• A. Introduction

  Enacting legislation is not a simple task. Intricacies become double-fold in jurisdictions where legislation is enacted in multiple languages. It gives rise to complicated questions of statutory interpretation, which is amplified by discrepancies between various versions of a piece of legislation.¹xD.L. Revell, ‘Authoring Bilingual Laws: The Importance of Process’, Brook. J. Int’l L., Vol. 29, 2003-2004, p. 1085 <http://heinonline.org>.
  To interpret legislation in general is to assign the meaning to a word, phrase or legislative sentence where there is a possibility of having two or more meaning, to decide between them or to declare the definite meaning. It is the process of taking decision on the basis of applicability or non-applicability of a piece of legislation to a particular situation.²xAttorney General’s Department Report, Symposium on Statutory Interpretation, Australian Government Publishing Service, Canberra, 1983, p. 96. Judges and practicing lawyers are most specifically involved in the interpretation of legislation.³xE. Alcaraz & B. Hughes, Legal Translation Explained, St. Jerome Publishing, Manchester, 2007, p. 24. A part from judges and lawyers, statutory interpretation directly or indirectly involves all those who are affected by the legislation.⁴xAttorney General’s Department Report, 1983, p. 96. Interpreting multilingual legislation involves matching the closest possible linguistic equivalent in all the versions of a piece of legislation.
  The problem arises because words change or loose meaning in the process of translation. As Alcaraz states, “Words strain, crack and sometimes break under the burden, under the tension, slip, slide perish, decay with imprecision, will stay in place, will not stay still.”⁵xAlcaraz & Hudges, 2007. In one jurisdiction, only one version may be authentic, while others are merely official translations. In another jurisdiction, all versions of the legislation may be authentic. The rationale of more than one authentic language is brought about by the fact that in some jurisdictions there is a presumption that language texts are identical in meaning.⁶xC.B. Kuner, ‘The Interpretation of Multilingual Treaties: Comparison of Texts Versus the Presumption of Similar Meaning’, ICLQ, Vol. 40, No. 4, 1991, pp. 953-964 <http:login.westlaw.co.uk.catalogue.ulrls.lon.ac.uk>.
  The approach of interpreting legislation expressed in two or more not-so-compatible versions is more complex than the approach of a unilingual construed legislation.⁷xM. Beaupre, Interpreting Bilingual Legislation, Carswell, Toronto, 1986, p. 4. In multilingual drafting environment, problems stem from the tendency to use the same word in several languages even when the meaning of the word significantly differs from one language to another.⁸xD. Shelton, ‘The Interpretation of Multilingual Treaties’, Hastings Int’l & Comp. L. Rev, Vol. 20, 1996-1997, pp. 611, 620 <http://heinonline.org>. Other challenges in the multilingualism include the fact that when interpreting the legislation, there should be no distinction between words in the text of the statute and the words in the legislative history.⁹xA. Scalia, A Matter of Interpretation, Federal Courts and the Law, Princeton University Press, Princeton, NJ, 1970, p. 31 <http://heinonline.org>.
  Since translators translate legal ideas from one language to another, interpreters of multilingual legislation are bound to consider that knowledge of one version alone is insufficient point of reference to the idea in question. They ought to grasp the meaning of words or phrases at both textual and contextual levels of each of the versions.¹⁰xR.A. Macdonald, ‘Legal Bilingualism’, McGill L. J., Vol. 42, 1996-1997, pp. 126, 160 <http://heinonline.org>.
  Looking back in the early years, the drafting of laws in Rwanda was done in French and then translated to Kinyarwanda (Kinyarwanda is the local language spoken by most of the Rwandans), and thus the legislation was composed of two versions, namely, Kinyarwanda and French. From 2003, the Government bills are drafted by lawyers of Legislative Drafting Unit in the Ministry of Justice. The Legislative Drafting Unit is composed of legislative drafters and translators.¹¹xV. Nsanze, ‘Challenges of Drafting Laws in One Language and Translating Them: Rwanda’s Experience’, The Loophole, No. 1, January 2012, p. 46. While the drafters draft laws, the translators translate them, and they have the duty to ensure harmony in all the three language versions. Translators of the Unit are lawyers and have to verify drafting issues as well.¹²xIbid. Various public offices also have legislative drafters, and their draft bills are sent to the Legislative Drafting Unit in the Ministry of Justice for polishing up all the drafting versions.¹³xNsanze, 2012. The bill is then forwarded to the Parliament in the three language versions aforesaid.¹⁴xOrganic Law No. 06/2006 of 15 February 2006 Establishing Internal Rules of Procedure of the Chamber of Deputies in the Parliament, Art. 107.
  The Rwandan Constitution provides that the official languages are Kinyarwanda French and English.¹⁵xConstitution of the Republic of Rwanda of 4 June 2003 as amended to date, Art. 5. The Constitution provides that each law shall be considered and adopted in Kinyarwanda or in the language of preparation in respect of any of the official languages. In case of conflict between the three official languages, the prevailing language shall be the language in which the law was adopted.¹⁶xIbid., Art. 93. From this context, it would be easier for anyone to assume that the language version in which legislation is adopted is the one that is authentic. There is a ministerial ordinance, however, which provides that all draft bills should be submitted to the cabinet for approval in three official languages.¹⁷xInstructions of the Minister of Justice No. 01/11 of 14 November 2006 relating to the Drafting of the Texts of Laws, Art. 3. Irrespective of the fact that the draft bill may have been conceived in one of the three official languages, all the three versions have equal value.¹⁸xIbid., Art. 4. The authentic interpretation of Laws in Rwanda is done jointly by both Chambers of the Parliament. The Supreme Court has to give its opinion on the issue that needs interpretation. Authentic Interpretation may be requested by a Member of Parliament, the Government or by the private practice association known as the ‘Bar Association’. Any other interested person may request for authentic interpretation through a Member of Parliament or the Bar Association.¹⁹xConstitution of the Republic of Rwanda, 2003, Art. 96. This trend of statutory interpretation is quite rare. In most countries, statutory interpretation is the prerogative of the courts.²⁰xM. Arden, ‘The Impact of Judicial Interpretation on Legislative Drafting’, The Loophole, 2008, p. 5.
  The success of an authenticated translation depends on its interpretation and application in practice. It is important to foresee the interaction between translation and interpretation of legislation. Unlike other jurisdictions that have interpretation Act, Rwandan legislature has not attempted to make any rules regarding the approach to the interpretation of multilingual legislation. Only one provision in the Constitution talks about the issue. This lacuna is nevertheless lessened by the fact that once interpretive rules are set, the judicial perspective may be impeded by the naive rules set to resolve conflicts between the equally authentic versions.²¹xBeaupre, 1986, at 161.
  Beside multilingual natural complexities and those inherent in the clauses regulating interpretation, statutory interpretation in Rwanda is further complicated by blending Civil Law and Common Law. Assuming that a bill is drafted in English, English and Common Law are so much tied together that the terminology used exclusively belongs to Common Law. The draft bill is then translated to French and Kinyarwanda. French relates a lot to Civil Law to an extent that a text as finally produced belong to civil law of France.²²xD.L. Revell, ‘Bilingual Legislation. The Ontario Experience’, Statute Law Review, Vol. 19, No. 1, 1998, pp. 32, 38. For the lack of appropriate terminology related to English or French, the Kinyarwanda version is in most instances translated by a couple of words to match up with the meanings of words in English or French. Lack of appropriate terminology is also prevalent between English and French texts. The challenge directly emanates from interpreting legislation, which is a blend the Civil Law and Common Law.²³xNsanze, 2012, at 50. As it is be discussed in the proceeding parts, the probability of inconsistencies in the three language versions is so profound that it would not be surprising if the entire piece of legislation would be subject repeal or amendment over the weight of inconsistencies.
  The aim of this paper is to prove that Ruth Sullivan’s theory on the factors of complexity of statutory interpretation in multilingual environment holds true in the case of Rwanda. The paper proves the hypothesis by applying the criteria set out by Ruth Sullivan in her article: ‘Challenges of Multilingual and Multijural Statutory Interpretation’.²⁴xR. Sullivan, ‘Challenges of Statutory Interpretation Multilingual Multijural Legislation’, Brook. J. Int’l L., Vol. 29, 2003-2004, pp. 993, 985, 1005-1018 <http://heinonline.org>. They are the Legal Status, Equal Authenticity, The Shared Meaning Rule and Application of the Shared meaning Rule. It suffices to note that there is no standard way of classifying the rules of statutory interpretation nor is there a standard way of ranking them.²⁵xR. Sullivan, Statutory Interpretation, 2nd edn, Irwin Law, Toronto, 2007, p. 42. I have chosen to apply these criteria because they were developed primarily in relation to interpreting legislation in multilingual jurisdictions. Rwanda is one of the jurisdictions that use multilingual system of drafting bills and consequently publish legislation in three official languages. Sullivan used these criteria to show the complexities of interpreting multilingual legislation with special reference to Canada.
  Although Sullivan’s theory is about multilingual drafting environment, she has specifically used examples of Canada where drafting is bilingual and bijural. For the purpose of this article, multilingual drafting environment includes examples of bilingual drafting jurisdictions.
  The paper is introduced by giving the broad problems that surround the process of multilingual interpretation. It introduces how statutory interpretation comes into play and who does what it in the due process. The paper then takes a quick look into the drafting history of Rwanda and the procedure of statutory interpretation. In order to fully understand the process by which bilingual legislation is authored, it is important to comprehend the larger picture of the entire legislative process in Rwanda.
  Section B explains about the ‘Legal Status’. It is further subdivided into two subsections. Section I examines relation between official and authentic versions. It seeks to analyse whether the official languages are consequently authentic versions of the legislation. Section II goes farther to establish the relationship between translation and legislated multilingualism. The fact that legislation is drafted in one language and translated into the other two languages in Rwanda gives rise to the question as to whether translated versions are legislated or not. Section C is about ‘Equal Authenticity’. This section is subdivided into two subsections. Section I analyses ambiguity in one version or two versions while other versions are clear citing some examples of the Rwandan legislation. Section II analyses the situation when the versions are clearly in conflict with an example from a piece of legislation. Section D deals with the ‘Shared Meaning Rule’. It is further subdivided into two. Section I examines with instances of provisions of legislation the ambiguity shared meaning and how the interpreter ought to resolve it. And then Section 3.2 seeks to analyse the breadth versus narrow shared meaning. Section E is about the application of the shared meaning rule. This section analyses how the shared meaning rule is applied and the associated difficulties. It is followed by the Conclusion, indicating the gaps and overlaps, thereby showing the inherent complexities that give way to challenges of interpreting multi-bilingual legislation with particular reference to Rwanda.
  So far, there has been no court case arising out of translation discrepancies of different versions of a piece of legislation. Nonetheless, as time goes, litigants become aware that they would get some advantages out of the discrepancies. It is thus a fact that must not be simply ignored, as there may be cases of alleged discrepancies in the three language versions of legislation. For lack of decided cases on the issues of divergence of language versions, in as far as this paper goes; it does not make any reference to any case decided by Rwandan courts.

• B. Legal Status

  Whatever the legal status it takes in the multilingual perspective, it is appreciated that readers of the law refrain from the presuming that language precedes law and that language fully captures law.²⁶xMacdonald, 1996-1997, at 128. Readers of law authored in more than one language versions ought not to be transfixed by language or even examples drawn from certain language expressions.²⁷xIbid., p. 131.

