1 Introduction

In any discussion about interpretation in international law, the rules of interpretation enshrined in Articles 31 and 32 of the Vienna Convention on the Law of Treaties (‘VCLT’) loom large. Yet, quite what role those rules play in the process of interpretation is unclear: to some, they constrain the interpreter’s discretion, directing them to certain elements that must be borne in mind when interpreting a text; to others, reference to the rules plays little more than serving as an indication of lip-service paid by the interpreter to the conventional rhetoric of interpretation in international law;Footnote 1 yet another point of view considers them to play an altogether more restrictive role, dictating a particular interpretive method.Footnote 2 This divergence belies a fundamental disagreement about the status and function of the VCLT rules in the interpretive process.

In this article, I explore the extent to which the VCLT rules could be considered to be ‘disciplining rules’, to borrow a term initially coined by Owen Fiss.Footnote 3 Disciplining rules act to constrain the interpreter and furnish standards against which the correctness of an interpretation can be judged,Footnote 4 functions which I term the ‘directive’ and ‘evaluative’ functions, respectively. The existence of disciplining rules allows us to say whether an interpretation is correct or not, and whether an interpreter has crossed the bounds into the impermissible or illegal. These functions are no doubt useful in trying to curtail the discretion of interpreters and hold them accountable. Yet, disciplining rules’ operation depends on the presence of certain elements: first, the support of a community that recognises those rules as conventionally-valid within that community; and, second, sufficient clarity of the supposed disciplining rules to serve as a benchmark against which to judge the correctness of an interpretation. Using the case study of Argentina’s proposed interpretation of Article 54 of the Convention on the Settlement of Investment Disputes between States and Nationals of other States (the ‘ICSID Convention’), I explore whether the VCLT rules could be thought of as disciplining rules, arguing that the pushback against Argentina’s interpretation tells us something about the ‘ruleness’ of the Vienna Convention provisions.

The article proceeds in three sections. The first section outlines the concept of disciplining rules and criticisms made of that concept. It places this debate in its historical context, highlighting the importance of the then-emerging critical legal studies (CLS) movement to the creation of the concept, and draws a parallel between the (social) conventional basis of disciplining rules and H.L.A. Hart’s secondary rules of recognition, change, and adjudication. The second section moves to examine whether the VCLT rules could be considered to be discipline rules from two perspectives. The first, historical perspective analyses whether the drafters of Articles 31 and 32 ever intended those rules to have the normative, ‘benchmark’ character that is characteristic of disciplining rules. The second part moves to examine whether the VCLT rules operate in practice as a benchmark in accordance with which we can determine an interpreter’s compliance, using the above-mentioned example of Argentina’s interpretation of Article 54 ICSID as a case study. The final section draws a distinction between the ‘thick’ and ‘thin’ evaluative dimension of rules, and suggests that the VCLT rules are best not thought of as acting as disciplining rules à la Fiss but could be thought of as a way for interpreters to manifest their membership of the ‘invisible college’ of international lawyers.

2 The Concept of Disciplining Rules

Articles 31 and 32 of the Vienna Convention enshrine what are commonly thought to be the principal elements of the customary international rules of interpretation.Footnote 5 Students, academics, and practitioners alike are well-versed in concepts that those articles mention, such as ordinary meaning, context, subsequent practice, and travaux préparatoires.Footnote 6 To be able to explain and justify one’s proposed interpretation in terms of Articles 31 and 32 is to demonstrate membership of a group, to show that one is an international lawyer properly so-called.

However, all but the most dogmatic commentators recognise that these provisions provide,Footnote 7 if not infinite, at least significant flexibility to the interpreter.Footnote 8 What counts as ordinary meaning, context, or object and purpose is not dictated by the rules of interpretation but is left to the interpreter to resolve themselves. Moreover, if and how one of these elements should take priority over another—object and purpose trumping ordinary meaning, for exampleFootnote 9—is left to the good judgment of the interpreter. The VCLT rules seem not to dictate one particular interpretive method but rather recognise, and perhaps structure, numerous different elements that may or not may be relevant for interpretation in any particular case.

This leaves us with a paradox: how can provisions that are recognised not to constrain an interpreter’s discretion in any significant way be referred to as rules? In what way do they possess ‘ruleness’?Footnote 10 From a formal point of view, Articles 31 and 32 are clearly conventional rules insofar as they are provisions of an agreement constituted between states. However, I want to dig a bit deeper into the ‘ruleness’ of the VCLT provisions and ask what it means for a formally-valid legal provision to be considered as a ‘rule’ of interpretation.

