Linguistic Nihilism

One common line of attack against textualism—the idea that “the words of a governing text are of paramount concern, and what they convey, in their context, is what the text means (Scalia & Garner, at 56)—is that language is never clear, or put differently, hopelessly vague or ambiguous. Put this way, the task of interpretation based on text is a fool’s game. Inevitably, so the argument goes, courts will need to resort to extraneous purposes, “values,” social science evidence, pre or post-enactment legislative history, or consequential analysis to impose meaning on text that cannot be interpreted.

I cannot agree with this argument. For one, the extraneous sources marshalled by anti-textualists bristle with probative problems, and so are not reliable indicators of legislative meaning themselves. More importantly, an “anything goes” approach to interpretation offers no guidance to judges who must, in tough cases, actually interpret the law in predictable way. In this post, I will explore these arguments. My point is that a sort of linguistic nihilism that characterizes anti-textualist arguments is not conclusive, but merely invites further debate about the relative role of text and other terms.

**

Putting aside frivolous arguments one often hears about textualism (ie: “it supports a conservative agenda” or “it is the plain meaning approach”), one clear criticism of textualism is that interpretation is not self-executing. Jorge Gracia, for example, writes:

…texts are always given in a certain language that obeys rules and whose signs denote and connote more or less established meanings. In addition, the audience cannot help but bring to the text its own cultural, psychological, and conceptual context. Indeed, the understanding of the meaning of a text can be carried out only by bringing something to the text that is not already there…

Gracia, A Theory of Textuality: The Logic and Epistemology, at 28

Sullivan calls this situation the “pervasive indeterminacy of language” (see here, at 206). Put this way, as Sullivan notes, it is impossible to interpret text in its linguistic context:

It is not possible for judges  who interpret a provision of the Criminal Code or the Income Tax Act to wipe out the beliefs, values and expectations that they bring to their reading. They cannot erase their knowledge of law or the subject of legislation. They cannot case aside legal culture, with its respect for common law and evolving constitutional values…Like any other readers, if they want to make sense of a text, judges must rely on the context that they themselves bring to the text (see 208).

This form of linguistic nihilism is highly attractive. So goes the argument, if texts cannot be interpreted on their own, judges should and must bring their own personal biases and values to the text, as a desirable or inevitable result of the unclear text. And if that’s the case, we should adopt another type of interpretive record—perhaps one that centres what a judge in a particular case thinks the equities ought to be.

**

This argument aside, I find it hard to accept. First, the tools that are inevitably supposed to resolve these ambiguities or vagueness themselves are ambiguous and vague; so it is hard to hold them up as paragons of clarity against hopelessly clear text.

Let’s consider, first, the tools often advanced by non-textualists that are supposed to bring clarity to the interpretive exercise. Purpose is one such tool. In Canadian statutory interpretation, purpose and context must be sourced in every case, even when the text is admittedly clear on first blush (ATCO, at para 48). Put together, text, context, and purpose must be read together harmoniously (Canada Trustco, at para 47). But sometimes, purpose is offered by anti-textualists as an “out” from ambiguity or vagueness in the text itself. The problem is that sourcing purpose is not self-executing either. Purpose can be stated at various levels of abstraction (see here, and in general, Hillier). In other words, purpose can be the most abstract purpose of the statute possible (say, to achieve justice, as Max Radin once said); or it could be the minute details of particular provisions. There can be many purposes in a statute, stated in opposite terms (see Rafilovich for an example of this). Choosing purposes in these cases can be just as difficult as figuring out what words mean. This is especially so because the Supreme Court has never really provided guidance on the interaction between text and purpose, instead simply stating that these things must be read “harmoniously.” What this means in distinct cases is unclear. This is why it is best to source purpose with reference to text itself (see here).

Legislative history also presents well-known problems. One might advance the case that a Minister, when introducing a bill, speaks to the bill and gives his view of the bill’s purpose. Others may say differently. In some cases, legislative history can be probative. But in many cases, legislative history is not useful at all. For one, and this is true in both Canada and the US, we are bound by laws; not by the intentions of draftspeople. What a Minister thinks is enacted in text does not necessarily equate to what is actually enacted (see my post here on the US case of Bostock). There may be many reasons why bills were drafted the way they were in particular cases, but it is not probative to think legislative history (which can be manipulated) should be some cure-all for textual ambiguity or vagueness.

Finally, one might say that it is inevitable and desirable for judges to bring their own personal values and experiences to judging and interpreting statutes. This is a common refrain these days. To some extent, I agree with those who say that such value-based judging is inevitable. Judges are human beings, and are not robots. We cannot expect them to put aside all implicit value judgments in all cases. But one of the purposes of law, and of the rules of interpretation, is to ensure that decisions are reasoned according to a uniform set of rules applicable across the mass of cases. We have to limit idiosyncratic reasoning to the extent we can/ If we give up on defining with clarity such rules—in order to liberate judges and their own personal views—we no longer have a system of interpretation defined by law. Rather, we have a system of consequences, where judges reach the results they like based on the cases in front of them. This might sound like a nice idea to some, but in the long run, it is an unpredictable way to solve legal disputes.

**

If all of the tools of interpretation, including text, are imperfect, what is an interpreter to do? One classic answer to this problem is what I call the “anything goes” approach. Sullivan seems to say that this is what the Supreme Court actually does in its statutory interpretation cases (see here, at 183-184). While I question this orthodox view in light of certain cases, I take Sullivan’s description to be indicative of a normative argument. If the Supreme Court cannot settle on one theory of interpretation, perhaps it is best to settle on multiple theories. Maybe, in some cases, legislative history is extremely probative, and it takes precedence over text. Maybe, in some cases, purpose carries more weight than text. This is a sort of pragmatic approach that allows judges to use the tools of interpretation in response to the facts of particular cases.