  I. Relation between Official and Authenticity of Language Versions

  While interpreting a multilingual legislation, it is crucial to establish legal status in terms of the official language and authenticity of language versions. The regulatory provisions that are contained the Constitution of the Republic of Rwanda merely refines the important issue of the status of the official languages compared with the authentic languages. This concept generates debate of interest in the Rwanda case. Article 5 of the Constitution provides that there are three official languages²⁸xConstitution of the Republic of Rwanda, 2003, Art. 5.; however, it does not expressly recognize their equal authenticity. In fact Article 5 of the Constitution per see does not distinguish between strong and weak multilingualism. That is to say, strong multilingualism where all official language versions of a law are equally authentic and weak multilingualism as where only one language version of legislation is authentic, the rest being official translations.²⁹xT. Schilling, ‘Multilingualism and Multijuralism: Assets of EU Legislation and Adjudication?’, German L.J., Vol. 12, 2011, pp. 1462, 1463 <http://heinonline.org>. The distinction is even more blurred since on the other hand, on the procedure of simultaneous interpretation, Article 93 of the same Constitution provides that in case of a conflict between versions, the language in which legislation was adopted prevails.³⁰xConstitution of the Republic of Rwanda, 2003, Art. 93.
  Reading from Article 93, the authenticity of all the three versions is in question. The article on the face value suggests that interpretation must respect the version in which legislation is adopted. It seems to automatically deprive the official versions of their authoritative value before even the context of their provision is analysed.³¹xBeaupre, 1986, at 159. One cannot assert that all languages are official (equally authentic in other words) and then suggest that one of the versions shall prevail in case of conflict.³²xIbid. In some jurisdictions translation to several language versions may be for convenience and use with no legal force.³³xSullivan, 2003-2004, at 1005. Having one language prevail over others, which presumably serves to prevent inconsistency in the interpretation, would not be a true multilingualism as the versions in other languages may only exist as a reference rather than authentic.³⁴xRevell, 1998, at 39.
  The issue of Rwanda is unsettled because legislation may be adopted in a language which is not the one in which it was drafted. For example, the law on gaming activities was drafted in English, translated into French and Kinyarwanda, considered and adopted in Kinyarwanda.³⁵xArticle 40 of Law No. 58/2011 of 31 December 2011 Governing the Gaming Activities in Rwanda. There are two presumptions from the onset; one is that the language of conception is the authentic and the other that the language in which it was adopted is the authentic. Ruth highlights that others may be official subject to interpretation rule that gives paramount to one or more other languages.³⁶xSullivan, 2007, at 106. To a certain extent, it may be assumed that for the purpose of interpretation, official languages stand equivalent to the equal authenticity of versions and as stated in Article 5 of the ministerial instructions.³⁷xMinister’s Instructions, supra note 17, Art. 5. The analogy must be however treated with maximum precaution considering that equal authenticity is a principle by its own and it cannot be imposed by instructions.³⁸xS. Rosenne, ‘On Multi-Lingual Interpretation’, Isr. L. Rev., Vol. 6, 1971, pp. 357, 361 <http://heinonline.org>. Equal authenticity of the different versions of the different language versions should be ratified at the constitutional level. Existence of Articles 5 and 93 of the Constitution creates opportunity of forum shopping, where one litigant would argue that all languages are equally authentic on basis of Article 5 of the Constitution. This is arguably an indication that the legislator intended to give equal authenticity of all versions of the legislation as they are enacted in Kinyarwanda, English and French. On the other hand, the other party may contest it on the basis of Article 93 and challenge the authenticity of other versions. At the constitutional level inter se, there is no established hierarchy as all provisions are of the same footing as to priority, deference or other solution. It would therefore be difficult to apply any rule to resolve the conflict based on Articles 5 and 93 of the Constitution.
  In the essence of multilingual interpretation, it is of great importance to recognize availability of the three versions and establish the textual and contextual meaning of words.³⁹xG.R. Sneath, ‘Bilingualism and Interpretation. Examples of Two Jurisdictions’. Statutory Law Review, Vol. 5, No.2, 1984, p. 49. In reality, courts normally consult one version, that is the one in which it conducts its business.⁴⁰xKuner, 1991, at 3. In the Rwandan context, courts most of the time consult Kinyarwanda version since many of the proceedings are held in Kinyarwanda. This may be dangerous because users are attempted not to consult other versions. Christopher B. Kuner explains that despite the New York Arbitration convention of 1958 containing several discrepancies between its five authentic texts, the review on the US leading cases that were construed on the basis of this convention do not make any mention on other versions other than the English one.⁴¹xIbid., p. 2. In the case of R. v. Compagne Immobiliere BCN Ltee, the Supreme Court of Canada disregarded Subsection 8(2) of the official languages act by describing it as a mere guide to interpretation because it collided with the constitutional clause on the equal authenticity.⁴²xBeaupre, 1986, at 160. In a nutshell, there is no express provision determining the status of different three language versions in as regards their authenticity.

  II. Relationship between Translation and Legislated Multilingualism

  Legislation in Rwanda is adopted by the Parliament in any of the three languages.⁴³xConstitution of the Republic of Rwanda, 2003, Art. 93. Is this an implication that the other two versions are legislated? Apparently, there is no clear provision which indicates whether all versions are legislated or not. Article 4 of the ministerial ordinance provides that translation must be done before a draft is submitted to the cabinet.⁴⁴xMinister’s Instructions, supra note 17, Art. 4. This is arguably an indication that all other versions that are enacted alongside with the language in which the legislation was adopted are equally legislated. This would seem to be a symbolic gesture. In other words, it implies that neither of the text takes precedence over the other, and any divergences between the three versions of the legislation are to be resolved according to rules of interpretation.
  The most important issue is not whether one version is a translation of the other, but whether all the versions are legislated and enacted by the Parliament.⁴⁵xSullivan, 2003-2004, at 1006. As far as most pieces of the legislation are concerned, this issue seems to have skipped the minds of drafters and legislators because the provision only refers to the language in which a given legislation was drafted and adopted. For example, Article 374 provides that “This Law was drafted in French, considered and adopted in Kinyarwanda.”⁴⁶xLaw No. 21/2012 of 14 June 2012 Relating to the Civil, Commercial, Labour and Administrative Procedure in Rwanda, Art. 374. Should it then be assumed that the English version is not legislated? Article 107 of the internal rules of Chamber of Deputies states that “Government bills sent to the Chamber of Deputies as well as their explanatory statements shall be written in the three (3) official languages recognized by the Constitution. The Bureau of the Chamber of Deputies distributes the bills to Deputies within seven (7) days from the day of their reception.”⁴⁷xOrganic Law No. 06/2006 of 15 February 2006, Supra note 14, Art. 107. And, Article 115 of the same internal rules provides that “the bill examined by the Committee is transmitted to Deputies in the three (3) languages recognized by the Constitution before the Plenary Sitting examines it.”⁴⁸xIbid., Art. 115.
  Apparently, most of the pieces of legislation do not make any mention of the language versions in which legislation is published. One can simply assume that all versions are legislated by the fact that they bear the signature of the promulgation figures. Some multilingual jurisdictions may provide for a version in which legislation is accented and published. For instance, in South Africa, statutes were alternatively signed in either English or Afrikaans, and signing of a particular version was matter of chance.⁴⁹xM. Loubser, ‘Linguistic Factors Into the Mix: The South African Experience of Language and the Law’, Tul. L. Rev., 2003-2004, Vol. 78, pp. 127, 128 <http://heinonline.org>. To resolve any conflict that would arise from such a situation, Section 35 of the 1983 South African Constitution,⁵⁰xThe Constitution of the Republic of South Africa of 1983 is abrogated by the Constitution of 1993. provided that “in instances of conflict between the English and Afrikaans versions of an Act, the copy signed by the State President (when he or she assented to the Act) prevailed.”⁵¹xLoubser, 2003-2004, at 128. This constitutional solution was not, however, practical and courts usually applied the mechanism of comparing all versions of the statute to clarify each other, invoking the constitutional provision as a result resort.⁵²xIbid., p. 129. In the absence of the clause to determine which version prevails, the signed version may not carry more weight because it was signed. Attempts are made and the texts are read together to establish the common denominator. To this far, since all the versions are regarded as complementing each other, all versions are then considered as legislated.⁵³xB. Bekink & C. Botha, ‘Aspects of Legislative Drafting: Some South African Realities (or Plain Language Is Not Always Plain Sailing)’, Statute Law Review, Vol. 28, 2007, p. 15 <http:login.westlaw.co.uk.catalogue.ulrls.lon.ac.uk>.
  Another example is in the Canadian legislation, in the case of King v. Dubois Justice Duff contended that “the states of the Parliament of Canada in their French version pass through the two houses of Parliament and receive the assent of His Majesty at the same time and according to the same procedure as those in the English version.”⁵⁴xBeaupre, 1986, at 18. In the same essence, all legislations bear the signature of accent on all the three language versions in Rwanda. When we apply the Canadian and South African precedents to Rwanda, the assumption is that all the three versions of legislation are legislated.

• C. Equal Authenticity

  When all versions of a piece of legislation are declared to be authentic, it means that the legislature recognizes all versions as accurate statements of the law and the people it is addressed to can safely rely on any of the versions.⁵⁵xP.-A. Côté, ‘Bilingual Interpretation of Enactments in Canada: Principles V. Practice’, Brook. J. Int’l L., Vol. 29, 2003-2004, pp. 1067, 1069 <http://heinonline.org>. The rule of equal authority carries symbolic significance. Theoretically, it ensures that none of the users of any language are to be considered as first class citizens while others are regarded as second-class citizens. On the practical side of multilingualism, all versions equally contribute to the meaning of any given provision.⁵⁶xIbid., p. 1069. The substantive effect of the rules is that all language versions of legislation are equally authentic and divergences in the language versions are not to be resolved by the predominance of one version over the other.⁵⁷xT. Scassa, ‘Language of Judgment and the Supreme Court of Canada’, UNBLJ, Vol. 43, 1994, pp. 178, 189 <http://heinonline.org>. The characteristic nature of equal authenticity is in most instances achieved by translating the version in which a piece of legislation was drafted into all languages and then declaring all translated versions to be authentic versions.⁵⁸xJ. Engberg, ‘Statutory Texts as Instances of Language(s): Consequences and Limitations on Interpretation’, Brook. J. Int’l L., Vol. 29, 2003-2004, pp. 1135, 1154 <http://heinonline.org>.
  In the Rwandan case like in other multilingual drafting jurisdiction, this is in line with keeping up with the rule of law so that people have access to the law they can read and understand in their own language or the language they feel at ease when reading.⁵⁹xJ. Hardy, ‘The Interpretation of Plurilingual Treaties by International Courts and Tribunals’, Brit. Y. B. Int’l L., Vol. 37, 1961, pp. 74, 75 <http://heinonline.org>. Presumably, everyone would be able to know the legal consequences of their conduct and will not be unfairly taken by surprise.⁶⁰xR. Sullivan, ‘Some Problems With the Shared Meaning Rule as Formulated in R. v. Daoust and the Law of Bilingual Interpretation’, Ottawa L. Rev., Vol. 42, 2010-2011, pp. 75, 76 <http://heinonline.org>. Despite the fact that legislation may contain special provisions, discrepancies still exist between the versions of legislation. It invites the question of attitude the interpreters are likely to adopt.