2.1 Fiss’ Conception of Disciplining Rules

International lawyers are not the only ones that have had to broach this question. Formal rules of statutory interpretation are found relatively frequently in domestic jurisdictions,Footnote 11 and debate regarding both the effectiveness and the character of these rules has been common in domestic law scholarship.Footnote 12 Some of the early work on this topic took place against the backdrop of the emergence of CLS, a new branch of legal scholarship that developed in the 1970s, the effects of which would spread across the world and spillover into myriad sub-fields of law.Footnote 13 CLS scholars argued that judges (for judicial interpretation was their main concern) were not in any way constrained by the text in front of them; interpretation was a radically-subjective operation that allowed the expression of any preference that the judge wanted to privilege in a certain situation.Footnote 14 From this standpoint, rules of interpretation are neither here nor there—they fail to constrain the interpreter in any real way because interpretation is inevitably and unavoidably political; rules of interpretation simply provide another layer of formal subterfuge that judges could throw over their inherently-subjective decision-making.

It was against this backdrop that Owen Fiss argued for the use of disciplining rules for interpretation. His argument was that interpretation both can and should be an objective activity, one that is capable of being constrained by outside norms as well as being evaluated against objective benchmarks. In his words, objectivity in the law:

connotes standards. It implies that an interpretation can be measured against a set of norms that transcend the particular vantage point of the person offering the interpretation. Objectivity implies that the interpretation can be judged by something other than one's own notions of correctness. It imparts a notion of impersonality. The idea of an objective interpretation does not require that the interpretation be wholly determined by some source external to the judge, but only that it be constrained.Footnote 15

In order to have the desired objectivity, Fiss argued that two things needed to be in place. The first are disciplining rules, which operate both as the normative guidelines for interpretation (the ‘directive function’) as well as the benchmark against which the correctness of an interpretation can be assessed (the ‘evaluative function’). The latter point echoes what Gerald Postema has called the ‘evaluative dimension’ of rules, the attribute according to which ‘law provides standards by which law-subjects evaluate their behaviour and that of others (including officials)’.Footnote 16 The VCLT provisions would seem prima facie to constitute disciplining rules par excellence, but the following section explains why from a historical and practical perspective there are reasons to be sceptical about this.

The second element that Fiss considered to be necessary was the presence of an interpretive community that recognised those disciplining rules to be authoritative. In Fiss’ conception of interpretive community (unlike the broader conception subsequently adopted in the context of international law by Ian Johnstone),Footnote 17 the crucial point is that there is agreement within that community on what the authoritative disciplining rules that should guide interpretation are.Footnote 18 This agreement within the interpretive community permits an observer to adopt criticism from a perspective that would otherwise not be available: the ‘internal’ point of view. Put another way, the interpretive community, by recognising that certain rules constitute disciplining rules, provides a benchmark against which an external observer can assess whether an interpretation is correct according to the standards set by the community itself. This is to be contrasted to the external point of view for evaluating interpretation, according to which the evaluator could have recourse to any political, social, or extra-communal legal benchmarks of what they consider to be an appropriate interpretation, whether that be justice, stability, equity or some other factor. The important point is that, in the latter case, it is the evaluator and not the interpretive community that determines the relevant metric against which to assess the interpretation.Footnote 19

In distinguishing between internal and external points of view, Fiss’ argument echoes H.L.A. Hart’s distinction between the internal and external points of view of a legal system. Hart’s argument was that in order to understand how the law operates, we need to conceive of a legal system not as a top-down system of commands backed with coercive force but rather as a system in which the actors themselves understand certain rules as imposing obligations upon them.Footnote 20 In language that is strikingly similar to Fiss’, Hart argues that rules recognised within the legal system as valid act ‘both as guides to their own future conduct and as standards of criticism which may legitimate demands and various forms of pressure’,Footnote 21 and that understanding why those rules are recognised as obligation-imposing (in particular, by identifying and describing the rule of recognition in the legal system) constitutes an important task of descriptive legal theory.Footnote 22

The concept of disciplining rules (and the related concept of an interpretive community) is useful for us when thinking about whether the VCLT rules bring any measure of objectivity to the interpretive process. Understanding the ‘ruleness’ of disciplining rules as stemming from the role, or function, those rules play in an interpretive community—acting as a guide for action and as a standard for evaluation and criticism—rather than from their formal status as treaty provisions, acts as an extra layer of what it means for something to be usefully considered as a ‘rule’ in the context of interpretation, a layer that moves beyond the barebones approach of formalism.

