Introduction
Should judges of the Supreme Court of Canada be bilingual? More bluntly, since everyone expects them to speak English, should they also speak French? Since the mid-2000s, judicial bilingualism has become increasingly contentious. Opposition members of parliament (MPs) have fueled the controversy by tabling bills to amend either the Official Languages Act Footnote 1 (OLA) or the Supreme Court Act Footnote 2 (SCA) (for a summary, see Jimenez-Salcedo Reference Jiménez-Salcedo2020 and Bédard-Rubin Reference Bédard-Rubin2021a). Amending the former, which the Liberals promised to do in their comprehensive reform of the OLA, would require the Supreme Court to ensure justices were functional in the official language chosen by the parties, like all other federal courts. The Court could thus probably function with seven or eight bilingual justices. Amending the SCA, by contrast, would impose a blanket requirement that all future Supreme Court justices master both official languages before their appointment. Opponents of the bilingualism requirement, in either form, have decried its incompatibility with regional representation (e.g., Doughart Reference Doughart2016), its hindrance on the promotion of diversity on the bench (e.g., Ha-Redeye Reference Ha-Redeye2016), its hampering reconciliation with Indigenous peoples in reasserting the primacy of colonial languages and legal traditions (e.g., Nasager Reference Nasager2019; Matthews Reference Matthews2019), or its incompatibility with the prioritization of judicial competence in general (for a summary, see Bédard-Rubin Reference Bédard-Rubin2021b). With the implementation of a more open judicial appointment process since 2004, judges have been questioned extensively about their linguistic capacities (Lawlor and Crandall Reference Lawlor and Crandall2015). Language has become one of the main features of the media coverage of the Supreme Court appointment process (Schneiderman Reference Schneiderman2015; Crandall and Lawlor Reference Crandall and Lawlor2015). As we write these lines, in the middle of a federal election campaign, Supreme Court bilingualism is being used by journalists as a barometer for the commitment of federal parties to official bilingualism.
Yet, judicial bilingualism is seldom discussed in depth. It is often presented as a partisan issue and has had little impact on the legalist frame used to discuss the Court’s work (Bédard-Rubin Reference Bédard-Rubin2021b). Difficult questions are overlooked such as the definition of “functional bilingualism,” the level of fluency in French that this entails, and the methods that should be used to assess candidates for the highest judicial office in Canada.
Research on the linguistic dimension of judicial behaviour is scant. In the first large-N empirical study on the impact of unilingualism in the Supreme Court of Canada, we tried to assess whether unilingualism makes a difference at all (Bédard-Rubin and Rubin Reference Bédard-Rubin and Rubin2018). Our research findings reveal that unilingual Anglophone justices sit less frequently on federal law cases argued in French, that they voice their opinion less often in those cases, and that, ceteris paribus, most of their federal law caseload is composed of opinions written in cases argued in English. The findings also show that unilingual Anglophone justices vote to overturn Francophone lower court decisions at the same rate as their bilingual colleagues.
Our primary goal in the present article is to clarify and disambiguate the linguistic expectations underlying the main arguments in favour of judicial bilingualism, to provide a finer picture of the level of proficiency in French of individual justices across time, and to construct a score composed of four indicators tracking the use of French in the Court. Our score suggests that there is wide variation between justices’ level of bilingualism not always matching the qualitative assessment done by parliamentarians or presented in media coverage. These findings underscore that better and more transparent testing of judges is necessary. Building on the literature on panel effects, we also discuss hypothetical explanations for the variation in the levels of bilingualism observed across time. While the overall story remains to be further clarified, our results suggest that more attention should be paid to the level of bilingualism required of judges and to the institutional rather than the individual dimension of bilingualism, which has monopolized the debate so far. If we want to have a better picture of the overall linguistic capacity of justices and the ability of the Court to discharge its duty in both official languages, the influence of judges on one another (or panel effects) needs to be foregrounded. We think that these results warrant further research on the relationship between language proficiency and judicial behaviour and that they have the potential to reorient and enrich the normative debate with a better positive theory of judicial behaviour that takes language seriously.
The first section (I) disentangles the different components in the debate about mandatory bilingualism. The second section (II) presents the data, the methodology, and the indicators we use in the construction of our score. The third section (III) discusses our results. The fourth section (IV) briefly discusses panel effects and their usefulness in conceptualizing the role of language in the institutional dynamic of the Court.
I. French, Sure, But How Much?
Official bilingualism generates different kinds of arguments from its proponents, who stress its symbolic importance, and its opponents, who underscore pragmatic challenges in its implementation (Charbonneau Reference Charbonneau2015). Both camps generally talk past each other, and the issue of judicial bilingualism is no exception (Bédard-Rubin Reference Bédard-Rubin2021b). The ideas, ideals, and assumptions about judicial behaviour that form the background of the conversation rarely get tested or developed. This first section disentangles the normative arguments in the debate from the positive theories of judicial behaviour on which they are based and their epistemic presuppositions.
1. Media Coverage, Parliamentary Hearings, and Public Perception
Before 2006, when public hearings for Supreme Court nominees were first implemented, there were no direct ways of assessing the fluency of Supreme Court justices in both official languages. When it mentioned language at all, media coverage was inconsistent and relied mostly on anecdotal evidence.