  I. Ambiguity in One Version while Other Versions Are Clear

  The Rwandan Constitution clearly regulated that the language in which legislation is adopted prevails over the other.⁶¹xConstitution of the Republic of Rwanda, 2003, Art. 93. We are yet to analyse what happens if the language in which it was adopted is ambiguous or vague. This position does not provide any solution as to what happens if the application of Article 93 does not remove the ambiguity or give a meaning that reconciles all versions in regard to the object and purpose of the legislation.⁶²xG. Schwarzenberger, ‘Myths and Realities of Treaty Interpretation Articles 27-29 of the Vienna Draft Convention on the Law of Treaties’, Va. J. Int’l L., Vol. 9, 1968-1969, pp. 3, 4 <http://heinonline.org>. It is argued that where the degree of divergence of multilingual legislation is inevitable, then it is not proper for the interpretation to rely on solely a single language version, it would be ideal to compare several of them.⁶³xKuner, 1991, at 3. When the text under consideration contains ambiguous provisions but the ambiguity can be resolved by referring to unambiguous words or expressions in the other text then the latter meaning can be adopted.⁶⁴xS. Yuen-Ching Fung, ‘Interpreting the Bilingual Legislation of Hong Kong’, Hong Kong L.J., Vol. 27, 1997, pp. 206, 218 <http://heinonline.org>. When the meaning of one version is ambiguous and the other meaning is plain, the plain meaning is adopted. In fact this as a matter of constitutional law where all versions are equal, how can an interpreter reject the meaning found in both versions for the one that is found in one.⁶⁵xSullivan, 2003-2004, at 1014. This is what an interpreter of legislation may resort to if other means of resolving ambiguity cannot reach a fine interpretation. The interpreter opts for a clear version for the ambiguous version.⁶⁶xCôté, 2003-2004, at 1071.
  Equal authenticity of language versions means that all versions of legislation are equally official and authoritative enactments of the Parliament.⁶⁷xSullivan, 2010-2011, at 76. By assuming that one of the multilingual versions is truly authentic and that others are mere translations or only for the purpose of reference, it simply defeats the principle of equal authenticity.⁶⁸xRevell, 2003-2004. The rule of equal authenticity means reconciling discrepancies among Kinyarwanda, English and French texts. It basically overrides the tendency of choosing to have one language version prevail over others.⁶⁹xConstitution of the Republic of Rwanda, 2003, Art. 93. Equal authenticity in other words is an attempt to finding a solution to the problem of accommodating all languages spoken by the communities within a single political entity. It provides members of the communities with direct access to the law.
  The challenge lies into this fact, where the clear version is considered to clarify the ambiguous version; it undermines the linguistic security, one of the key functions that the equal authenticity ought to serve.⁷⁰xSullivan, 2010-2011, at 82. Reconciling versions when one of them is clear while others are ambiguous is facilitated if the ambiguous text means the same thing as the version that expresses it clearly. That is to say, all versions read together, they point to the one conclusion. Here are some examples. “Any instrument for house breaking” and the French version is “un instrument pouvant servir aux effractions de maisons.”⁷¹xBeaupre, 1986, at 20. The French version makes it clear. In English version, an instrument for house breaking is one capable of being used for house breaking.⁷²xIbid. The main idea here is to read one version in light of another and reconcile the versions wherever possible. In cases of clear and ambiguous versions involving litigants, if versions are equally authentic and it is not possible to reconcile them, courts are likely, especially in penal cases, to resolve it by giving effect to the version in favour of the defendant (accused).⁷³xYuen-Ching Fung, 1997, at 207.
  Article 3 of the Rwandan criminal procedure⁷⁴xLaw No. 13/2004 of 17 May 2004 Relating to the Code of Criminal Procedure, O.G Special No. of 30 July 2004, Art. 3 <www.amategeko.net> is one of the examples of the ambiguity of version while another is clear. There is ambiguity in Kinyarwanda and English while French version seems to clear the ambiguity in other versions. Article 3 of the penal procedure, The English version reads, “A criminal action abates upon death of the offender, in case of prescription of offence, when there is amnesty, when a law is repealed or following a court’s final judgment on a particular offence. In case the law provides otherwise, the action can also be extinguished if the defendant accepts to pay a fine without trial or in case a complainant withdraws his or her claim,”⁷⁵xIbid.The French version can be translated as “a criminal action abates upon death of the defendant, in case of prescription of offence, when there is amnesty, when a law is repealed or following a court’s final judgment on a particular offence. In case the law provides otherwise, the action can also be extinguished if the defendant accepts to pay a fine without trial or in case a complainant withdraws his or her claim.”⁷⁶xCode of Penal Procedure of Rwanda, supra note 74, Art. 3. The Kinyarwanda translates as “a criminal action abates upon death of the offender, in case of prescription of offence, when there is amnesty, when a law is repealed or following a court’s final judgment on a particular offence. In case the law provides otherwise, the action can also be extinguished if the defendant accepts to pay a fine without trial or in case a complainant withdraws his or her claim.”⁷⁷xIbid.
  There is ambiguity in Kinyarwanda and English versions. English version cites offender and this term is a conceptual ambiguity. There is an indication that the conditions stated in the article would apply after conviction. Kinyarwanda version also refers to ‘uwakoze icyaha’, which translates to ‘offender’ in English. The French however refers to ‘prevenu’, which translates to the ‘defendant’ in English. The English and Kinyarwanda versions at some point use the words ‘offender’ and ‘defendant’ interchangeably. On the preceding paragraph of both versions, the person is referred to as defender, and it makes us wonder if the word defendant and offender carry the same meaning.
  So how does the concept of ambiguity in one text while another one is clear help the interpreter to resolve this issue? There are two scenarios that led to this error. The legislation was drafted in French, translated to Kinyarwanda and then English. The English version in this case inherited translation error from Kinyarwanda where the word ‘prevenu’ was translated from French to Kinyarwanda as ‘uwakoze icyaha’ and then to English as ‘offender’. The other possibility is that it was translated from French to English. A translation error was made at this stage and then transferred to Kinyarwanda version. What would occur in case of discovering an error or omission in the translated version? If the reader chose to read the French version, there would not be any problem of ambiguity, but since all the versions are equally authoritative, other version have to be read as well.
  In this case, it is suggested that the clear version must be in reasonable construction of the unclear one.⁷⁸xBeaupre, 1986, at 24. While the English and Kinyarwanda version refer to the person as the offender, the conditions contained in the two versions are similar to those in French. From the general principles of criminal law, a person can be referred to as offender only after conviction; Offender is defined as a person who has been found guilty of an offence and is liable to punishment pursuant to criminal law.⁷⁹xP. Butt & D. Hamer, Concise Australian Legal Dictionary, 4th edn, LexisNexis Butterworths, Australia, 2011, p. 411. It actually competes with the principle of innocence presumption. The English and Kinyarwanda version clearly conflicts with the Constitution out right. It is in other words ‘unconstitutional’ because one is presumed innocent until proven guilty.⁸⁰xConstitution of the Republic of Rwanda, 2003, Art. 19. It would then be eminent that conditions of death, amnesty, repeal and the final judgement cease to apply. In light of comparing the unclear version to the context, then the second paragraph would be enough to suggest that the word is the ‘defendant’ and not ‘offender’ as it is the word used in the next paragraph and therefore the French version applies. This is in other words referred to as relative ambiguity, that is to say where ambiguity is in relation to certain facts.⁸¹xF. Bennion, Bennion on Statutory Interpretation, 5th edn, LexisNexis, London, 2008, p. 447. In principle, the word offender would not be open to diverse meanings, but the determination of the application of those words to particular circumstances⁸²xIbid., p. 448. gives rise to difficulties M. Beaupre says that “where possible, justified, one must attempt to extract a mutually compatible rendering. If that is impossible, the context naturally rules the inevitable choice of the version to be preferred.”⁸³xBeaupre, 1986, at 26. When we apply Article 93 of the Constitution,⁸⁴xConstitution of the Republic of Rwanda, 2003, Art. 93. it would mean that the Kinyarwanda version prevails because the penal procedure was adopted in Kinyarwanda.⁸⁵xCode of Penal Procedure of Rwanda, supra note 74, Art. 275. The Kinyarwanda cannot help to resolve the conflict because it undermines French version, which is clear and carries the purpose of the provision as well. It is a true reflection of the failure of the constitutional provision of predominance of one language over another.