2.2 Criticism of Disciplining Rules

The idea of disciplining rules was not uncontentious in the literature. One particularly vociferous critic was Stanley Fish, who argued that such rules neither successfully function to constrain the interpreter, as Fiss had argued, nor were they in fact necessary.Footnote 23

Fish’s point regarding the necessity of disciplining rules relates to the situation of the text and the interpreter in their particular professional and historical context, and whether those factors obviate the discretion that the interpreter might otherwise have. Put another way, given judges’ professional training, their socialisation within the interpretive community, and the context in which the law is embedded, in Fish’s view, the flexibility that Fiss tries to constrain with disciplining rules simply does not exist: ‘Just as rules can be read only in the context of the practice they supposedly order, so are those who have learned to read them constrained by the assumptions and categories of understanding embodied in that same.’Footnote 24

The problem with this argument is that interpreters that are socialised into the same interpretive community, reading the same text, in the same (or very similar) historical, political, and legal contexts sometimes come to different legal conclusions. A case in point on the international plane is a series of investment treaty arbitrations brought by investors against Argentina in the wake of the latter’s 2001 financial crisis. In those cases, investment treaty arbitral tribunals were called upon to interpret Article XI of the US–Argentina bilateral investment treaty, which stated that the treaty shall not preclude measures inter alia ‘necessary for the […] protection of its own essential security interests’, in light of Argentina’s argument that the emergency measures that were the root cause of the disputes were indeed necessary to prevent economic and social collapse. Presented with the same facts, the same parties, and the same arguments, different tribunals came to divergent conclusions regarding the interpretation of Article XI and its application to the case at hand, with the tribunal in LG&E upholding Argentina’s argument whilst other tribunals (including those in CMS, Enron, and Continental Casualty) rejected the very same argument.Footnote 25 One might argue that the arbitrators on those panels in fact belonged to different interpretive communities as the result of their different professional experiences.Footnote 26 But that, however, begs the question: how similar does one’s educational, socioeconomic, and professional background need to be to constitute a member of an interpretive community? Indeed, on the international level we might, if anything, expect interpreters to be less constrained by context than adjudicators on the domestic level due to their experiential differences. Nevertheless, the point that context, broadly understood, acts to limit an interpreter’s discretion clearly is valid—the interpretive possibilities open to a judge at the European Court of Human Rights are different to those open to a member of the World Trade Organization’s Appellate Body—however, Fish’s argument goes too far in denying that any interpretive discretion exists at all.

The more relevant point to our enquiry is Fish’s argument that disciplining rules are not in any case capable of constraining an interpreter. In international law scholarship, others have noted the seeming paradox that results from trying to limit interpretation by written rules of interpretation.Footnote 27 But Fiss deflects this criticism by arguing that disciplining rules do not necessarily have to be absolutely determinate to serve the purpose of providing objectivity to interpretation, but rather that, if there is a disagreement regarding the interpretation of the disciplining rules themselves, ‘[t]o resolve this dispute, the disciplining rules must be interpreted, and the process of interpreting those rules must itself be constrained by norms further along or higher up the spiral [of reasoning]’.Footnote 28 Fiss himself recognises the limitations to this (‘if the dispute about any norm is so pervasive as to return one to the previous level of constraint, then we have made no progress’),Footnote 29 but nevertheless maintains that ‘[d]isputes over the meaning of a text deny neither the existence of a text nor that it has a meaning which can inform, guide or constrain intellectual processes’.Footnote 30

The problem with this argument is that it assumes that there are higher-level rules that may inform—but which have not been integrated into—disciplining rules in case of contested interpretation. This point, at least to me, is unclear. Consider the rule of UK statutory interpretation that an interpreter should search for the intention of Parliament when it passed that law.Footnote 31 That does not tell us whose intentions are to count (those that introduced the legislation? Members of Parliament as a whole?), how those intentions are to be evidence (are we limited to Parliamentary debates? Or might legislative preparatory work also be probative?), and what background assumptions can we make about those intentions (do democratic legislatures presumptively intend to protect human rights?).Footnote 32 If a disciplining rule fails to give guidance on these matters, it would seem to be unlikely that a ‘higher order’ rule would be capable of providing clearer direction.