As linguistic expectations for high public officials evolved, reports on individual judges changed over time. For example, in 1977, the Globe and Mail wrote that “all members of the Court now understand or speak some French” (Canadian Press, 1977). Yet, eight years earlier Peter H. Russell had concluded in a thorough study on the place of French and English in the Court that “by any reasonable measure of bilingualism, the Court has failed” (Russell Reference Russell1969, 213). In this context, “some French” sounds unfortunately vague. An article published in La Presse in 1984 said that Brian Dickson was “by far the most bilingual of Anglophone judges” (our translation, Auger Reference Auger1984a). A couple of weeks later the same newspaper published an article saying that he was “perfectly bilingual” (Auger Reference Auger1984b). His biographers are much more skeptical about his “modest proficiency in French” that “never quite matched his genuine enthusiasm for bilingualism” (Sharpe and Roach Reference Sharpe and Roach2003, 413). In recent years, Dickson and Bertha Wilson have become go-to examples for opponents of mandatory bilingualism. They argue that fluency in French and English is too high a threshold and that it would prevent excellent judges from being elevated to the Supreme Court (e.g., Gardner Reference Gardner2010; Heuser Reference Heuser2016). Others claim that the former Chief Justice, Beverly McLachlin, is living proof that judges can learn while they are on the bench and that requiring bilingualism at the time of appointment is too strict a standard.
Proponents of mandatory bilingualism reply by saying that we should dispel “the illusion that a unilingual Canadian can always go to night school and learn the other language in a few months” (Slayton Reference Slayton2011, 252). For example, commentators say that Marshall Rothstein failed in his endeavour to learn French on the bench (McCharles Reference McCharles2015), and his score tends to confirm this assessment.
The devil is in the details. Extra-judicially, Francophone justices have said that their Anglophone colleagues have a very modest knowledge of the French language (Slayton Reference Slayton2011, 130; Buzzetti Reference Buzzetti2010). Anglophone justices have responded that real bilingualism in the Court has always been rare and that Francophone justices would fail to meet, in English, the threshold that they expect Anglophones to meet in French (Tibbetts Reference Tibbetts2010). Marion Buller, a First Nations judge from British Columbia, recently said: “[Supreme Court justices] should have some functionality in both languages, but I don’t think that there should be a requirement that they be bilingual” (emphasis added) (Stefanovich Reference Stefanovich2021). As Buller’s comments show, no one seems to know exactly what is or should be the linguistic standard for Supreme Court appointments.
The Liberals promised during the 2015 federal electoral campaign to only appoint bilingual justices to the Supreme Court of Canada (Liberal Party of Canada 2015). The Liberals have thus included “functional bilingualism” as an essential criterion in the new selection procedure, headed by a seven-member independent advisory board, that they have used for the selection of Supreme Court justices since 2016 (Canada 2016). The “functional bilingualism” requirement refers to the definition given by the Office of the Commissioner for Federal Judicial Affairs (OCFJA):
It is expected that a Supreme Court judge can read materials and understand oral argument without the need for translation or interpretation in French and English. Ideally, the judge can converse with counsel during oral argument and with other judges of the Court in French or English.
(OCFJA 2016)In 2016, Malcolm Rowe, the first candidate to be selected through this new administrative procedure, wrote in his application questionnaire that he could, without further training, read and understand court materials, discuss legal matters with colleagues, converse with counsel in court, and understand oral submission in court in both official languages. In addition, Rowe underwent an ad hoc French-language test the results of which were never made public. Commentators generally recognized that Rowe’s appointment set the linguistic bar quite high (Hébert Reference Hébert2016). However, after years of discussion about judicial bilingualism, lack of clarity and transparency with regard to linguistic expectations made it difficult for the public to know what was deemed an appropriate level of French.
The Official Languages Act provides some guidance. It requires all federal courts to provide, if the situation so necessitates, judges or officers “able to understand both languages without the assistance of an interpreter.”Footnote 3 Until the reform of the OLA is adopted,Footnote 4 the Supreme Court of Canada is the only federal court exempted from that obligation. But, even in lower courts, this provision has been insufficient to ensure that litigants can always be heard by a judge fluent in the official language of their choice (House of Commons 2017). Without uniform language tests, unreliable linguistic self-assessment is the only basis chief justices have when assigning judges to cases argued in French or English.
When the Selection committee resumed its work in 2017, 2019, and 2021, Sheilah Martin, Nicholas Kasirer, and Mahmud Jamal, who all ultimately got appointed, self-identified as “functionally bilingual” in their application questionnaires. The language test was formalized and now consisted of three parts involving “reading legal text,” “legal pleading,” and conversation about legal issues in the second official language (Canada, 2018).
After Justice Jamal’s appointment, Chief Justice Wagner weighed in on the issue. He said in an interview that he “has no hesitation in affirming that Canadian citizens have the right to argue and present their case before the Supreme Court of Canada in the official language of their choice, either English or French. It is thus normal that all justices of the Supreme Court of Canada would understand, hear, discuss, write, read in Canada’s two official languages” (our translation, RDI, 2021).
Again, the devil is in the details. The advisory board recommended in its report following Justice Jamal’s appointment that, “on the question of language proficiency, candidates should be able to describe their level of language skills beyond a ‘yes or no’ question.” (Canada 2021). Yet providing the advisory board with more information only compounds the question: what degree of linguistic competence is expected? Moreover, why? Knowing why judges should speak both official languages would help determine how much is required of them and how to test whether that standard is met. The public debate has confused matters further because normative arguments over mandatory bilingualism do not reflect a shared standard.
