Symposium on the VCLT, ILC and Investor-State Disputes: Treaties and the Tower of Babel

Treaty interpretation would be so much easier if we all spoke the same language. According to the Biblical story, once we all did. But divine intervention at the Tower of Babel resulted in the multitude of languages that we speak today. This complicates things considerably, treaty interpretation included.

In many instances, members of a delegation negotiating a treaty have no knowledge of the language of the other delegation. An example we like to cite is that of the 2007 Mexico-Slovak Republic bilateral investment treaty. No member of the Slovak delegation negotiating the treaty spoke Spanish. No member of the Mexican delegation spoke Slovak. The delegations conducted their negotiations in English. Yet the treaty also provides for authentic versions in Slovak and Spanish.

This example illustrates a key policy question: why should any weight at all be accorded to authentic versions such as these? After all, the Slovak and Spanish versions could not have been prepared or reviewed by any member of the delegations negotiating the treaty. The nuances in language reflected in those versions necessarily are those of translators who, no matter how skilled, were not in the negotiating room and could not know what was said. 

For treaties such as the Mexico-Slovak BIT, Article 33 of the Vienna Convention on the Law of Treaties (“VCLT”) gives equal weight in terms of authoritativeness to each authentic version. This aspect of Article 33 was reproduced verbatim from the ILC Draft Articles on the Law of Treaties of 1966. Did the UN International Law Commission get it right in giving authentic versions like these presumptively equal weight to the version in the language of the negotiations?

Our considered view is “yes”. Article 33 did indeed make the right policy call. A treaty is entered into by the State parties to the treaty. It is not entered into by the individuals comprising the negotiating teams. Articles 31 and 32 give primordial importance to the text of the treaty adopted by the States, with only a limited role for the negotiating history. It is fully coherent for Article 33 to do the same.

And this makes sense. The reality is that when a treaty text is brought back for consideration by the legislators and executive branch officers responsible for ratification (in the domestic sense), those officers overwhelmingly review the version of the treaty in the State’s official language. While a State’s ministry of foreign affairs may be conversant in diplomatic languages such as English or French, legislators and other decision makers often are not. 

It is the authentic version in Spanish that the Senate of the Mexican legislature considered and approved. There is no reason to believe that the Slovak legislators considered anything but the Slovak version.

While the treaty may have been negotiated in English, the authentic versions in Slovak and Spanish likely best represent the respective understanding of the Slovak Republic and the United Mexican States as to the content of the treaty they approved.

The ILC and Article 33 were right, in our view, to accord those versions the weight that it does.

A separate question is whether the authentic treaty texts can be uniformly interpreted and applied by arbitrators who do not read or understand the languages in which these treaty texts have been authenticated. How this is done in practice may have direct consequences for the outcome of the dispute. 

Three separate ICSID arbitrations filed against Turkmenistan on the basis of the Turkey-Turkmenistan bilateral investment treaty offer a rare example of different tribunals interpreting the same treaty provision in a language that none of the tribunal members spoke—and coming to three different conclusions.

The tribunals had to determine whether the Turkey-Turkmenistan BIT required an investor to first bring a dispute before local courts as a mandatory precondition to investor-State arbitration. To do so, the tribunals needed to interpret perceived discrepancies in the English and Russian texts of the relevant treaty provision, both texts being equally authoritative. 

When a discrepancy is identified, and as explained by the ILC Special Rapporteur at the time of drafting of the VCLT, Article 33(3) “requires that every effort should be made to find a common meaning for the texts before preferring one to another.”[1] None of the three tribunals was able to distill a common meaning.

In Kiliç, the tribunal concluded that the ordinary meaning of the Russian text, as translated for the tribunal’s benefit into English, was clear: it required the submission of the dispute to local courts prior to the initiation of investor-State arbitration.

In Muhammet Çap, however, a different tribunal reached the opposite conclusion. That tribunal first determined that the ordinary meanings of the English and Russian texts were equally ambiguous. The tribunal proceeded to interpret the ambiguous provisions in light of the treaty’s object and purpose and, ultimately, decided that the treaty offered an option between arbitration and local courts.

The İçkale tribunal ruled that the disputed clause required mandatory prior recourse to domestic courts. In doing so, the tribunal relied on an English translation of the Russian treaty text that was different from the translation submitted to the Kiliç tribunal, as well as testimony of the linguistic experts. The tribunal’s decision was accompanied by two partially dissenting opinions, each of which disagreed with separate findings in relation to the interpretation of the disputed clause. 

In each of these cases the tribunals relied on the English translations of the Russian text, submitted by the parties to the dispute. To the extent that the parties and their counsel agreed on the use of a specific translation—which was different in each case—and framed their arguments in a specific manner, each tribunal was inevitably influenced by the disputing parties’ choices and litigation strategy decisions.

While tribunals may not necessarily be bound by the parties’ arguments if such arguments are inaccurate as a matter of law, an arbitral tribunal cannot easily disregard the translation of an authentic treaty text on which the parties agree and which is presented before it, in particular where the tribunal members cannot themselves read the original text.  

However, in our view, the above examples do not undermine the application of the rules of interpretation in Article 33. The Convention leaves room for discretion, which is consistent with the approach contemplated by the ILC. For the drafting committee, Article 33 resulted not from a codification exercise but rather represented a fresh approach to expressing what was, in the ILC’s view, a “matter of common sense.”[2] Indeed, Special Rapporteur Waldock “had some hesitation in formulating … as a general rule” some of the underlying principles laid down in Article 33.

Additionally, any discrepancy between the interpretations applied by different investor-State arbitral tribunals has limited effect. First, and as seen from the above examples, there is no precedent in investor-State arbitration. Therefore, the decision of one investor-State tribunal, if it is perceived to be erroneous by the parties to the treaty, would not bind future tribunals. Second, bilateral investment treaties typically provide for State-to-State dispute settlement regarding issues of treaty interpretation. Thus, should the parties to the treaty disagree on the interpretation of the treaty clause and consider that an interpretation error has been made by an investor-State tribunal, they can seek to correct such error through State-to-State dispute resolution. 

The limited examples above show that the concerns identified by the ILC at the time of drafting Article 33 remain current. Interpretation necessarily comports an element of subjectivity and discrepancies in the interpretation of the authentic treaty texts cannot be avoided entirely. However, Article 33 leaves sufficient room for the international courts and tribunals to reconcile the authentic treaty texts in a manner that ensures the treaty’s application in practice. In that sense, solutions offered in the VCLT remain as topical today as they were at the time of the VCLT drafting.

The promo code 20VCLT22 offers a 20% discount on the retail price of The Vienna Convention on the Law of Treaties in Investor-State Disputes: History, Evolution and Future in the Wolters Kluwer eStore. The code is valid through to 28 February 2023. 

Anna Crevon is a Partner in Dentons’ Paris office and Global Co-Head of Dentons’ International Arbitration Group. She focuses on contentious and advisory work for corporate clients, States and State-owned entities on issues of international law, including in investor-State and commercial arbitration proceedings. Barton Legum is founding Partner at Honlet Legum. He concentrates on international arbitration, both commercial and investor-State. 

[1] Special Rapporteur Humphrey Waldock, Third Report on the Law of Treaties, YILC (1964), Vol. II, at 63-64, para. 6.

[2] Special Rapporteur Humphrey Waldock, Third Report on the Law of Treaties, YILC (1964), Vol. II, at 65, para. 9.