Here, I think that Fiss does himself a disservice by failing to emphasise the conventional constraints on this kind of questions that might result from the interpretive community itself. It is conventionally-accepted in international law, for example, that the work of the International Law Commission (ILC) that preceded an international conference may be referred to as part of the travaux préparatoires of a treaty.Footnote 33 This neither comes from the text of Article 32 of the Vienna Convention nor from a higher norm that points one to this conclusion; rather, it is the practice of the interpretive community itself that ‘fills the gap’.

3 The VCLT as Disciplining Rules

The previous section introduced the concept of disciplining rules, identified their directive and evaluative functions, and highlighted two issues regarding their ability (or necessity) to provide objectivity to interpretation. This section moves to look at the VCLT rules in particular, asking whether they might be qualified as disciplining rules in the sense intended by Fiss. The first part of the section takes the historical approach to the question; that is to say, were the VCLT rules ever intended by their drafters to fulfil the functions of disciplining rules? The second section then continues to consider whether the VCLT rules have in practice been used as a benchmark against which to evaluate interpretation.

3.1 Were the VCLT Rules Intended To Be Disciplining Rules?

Fiss’ argument is based on the premise that the enactment of disciplining rules is an intentional, and desirable, action taken to curtail the discretion of the interpreter and to provide objectivity to interpretation.Footnote 34 But when one looks at the drafting history of Articles 31 and 32 of the Vienna Convention, this seems to be far from the principal motivation that underpinned the elaboration of those provisions. Instead, on the one hand, there was general acknowledgement of the undesirability (if not impossibility) of constraining the interpretative process with rules; whilst, on the other hand, there was a sentiment that any convention on the law of treaties should include provisions on interpretation.

The ILC started work on the law of treaties in 1950, a topic to which it had accorded priority in its first session in 1949,Footnote 35 although it was only when the project reached the hands of the final rapporteur, Sir Humphrey Waldock, that the ILC considered the topic of treaty interpretation. In his Third Report (the first to address treaty interpretation), Waldock made several decisions that indelibly influenced the direction that the ILC’s work took. The first was to define the scope of the rules that the ILC codified. According to Waldock, many of the principles and maxims of interpretation frequently cited by tribunals are:

for the most part, principles of logic and good sense valuable only as guides to assist in appreciating the meaning which the parties may have intended to attach to the expressions which they employed in a document. Their suitability for use in any given case hinges on a variety of considerations which have first to be appreciated by the interpreter of the document […] [i]n other words, recourse to many of these principles is discretionary rather than obligatory.Footnote 36

Waldock thought it inadvisable to codify such principles and maxims because their application was indelibly context-dependent, and that ‘detached from that context they retain a certain fictitious ring of unassailable truth’.Footnote 37 In contrast, he thought that ‘the comparatively few rules which appear to constitute the strictly legal basis of the interpretation of treaties’Footnote 38 could be usefully codified. Waldock proposed four draft rules of interpretation that were inspired by the 1956 Resolution of the Institut and Gerald Fitzmaurice’s articles in the British Yearbook on the practice of the ICJ.Footnote 39 The cornerstone of Waldock’s approach—and the subsequent leitmotif of the VCLT project—was that:

the basic rule of treaty interpretation [is] the primacy of the text as evidence of the intentions of the parties […] the text must be presumed to be the authentic expression of the intentions of the parties; and that, in consequence, the starting point and purpose of interpretation is to elucidate the meaning of the text, not to investigate ab initio the intentions of the parties.Footnote 40

Whilst the records show that the reaction of the ILC members was ‘splintered and somewhat uncertain’,Footnote 41 a twist of fate gave the membership the opportunity to reject almost unanimously more constrictive draft articles. During the latter stages of the 1964 session, Waldock presented revised draft articles that purported to limit the scope of recourse to the preparatory work, seemingly in the mistaken belief that he had been instructed to do so by the members of the Commission.Footnote 42 Crucially, his changes eliminated recourse to the travaux to confirm a meaning arrived at by application of the general rule. Such an approach would have given significantly more weight to the text of the treaty by curtailing the interpreter’s ability to examine wider evidence of the parties’ intentions. However, faced with the new draft, the members were nearly unanimous in their desire to reinstate the ‘fairly wide use of the travaux préparatoires’ permitted by Waldock’s previous draft,Footnote 43 emphatically rejecting any suggestion that the text was in itself sufficient to discern the intention of the parties.