2. Normative Arguments and Their Implications
Advocates of mandatory judicial bilingualism provide a variety of arguments in support of their position (e.g., Grammond and Power, Reference Grammond and Power2011; St-Hilaire et al. Reference St-Hilaire, Alexis, Chouinard and Gervais2017). However, not only do the different arguments point to varying degrees of bilingualism, there is also virtually no discussion of the ways in which actual linguistic abilities could be tested or monitored.
The most straightforward argument for mandatory judicial bilingualism is a legal one. Some scholars argue that section 19(1) of the Charter, interpreted in light of the recent case law of the Supreme Court of Canada, gives litigants the right to argue and be heard directly in the official language of their choice at the Supreme Court of Canada (Tomkins Reference Tomkins2008; Power and Roy Reference Power and Roy2015; Harrington Reference Harrington2017). If it were recognized in practice, this would extend to the Supreme Court the standard of the Official Languages Act, currently applicable to all other federal courts, that every judge should be “able to understand both languages without the assistance of an interpreter” when litigants so require. Short of language tests, this might be more a cosmetic change than a substantial solution.
A second argument, put forth by Michel Doucet (House of Commons 2008; Doucet Reference Doucet2017), is that simultaneous interpretation does not convey the meaning of the oral arguments of Francophone litigants. According to Doucet, unilingual Anglophone justices who rely on simultaneous interpretation are at risk of misunderstanding important nuances, misinterpreting legal arguments and, ultimately, issuing wrong judicial decisions.
Critics of this position, like Dennis Baker, argue that “there is no evidence … that [judicial unilingualism has] led to substantial injustice to the litigants” or “problematic judicial results coming from these mistranslations or missteps in the translations” (TVO 2010). Both Doucet and Baker’s arguments rely on a common postulate: that it is possible to tell a right from a wrong decision. This position seems empirically difficult to falsify and is highly speculative, as law is by nature “essentially contested” (Dworkin Reference Dworkin1986). Moreover, asking for a smoking-gun-like proof that unilingualism has caused wrong decisions sits quite uneasily with a probabilistic conception of causation (King, Keohane, and Verba Reference King, Keohane and Verba1994).
A third kind of argument, put forth by Sebastien Grammond and Mark Power (Reference Grammond and Power2011), is that limited proficiency in French shrinks the pool of legal materials available to unilingual judges, such as lower court decisions or doctrinal texts written in French. They conclude from their review of the empirical literature that “unilingual judges … are unable to draw upon the rich body of Canadian literature written in French.” In a nutshell, “English-language books and articles overwhelmingly dominate, and French-language texts are mostly cited in judgments dealing with civil law or other issues peculiar to Quebec” (Grammond and Power Reference Grammond and Power2011, 9). Grammond and Power’s argument is complex, but this part of their study points to the ability to read written texts in French, a linguistic threshold different from the one found in the Official Languages Act.
A fourth argument is that unilingual judges cannot interpret bilingual legislation. Since all federal statutes and those of Quebec, Ontario, New Brunswick, and Manitoba are adopted in both French and English and that both versions have equal status, being able to read the law directly in both original and untranslated versions should be obligatory (Bastarache et al. Reference Bastarache, Metallic, Morris and Essert2008; Bastarache Reference Bastarache, Solan and Tiersma2012). Sometimes, being able to read both versions of a statute has important consequences for litigants (McLaren Reference McLaren2013). If judges need a translation to have access to the version of the statutory text in the other language, this defeats the whole purpose of bilingual interpretation. Whether bilingual interpretation should be seen as remedial—i.e. as a tool to interpret a single ambiguous legal provision—or as essential—i.e. as creating a bilingual meaning with the conjunction of both versions (Macdonald Reference Macdonald1997)—raises questions that we cannot address here (see e.g., Sullivan Reference Sullivan2010). If we take as a baseline the remedial understanding of bilingual interpretation, this argument requires a degree of fluency in French that could be quite modest.
A fifth argument builds on the same idea. Considering that Supreme Court justices not only interpret statutes, but also develop the law more generally in their judgments, they should be able to write in both official languages (Doucet Reference Doucet2016). Justice Lebel, who wrote his decisions in the language used by the parties (Slayton Reference Slayton2011), explained that he painstakingly read the translation of all his decisions to make sure that both the French and English versions conveyed the same meaning (Grammond and Power Reference Grammond and Power2011, 6). Developing the law requires a much higher level of mastery of both official languages than is required for the interpretation of discrete and ambiguous statutory provisions.
A final argument is put forward by former Commissioner of Official Languages Graham Fraser (Commissioner of Official languages 2010; Fraser Reference Fraser2016) and former Supreme Court Justices Louis Lebel (Slayton Reference Slayton2011) and Claire L’Heureux-Dubé (Buzzetti Reference Buzzetti2010), among others. They argue that judicial unilingualism puts an extra burden on Francophone justices, as English becomes the lingua franca of the Court. Graham Fraser has argued that this infringes on the right to work in either official language.Footnote 5 Putting this legal question aside, Justice L’Heureux-Dubé also said that she often wrote her judgements in English to get them circulated as early as possible to her colleagues in order to avoid delays caused by translation (Buzzetti Reference Buzzetti2010). This argument aimed at ensuring that Francophones can work in French suggests a more modest benchmark; a mere passive understanding of French among all justices would allow each to express themselves in the language in which they are most comfortable.