  II. When Versions Are Clearly in Conflict

  Rwanda’s Constitution provides for the case of conflict, but such solution may not be sufficient to solve language version divergence. From precedence, the Vienna convention adopted a solution to this issue. When a treaty has been authenticated in two or more languages, the text is equally authoritative in each language, unless the treaty provides or the parties agree that, in case of divergence, a particular text shall prevail.⁸⁶xB.J. Condon, ‘Lost in Translation: Plurilingual Interpretation of WTO Law’, Journal of International Dispute Settlement, 2010, p. 4 <http:login.westlaw.co.uk.catalogue.ulrls.lon.ac.uk>. Discrepancies in the meaning of all versions should be treated as any other ambiguity. In this situation, none of the rules of the multilingual legislation would be applicable.⁸⁷xCôté, 2003-2004, at 1070. Even though there could be three linguistic versions, the rule of interpreting them has to be based on particular facts. A choice has to be made in order to reach the relevant meaning of words used in the legislation.
  Any piece of legislation is drafted in a manner that aspires to have all versions mirror each other. In practice, the three versions are often open for different interpretations, a situation that can prove extremely problematic. It is not surprising to state that the complexity of multilingual statutory interpretation lies in the difference between or among the versions.⁸⁸xD.A. Wirth, ‘Multilingual Treaty Interpretation and the Case of Salt Ii’, Yale Stud. World Pub. Ord., Vol. 6, 1979-1780, pp. 429, 429 <http://heinonline.org>. The general principle is to analyse the difference between expressions of the three texts since each version is capable of giving a different meaning from other texts. Any version chosen for interpretation takes into account factors normally relevant in ascertaining the statutory meaning.⁸⁹xCôté, 2003-2004, at 1070. It is argued that the literal meaning in each of the three versions retains relevance and that the words used in all versions as interpreted will determine the semantic possibility of the provision.⁹⁰xIbid. Pierre André Côté underscores that “the ‘ordinary’ or ‘textual’ meaning, however, cannot be a factor in the selection of best interpretation because, in cases of divergence, both versions being of equal weight, counsel other out as it were, at least at the textual level.”⁹¹xCôté, 2003-2004, at 1070.
  It is possible that implicit and symbolic meanings of words in the version in which the legislation was originally drafted may be lost, ignored or compromised in translation to other texts. When equal authenticity is presupposed, it is challenging as to how the interpreter ought to react when other versions are a derivative translation of the other.⁹²xMacdonald, 1996-1997, at 148. The challenge is aggravated by the fact that each of the versions says things different from one another. The requirement for the equal authenticity is that all versions must equally express a single rule as intended by the drafter. When the versions express different but clear idea, from the drafting point of view, none of them is reliable. The basis other than the textual meaning has to be established in order to determine the version that is to be considered.⁹³xSullivan, 2010-2011, at 84. When all versions of the multilingual legislation express clear but different meaning, the state has made a mistake and all the audience of the legislation are subject to the risk of being taken by surprise. The duty of resolving the clear words in conflict is not an easy one, yet the interpreter has the daunting task to minimize the surprise.⁹⁴xIbid., p. 88. Under such circumstances, the ambiguity is not within the versions but is rather caused by the divergence of versions. It should be dealt with like any other question of ambiguity by seeking to ascertain the purpose of the provision.⁹⁵xBeaupre, 1986, at 22.
  Other thoughts suggest that such divergence should be treated by settled cannons of interpretation and construction. It is a presumption that the Parliament has spoken in more than one language with variance of meanings between versions. This presumption is however rebutted by the fact that it cannot apply when there is divergence between versions that are supposedly equally authentic. Beaupre explains that “Once the court determines that each text is clear but at odds with the other, we are no longer speaking of equal interpretation. The language in question requires no interpretation; the law needs only to be applied. And the impossibility of applying two clear texts that are at odds with one another is the essence of the problem which, for its solution may seem to require a legislative act on the part of the judiciary.”⁹⁶xIbid., p. 23. It is a general understanding that when the courts are called upon to interpret a statute, their main task is to discover the intention of the legislature. Whenever words used in legislation turn out to be clear and unambiguous, the interpreter should not be bothered of taking further steps to identify the legislative intent.⁹⁷xP.M. Perell, ‘Plain Meaning for Judges, Scholars and Practitioners’, Advoc. Q., Vol. 20, 1998, pp. 24, 27 <http://heinonline.org>.
  Arguments put forward for equal authenticity suggest that there is minimum impairment for linguistic equality. When resolving discrepancies between versions, it does not involve automatic preference of a version over another. All versions have equal opportunity of being picked up as the version that states well the legislative purpose.⁹⁸xSullivan, 2010-2011, at 77. One could simply apply the unilingual approach if there were no such discrepancies. But because they are present and unavoidable, the most prudent way is to consider all versions of the legislation.⁹⁹xCôté, 2003-2004, at 1069. Comparing all versions may facilitate interpretation since the presence of the three language versions increases the possibility that legislation may be interpreted in a way conforming to the legislative intent.¹⁰⁰xKuner, 1991, at 3.
  This attempt can, however, be realized if only all versions do state the same meaning, a condition that is realistically difficult to achieve.¹⁰¹xSullivan, 2010-2011, at 77. Discrepancies among these texts still have the potential to cause systematic problems.¹⁰²xCondon, 2010, at 2. The plurality of authentic texts of legislation always carries its own intricacies when it comes to its interpretation.¹⁰³xIbid., p. 4. Equal authenticity is to be attained where one version equally reflects one another. However, different languages cannot cover exactly the same semantic area. Realistically, it involves mapping a language into another and the outcome is usually closeness, not particularly identical in meaning.¹⁰⁴xScassa, 1994, at 179.
  One of the complexities of equal authenticity rule is that an interpreter cannot know the legislative intent by merely reading one version of the law. The interpreter has to consider all versions and resolve discrepancies, if any, before deciding on how to apply it.¹⁰⁵xSullivan, 2003-2004, at 1007. To the extent where one is forced to read both versions to understand the meaning, linguistic security is undermined. When a person reasonably relies on a version that turns out to be inaccurate statement of law, fairness is undermined.¹⁰⁶xSullivan, 2010-2011, at 77. Example of when language versions are clearly in conflict is Article 198 of the penal code that defines marital rape.¹⁰⁷xOrganic Law No. 01/2012/OL of 2 May 2012 Instituting the Penal Code of Rwanda, Art. 198.
  English version reads, “Marital rape is any act of sexual intercourse committed by one spouse on the other by violence, threat or trickery.”¹⁰⁸xIbid. The Kinyarwanda version can be textually translated as, marital rape is any act aimed at sexual intercourse committed by one of the spouses without consent of the other by using force, threat or trickery.¹⁰⁹xPenal Code of Rwanda, supra note 107, Art. 198. French version can be translated as “marital rape is any act aimed at sexual intercourse committed by one of the spouses against another without consent of the other by using force, threats or trickery.”¹¹⁰xIbid.
  The Kinyarwanda and French versions refer to ‘any act aimed at sexual intercourse’ the English version refers to ‘any act of sexual intercourse’. The English and Kinyarwanda or French versions do not correspond and the proposal to change any of the versions to conform to others would not bear any result that responds to resolving the divergence in meaning that is clearly expressed. On a general note, the drafter/translator did less to avoid the divergence in meaning and all versions ended up expressing different concepts.
  The major concern is that the Kinyarwanda and French version texts appear to be clear because they refer to acts aimed at sexual intercourse. The English is also clear but the text has its own meaning. If there is any ambiguity, it is because of the divergences in two versions, Kinyarwanda together with French against the English version, which clearly expresses different meanings. We have multi-faceted definition that is, versions that clearly state different forms of conduct, yet all versions are equally authoritative versions. Since the problems of interpreting legislation would not lie in the difficult to determine the meaning, but rather on how to implement the intended meaning of a provision,¹¹¹xR. Bigwood (Ed.), The Statute, Making and Meaning, LexisNexis, Wellington, 2004, p. 130. implementing the meaning of the given definition in divergent meanings is the main challenge.
  Although the definitions of terms are not to be limited to an equivalency of AB means A and B, it has to be taken into account even when it does fully be fit into the context.¹¹²xD.C. Pearce & R.S. Geddes, Statutory Interpretation in Australia, 6th edn, LexisNexis, Australia, 2006, p. 238.The absurd situation of unstable definition this article is aggravated by conflicting divergence of all the versions in an attempt to define marital rape. The problem arises because two versions compared with one version prescribe clearly contrasting forms of conduct. Here we have to examine a fact of two aspects conveyed differently by two language versions.
  When we apply the Kinyarwanda version because the initial drafting was done in Kinyarwanda as per Article 764¹¹³xPenal Code of Rwanda, supra note 107, Art. 764. of the same code, the problem may be solved. But would it mean that the English version is undermined. The English version of is as well clear and the pure intention of the drafter be underplayed by this decision. We must pay attention to the fact that the Parliament has spoken in three languages with variance in meanings in which case all versions are to be considered as authentic versions of the legislation. Here lies the dilemma of interpreting multilingual legislation. Sometimes, definition of terms can create more problems than they solve. Interpreting from the internal context requires that meaning is determined from the ordinary linguistic usage including any special technical meaning and the purpose for which a piece of legislation is passed.¹¹⁴xJ. Bell & G. Engle, Cross Statutory Interpretation, 3rd edn, Butterworths, London, 1995, p. 50. The legal certainty in this case would not be achieved by application of Article 93 of the Constitution. Elimination of linguistic discrepancies by the way of interpreting may, in certain circumstances, run counter to the concern for legal certainty.¹¹⁵xJ. Vaiciukaiti & T. Klimas, ‘Interpretation of European Union Multilingual Law’, Int’l J. Baltic L., Vol. 2, 2005, pp. 1, 3 <http://heinonline.org>. It is therefore preferable to explore the possibilities of solving the ambiguity at issue rather than employing the principle of one version superseding others. A consideration of purpose and context would help to settle the divergence between the versions.

• D. The Shared Meaning Rule

  Multilingual interpretation presupposes that the method of reading and interpreting clauses necessitates that interpreter takes all language versions into account and assign the same meaning to them all. It is presumed that the drafter intended the shared meaning.¹¹⁶xSullivan, 2007, at 43. Absence of this approach results into de facto legal dualism, by pretending that legislation can be understood by referring to only one of the other versions of the legislation.¹¹⁷xRevell, 2003-2004, at 1099. It cannot be claimed to be a correct way of interpreting legislation if the other part of the legislation is ignored. Multilingual legislation requires multilingual interpretation. It means that an interpretation which puts into consideration all versions of a piece of legislation.¹¹⁸xCôté, 2003-2004, at 1069. In Rwandan jurisdiction, those include Kinyarwanda, English and French versions.

  I. Ambiguity Shared Meaning

  This is termed as ambiguity shared meaning because one version lends itself to two or more possible meanings.¹¹⁹xSullivan, 2010-2011, at 78. Certainly, most of the approaches of statutory interpretation revolve around the problem of having the words of legislation capable of delivering more than one meaning.¹²⁰xPerell, 1998, at 58. Consequently, multiple approaches are used to determine the meaning of the ambiguous word or expression within the provision of the legislation.
  Shared ambiguity in the multilingual environment is more complex because almost every word is used in various senses and thus has more than one meaning. Any of the versions in multilingual environment can have different significations capable of multiple interpretations.¹²¹xR. Dickerson, ‘The Diseases of Legislative Language’, Harv. J. on Legis., Vol. 1, 1964, pp. 5, 6 <http://heinonline.org>. Indeed, shared ambiguity in the context of Rwandan multilingual drafting is exacerbated by translating words that carries ambiguity within themselves to other languages. It results into a difficult approach if the shared ambiguity is not capable of reconciling all the versions.
  By attaching clear and precise communication to the same meaning, ambiguity can be avoided. But exact communication is possible only if there is one-to-one relation between term and a concept. Certainly, concepts do not coincide and they do differ from one language to another.¹²²xBekink & Botha, 2007, at 14. Translation of these differing concepts leads to the shared ambiguity in the due course. The matter subject to debate in this case is how far the interpreter of legislation goes in examining analysing the circumstances, including the history of the matter and tips on the drafting of the legislation.¹²³xPerell, 1998, at 29. The ultimate step is to try and reconcile ambiguity of the versions within evolution drawn from the meaning of the entire legislation and the subject that it is intended to address.¹²⁴xM. Beaupre, ‘Litigating the Meaning of Bilingual Legislation’, Advoc. Q., Vol. 9, 1988, pp. 330, 331 <http://heinonline.org>. It requires that the interpreter evaluates the shared meaning from all the versions and compare it with the entire legislation and the general context in which it is meant to function.¹²⁵xIbid. It is a multi-faceted approach task and places the burden on the interpreter.
  An example of the shared ambiguity is in Article 32 of the Presidential Decree regulating road traffic.¹²⁶xPresidential Decree No. 85/01 of 2 September 2002 Regulating General Traffic Police and Road Traffic in Rwanda, Art. 32. <www.amategko.net>. In English, Article 32 reads, “Except for local regulations or particular lay-out of the areas, every vehicle or animal at stop or in parking must be pulled out to the side.....”¹²⁷xIbid. In French, the version can be translated as “Except for local regulations or particular lay-out of the areas, every vehicle or animal at stop or in parking must be pulled out to the side.”¹²⁸xPresidential Decree, supra note 126, Art. 32. The Kinyarwanda version can be translated as “Except for local regulations or particular lay-out of the areas, every vehicle or animal at stop or in parking must be pulled out to the side;...”¹²⁹xIbid.
  The Kinyarwanda version translates exactly like the French and the English versions. There are two drafting errors involved. A vehicle or animal stops and parks, therefore any person driving a vehicle or pulling an animal is not supposed to abide by the instructions. It does not indicate a person in that situation. A litigant may simply argue that the regulation does not apply to persons, but rather to vehicles or animals. The ambiguity is caused by words in the legislation construed with redundancy because it appears to have no subject. In fact, there is no identifiable person on whom an obligation or restriction is imposed. Otherwise, stating that a person who rides or drives... stops or parks near... Shall be fined a certain amount would be a simple way to prevent people from stopping or parking in the prohibited or restricted areas.
  The drafter uses the word ‘vehicles’ in all the three versions, which is a general term that refers to a class of things.¹³⁰xSullivan, 2010-2011, at 106. The term ‘vehicle’ is very broad and when used in a clause like this renders it ambiguous. The interpreter would have to find out whether the drafter intended to include new members of the class as they are discovered from time to time or intended to apply to specific types within the class of vehicles. If we are to say that a new member of the vehicle is invented later, we cannot possibly state that the new member could have been in contemplation¹³¹xJ. Evans, Statutory Interpretation, Problems of Communication, Oxford University Press, 1988, p. 18. at the time this provision was being drafted. Jim Evans stated that although a car is certainly a ‘vehicle’ for the purpose of a rule that excludes vehicles from a park, there is no conclusive answer, as far as linguistic conventions go, to the question whether a toy motor car or a sledge or a bicycle is included in this term.¹³²xIbid., p. 73. The use of the term vehicle in all the versions clearly lacks precision, which in turn leads to difficulty in interpretation and consequently in implementation.
  The provision does not indicate the subject; it only shows the object. Naturally, the reader or the addressee of the legislation is always trying to apply a possible meaning that is to his or her advantage regardless of the intention of the Parliament.¹³³xPearce & Geddes, 2006, at 3. The shared meaning rule states that where there is an overlap in meaning between the two language versions of a provision that are otherwise at variance, the meaning that is shared by both versions is to be preferred.¹³⁴xP. Salembeir, ‘Rethinking the Interpretation of Bilingual Legislation: the Demise of the Shared Meaning Rule’, Ottawa L. Rev. Vol. 35, No. 1, 2003-2004, p. 79 < htt://heinonline.org>. The question is how this shared ambiguity helps us to determine how this error occurred and how it can be ironed out. In the complexity of interpreting multilingual legislation, there appear to be few general rules for distinguishing these meanings; the only approach is to examine the context of an ambiguous word or phrase.¹³⁵xWirth, 1979-1980, at 436.