The draft articles were next debated by the ILC in 1966, a year after states’ representatives in the Sixth Committee of the UN submitted comments on the draft articles adopted by the Commission in 1964.Footnote 44 Of particular note were the comments of the United States (US), which foreshadowed the views expressed forcefully by Myres McDougal at the Vienna Conference.Footnote 45 The US questioned the wisdom of codifying principles of interpretation as ‘rules’ and criticised what it saw as the ‘apparent primacy given to the ordinary meaning’ and the undue restriction on recourse to the preparatory work that was reflected in the 1964 articles.Footnote 46 Waldock’s response to the comments of the US was unequivocal. In relation to whether principles of interpretation should be codified as ‘rules’, he understood the comment ‘primarily as a caveat against formulating the general principles for the interpretation of treaties in such a manner as to give them a rigidity which might deprive the process of interpretation of the degree of flexibility necessary to it’.Footnote 47 This was, in his view, a misunderstanding of the approach that the ILC had taken at the 1964 session:

The Commission was fully conscious in 1964 of the undesirability—if not impossibility—of confining the process of interpretation within rigid rules, and the provisions of [the 1964 draft articles] when read together, as they must be, do not appear to constitute a code of rules incompatible with the required degree of flexibility […] [i]n a sense, all ‘rules’ of interpretation have the character of ‘guidelines’ since their application in a particular case depends so much on the appreciation of the context and circumstances of the point to be interpreted.Footnote 48

Elsewhere in his Sixth Report, Waldock reiterated that it was the interpreter’s appreciation of the particular circumstances of the case at hand that guided the process of interpretation, and not any preordained method of interpretation that was codified in the draft articles. Most strikingly, he described the elements enumerated in the draft articles as non-hierarchical, stating that they would be, ‘so far as they are present in any given case, thrown into the crucible and their interaction would give the legally relevant interpretation’.Footnote 49 In the debates, members supported Waldock’s statement that the draft articles did not create a hierarchy amongst the various methods of interpretation,Footnote 50 a point that Roberto Ago expressed in particularly unequivocal terms: ‘[t]he separation of [the draft articles] […] did not in any way imply that the Commission was taking a position in favour of one theory rather than another’.Footnote 51 Similarly, the Brazilian member, Gilberto Amado, remarked that the Special Rapporteur had convincingly ‘shown that the various means of interpretation could be employed simultaneously […] [there] was no hierarchy and no precedence of one means of interpretation over another’.Footnote 52 The result of the near unanimity amongst members in the debates was, in the words of Waldock, ‘a return to the scheme of the 1964 text’,Footnote 53 which was adopted without dissent.Footnote 54

The final draft articles adopted by the ILC were sent to the General Assembly, which decided to convene an international conference to ‘embody the results of [the ILC’s] work in an international convention and such other instruments as it may deem appropriate’.Footnote 55 The Vienna Conference on the Law of Treaties considered the articles on interpretation in its first session over a short period of three days,Footnote 56 during which the debates were coloured by the very first intervention that took place—that of the delegate of the United States, Myres McDougal. McDougal had vociferously criticised the ILC draft articles in his private writings, arguing that ‘[t]he great defect, and tragedy, in the International Law Commission’s final recommendations about the interpretation of treaties is in their insistent emphasis upon an impossible, conformity-imposing textuality’.Footnote 57 At the Vienna Conference, he continued on the same line of attack, criticising the draft articles for embodying an ‘over-rigid and unnecessarily restrictive’ textual approach, which was based on the ‘obscurantist tautology’ that the text could be interpreted without reference to extraneous evidence of the parties’ intentions.Footnote 58 In addition to being futile (if not impossible), this approach was, in his eyes, at odds with the overwhelming body of arbitral and state practice, which ‘bore out the right of the interpreter to take into account any circumstance affecting the common intent that the parties had sought to express in the text’.Footnote 59