3. Empirical Literature and Judicial Behaviour
The empirical literature on judicial behaviour does little to clarify the debate. Except for some brief references (Slayton Reference Slayton2011), qualitative work on the Court generally does not tackle the issue of language (see e.g., Songer Reference Songer2008; Macfarlane Reference Macfarlane2013; McCormick Reference McCormick2014). From a quantitative perspective, the last two decades have seen a burgeoning literature assessing the impact of a range of factors on judicial behaviour in Canada: gender (Johnson and Songer Reference Johnson and Songer2009; Songer, Radieva, and Reid Reference Songer, Radieva and Reid2016; Johnson Reference Johnson2017), party affiliation and patronage (Stribopoulos and Yahya Reference Stribopoulos and Yahya2007; Hausegger, Riddell, and Hennigar Reference Hausegger, Riddell and Hennigar2013), institutional norms (Alarie and Green Reference Alarie and Green2007; Alarie and Green Reference Alarie and Green2017), ideology (Ostberg and Wetstein Reference Ostberg and Wetstein2007), societal value change (Wetstein and Ostberg Reference Wetstein and Ostberg2017), and various characteristics of litigants (Alarie and Green Reference Alarie and Green2010) or their legal teams (Kaheny, Szmer, and Sarver Reference Kaheny, Szmer and Sarver2011). Scholars analyzing judicial behaviour from a quantitative perspective have thus largely imported categories from the American scholarship on judicial politics where multilingualism is not a particularly salient concern, at least at the US Supreme Court level. Except for Peter Russell’s (Reference Russell1969) pioneering work and Bédard-Rubin and Rubin’s (Reference Bédard-Rubin and Rubin2018) article discussed above, language has remained largely out of the scope of the empirical work done on the judiciary in Canada, despite its political saliency. Although some exploratory work done on other multilingual jurisdictions, such as the European Court of Justice (Cheruvu Reference Cheruvu2019), has been conducted, there remains a paucity of empirical literature on the impact of language on judicial behaviour.
The debate on mandatory bilingualism thus proceeds on unstable foundations. It is sometimes unclear what advocates of mandatory bilingualism want to achieve and whether the means they choose are likely to produce the outcome they want. In these circumstances, we lack both conceptual clarity and valid and reliable empirical evidence. The purpose of our study is to fill this gap by providing an objective assessment of the level of bilingualism of Supreme Court of Canada justices between 1985 and 2013. Because there are so few studies on judicial multilingualism, our study is largely exploratory and should be complemented by other kinds of observational and comparative studies.
II. Data, Methodology, and Variables
To compute the bilingualism score, the data set compiled by Alarie and Green of all the decisions of the Supreme Court of Canada between 1954 and 2013 was used (Alarie and Green Reference Alarie and Green2017). It is the most comprehensive data set available of Supreme Court decisions in Canada.
Our inquiry was limited to cases of federal jurisdiction only: Aboriginal law, administrative law, citizenship, immigration and refugee law, civil rights and liberties and human rights, criminal law and procedure, division of powers, intellectual property law, and international law. All justices of the Supreme Court of Canada should have had somewhat equivalent training and knowledge of the law in these areas regardless of whether they were initially trained in civil law or common law.
Our study focuses on the period from 1985 to 2013. The reason is fourfold. First, until Gérard La Forest was appointed in 1985, there seems to have been an implicit linguistic divide on the Court (Bédard-Rubin Reference Bédard-Rubin2021a). The three Quebec justices were bilingual and the justices from the other provinces were, most of the time, unilingual. We want in part here to evaluate the variance between justices deemed to be “bilingual” but who are not from Quebec, and there were few of them before 1985. Second, citations of academic journals, used as one indicator in the composition of the score, only emerged in the late 1970s at the Supreme Court of Canada. It became more widespread with the adoption of the Canadian Charter of Rights and Freedoms in 1982 (McCormick Reference McCormick2004). Third, the format of Supreme Court decisions was standardized in 1985. From then onward, the cases cited in the decision are clearly identified in the heading. Since citation to French lower court decisions is one of our indicators, it was easier to identify them after 1985. Fourth, the data set, though the most comprehensive so far, ends in 2013.
The language of the case was determined in two steps. First, we looked at the decisions of the officially bilingual appellate courts in Canada, i.e. the New Brunswick Court of Appeal, the Federal Court of Appeal, and the Court Martial Appeal Court of Canada. Based on the statements of bilingual justices Louis Lebel and Nicholas Kasirer, indicating that they write their decisions in the language used by the parties, we used the original version of the decision appealed to the Supreme Court to identify the main language of the case. We verified whether the French or English version bore the mention “certified translation” or “This is the English version of the judgement rendered by Justice so-and-so.” Second, we used the language filter in Westlaw to identify which of the cases coming to the Supreme Court of Canada from the Quebec Court of Appeal had been written in English and which of those from other courts of appeal in Canada had been written in French. We are thus confident that our categorization is a good approximation of the entire population of French and English cases to have reached the Supreme Court of Canada in areas of federal law between 1985 and 2013. Our sample included 348 French federal law cases and 1634 English federal law cases for a total of 1982 cases (284 from Quebec 22 from New Brunswick, 42 from federal courts).
The bilingualism score combines four indicators: the linguistic distribution of absences, the proportion of cases heard in French that resulted in an opinion, the proportion of citation of reports or doctrinal writings in French, and the proportion of citation of Francophone lower court decisions.