  II. Breadth versus Narrow Shared Meaning

  Breadth shared meaning refers to a situation when the meaning expressed by one version is narrower than the meaning expressed by the other, and the narrow meaning is contained in the broader meaning.¹³⁶xSullivan, 2010-2011, at 78. This is a moment in legislation where the meaning conveyed in the narrow version constitutes a subset of the broader version.¹³⁷xSalembier, 200-2004, at 85. It may as well include an overlap between the versions to an extent that the range of reference of one version is enclosed within the range of reference of another.¹³⁸xSullivan, 2010-2011, at 79.
  It is suggested that where one version conveys a wider meaning and another version conveys a limited meaning that contains the legislator’s intent, it is better to adopt the limited meaning. The narrow meaning of the adopted version must, insofar as it goes, undoubtedly express the will of the Parliament.¹³⁹xChing Fung, 1997, at 218. In other words, breadth shared meaning is a form of absolute conflict.¹⁴⁰xSullivan, 2010-2011, at 79. This presumption seems to apply only in favour of the legislative rule based on the strict form of construction of the legislative expression. It is in this sense not easier to link the legislative intent with the narrow expression than with the broad one.¹⁴¹xIbid., p. 90. When the interpreter chooses to stick to the narrow meaning, there is a possibility that it may lead to the narrow result compared with the substantial controversies surrounding the legislation.¹⁴²xC. Birch, ‘The Connotation/Denotation Distinction in Constitutional Interpretation’, J. App. Prac. & Process, Vol. 5, 2003, pp. 448, 449 <http://heinonline.org>. Where a version is capable of two or more meanings and one version corresponds to those meanings, the common meaning prevails over the other alternatives. It is believed that such an approach would be proper only when applied to bilingual legislation. It is questionable to restrict the common meaning that relies to a greater extent on one version where there are more than two language versions.¹⁴³xShelton, 1996-1997, at 628.
  An example of shared ambiguity is drawn from Article 55 of the law relating to children infected or affected by HIV/AIDS.¹⁴⁴xLaw No. 54/2011 of 14 December 2011 Relating to the Rights and the Protection of the Child in Rwanda. <www.Premature.gov.rw.> The English version reads thus: “A child above the age of 12 years has the right to consult an authorized professional medical doctor or a nurse and to go through medical examination notwithstanding the opposition or prohibition of his/her parents or guardian.” ¹⁴⁵xIbid. The French version translates as “A child above the age of 12 years has the right to consult an authorized professional medical doctor or a nurse and to go through medical examination notwithstanding the opposition or prohibition of his/her parents or guardian.”¹⁴⁶xIbid. The Kinyarwanda version reads: “A child above the age of 12 years has the right to consult a government professional medical doctor or a nurse and to go through medical examination notwithstanding the opposition or prohibition of his/her parents or guardian.”¹⁴⁷xIbid.
  The first question is to ask whether there is a discrepancy between these versions, and surely there is. The second question is whether this is a breadth and narrow shared meaning. In the first place, this provision has vague statements in all the versions. All versions provide that a child above the age of 12 may consult a doctor or a nurse. Literally, above 12 is 12 + 1. The meaning of 12 years and above has a plain meaning and is clear only insofar as its general applicability or usage is concerned, but is vague in its specific application to particular circumstances. But was legislative intent meant to exclude a child who is only 12 years old? Here comes the ordinary meaning. The meaning of a word or phrase may be affected by the reader’s knowledge that he or she is reading a piece of legislation that deals with a particular subject and aims to solve a particular problem,¹⁴⁸xSullivan, at 2007, at 50., in which case, every child of 12 years and above may consult a doctor or nurse.
  With regard to broad versus narrow, the word Government doctor or nurse is wide and varies according to context. The Government doctor or nurse legislated in the Kinyarwanda version carries three meanings: a doctor or a nurse working for the Government; a doctor or a nurse authorized by the Government to practise; and a doctor or a nurse authorized to consult a child. The English and French versions also carry two meanings: the doctor or nurse authorized to consult a child, and the doctor or nurse authorized to practise. In other words, the term ‘authorized doctor or nurse’, as found in the English and French versions, is a subset of the Kinyarwanda version ‘government doctor or nurse’.
  The Kinyarwanda version carries an interesting reference from the English and French versions. The Kinyarwanda version refers to ‘umuganga cyangwa umuforomo w’umwuga wa Leta’, which translates as the Government doctor or nurse that includes authorized nurse or doctor as found in French and English versions. The English version refers to ‘authorised professional medical doctor or a nurse’, while the Kinyarwanda version carries a somewhat broad meaning. The French version refers to ‘médecin agréé ou infirmier(e)’, which translates as authorized doctor or nurse. It carries an ambiguous narrow meaning. The narrow meaning would imply that the doctor or nurse has to be granted permission to consult the child, and in the absence of such permission it would be illegal to consult the child; or a doctor or a nurse who is authorized to practise may consult a child. While the French and English versions are narrow, they are not precise and as a result not conclusive on the meaning of the authorized doctor or nurse.
  Article 68 of the same law provides that “This Law was drafted in French, considered and adopted in Kinyarwanda.”¹⁴⁹xLaw on the Protection of the Child, supra note 144, Art. 68. This may lead to the assumption that the broad meaning in the Kinyarwanda version is a result of a translation mistake, yet the French version that was translated also carried its own error. What may have happened during the drafting process is that the drafters of the French and English versions may have heard the word authorized, but paid no attention to how the ‘authorized’ applies.
  The Kinyarwanda version drafter may have simply misinterpreted the instruction and possibly thought that it was the Government doctor but was not careful enough to find out whether it is the doctor or nurse practicing in the Government hospital or authorized by the Government to practice. What we are sure of is that there is some commonality to do with the Government that is either a hospital or permission to practice. It is clear that neither drafter of the alternative versions gave due consideration to the two concepts. The concept of commonality in this case would not guide us in establishing how these two diverging situations occurred, and therefore fails the rule of shared meaning within the concept of broad and narrow shared meaning.
  If one may argue that the narrow meaning that is authorized in this case take precedence, it would not be enough to give a clear meaning. It is narrow in comparison with the broad but is not precise enough to reflect the true legislative intent. If one reads all versions with a presumption that the language in which it was adopted prevails, they would not catch the drafter’s intent. If one reads with a presumption that the language in which it was drafted prevails, there is some reasonable cause that the drafter intended to mean the person authorized to practise as a doctor or nurse or a doctor or nurse authorized to consult the child.
  By adopting the breadth shared meaning, readers relatively understand that the Kinyarwanda version would lend itself to two meanings, a doctor working for the Government, a doctor authorized by the Government. In fact, the word Government doctor or nurse is wide enough to include the authorized doctor or nurse. In the contrast between clear and narrow, the French and English versions render a narrow but ambiguous meaning. There is no reason to presume that the narrow meaning is the one intended by the drafter. Assuming that the narrow meaning is contained in the broad meaning, adopting it would be appropriate. But if the narrow meaning contains ambiguity, readers of the narrow version cannot rely on it unless they compare it with the broad version. For want of evidence, we are divided between applying the broad meaning, which we are reasonably certain will be accurate to some extent, and applying the narrow meaning, which we may have reasonable grounds to believe will deliver accurate results. We are actually gauging narrow versus broad at 50% each.
  This example indicates that the shared meaning rule fails to carry any predictive value in terms of broad versus narrow divergences.¹⁵⁰xSalembier, 200-2004, at 87. It would inevitably not make any sense to constrain judges to rely on the limited meaning if it did not deliver the legislative intent as well as upholding the rule of law. If three meanings are possible in the Kinyarwanda version and two meanings are possible in the French and English versions, it is difficult to determine which possibility is to be excluded. In fact, the broader version, by definition, may at times contain elements lacking in the narrow version. The overlap between the versions does not provide room for determining the shared meaning in the sense of broad and narrow. It would be inappropriate if it leads to a conclusion that denies the legitimacy of the judiciary or whoever interprets the law in a democratic society.¹⁵¹xPerell, 1998, at 25. From the interpretive view, it is absurd to premise around the narrow meaning. A contextual analysis would be paramount in order to determine who is an authorized doctor or nurse, or a Government doctor or nurse. Note that neither the term authorized nurse or doctor nor Government nurse or doctor is defined in the terms of the legislation.