Whilst some were sympathetic to McDougal’s reading of the draft articles,Footnote 60 the majority recognised that his characterisation of the ILC draft articles bore no resemblance to reality. In response to McDougal’s charge of rigid textualism, the delegate of the United Kingdom, Sinclair stated that it was inconceivable that the ILC had intended to adopt a literal approach to interpretation which advocated that the ‘interpreters of treaties should arbitrarily select dictionary meaning when construing treaty texts’.Footnote 61 This understanding was shared by the majority of delegates at the conference, which commended the Commission for preparing articles that were ‘flexible enough to become a most useful instrument of treaty interpretation’.Footnote 62 Indeed, such overwhelming support for the ILC’s draft articles is hardly surprising considering that 14 members of the Commission were present at the Vienna Conference as representatives of their countries.Footnote 63

The discussion of the rules of interpretation at the First Session of the Vienna Conference came to a close with an intervention by Sir Humphrey Waldock in his capacity as Expert Consultant to the Conference, who aimed to lay to rest any misconception that the draft articles embodied a strict methodology. Along with other members of the ILC that were present at the Vienna Conference,Footnote 64 he rejected squarely McDougal’s claim that the ILC had advocated a literal approach to interpretation, emphasising that ‘nothing could have been further from the Commission’s intention than to suggest that words had a “dictionary” or intrinsic meaning in themselves’.Footnote 65 He also took aim at the contention that the ILC’s draft articles prioritised the text to the detriment of other indications of the Parties’ intentions, noting that any arrangement was based on matters of practical consideration rather than an hierarchical ordering.Footnote 66 The ILC draft articles on interpretation were subsequently referred to the drafting committee by the Committee of the Whole with only stylistic amendments. The articles were subsequently adopted at the Second Session of the Vienna Conference, held in April and May 1969, with no further substantive discussion.Footnote 67

Although often defining their positions in juxtaposition to those of their colleagues, those involved in the codification efforts took a remarkably similar view of interpretation, which recognised that the highly-context specific nature of the activity and the impossibility of determining a priori what constitutes an appropriate interpretation in any given context. As a result, the rules codified in the VCLT are permissive, non-hierarchical and eminently capable of being invoked in a wide range of different scenarios. What they do not do, however, is dictate how to interpret a text. The VCLT articles were never intended to provide objectivity to interpretation in the manner that Fiss suggested disciplining rules might. But whilst the debates certainly paint that picture, the reality of how the VCLT rules operate is more nuanced. For the VCLT rules do seem to act as some kind of benchmark against which we evaluate interpretation, albeit perhaps not in the same way as Fiss understood disciplining rules to do so.

3.2 Do the VCLT Rules Operate As Disciplining Rules? Argentina’s Interpretation of Article 54 ICSID

The previous section argued that, from a historical perspective, the VCLT rules were not intended to constrain the interpreter in the manner envisaged by Fiss’ disciplining rules. But how should we explain the fact that Articles 31 and 32 are almost invariably invoked by any interpreter of an international treaty? In a decentralised system, such as the international legal order, one actor’s evaluation of an interpretation proffered by another actor is relatively less systematised and formalised than in domestic legal systems, where a superior court commonly reviews the interpretation of an inferior court. Yet, where judicial or arbitral hierarchies have been created on the international plane, we can gain an insight into whether the VCLT rules operate as an evaluative benchmark.

The illustrative example I will use is Argentina’s proposed interpretation of Article 54 of the ICSID Convention and its review by ad hoc annulment committees established under that Convention. The Argentinian example serves as an instructive case study because it is one of the few instances in which the same proposed treaty interpretation has been analysed and adjudicated upon by numerous different tribunals. This resulted from the fact that Argentina faced over 40 investor-state dispute settlement cases brought by claimant investors, the majority of which were rendered within the framework of the ICSID Convention and related to the same set of facts. The substantive defence of Argentina on the merits in these cases centred around the severe economic and social crisis from which the emergency measures at issue resulted. However, once final awards in some early cases were rendered, Argentina refused to comply voluntarily with the compensation awarded on the basis of an argument based on its interpretation of Articles 53 and 54 of the ICSID Convention.

At issue was Article 54(1), which in relevant part reads:

Each Contracting State shall recognize an award rendered pursuant to this Convention as binding and enforce the pecuniary obligations imposed by that award within its territories as if it were a final judgment of a court in that State [emphasis added].