First, we use an indicator representing the linguistic distribution of absences. This indicator measures the proportion of Francophone federal law cases in the total number of federal law cases for which a Supreme Court justice was absent in a given year. The higher the proportion, the lower we assume their level of bilingualism to be. The underlying assumption builds on Bédard-Rubin and Rubin’s finding (Reference Bédard-Rubin and Rubin2018) that unilingual Anglophone judges are more likely to sit out Francophone cases than their bilingual colleagues. By using a proportion instead of an absolute number, we capture the increase of the average panel size of the Court since 1985 and the fact that some justices tend to hear fewer cases towards the end of their careers.
The second indicator is what we call the “linguistic assertiveness” of judges. This indicator represents the relative frequency with which justices make their voice heard in French and English cases. It is calculated by dividing the number of opinions written for federal law cases argued in French by the number of French federal law cases heard. When hearing a federal law case argued in French, writing the opinion of the majority, a concurrence or a dissent, including co-authored opinions, all increased a justice’s “linguistic assertiveness” .
The third indicator of the score is the linguistic composition of citations of official documents or doctrinal sources in federal law opinions. To create this variable, we identified in the heading of the English version of the decision the texts that were cited in French. Many official documents (e.g., official reports, Hansards, etc.) and some doctrinal texts (e.g., Pierre-André Côté’s Interprétation des lois) are published in both official languages. However, when the work was cited in French in the English version of the decision, we assumed that justices used the French version themselves. Thus, this variable reflects the proportion of unequivocally French texts cited as compared with other texts (almost all English texts) cited by individual justices.
The fourth indicator is the proportion of citations to Quebec lower court decisions in all the decisions referenced by a justice in his or her opinions in federal law cases. For this indicator, we use the Quebec courts as a proxy for Francophone courts.Footnote 6 Because the decisions of the Supreme Court of Canada, the Court of Appeal of New Brunswick, and the Federal Court of Appeal are systematically translated, we cannot know which of their French or English versions were used. Using the decisions of the Quebec Court of Appeal is thus an approximation of the decisions only accessible in French. Since decisions of Quebec courts are used as a proxy for citations to French lower court decisions, this tends to inflate the score of Quebec justices, who are more familiar with these cases.
We then combined the z-scores of each indicator to create the aggregate score. The z-scores, indicating deviations from the mean, standardized all indicators. The score thus measures how far above or below, in terms of standard deviation, a justice is relative to the Court’s mean for the whole period. Combining z-scores allows us to capture the evolution of individual justices and the whole Court across time. Because the caseload of the Supreme Court in federal law cases is rather small, there is a lot of year-by-year volatility, making it difficult to draw inferences. A three-year moving average is used in Table II and Figures 1, 2, and 3 to help tame that volatility. The mathematical details about the construction of the score are given in Appendix A.
Average bilingual score per linguistic group, three-year moving average, 1985–2012. RoC: rest of Canada.
Average score of the court and median justice’s score, 1985–2012.
Highest and lowest bilingualism score, 1985–2012.
The underlying premise is that a judge who almost never sits on Francophone cases, and, even when he or she does, never writes opinions in Francophone cases, never cites official or doctrinal publications written in French, and never cites lower courts’ decisions in French, is more likely to be unilingual—or, at least, less “functionally bilingual.”
To determine the reliability of our score, we used Cronbach’s Alpha (α), a common statistical test in this kind of study. Cronbach’s Alpha (α) for the aggregate bilingualism score is 0.78. According to Nunnally (Reference Nunnally1978), an Alpha of 0.70 or above is acceptable, especially for exploratory research like ours (see also Lance, Butts, and Michels Reference Lance, Butts and Michels2006).
Table I presents the score with percentages corresponding to the four variables used in its construction. Table II provides a dynamic picture of the yearly evolution of the bilingualism score of each justice across time. Figure 1 uses the categorization of Bédard-Rubin & Rubin (Reference Bédard-Rubin and Rubin2018) to compare Francophone and Anglophone bilingual justices from Canada outside of Quebec. It also gives information about the evolution of Quebec justices and unilingual Anglophone justices. Figures 2 and 3 seek to capture the evolution of the Court across time. Figure 2 presents the median justice by aggregate score, and the Court’s average score for every year. Figure 3 presents the highest and lowest scores observed in a three-year window.
Bilingualism Score and Individual Indicator Scores for Justices of the Supreme Court of Canada, 1985–2013
* Results for Justices Wagner, Chouinard, Stevenson, Karakatsanis, and Moldaver are based on very small samples. Results should be interpreted with caution.
Bilingualism score for justices of the Supreme Court of Canada, three–year moving average, 1985–2012
III. Findings and Discussion
The main goal of our score is to provide a means of assessing the extent to which justices use French when they sit on the Court. The score provides two kinds of information. First, from a cross-sectional perspective, it allows us to compare justices individually with one another, to compare the various linguistic groups, and to contrast the level of bilingualism measured here with the information provided by anecdotal sources, such as newspaper coverage. Second, from a longitudinal perspective, the score allows us to see evolutions and patterns in the use of French across time by individual justices and the Court as a whole. In this latter case, it can also provide some useful information for future studies of judicial behaviour in which language is a dependant or an independent variable. We take these two kinds of information in turn.