• E. Application of the Shared Meaning Rule

  When a shared meaning can be established, it contributes as a factor in finding a solution to the interpretation. It cannot, however, be solely relied on and may be ignored if it does not reflect the legislative intent.¹⁵²xCôté, 2003-2004, at 1071. In most circumstances, when applied, it only results in no more than what a random chance would predict.¹⁵³xSalembier, 200-2004, at 80. It has been suggested that the choice has to be made between the legislative intent and the meaning of words. In principle, there has to be a link between what the law makers intend and what the reader of the legislation perceives. ¹⁵⁴xAttorney General’s Department, 1983, at 96. The application of the shared meaning requires fidelity to the shared meaning and consistency with the idea in the legislative versions as originally constructed.¹⁵⁵xSullivan, 2007, at 86. In the light of the shared meaning rule, the interpreter has to consider the provision in context, taking into account the legislative intent and may go farther to policy concerns as well as external evidence. The interpreter may find that any of the versions was improperly drafted and prefer the version that corresponds to the factors of the legislative intent and policy issues.¹⁵⁶xIbid. In that case the shared meaning may be rejected in favour of appropriate interpretation.
  The difficulty with this approach is that the word’s common meanings have several meanings attached to it. It may as well mean the common intent, that is to say, the common intent the legislators had in mind when the legislation was being adopted. Common meaning may as well entail all elements common to the versions.¹⁵⁷xHardy, 1961, at 79. Each of the versions is likely to consider a common meaning it attaches to a particular word as ordinary meaning in the light of the object and purpose of the legislation, which may differ across all versions of a piece of legislation.¹⁵⁸xSchwarzenberger, 1968-1969, at 13. Moreover, it does not provide for a point in time when an extraordinary meaning has to be attached to the word in the context of legislation. Common meaning is viewed as when a Legislature speaks in two languages in the same breath. It seems that the meaning that is common to both versions must be regarded as the legislative intent. The surplus on one side must be regarded as being due to incautious expression.¹⁵⁹xChing Fung, 1997, at 218. This rule has been developed basically to govern interpretation of multilingualism. In cases where the alternative versions of the legislation do not declare the same meaning, the meaning contained in all the versions is presumed to be the legislative intent. It is based on the assumption that all versions of the legislative text must declare the same law. The unanswered question then is, if the shared meaning must be adopted, there would be no purpose of looking at the legislative intent.¹⁶⁰xSullivan, 2003-2004, at 1012. Lord Reid in Westminister Bank Ltd V Zang said, “But no principle of interpretation of statutes is more formally settled than the rule that the court must deduce the intention of the Parliament from the words used in the Act. If those words are in any way ambiguous, if they are reasonably capable of more than one meaning or if the provision in question is contradicted by or is incompatible with one or any other provision in the Act, then the Court may depart from the natural meaning of the words in question.”¹⁶¹xE. Driedger, Manual of Instructions for Legislative and Legal Writing, Department of Justice Canada (books 1-6), p. 547.
  Salembier suggests that “one of the currently accepted interpretive approaches, known as the ‘shared meaning rule’, is largely ineffective and potentially misleading.”¹⁶²xSalembier, 200-2004, at 78. He asserts that the shared meaning rule rests on a thin veneer of logic that does not withstand reasoned scrutiny. It does not provide a coherent account for origins of linguistic divergences.¹⁶³xIbid. It should be noted that the shared meaning rule applies only when all the versions are equally authentic. There is no such presumption in the case of one authentic version and an official translation.¹⁶⁴xKuner, 1991, at 2. Arguments against the shared meaning rule allege that it is nothing other than a rule of convenience designed to reconcile the practice of all authentic versions of legislation.¹⁶⁵xIbid., p. 5. They opine that a mandatory comparison of versions in the search for shared meaning blocks the opportunity to rely on the meaning as may be found in one version that reflects well the drafter’s intent.¹⁶⁶xIbid.
  In a multilingual environment, interpretation cannot end once a shared meaning is found. On the contrary, it has to continue until all legitimate evidence of legislative intent is exploited.¹⁶⁷xSullivan, 2010-2011, at 79. Once the existence of the shared meaning invokes the presumption that favours shared meaning, it should turn into an interpretive question as to “what is the best or most plausible meaning having regard to the purpose and context?” and would there be any other evidence as regards legislative intent that would oust the presumption. The presumption would be taken as determinant if any other evidence for legislative intent and purpose does not hold or is inconclusive.¹⁶⁸xIbid. The common meaning must be drawn from all the versions compared with the entire enactment and the system in which it is intended to operate. If the preliminary common meaning clashes with such a system or with the enactment as a whole, then it is quite likely and reasonable that the court will have to reject it. It would not be a proper solution to the problem. The court will at least be persuaded to look elsewhere for the rational choice.¹⁶⁹xM. Beaupre, ‘Judicial Approaches to the Interpretation of Bilingual Legislation’ – Presentation to the Canadian Institute for the Administration of Justice’ 21 August 1987<http://heinonline.org>.
  Reading from R. Dickerson’s model of successful communication, it is imperative that the drafter and the audience share a common understanding of the common meanings of words in legislation. They must have a common understanding of ranges of meanings to certain words and how context limits meanings. It basically implies that in a multilingual environment, the common meaning of words must apply to all versions and those versions must have the same meaning, a task that may not be easily realized.¹⁷⁰xP.J. Kelley ‘Advice from the Consummate Draftsman: Reed Dickerson on Statutory Interpretation’ S. Ill. U. L.J., Vol. 16, 1991-1992, pp. 592, 593 <http://heinonline.org>. Shared meaning does not mean simple reproduction of words from the original version. It is rather the construction of words into another language version in a way that it carries the same significance as the other versions.¹⁷¹xR. Edgardo, ‘Inherent Problems of Legal Translation Theoretical Aspects’ Ind. Int'l & Comp. L. Rev, 1995-1996, pp. 187, 193 <http://heinonline.org>. Taking an example of the term ‘gender’ from the Rwandan Constitution, it is not easy to get its equivalent in Kinyarwanda and French. It was translated in Kinyarwanda as ‘ equality between women and man in development’ and in French as ‘genre’.¹⁷²xConstitution of the Republic of Rwanda, 2003, Art. 185. The word ‘genre’ then had to be ignored because it conveys a different meaning and was simply inserted in the French version as ‘gender’.¹⁷³xIbid. This is because what one language system conceptualizes in one way is not conceptualized in the same way in all other language systems. It is naturally sometimes impossible owing to historically grown meaning to take a word or a phrase from one language and substitute the meaning in another language. ¹⁷⁴xEngberg, 2003-2004, at 1164.In the Rwandan context of multilingualism, there is an underlying risk for the Parliament to adopt the errors of translation, a situation that occurs when the legislation is drafted in English or French (currently, most of the business laws termed as ‘doing business’ are drafted in English, and most of the labour laws are drafted in French).¹⁷⁵xConstitution of the Republic of Rwanda, 2003, Art. 93. The laws are then translated into Kinyarwanda; to this effect using equivalent terms or words that deliver the same meaning is indeed a hassle.
  There have been arguments that similar/shared meaning does not exist. There is no presumption of shared meaning between the authentic and official translated versions in the first place. Secondly, as soon it is established that the authentic versions present a difference in meaning, the presumption of the shared meaning ceases to carry any effect.¹⁷⁶xKuner, 1991 at 2. It has been suggested that perhaps the biggest mistake when interpreting legislation is to select any preferred rule and then apply it to the exclusion of all other rules. All that should be done is to consider relevant criteria in order to reach a balanced decision.¹⁷⁷xBennion, 2008 at 547. If the shared meaning removes doubt owing to lack of the necessary clarity or transparency required for the application of the law, it can be applied. It must, on the other hand, be ignored if it is to lead to a situation where justice or fairness is to be stalled.¹⁷⁸xEngberg, 2003-2004, at 1136. Engberg believes that any legal interpretation that relies heavily on the fixed interpretive rules cannot in many instances achieve the basic requirements of justice without considering more subjective factors. An element of more free and subjective interpretation has to be invoked.¹⁷⁹xIbid., p. 1148
  In the light of the shared meaning rule, comparison of texts to get the shared meaning may on the one hand help to resolve the ambiguity inherent in the term or phrase used in one language, making clearer the intention of the drafter.¹⁸⁰xShelton, 1996-1997, at 612. On the other hand, lack of precise linguistic equivalents and differences in the legal systems and terminology make it difficult to have shared meaning in certain multiple language versions, leading to conflicting interpretation of legislation.¹⁸¹xIbid. Just like judges in the interpretation of multilingual treaties aim to apply the closest approximation of the parties’ shared expectations,¹⁸²xP.J. Liacouras, ‘The International Court of Justice and Development of Useful “Rules of Interpretation” in the Process of Treaty Interpretation’, Am. Soc'y Int'l L. Proc., Vol. 59, 1965, pp. 162, 163 <http://heinonline.org>. the same should apply to interpreting multilingual legislation. Interpreters should strive to apply the shared meaning to the extent that it corresponds in the least to the closest expectation of the legislature and to the audience of the legislation. It is natural to think that citizens will read and observe laws in the plain meaning of the words in the legislation.
  It is worth noting that whereas supremacy of clear version over the ambiguous one is reasonably acceptable, the same cannot apply for the narrow version over the broad one. There is no justification as to why the narrow meaning should prevail over the broad if it does not fully embrace the will of the Parliament.¹⁸³xCôté, 2003-2004, at 90. Salembier proposed that courts should disregard the shared meaning rule and begin with the presumption of favouring clarity and interpret each version by applying “the standard techniques of statutory interpretation.”¹⁸⁴xSalembier, 200-2004, at 75. He suggests standard techniques as “looking to the purpose of the Act, its internal consistency and legislative evolution, and the relevant presumptions of legislative intent – to determine which language version produces the most coherent legislative scheme.”¹⁸⁵xIbid.
  The shared meaning rule may for some reasons be unacceptable or may be inconsistent with the provisions of the same sort. Corcoran and Bottomley rightly suggested that statutory interpretation cannot be divorced from general principles of justice and fairness.¹⁸⁶xS. Corcoran & S. Bottomely. Statutory Interpretation, The Federation Press, 2005, p. 50.

• F. Conclusion

  The wording of legislation is not an easy task, and it becomes more difficult where more than one language is used to carry the legislator’s message. Consequently, it makes for difficult reading and interpretation.¹⁸⁷xW. Voermans, ‘Styles of Legislation and their Effects’, Statute Law Review, Vol. 32, No. 1, 2011, p. 3 <http:login.westlaw.co.uk.catalogue.ulrls.lon.ac.uk>. Realistically, multilingual legislation contains discrepancies in the meaning of different language versions that may either complicate or facilitate the interpretation.¹⁸⁸xCondon, 2010, at 5. When coordinating the text’s results into discrepancies in the meaning of different versions of legislation, on the one hand it leads to an additional source of ambiguity or obscurity of legislation; on the other hand, when the meaning of terms is ambiguous or obscure in one language, but clear in another, the multilingual character of the legislation facilitates interpretation.¹⁸⁹xIbid.
  Some opinions are in favour of multilingualism as regards the challenges of statutory interpretation not only because the text has been pre-interpreted by the translator, but also the drafting errors may be easily checked, and the text becomes more understandable by comparison of different versions by judges and lawyers.¹⁹⁰xJ.C. Gemar & N. Kasirer, Jurilinguistique: entre Langues et Droit, Bruylant, 2005, pp. 354-355. In this situation, an unclear version is interpreted by reference to other versions and reconciled with them, and the clear version is preferred as the objective one.¹⁹¹xBeaupre, 1987, at 17-18. However, this solution cannot apply when two versions are clearly in contradiction, and the ambiguity arises not from ambiguity in one or other version of the legislation, but from the fact that the same versions say the same thing clearly but differently.¹⁹²xIbid., p. 24. On the whole it is inaccurate to assume that interpreting multilingual legislation in Rwanda or any other multilingual legislation can be catered for by equal authenticity or shared meaning rules. All rules of interpretation must continue to apply until all legitimate evidence of legislative intent can be fairly conversed.¹⁹³xSullivan, 2010-2011, at 78.
  Whichever way the interpretation provisions choose to go, it is suggested that in any multilingual jurisdiction, and particularly in Rwanda, it is essential that the process used in interpreting legislation be credible if all versions of the legislation are to obtain equal authenticity.¹⁹⁴xRevell, 2003-2004 at 1105. Nevertheless, there should not be an explicit clause that a certain version must prevail. The reason is that it may not disclose the drafter’s intent or provide unambiguity, in which case any version that provides for such should be considered as it was the solution in the Vienna convention, where Article 33(4) provided for resolving any divergence in versions with regard to the object and purpose of the treaty. ¹⁹⁵xChing Fung, 1997, at 216. Some weight may be given to the language prevailing over others only if it is apparent from the ‘travaux prepartoires’that it corresponds to the legislative object and purpose and that other versions are mere translations. Moreover, the version that is preferred to prevail over others may lead to a result that is manifestly absurd or unreasonable.¹⁹⁶xIbid., 217. Some of the interpretation difficulties are inherent in the interpretation procedure regulated by Articles 5 and 93 of the constitution.¹⁹⁷xSee Arts. 5 and 93 of the Constitution of the Republic of Rwanda. If equal authenticity has to apply, there is no need to regulate which version prevails. Preference of one version over another gives the impression that the language that prevails is the source text for the true meaning and others are simply translations.¹⁹⁸xRevell, 1997, at 1085. It defeats the principle of equality of versions; if a Government acts in more than one language, then its acts should be taken as authentic in all the languages in which it acts.¹⁹⁹xIbid. It is essential that interpretation rules are applied in a sense that seeks meaning that gives effect to the intended purpose.
  The interpreter needs to strike a balance between the two systems considering that the country fundamentally relied on a Civil Law system and is now moving to a hybrid system. R.A. Macdonald stated that “‘common law in French’ is fatally compromised; and the civil law and the French language are intimately connected, such that the possibility of ‘civil law in English’ is also fatally compromised.”²⁰⁰xMacdonald, 1996-1997, p. 151. The two legal traditions rest on a sophisticated view of the relation of language to meaning, and in doing so, they hinder the smooth interaction and practice of multilingualism.²⁰¹xIbid. There is an inherent difficulty in attempting to bridge the gap between the two legal traditions. The relativity of legal terms, the inconsistency of categorizations and classifications between different cultures of languages are distinguished at the level of terms and concepts that comes with the complexities of statutory interpretation.²⁰²xB. Viorica & F. Delia-$tefania, ‘Linguistic Rules Applied in the Interpretation of Legal Norms’, AGORA Int'l J. Jurid. Sci. [clx], 2010, p. 2 <http:login.westlaw.co.uk.catalogue.ulrls.lon.ac.uk>. Very often, creation of legal meaning always takes place through an essentially cultural medium of any given language.²⁰³xIbid. Clarity and precision is paramount when translating a text from one language to another. But there is likelihood of using bound abstractions whose meaning is derived from social and cultural contexts. This may generate ambiguity, especially where legal traditions are different, that is common law and civil law. ²⁰⁴xEdgardo, 1995-1996, at 189. It is believed that persons speaking the same language but from different legal systems are prone to translational problems than the persons speaking different languages but under one legal system.²⁰⁵xIbid. The interpreters of legislation in Rwanda today are embedded in the legal culture that is not Kinyarwanda, French nor English. A legal culture that is neither common law nor civil law, it is all of these combined together with the ambiguities and complexities surrounding each legal tradition.
  As regards the interpretation of legislation by the Parliament of Rwanda, some scholars suggest that it is a prerogative of courts. According to Pierre–Andre Cote, “The Canadian Parliament enacts legal texts, not legal norms.”²⁰⁶xCôté, 2003-2004, at 1068. In Rwanda, the legislature enacts legislation and interprets it.²⁰⁷xSee Art. 96 of the Constitution of the Republic of Rwanda. From the drafting point of view, this approach is dysfunctional because the legislature is not practically involved in application of the legislation. This procedure seems to be rather cyclical because it goes back to the legislators to ask them what they meant or desired when they enacted the legislation, which must otherwise be embedded in the language of the statutory instrument. Perrel emphasizes that the legislators make the law and the interpreters discover it.²⁰⁸xPerell, 1998, at 30. He denotes that a democratic society should be governed by the rule of law. In principle, the legislature is there to make the law, and then the judiciary finds the legislator’s intent about the law as expressed in the language of the law.²⁰⁹xIbid., p. 55. The norms should be constructed by the readers of legislation but not the legislature. Construction of legal norms in this case refers to interpretation.²¹⁰xCôté, 2003-2004, at 1068. In any case, it provides an opportunity for future reference on the decided cases instead of going back and forth to the Parliament.
  Considering that Rwanda has not legislated much on interpretation is also a setback. Despite very many theories that have developed over centuries, it is worth noting that there seems to be a significant tension between practice and theory of statutory interpretation.²¹¹xG.R. Hall, ‘Statutory Interpretation in the Supreme Court of Canada: The Triumph of a common Law Methodology’, Advoc. Q. 21, 1998-1999, pp. 38, 39 <http://heinonline.org>. “Legal process” school of legal theory which developed at Harvard Law School suggested that “a coherent theory of statutory interpretation simply did not exist”.²¹²xHall, 1998-1999, at 39. As a result, there may be no theory that is expected to be an accurate statement of what the courts would actually do with the legislation.²¹³xIbid., p. 40. It is argued that “Rules do not guide. Rules are always open to several interpretations. Which one we choose hangs upon the ‘disposition’ of our community. It is entirely ‘arbitrary’ how we choose to interpret the rule.”²¹⁴xPerell, 1998, at 32.
  As we conclude, we draw attention to some of the steps that can be taken to lessen the challenges of interpreting the multilingual legislation, in particular, and the quality of legislation in Rwanda. Significant steps have already been taken, and other steps are underway to address some of the challenges discussed in this article.
  The drafting and translation teams need to evolve to the point where the quality of translation is good enough so as to have all versions serve the equal authenticity purpose. Learning from the example of Canada, to avoid translational errors, co-drafting can be tried in Rwanda. In co-drafting, English and French drafters are brought together in Canada.²¹⁵xRevell, 1997, at 1098. In Rwanda, where drafting is in three versions, it may involve either three drafters who are fluent in each language or two drafters, because the Kinyarwanda language is understood by most of the drafters and translators. Where co-drafting may not be possible, the drafter and the translator work closely together in order to reduce divergent meanings, ambiguity and vagueness.
  Another step is training in both Common Law and Civil Law systems in order to avoid the clash.²¹⁶xNsanze, 2012, at 51. Since translation demands precision and certainty, it is possible to derive meaning from changing cultural and social contexts that generate ambiguity, especially when legal traditions are different from each other.²¹⁷xEdgardo, 1995-1996, at 189. Quality in legislation requires training in drafting and in research skills.²¹⁸xL. Dushimimana, ‘Aspects of Legislative Drafting: Some African Realities’, The Loophole, 2012, 50. Lawyers trained and skilled in drafting discipline are not many in all the drafting departments. Lack of drafters who are knowledgeable in basic concepts of both civil law and common law may lead to legislation that appears to be serving two masters simultaneously with a cautious omission to state which legal system is meant to be served by which version.²¹⁹xBeaupre, 1987, at 154. This implies that there is an urgent need to train drafters in both common and civil law.
  Another solution may come with the revision of Article 93 of the Constitution, which regulates interpretation. In Rwanda, English or French is often used as in many other African countries to draft laws. These languages are very difficult to translate into plain language of a particular local language (Kinyarwanda). Thus the Kinyarwanda version, which is seemingly resorted to in case of conflict, carries most of the translation quibbles.²²⁰xBekink & Botha, 2007, at 14. The degree of divergence may depend on a number of factors, including the distance between languages and cultures,²²¹xScassa, 1994, at 179. reviewing Article 93 of the Constitution to resolve the problem of predominance of one version over the others and thus the language that carries a clear reflection of the legislative intent would prevail.
  At the end of the day, multilingual interpretation is very challenging, and all the rules developed to this effect carry deficiencies with them. To sum it up, I agree with Dreidger: “There is only one rule in modem interpretation, namely, courts are obliged to determine the meaning of legislation in its total context, having regard to the purpose of the legislation, the consequences of proposed interpretations, the presumptions and special rules of interpretation, as well as admissible external aids. In other words, the courts must consider and take into account all relevant and admissible indicators of legislative meaning. After taking these into account, the court must then adopt an interpretation that is appropriate. An appropriate interpretation is one which can be justified in terms of (a) its plausibility, that is, its compliance with the legislative text; (b) its efficacy, that is, its promotion of the legislative purpose; and (c) its acceptability, that is, the outcome is reasonable and just.”²²²xSullivan, 2007, at 131.
  Multilingual legislation is good for interpretation, where one version is helpful to interpret another and disastrous if one version raise doubts about another, as echoed by the saying, “A person with one watch always knows what time it is. A person with two watches is never sure.”²²³xSalembier, 200-2004, at 78. I have analyzed Ruth’s theory on application of rules of interpretation developed by scholars for multilingual legislation, applying them to the Rwandan context. In the instance of the clear and ambiguous version, it works, but in the other three, it does not work. The examples given are a clear indication of the challenges of statutory interpretation in a multilingual environment. The theory of interpreting multilingual legislation carries complexities, particularly in Rwanda and in other multilingual jurisdictions.