Argentina argued that Article 53 of the ICSID Convention, which obliges parties to the arbitration to comply with the final award of a tribunal,Footnote 68 must be read in conjunction with Article 54, which, it argued, placed award creditors were in exactly the same situation as individuals in whose favour an Argentinian domestic court judgment had been issued; that is, they had no automatic right to payment of compensation due and were required to go before domestic Argentinian courts to enforce their award.Footnote 69 In its view, permitting enforcement without requiring the award creditor to go before domestic courts would in effect discriminate in favour of foreign investors, placing ICSID awards on a level superior to judgments of Argentine courts, something that was ‘certainly never intended by Article 54’.Footnote 70

Argentina’s interpretation was at-odds with the prevailing view regarding Article 54 at the time (and now), which considers the provision to provide an award creditor with an automatic right to compliance that bypasses any review that may otherwise have been exercised by domestic courts (aside from verification of the authenticity of the award).Footnote 71 From our perspective, the interesting element is that the ad hoc annulment committees formed to decide Argentina’s requests for annulment in two cases—Sempra and Enron—engaged with Argentina’s proposed interpretation at some length in response to Argentina’s request to stay the enforcement of the award in both cases.

The question that we are interested in is not the interpretation of Articles 53 and 54 per se, but rather whether those committees used the VCLT provisions as a benchmark against which to evaluate Argentina’s interpretation. This was not the case. Both tribunals acknowledged Argentina’s proposed interpretation when recounting the arguments of the parties but then moved to outline their own interpretation of Articles 53 and 54 in reference to the VCLT provisions, emphasising the fact that the terms, context, and object and purpose of those provisions suggested that the obligation to comply with an award was not conditioned on enforcement through domestic courts.Footnote 72 They did not state that Argentina’s understanding of ‘ordinary meaning’ or ‘context’ or ‘preparatory work’ of Article 54 was in some way ‘wrong’ under the VCLT, or suggest that Argentina had contravened in any manner the provisions of the Vienna Convention.Footnote 73

This example provides some useful background when thinking about how the VCLT rules operate in practice. First, it seems clear that the VCLT provisions do not provide a ‘strong’ directive function as Fiss claimed should be the case with disciplining rules. Recall that he acknowledged that disciplining rules need not be unambiguous, but that their directive function could only be fulfilled if the lack of clarity may be resolved by reference to a higher order norm (instructing the interpreter, for example, to privilege a certain body’s intention).Footnote 74 Looking at how the ad hoc committees in Sempra and Enron dealt with the question of ordinary meaning nicely illustrates the point: both committees placed importance on the text of Articles 53 and 54 yet neither committee substantiated their proposed ordinary meaning with reference to any external source; it was simply asserted.Footnote 75 The ambiguity inherent in the terms of Article 31(1) VCLT itself were not resolved by reference to another norm, but were rather resolved as if by reference to common sense. This is not intended to be a criticism of those committees; rather, it demonstrates the limits of the directive function of the VCLT rules.

Second, what is striking about the committee’s decisions is that the VCLT rules do not seem to play the role of an evaluative benchmark—at least not as understood in the common meaning of the term. Argentina’s interpretation is not contrasted to the ‘correct’ interpretation according to the VCLT rules with the result that Argentina’s interpretation was deemed to be, in some way, in contravention of Articles 31 and 32 of the Vienna Convention. Put another way, the committees did not chastise Argentina for incorrectly applying the rules of interpretation, but rather they approached the interpretation of Articles 53 and 54 ICSID as if a tabula rasa, arriving at an interpretive conclusion that was different from that of Argentina.

Third, whilst the VCLT rules might not have provided the specificity necessary to direct the committees’ interpretations nor did they fulfil an evaluative function, it is notable that both Sempra and Enron committees start their process of interpretation by acknowledging expressly that they were bound by the VCLT rules.Footnote 76 This reflects the practice of the vast majority of other international courts and tribunals. In this context, it is useful to distinguish between two different kinds of way that a provision could play an evaluative role.

The first, ‘thick’ evaluative dimension is that envisaged by Fiss and Postema; that is, the rule permits us to say whether a particular act or omission—with relative certainty—is consistent with that rule or not, and allows us to act accordingly. A speed limit, for example, has a ‘thick’ evaluative dimension because we can say unequivocally that someone driving at more than 120 km/h on a Dutch highway is breaking the law. This thick evaluative dimension also provides clear guidance to actors within the legal system about permissible actions and omissions, allowing them to modify their behaviour accordingly.