First, our score shows that Anglo-bilingual justices generally do not use French as much as Franco-bilinguals from outside Quebec do. If we compare justices deemed to be bilingual as identified by Bédard-Rubin and Rubin (Reference Bédard-Rubin and Rubin2018), shown in Figure 1, we see that Franco-bilinguals, on average, use French more often in their work than their Anglo-bilingual colleagues. This is an important finding as it signals a potential hurdle for the advocates of mandatory bilingualism who seek to create a genuinely bilingual legal culture. Having Anglo-bilingual justices appointed might not yield the desired outcome or, at least, might not be as effective as having a balance between Francophones and Anglophones in the Court.
Our score also suggests that some justices might be less bilingual than they are deemed to be. For example, when Michael Moldaver and Andromache Karakatsanis were appointed to the Supreme Court, Moldaver was criticized for his unilingualism. Karakatsanis, for her part, was praised for her mastery of three languages (English, Greek, and French). However, their behaviour does not reflect the general assessment that emerged from the media coverage and their parliamentary hearing. Moldaver’s bilingualism score (-0.68) is just slightly below Karakastanis’s (-0.57) (Table I). We only have two years of data for Moldaver and Karakatsanis so it is hard to draw inferences, but their score suggests that parliamentary hearings might be of limited use when it comes to assessing linguistic capacities.
The case of Justice Major is also particularly telling. In 2008 and 2010, Major fuelled the controversy about bilingualism in a series of interviews in which he repeatedly said that language had made no significant difference in his career (e.g., Boesveld Reference Boesveld2010). Major claimed that simultaneous interpretation and bilingual clerks had allowed him to never feel impeded by his unilingualism. Our score suggests otherwise. Major, like Marshall Rothstein who replaced him, ranked consistently below his eight colleagues every single year of his tenure on the Court. Major also wrote ninety-one solo opinions in federal law cases during his tenure on the Supreme Court, only three of which were in cases argued in French, and one of these consisted of a single sentence.Footnote 7 He did not cite a single doctrinal source written in French throughout his career in federal law cases, nor did he ever refer to a decision of the Quebec Court of Appeal in those cases.
In general, our score also casts doubt on the idea that unilingual Anglophone justices can learn French while they sit on the bench. For example, Justice Rothstein, who promised to learn French on the bench, does not seem to have improved during his tenure (Table II).
Finally, McLachlin’s score is interesting because it illustrates the tension between two aspects of language on the Court. McLachlin was the median bilingual justice of the Court twice during that period and the median justice of the whole period studied (1985–2013), ranking fifteenth out of thirty justices. Her statements also suggest that she felt increasingly at ease in French throughout her career (Slayton Reference Slayton2011). However, as Table II shows, no discernible trend can be inferred from her annual scores, which varied greatly from year to year with no real difference between the beginning and the end of her career. The explanation might be that language was never a significant impediment to her work. Meanwhile, French probably never was her special “expertise” on the Court either. Compared with her other colleagues more fluent in French, she might have never felt bound to hear cases argued in French out of institutional duty, or to cite French texts to signal to Francophone litigants that their arguments had been heard. In fact, McLachlin’s case might be a good illustration of the difference between the institutional and the deliberative aspects of panel effects that we discuss below. McLachlin might have been deliberatively bilingual, though not bilingual enough for her language capacities to make an institutional difference on her or her colleagues’ work.
This brings us to the second kind of information that our score provides. If we look at our findings from a longitudinal perspective, the Court as a whole seems to have increased its level of bilingualism since 1985. Figure 1 shows that while there was a sharp difference between Quebec justices and bilingual justices from the rest of Canada in the 1980s and early 1990s, the gap narrowed in the mid-1990s. Not only did more justices deemed to be bilingual get appointed, but the level of bilingualism of those justices increased significantly as well. Of the ten most bilingual non-Quebec justices, seven were appointed after 1998. Of the ten least bilingual non-Quebec justices, seven were appointed before 1998. Another way of seeing this trend is to look at the Court average for every year, as shown in Figure 2. As we can see, the Court average was ‑0.24 in 1985 and increased steadily to 0.02 in 1993. It dropped again in 1997–1998 but hovered around 0 for the rest of the period under study, showing a rise in the early 2000s when Francophones were a majority on the Court and again in 2009, just after judicial bilingualism became highly politicized. Institutional norms might have changed at that point. The debate about judicial bilingualism may have provoked a shift in the justices’ institutional self-perception and have encouraged justices nominated after the emergence of the debate in 2008 to pay more attention to the integration of French in their work. Justices Cromwell and Moldaver exemplify this shift. Justice Cromwell, appointed in the midst of the controversy in 2008 (Bédard-Rubin Reference Bédard-Rubin2021b), obtained a significantly higher bilingualism score than all his Anglophone colleagues. And Justice Moldaver took steps to signal his progress in French at the hearing of the Supreme Court Reference, which revolved around the question of Quebec representation on the Court. Moldaver conspicuously heard the case without simultaneous interpretation (Dodek Reference Dodek2014) and, as if to make the signal even clearer, he used the techniques of bilingual interpretation, such as the shared meaning rule developed in Daoust, Footnote 8 to interpret sections 5 and 6 of the Supreme Court Act in his lone dissent. We also know from anecdotal sources that language plays a role in the hiring process of Supreme Court clerks (see, e.g., Sossin Reference Sossin1996; Macfarlane Reference Macfarlane2013). Future observational studies about institutional norms should try to understand which institutional mechanisms have been put in place to counter-balance the linguistic limitations of some justices and how they contribute to the overall linguistic capacity of the Court.