Noten

• 1 D.L. Revell, ‘Authoring Bilingual Laws: The Importance of Process’, Brook. J. Int’l L., Vol. 29, 2003-2004, p. 1085 <http://heinonline.org>.

• 2 Attorney General’s Department Report, Symposium on Statutory Interpretation, Australian Government Publishing Service, Canberra, 1983, p. 96.

• 3 E. Alcaraz & B. Hughes, Legal Translation Explained, St. Jerome Publishing, Manchester, 2007, p. 24.

• 4 Attorney General’s Department Report, 1983, p. 96.

• 5 Alcaraz & Hudges, 2007.

• 6 C.B. Kuner, ‘The Interpretation of Multilingual Treaties: Comparison of Texts Versus the Presumption of Similar Meaning’, ICLQ, Vol. 40, No. 4, 1991, pp. 953-964 <http:login.westlaw.co.uk.catalogue.ulrls.lon.ac.uk>.

• 7 M. Beaupre, Interpreting Bilingual Legislation, Carswell, Toronto, 1986, p. 4.

• 8 D. Shelton, ‘The Interpretation of Multilingual Treaties’, Hastings Int’l & Comp. L. Rev, Vol. 20, 1996-1997, pp. 611, 620 <http://heinonline.org>.

• 9 A. Scalia, A Matter of Interpretation, Federal Courts and the Law, Princeton University Press, Princeton, NJ, 1970, p. 31 <http://heinonline.org>.

• 10 R.A. Macdonald, ‘Legal Bilingualism’, McGill L. J., Vol. 42, 1996-1997, pp. 126, 160 <http://heinonline.org>.

• 11 V. Nsanze, ‘Challenges of Drafting Laws in One Language and Translating Them: Rwanda’s Experience’, The Loophole, No. 1, January 2012, p. 46.

• 12 Ibid.

• 13 Nsanze, 2012.

• 14 Organic Law No. 06/2006 of 15 February 2006 Establishing Internal Rules of Procedure of the Chamber of Deputies in the Parliament, Art. 107.

• 15 Constitution of the Republic of Rwanda of 4 June 2003 as amended to date, Art. 5.

• 16 Ibid., Art. 93.

• 17 Instructions of the Minister of Justice No. 01/11 of 14 November 2006 relating to the Drafting of the Texts of Laws, Art. 3.

• 18 Ibid., Art. 4.

• 19 Constitution of the Republic of Rwanda, 2003, Art. 96.

• 20 M. Arden, ‘The Impact of Judicial Interpretation on Legislative Drafting’, The Loophole, 2008, p. 5.

• 21 Beaupre, 1986, at 161.

• 22 D.L. Revell, ‘Bilingual Legislation. The Ontario Experience’, Statute Law Review, Vol. 19, No. 1, 1998, pp. 32, 38.

• 23 Nsanze, 2012, at 50.

• 24 R. Sullivan, ‘Challenges of Statutory Interpretation Multilingual Multijural Legislation’, Brook. J. Int’l L., Vol. 29, 2003-2004, pp. 993, 985, 1005-1018 <http://heinonline.org>.

• 25 R. Sullivan, Statutory Interpretation, 2nd edn, Irwin Law, Toronto, 2007, p. 42.

• 26 Macdonald, 1996-1997, at 128.

• 27 Ibid., p. 131.

• 28 Constitution of the Republic of Rwanda, 2003, Art. 5.

• 29 T. Schilling, ‘Multilingualism and Multijuralism: Assets of EU Legislation and Adjudication?’, German L.J., Vol. 12, 2011, pp. 1462, 1463 <http://heinonline.org>.

• 30 Constitution of the Republic of Rwanda, 2003, Art. 93.

• 31 Beaupre, 1986, at 159.

• 32 Ibid.

• 33 Sullivan, 2003-2004, at 1005.

• 34 Revell, 1998, at 39.

• 35 Article 40 of Law No. 58/2011 of 31 December 2011 Governing the Gaming Activities in Rwanda.

• 36 Sullivan, 2007, at 106.

• 37 Minister’s Instructions, supra note 17, Art. 5.

• 38 S. Rosenne, ‘On Multi-Lingual Interpretation’, Isr. L. Rev., Vol. 6, 1971, pp. 357, 361 <http://heinonline.org>.

• 39 G.R. Sneath, ‘Bilingualism and Interpretation. Examples of Two Jurisdictions’. Statutory Law Review, Vol. 5, No.2, 1984, p. 49.

• 40 Kuner, 1991, at 3.

• 41 Ibid., p. 2.

• 42 Beaupre, 1986, at 160.

• 43 Constitution of the Republic of Rwanda, 2003, Art. 93.

• 44 Minister’s Instructions, supra note 17, Art. 4.

• 45 Sullivan, 2003-2004, at 1006.

• 46 Law No. 21/2012 of 14 June 2012 Relating to the Civil, Commercial, Labour and Administrative Procedure in Rwanda, Art. 374.

• 47 Organic Law No. 06/2006 of 15 February 2006, Supra note 14, Art. 107.

• 48 Ibid., Art. 115.

• 49 M. Loubser, ‘Linguistic Factors Into the Mix: The South African Experience of Language and the Law’, Tul. L. Rev., 2003-2004, Vol. 78, pp. 127, 128 <http://heinonline.org>.

• 50 The Constitution of the Republic of South Africa of 1983 is abrogated by the Constitution of 1993.

• 51 Loubser, 2003-2004, at 128.

• 52 Ibid., p. 129.

• 53 B. Bekink & C. Botha, ‘Aspects of Legislative Drafting: Some South African Realities (or Plain Language Is Not Always Plain Sailing)’, Statute Law Review, Vol. 28, 2007, p. 15 <http:login.westlaw.co.uk.catalogue.ulrls.lon.ac.uk>.

• 54 Beaupre, 1986, at 18.

• 55 P.-A. Côté, ‘Bilingual Interpretation of Enactments in Canada: Principles V. Practice’, Brook. J. Int’l L., Vol. 29, 2003-2004, pp. 1067, 1069 <http://heinonline.org>.

• 56 Ibid., p. 1069.

• 57 T. Scassa, ‘Language of Judgment and the Supreme Court of Canada’, UNBLJ, Vol. 43, 1994, pp. 178, 189 <http://heinonline.org>.

• 58 J. Engberg, ‘Statutory Texts as Instances of Language(s): Consequences and Limitations on Interpretation’, Brook. J. Int’l L., Vol. 29, 2003-2004, pp. 1135, 1154 <http://heinonline.org>.

• 59 J. Hardy, ‘The Interpretation of Plurilingual Treaties by International Courts and Tribunals’, Brit. Y. B. Int’l L., Vol. 37, 1961, pp. 74, 75 <http://heinonline.org>.

• 60 R. Sullivan, ‘Some Problems With the Shared Meaning Rule as Formulated in R. v. Daoust and the Law of Bilingual Interpretation’, Ottawa L. Rev., Vol. 42, 2010-2011, pp. 75, 76 <http://heinonline.org>.

• 61 Constitution of the Republic of Rwanda, 2003, Art. 93.