But one could also imagine a ‘thin’ evaluative dimension, one that does not provide directive guidance in the same manner as a fully-fledged disciplining rule but rather stakes out the boundaries of permissible behaviour of actors. The VCLT provisions may serve this thin evaluative function, setting the outer limits of the interpretative enquiry by prescribing the materials that should be taken into account by the interpreter, where those materials are present. The text of the treaty, for example, cannot normally be ignored in a free-ranging search for the intentions of the parties, nor can one disregard any relevant subsequent practice of the parties in the application of treaty. Within these broad limits, one might say that the VCLT articles provide the standard for criticism of a particular interpretation.Footnote 77

What does the absence of a thick evaluative dimension—both as a result of their design and as a matter of practice—mean for the ‘ruleness’ of the VCLT rules? It is clear that we are not talking about the same kind of rules of interpretation as Fiss had in mind when he developed the concept of disciplining rules: they provide neither the guidance characteristic of such rules nor the clear benchmarks that allow for objective external assessment. Rather, it would seem that Articles 31 and 32 are widely and consistently referred to as ‘rules’ most probably by virtue of their inclusion in a formal legal agreement—the VCLT—rather than by any functional purpose that they serve.

That is not to say, however, that the provisions of the VCLT serve no function. Importantly, they structure how international lawyers talk about interpretation, and how they justify their proposed interpretations. For example, the text of a treaty might not ordinarily be ignored, but if an interpreter can find convincing evidence that the parties to a treaty intended to give a term a special meaning, that meaning might be given effect by reference to Article 31(4) of the Vienna Convention. Diverging from the ordinary meaning of the text of the treaty would, I suggest, be significantly less convincing to an audience of international lawyers if it were not justified by reference to the provisions of the VCLT.

Understanding how to ‘talk the talk’ of the VCLT rules is part of one’s training as an international lawyer, and distinguishes those within from those outside the discipline. In this way, the VCLT rules might not operate as disciplining rules, but are more akin to what were referred to by Wittgenstein as the rules of grammar: they are ways that we learn to talk about (or to do or to play) something in accordance with the conventionally-accepted practices of that activity in a particular society, and according with which we must abide in order to be identified as doing that activity.Footnote 78 They are analogous to the rules of a game: they constitute what actions are understood as engaging in an activity (whether that be interpretation in international law, a game of chess, etc.); they set the outer limits of what is possible, and thus constitute, in a broad sense, benchmarks for criticism; but they dictate no particular strategy, instead leaving their deployment to those engaged in the game (even if some tactics are recognised as being better and others worse).Footnote 79 The VCLT provisions might operate as rules, but in a manner completely unlike the preconceived notions of many international lawyers.

4 Conclusion: Disciplining Rules or Rules of a Discipline?

In what way are Articles 31 and 32 of the Vienna Convention ‘rules’ of interpretation? My aim in this piece was to explore what we mean by ‘rules’ in this context, and whether our use of that term, as opposed to similar concepts, like ‘maxims’ or ‘canons’ of interpretation, belies a qualitative difference.

A natural starting point for this enquiry was Owen Fiss’ concept of disciplining rules—intentionally-enacted interpretative directives which operate to guide the interpreter and to provide an objective benchmark for external criticism for interpretation. From both a historical and a practical perspective, the VCLT provisions seem ill-suited to fall within the definition of such rules. A review of the debates that led to the adoption of the VCLT show that none of the main protagonists thought that what would become Articles 31 and 32 dictated a particular method of interpretation, and that they were never intended to have what we have termed a ‘directive function’. Nor does practice seem to show that the provisions have an ‘evaluative function’, operating as an objective benchmark against which the success of a particular interpretation is assessed. Accordingly, it would be hard to attribute the ‘ruleness’ of the VCLT provisions to their function as disciplining rules.

Perhaps the most straightforward (and most likely) manner of explaining why the VCLT provisions are referred to by virtually all international lawyers as rules is simply because of their inclusion in a treaty, an explanation that, although clearly correct, seems unsatisfactory. I argue that the role that the VCLT provisions play in the structuring and justification of the act of interpretation in international law could also be viewed as a characteristic of rules, albeit one that stems from an avowedly non-legal sense of the term. The ambiguity surrounding the ‘ruleness’ of the VCLT provisions stems from the lack of clarity of the very concept of rules itself, with formal and functional conceptions of rules operating implicitly in international legal discourse. Whilst this article does not claim to have resolved these issues, it does aim to have promoted greater reflection on use of the everyday terminology of international lawyers.