IV. Future Research: Institutional and Deliberative Panels Effects
Our score provides data that could be used to evaluate theories of linguistic behaviour, whether language is studied as a dependant variable (what factors influence the level of bilingualism of individual justices?), or an independent variable (what is the relative impact of bilingualism on judicial behaviour?). One can note, for example, that justices appointed by Liberal governments were on average more bilingual (0.06 on average, -0.18 if Quebec justices are excluded) than justices appointed by Conservative governments (-0.20 on average, -0.35 if Quebec justices are excluded).
Also, our score sheds light on two interesting and contrasting panel effects (on panel effects, see Farhang and Wawro Reference Farhang and Wawro2004; Sunstein et al. Reference Sunstein, Schkade, Ellman and Sawicki2006; Kim Reference Kim2009). Panel effects occur when the decisions of individual judges are influenced by the composition of the panel or the Court of which they are part. For example, Songer, Radieva, and Reid (Reference Songer, Radieva and Reid2016) find that women judges on Canadian appellate courts are more likely to vote conservatively in criminal cases than their male colleagues, but that male judges tend to be more conservative as well when they sit on panels with women judges.
In the case of language, we can distinguish two positive (as opposed to normative) theories of judicial behaviour: a deliberative and an institutional one. The institutional effect, a reflection of the ordinal ranking of individual justices on the court, seems to move justices in opposite directions. The deliberative effect, by contrast, seems to move all justices in the same direction.
The observed level of bilingualism of individual justices might thus reflect an ordinal rank as well as a cardinal value. In other words, the use of French by justices might be influenced by their fluency in French relative to their colleagues. For example, between 1999 and 2004, the Court was composed of five Francophone justices (Arbour, Bastarache, Gonthier, L’Heureux-Dubé/Deschamps, Lamer/Lebel), three relatively bilingual Anglophone justices (Binnie, Iacobucci, McLachlin), and only one unilingual Anglophone justice (Major). As shown in Figure 2, Louise Arbour was, between 1999 and 2003, the most bilingual median justice the Court had ever had. Figure 1 also shows that, towards the end of this period, the behaviour of Francophone and Anglophone bilingual justices from the rest of Canada and from Quebec was not ostensibly different. In fact, the level of bilingualism of Quebec justices decreased. It looks as if, as the linguistic capacities of most justices increased, the linguistic burden put on individual justices decreased. The increased bilingualism of justices from the rest of Canada seems to have relieved Quebec justices of their linguistic burden, which was more evenly shared. This narrowed the gap between the scores of Quebec justices and those of their bilingual colleagues from the rest of Canada, as seen in Figure 1 around 2003. Figure 1 also shows that, as the level of bilingualism of justices from the rest of Canada decreased from 2009 onward, the level of bilingualism of Quebec justices increased again.
At the same time, the trajectory of some individual justices tells a different story. For example, when Gérard La Forest was appointed to the Court in 1985, he was the first Francophone from outside Quebec ever to be appointed. When he joined the Court, justices from outside of Quebec made little use of French in their work. La Forest followed the norm and Table II shows that his bilingualism score during his first years on the Court was rather low. As increasingly bilingual justices got appointed during his tenure, his score increased steadily. As Figure 3 shows, there might have been a general upward pull between 1985 and 1989, when the Court environment became more bilingual. This might have had a positive effect on the use of French by individual justices, even Francophone ones like La Forest. This raises an important point for the interpretation of the score, i.e. it is not a measure of the actual level of fluency in French, only a measure of the use of French in the work of every justice used as proxy for their fluency in French.
According to the deliberative theory illustrated by La Forest’s trajectory, bilingual justices joining the Court would have the effect of moving the whole Court in the same direction. The Court would then become an increasingly bilingual institution, from a deliberative standpoint, where French texts are cited, ceteris paribus, as frequently as English texts by all judges. Special legal expertise aside, justices would sit and write opinions as frequently in cases argued in French or English, all else held constant.
The institutional theory, by contrast, sees language itself as a form of expertise that affects the way the Court discharges its institutional duties. Given the unequal distribution of linguistic capacities among its members, the Chief Justice is likely to assign the most bilingual justices to the cases argued in French. Alternatively, the justices themselves might self-select in or out of those cases depending on their individual linguistic capacities. Thus, when justices have different capacities in French, the ablest might be asked to use their linguistic expertise in cases argued in French. Even if the most bilingual justices cite French texts as frequently as English texts, ceteris paribus, their sitting disproportionately in French cases will lead to the situation outlined by Grammond and Power (Reference Grammond and Power2011), where French texts are cited mostly in French cases. By contrast, their colleagues less fluent in French will sit more often in English cases and will have fewer opportunities to become familiar with the French literature and French decisions of lower courts, thus decreasing the probability that the Court will progress towards more bilingualism. According to the institutional theory, short of having a Court composed of nine justices equally competent in French—as was almost the case in the early 2000s—ordinal differences would have a recurring impact on the justices’ use of French.
The reality probably lies in a combination of both dimensions, and our score captures part of that complexity. It combines two variables that are rather group-dependent (linguistic distribution of absences, and linguistic assertiveness) and two variables that are not (citation of official and doctrinal sources, and citation of lower court decisions). For clarity’s sake, we also provide in Table I a score composed of the two group-dependent variables only (absences + assertiveness) and one for the two non-group-dependent variables only (citation of French doctrinal texts and official documents + citations of French lower court decisions).