• 62 G. Schwarzenberger, ‘Myths and Realities of Treaty Interpretation Articles 27-29 of the Vienna Draft Convention on the Law of Treaties’, Va. J. Int’l L., Vol. 9, 1968-1969, pp. 3, 4 <http://heinonline.org>.

• 63 Kuner, 1991, at 3.

• 64 S. Yuen-Ching Fung, ‘Interpreting the Bilingual Legislation of Hong Kong’, Hong Kong L.J., Vol. 27, 1997, pp. 206, 218 <http://heinonline.org>.

• 65 Sullivan, 2003-2004, at 1014.

• 66 Côté, 2003-2004, at 1071.

• 67 Sullivan, 2010-2011, at 76.

• 68 Revell, 2003-2004.

• 69 Constitution of the Republic of Rwanda, 2003, Art. 93.

• 70 Sullivan, 2010-2011, at 82.

• 71 Beaupre, 1986, at 20.

• 72 Ibid.

• 73 Yuen-Ching Fung, 1997, at 207.

• 74 Law No. 13/2004 of 17 May 2004 Relating to the Code of Criminal Procedure, O.G Special No. of 30 July 2004, Art. 3 <www.amategeko.net>

• 75 Ibid.

• 76 Code of Penal Procedure of Rwanda, supra note 74, Art. 3.

• 77 Ibid.

• 78 Beaupre, 1986, at 24.

• 79 P. Butt & D. Hamer, Concise Australian Legal Dictionary, 4th edn, LexisNexis Butterworths, Australia, 2011, p. 411.

• 80 Constitution of the Republic of Rwanda, 2003, Art. 19.

• 81 F. Bennion, Bennion on Statutory Interpretation, 5th edn, LexisNexis, London, 2008, p. 447.

• 82 Ibid., p. 448.

• 83 Beaupre, 1986, at 26.

• 84 Constitution of the Republic of Rwanda, 2003, Art. 93.

• 85 Code of Penal Procedure of Rwanda, supra note 74, Art. 275.

• 86 B.J. Condon, ‘Lost in Translation: Plurilingual Interpretation of WTO Law’, Journal of International Dispute Settlement, 2010, p. 4 <http:login.westlaw.co.uk.catalogue.ulrls.lon.ac.uk>.

• 87 Côté, 2003-2004, at 1070.

• 88 D.A. Wirth, ‘Multilingual Treaty Interpretation and the Case of Salt Ii’, Yale Stud. World Pub. Ord., Vol. 6, 1979-1780, pp. 429, 429 <http://heinonline.org>.

• 89 Côté, 2003-2004, at 1070.

• 90 Ibid.

• 91 Côté, 2003-2004, at 1070.

• 92 Macdonald, 1996-1997, at 148.

• 93 Sullivan, 2010-2011, at 84.

• 94 Ibid., p. 88.

• 95 Beaupre, 1986, at 22.

• 96 Ibid., p. 23.

• 97 P.M. Perell, ‘Plain Meaning for Judges, Scholars and Practitioners’, Advoc. Q., Vol. 20, 1998, pp. 24, 27 <http://heinonline.org>.

• 98 Sullivan, 2010-2011, at 77.

• 99 Côté, 2003-2004, at 1069.

• 100 Kuner, 1991, at 3.

• 101 Sullivan, 2010-2011, at 77.

• 102 Condon, 2010, at 2.

• 103 Ibid., p. 4.

• 104 Scassa, 1994, at 179.

• 105 Sullivan, 2003-2004, at 1007.

• 106 Sullivan, 2010-2011, at 77.

• 107 Organic Law No. 01/2012/OL of 2 May 2012 Instituting the Penal Code of Rwanda, Art. 198.

• 108 Ibid.

• 109 Penal Code of Rwanda, supra note 107, Art. 198.

• 110 Ibid.

• 111 R. Bigwood (Ed.), The Statute, Making and Meaning, LexisNexis, Wellington, 2004, p. 130.

• 112 D.C. Pearce & R.S. Geddes, Statutory Interpretation in Australia, 6th edn, LexisNexis, Australia, 2006, p. 238.

• 113 Penal Code of Rwanda, supra note 107, Art. 764.

• 114 J. Bell & G. Engle, Cross Statutory Interpretation, 3rd edn, Butterworths, London, 1995, p. 50.

• 115 J. Vaiciukaiti & T. Klimas, ‘Interpretation of European Union Multilingual Law’, Int’l J. Baltic L., Vol. 2, 2005, pp. 1, 3 <http://heinonline.org>.

• 116 Sullivan, 2007, at 43.

• 117 Revell, 2003-2004, at 1099.

• 118 Côté, 2003-2004, at 1069.

• 119 Sullivan, 2010-2011, at 78.

• 120 Perell, 1998, at 58.

• 121 R. Dickerson, ‘The Diseases of Legislative Language’, Harv. J. on Legis., Vol. 1, 1964, pp. 5, 6 <http://heinonline.org>.

• 122 Bekink & Botha, 2007, at 14.

• 123 Perell, 1998, at 29.

• 124 M. Beaupre, ‘Litigating the Meaning of Bilingual Legislation’, Advoc. Q., Vol. 9, 1988, pp. 330, 331 <http://heinonline.org>.

• 125 Ibid.

• 126 Presidential Decree No. 85/01 of 2 September 2002 Regulating General Traffic Police and Road Traffic in Rwanda, Art. 32. <www.amategko.net>.

• 127 Ibid.

• 128 Presidential Decree, supra note 126, Art. 32.

• 129 Ibid.

• 130 Sullivan, 2010-2011, at 106.

• 131 J. Evans, Statutory Interpretation, Problems of Communication, Oxford University Press, 1988, p. 18.

• 132 Ibid., p. 73.

• 133 Pearce & Geddes, 2006, at 3.

• 134 P. Salembeir, ‘Rethinking the Interpretation of Bilingual Legislation: the Demise of the Shared Meaning Rule’, Ottawa L. Rev. Vol. 35, No. 1, 2003-2004, p. 79 < htt://heinonline.org>.

• 135 Wirth, 1979-1980, at 436.

• 136 Sullivan, 2010-2011, at 78.

• 137 Salembier, 200-2004, at 85.

• 138 Sullivan, 2010-2011, at 79.

• 139 Ching Fung, 1997, at 218.

• 140 Sullivan, 2010-2011, at 79.

• 141 Ibid., p. 90.

• 142 C. Birch, ‘The Connotation/Denotation Distinction in Constitutional Interpretation’, J. App. Prac. & Process, Vol. 5, 2003, pp. 448, 449 <http://heinonline.org>.

• 143 Shelton, 1996-1997, at 628.

• 144 Law No. 54/2011 of 14 December 2011 Relating to the Rights and the Protection of the Child in Rwanda. <www.Premature.gov.rw.>

• 145 Ibid.

• 146 Ibid.

• 147 Ibid.

• 148 Sullivan, at 2007, at 50.

• 149 Law on the Protection of the Child, supra note 144, Art. 68.

• 150 Salembier, 200-2004, at 87.

• 151 Perell, 1998, at 25.

• 152 Côté, 2003-2004, at 1071.

• 153 Salembier, 200-2004, at 80.

• 154 Attorney General’s Department, 1983, at 96.

• 155 Sullivan, 2007, at 86.

• 156 Ibid.

• 157 Hardy, 1961, at 79.

• 158 Schwarzenberger, 1968-1969, at 13.

• 159 Ching Fung, 1997, at 218.

• 160 Sullivan, 2003-2004, at 1012.

• 161 E. Driedger, Manual of Instructions for Legislative and Legal Writing, Department of Justice Canada (books 1-6), p. 547.

• 162 Salembier, 200-2004, at 78.

• 163 Ibid.

• 164 Kuner, 1991, at 2.

• 165 Ibid., p. 5.

• 166 Ibid.

• 167 Sullivan, 2010-2011, at 79.

• 168 Ibid.

• 169 M. Beaupre, ‘Judicial Approaches to the Interpretation of Bilingual Legislation’ – Presentation to the Canadian Institute for the Administration of Justice’ 21 August 1987<http://heinonline.org>.

• 170 P.J. Kelley ‘Advice from the Consummate Draftsman: Reed Dickerson on Statutory Interpretation’ S. Ill. U. L.J., Vol. 16, 1991-1992, pp. 592, 593 <http://heinonline.org>.

• 171 R. Edgardo, ‘Inherent Problems of Legal Translation Theoretical Aspects’ Ind. Int'l & Comp. L. Rev, 1995-1996, pp. 187, 193 <http://heinonline.org>.

• 172 Constitution of the Republic of Rwanda, 2003, Art. 185.

• 173 Ibid.

• 174 Engberg, 2003-2004, at 1164.

• 175 Constitution of the Republic of Rwanda, 2003, Art. 93.

• 176 Kuner, 1991 at 2.

• 177 Bennion, 2008 at 547.

• 178 Engberg, 2003-2004, at 1136.

• 179 Ibid., p. 1148

• 180 Shelton, 1996-1997, at 612.

• 181 Ibid.

• 182 P.J. Liacouras, ‘The International Court of Justice and Development of Useful “Rules of Interpretation” in the Process of Treaty Interpretation’, Am. Soc'y Int'l L. Proc., Vol. 59, 1965, pp. 162, 163 <http://heinonline.org>.

• 183 Côté, 2003-2004, at 90.

• 184 Salembier, 200-2004, at 75.

• 185 Ibid.

• 186 S. Corcoran & S. Bottomely. Statutory Interpretation, The Federation Press, 2005, p. 50.

• 187 W. Voermans, ‘Styles of Legislation and their Effects’, Statute Law Review, Vol. 32, No. 1, 2011, p. 3 <http:login.westlaw.co.uk.catalogue.ulrls.lon.ac.uk>.

• 188 Condon, 2010, at 5.

• 189 Ibid.

• 190 J.C. Gemar & N. Kasirer, Jurilinguistique: entre Langues et Droit, Bruylant, 2005, pp. 354-355.

• 191 Beaupre, 1987, at 17-18.

• 192 Ibid., p. 24.

• 193 Sullivan, 2010-2011, at 78.

• 194 Revell, 2003-2004 at 1105.

• 195 Ching Fung, 1997, at 216.

• 196 Ibid., 217.

• 197 See Arts. 5 and 93 of the Constitution of the Republic of Rwanda.

• 198 Revell, 1997, at 1085.

• 199 Ibid.

• 200 Macdonald, 1996-1997, p. 151.

• 201 Ibid.

• 202 B. Viorica & F. Delia-$tefania, ‘Linguistic Rules Applied in the Interpretation of Legal Norms’, AGORA Int'l J. Jurid. Sci. [clx], 2010, p. 2 <http:login.westlaw.co.uk.catalogue.ulrls.lon.ac.uk>.

• 203 Ibid.

• 204 Edgardo, 1995-1996, at 189.

• 205 Ibid.

• 206 Côté, 2003-2004, at 1068.

• 207 See Art. 96 of the Constitution of the Republic of Rwanda.

• 208 Perell, 1998, at 30.

• 209 Ibid., p. 55.

• 210 Côté, 2003-2004, at 1068.

• 211 G.R. Hall, ‘Statutory Interpretation in the Supreme Court of Canada: The Triumph of a common Law Methodology’, Advoc. Q. 21, 1998-1999, pp. 38, 39 <http://heinonline.org>.

• 212 Hall, 1998-1999, at 39.

• 213 Ibid., p. 40.

• 214 Perell, 1998, at 32.

• 215 Revell, 1997, at 1098.

• 216 Nsanze, 2012, at 51.

• 217 Edgardo, 1995-1996, at 189.

• 218 L. Dushimimana, ‘Aspects of Legislative Drafting: Some African Realities’, The Loophole, 2012, 50.

• 219 Beaupre, 1987, at 154.

• 220 Bekink & Botha, 2007, at 14.

• 221 Scassa, 1994, at 179.

• 222 Sullivan, 2007, at 131.

• 223 Salembier, 200-2004, at 78.