), + (broadening effect), -(constraining effect) and Null (no effect). As for LEX shifts, there are slightly more shifts without effect, but the distribution is more equal: 8.7% without effect, 5.9% with a constraining effect and 5.4% with a broadening effect.

Trends: Object and Effect of Shifts by Context
In the next figures, data on shifts and effect observed in the 3 dominant groups of States (as reported in Figure 2, that is: CIV+1L/SP; CL+1L/EN; CIV/CUST+1L/FR) are compared against the overall distribution of shifts by object and effect on meaning (Figure 3). Generally speaking, this group replicates trends identified in the overall account. Perhaps to some extent, this group of 17 States defines those general trends. Indeed, the only significant discrepancy is related to STY shifts. In the overall account, most STY shifts had no effect on the meaning (15.9%), whereas in trend 1, STY mostly have a constraining effect (11.2%).
What is more surprising thus is the number of shifts in trend 1. Indeed, this group represents 22.6% of the corpus (17 definitions out of 75), yet it covers 49% of all shifts identified (500 shifts out of 1,021). There is an average 29 shifts per definition, compared to 13.6 for the base of comparison. Given this high number of shifts, we conclude that this group of States has largely and for the most part adapted the definition of genocide 17 . This second group of States represents 17.3% of the corpus (13 definitions out of 75). We identified 87 shifts 18 for an average of 6.7 shifts per definition. It is a significant drop from the average number of shifts per definition set at 13.6 for the base of comparison and that of the previous trend (29). This group follows more or less the first level of division (LEG and LING), but in the LING breakdown, we observe fewer SYN shifts (-10.7 points) and more LEX and STY shifts (+5.3 and +9.1 points, respectively).
At the effect level (external circle), it is worth mentioning that 60.9% of all shifts have no effect (all 'Null' sections). They are dominant in all groups of shifts, except in the SYN group, where there is a tied with shifts with constraining effect (10.3%). Additionally, it is worth noting that most LEG shifts have no effect 17 With the exception of Argentinatwo shifts (Ley 26.200, Corte Penal Internacional, 13 de diciembre 2006, art. 2)  Overall, we could conclude that this group has integrated a fair number of shifts, but that a vast majority do not have any effect on the meaning. In this last analysis, we identified 132 shifts in 10 definitions 20 . It is an average of 13.2 shifts per definition, which is similar to the overall average (13.6). At the first level (LEG and LING), we observe an increased number of LEG shifts (+6.5 points), which translated primarily by an increased number of LEG shifts with constraining effect (+4.1 points). On the LING side, there are fewer LEX shifts (-7.1 points), slightly fewer STY shifts (-3 points) and more SYN shifts (+4.6 points).
On the effect side, 42.5% of all shifts have no effect (all 'Null' sections), 32.5% have a constraining effect (all '-' sections) and 24.9% have an expanding effect (all '+' sections), which corresponds to the base of comparison. Nonetheless, we identified slightly fewer SYN shifts with a constraining effect (-2.2 points), 19 There are nine States, but ten definitions because of Burkina Faso (see note 13). 20 Two of the nine definitions cover 31 and 26 shifts respectively: Loi N  fewer LEX shifts without effect (-4.9 points), much less STY shifts with a constraining effect (-6.2 points) and an increased number of STY shifts without effect (+4.2 points). Overall, this group stands somewhere in between a literal approach (i.e. fewer LEX shifts) and an adaptation approach, including a significant increase of LEG shifts with constraining effect.

Concluding Remarks
As highlighted by the analysis of the preliminary results of our study on the transposition of the legal concept of genocide into 131 national jurisdictions, the recommendation for an 'exact' integration of the crime of genocide ('exact' alignment [Van Sliedregt 2012: 849 -50]) is by no means the rule. Preliminary results demonstrate the scope and complex dimensions of the large-scale self-reliant transposition and translation process of the concept of genocide into domestic legal languages and systems and its effects on the global concept of genocide. By coupling shifts, effects and communicative situations, translational trends will be further analysed, taking into account specific political factors that may influence translational decisions, and the question of how judges will interpret domestic definitions and deal with discrepancies. Only then will we be able to know with more precision how, why and to which extent States have collectively transformed the meaning of genocide, which they have in principle all agreed upon.