The contrasts between the deliberative and the institutional theory should be borne in mind in future normative debates. If the deliberative theory is true, then every new bilingual justice joining the Court will help create a bilingual legal culture and have a positive impact on his or her colleagues. If the institutional theory is true, by contrast, the differences between the most bilingual (mostly Francophone) and least bilingual (mostly Anglophone) justices will remain, a consequence of a linguistic separation of labour that risks reproducing the linguistic gap rather than narrowing it.
Conclusion
Empirical studies on the linguistic dimension of judicial behaviour are still rare. Given that multilingualism plays an increasingly important role both domestically and internationally, understanding the multi-faceted impact of individual and institutional multilingualism on judicial institutions is crucial. But the normative discussion needs an empirical anchor. We tried here to provide some anchoring in the form of a bilingualism score.
Given our findings, it seems that a simple requirement that justices of the Supreme Court of Canada can “hear cases without the assistance of an interpreter” is not a guarantee in and of itself that the Court will be more bilingual as an institution. While it could help, there are reasons to believe that the Court is generally going in the right direction with or without a formal statutory bilingualism requirement. Nevertheless, factors other than individual bilingualism should also be considered when selecting Supreme Court justices if the Court is to become bilingual as an institution, e.g., the balance between Francophones and Anglophones, the familiarity of individual justices with the legal literature in both French and English, and the capacity of future justices to write decisions in both official languages. All these factors have a role to play in the creation of a bilingual legal culture (Macdonald Reference Macdonald1997). On a final note, the successive changes to the appointment process to the Supreme Court of Canada in the last fifteen years have greatly increased its transparency. This transparency should include the releasing of the formal language test scores of judicial applicants. The linguistic proficiency of individual justices should not be a matter of conjecture or speculation. Likewise, it is time for Canadians to ask, not if, but how much bilingualism should be required of the Court.
Appendix A
Absence indicator
The Absence indicator measures, for each justice, the fraction of skipped Supreme Court cases that are cases argued in French.
For justice j and year y, the formula is:
$$ {Absence\ Indicator}_y^j=\frac{{\left( Skipped\ French\ Cases\right)}_y^j}{{\left( Skipped\ Cases\right)}_y^j} $$
Assertiveness indicator
The Assertiveness indicator measures, for each justice, the fraction of French cases heard that resulted in a written opinion.
For justice j and year y, the formula is:
$$ {Assertiveness\ Indicator}_y^j=\frac{{\left( Writing\ for\ French\ Cases\right)}_y^j}{{\left( Heard\ French\ Cases\;\right)}_y^j} $$
Doctrine citations indicator
The Doctrine citations indicator measures, for each justice, the fraction of articles, official documents and books cited only available in French. In the published English decisions, out of 2507 texts cited, 152 were listed in French.
For justice j and year y, the annual indicator is obtained by iterating over each decision d:
$$ Doctrine\;{citations\ indicator}_y^j=\sum \limits_d{Attribution}^{j,d}\frac{{\left( French\ Texts\ Cited\right)}_y^{j,d}}{{\left( Texts\ Cited\right)}_y^{j,d}} $$
In decisions with many opinions, all referred texts are listed together in the heading. In those cases, an attribution rule was needed. The attribution is inferred by first calculating a French Solo indicator using only unanimous decisions authored by a single justice. For example, when Justice Lamer is the sole author of a unanimous court, 9% of texts cited are in French; McLachlin, 3%. Their ratio is 3 to 1. If the heading of a decision refers to 4 French texts and both Lamer and McLachlin wrote an opinion, he gets 3 texts, she gets 1. These numbers are then applied to their Doctrine citations indicator.
Lower court citations indicator
The Lower court citations indicator measures, for each justice, the fraction of cited cases that are from Quebec lower courts. It is constructed using the “Cases Cited” section in the heading of every Supreme Court decision in Federal law cases. Out of 28,570 cases cited, 264 are of Quebec lower courts.
For justice j and year y, the formula is:
$$ Lower\ court\;{citations\ indicator}_y^j=\frac{{\left( Cases\ Cited\ of\ Quebec\right)}_y^j}{{\left( Cases\ Cited\right)}_y^j} $$
The Bilingualism Score
The aggregate bilingualism score was obtained by combining the z-score of the four indicators.
Using the z-score helped in combining 4 indicators of different scale. The z-scores ensured that all variables had the same mean (0) and standard deviation (1).
$$ Z\left({x}_i\right)=\frac{x_i-{average}_x}{{standard\ deviation}_x} $$
Floor and ceiling values were placed at -2 and +2 for the z-scores.
The annual bilingualism score is an average of the four z-scores:
$$ {Score}_y^j={\displaystyle \begin{array}{l}\frac{1}{4}Z\left( Cases\;{Cited\ Indicator}_y^j\right)+\frac{1}{4}Z\left({Writing\ Indicator}_y^j\right)\\ {}+\hskip2px \frac{1}{4}Z\left( Authors\;{Cited\ Indicator}_y^j\right)-\frac{1}{4}Z\left({Absence\ Indicator}_y^j\right)\end{array}} $$
In order to smooth out large yearly variations, the annual score was constructed using a 3-year moving average. For instance, the score for 2005 is composed of values from 2004, 2005, and 2006.