This is attractive because it does not put blinders on the interpreter. It also introduces “nuance” and “context” to the interpretation exercise. All of this sounds good. But in reality, I am not sure that the “anything goes” approach, where judges assign weight on various tools in various cases, is all that helpful. I will put aside the normative objections—for example, the idea that text is adopted by the legislature or its delegates and legislative history is not—and instead focus on the pragmatic problems. Good judicial decisions depend on good judicial reasoning. Good judicial reasoning is more likely to occur if it depends less on a particular judge’s writing prowess and more on sourcing that reasoning from precedential and well-practiced rules. But there is no external, universal rule to guide the particular weights that judges should assign to various tools of interpretation, and even further, what factors will guide the assignment of weights. At the same time, some people might argue that rules that are too stringent will stymie the human aspect of judging.

In my view, an answer to this was provided by Justice Stratas in a recent paper co-authored with his clerk, David Williams. The piece offers an interesting and well-reasoned way of ordering tools of interpretation. For Stratas & Williams,  there are certain “green light” “yellow light” and “red light” tools in statutory interpretation. Green light tools include text and context, as well as purpose when it is sourced in text. Yellow light tools are ones that must be used with caution—for example, legislative history and social science evidence. Red light tools are ones that should never be used—for example, personal policy preferences.

I think this is a sound way of viewing the statutory interpretation problem. The text is naturally the starting point, since text is what is adopted by the legislature or its delegates, and is often the best evidence of what the legislature meant. Context is necessary as a pragmatic tool to understand text. Purpose can be probative as well, if sourced in text.

Sometimes, as I mentioned above, legislative history can be helpful. But it  must be used with caution. The same goes with social science evidence, which might be helpful if it illustrates the consequences of different interpretations, and roots those consequences back to internal statutory tools like text or purpose. But again, social science evidence cannot be used to contradict clear text.

Finally, I cannot imagine a world in which a judge’s personal views on what legislation should mean should be at all probative. Hence, it is a red light tool.

In this framework, judges are not asked to, on a case-by-case basis, assign weights to the tools that the judge thinks is most helpful. Instead, the tools are ranked according to their probative value. This setup has the benefit of rigidity, in that it does assign objective weight to the factors before interpretation begins. At the same time, it keeps the door open to using various tools that could deal with textual ambiguity or vagueness.

The point is that textualism cannot be said to be implausible simply because it takes some work to squeeze meaning out of text. The alternatives are not any better. If we can arrange text at the hierarchy of a list of other tools, that may be a solid way forward.

On Canadian Statutory Interpretation and Recent Trends

I have had the pleasure of reading (for the first time front-to-back) the legal interpretation classic, Reading Law by Justice Scalia and Bryan Garner. For Canadian courts struggling with how to source and use purpose when interpreting statutes, Reading Law provides valuable assistance. It does so by outlining two schools of thought on how to source purpose, schools of thought that are prevalent in Canadian debates and recent decisions over statutory interpretation. On the one hand is purposivism; on the other hand is textualism. While these schools do not actually differ about whether purpose should form part of the interpretive exercise, they do differ about how to actually determine what purpose governs. Canada’s recent statutory interpretation cases point to the textualist direction.

The first school of thought, broadly known as purposivism, is apparently Canada’s leading approach to statutory interpretation.  Purposivism “acknowledges that the meaning of language is imprecise and measures words against contextual, schematic, and purposive considerations” (see Hutchison, here, at 8). Aharon Barak claims that:

[a]ccording to purposive interpretation, the purpose of a text is a normative concept. It is a legal construction that helps the interpreter understand a legal text. The author of the text created the text. The purpose of the text is not part of the text itself. The judge formulates the purpose based on information about the intention of the text’s author (subjective purpose) and the “intention” of the legal system (objective purpose) (Barak, Purposive Interpretation, at 110).

The motivation behind purposivism is a sort of legal realism that queries whether text can ever truly be clear enough to be a dominant force in legal interpretation (see, for a characteristic example of this line of thinking, the opinion of Breyer J in FCC v NextWave Personal Communications Inc, 537 U.S. 293, 311). Purpose is thus a way to deal with latent ambiguities that may naturally arise in text. And importantly, purpose is focused on the “ends” a statute is designed to achieve, perhaps at a high level of abstraction or generality. On a radical purposive account, the goal of interpretation is to effectuate whatever the court determines the purpose(s) to be; text is merely a means to the end of purpose. Put differently, text is derived from purpose under the purposive account.

On the other hand is “textualism.” Textualism receives a bad rap in Canada, but that is probably more due to caricature than a real appraisal of the merits and demerits of the textualist method. Here Scalia & Garner have much to say. While the central feature of textualism is the idea that “if the text…is clear, interpreters should not impeach the text using extrinsic evidence of statutory purpose…” (Manning & Stephenson, Legislation and Regulation, at 94), textualism does not ask a court to “put on blinders that shield the legislative purpose from view” (Scalia & Garner, at 20; see also William Popkin, “An ‘Internal’ Critique of Justice Scalia’s Theory of Statutory Interpretation,” 76 Minn L Rev 1133, 1142 (1992)).  Instead, purpose is “deduced from a close reading of the text” (Scalia & Garner, 20).  Put differently, purpose is derived from text on the textualist account.

Why are textualists concerned about purposes achieved without reference to the text? First, textualists are concerned about the generality problem (see Max Radin, “Statutory Interpretation,” 43 Harv L Rev 863, 876 (1930)). A court motivated by its own results-oriented reasoning could choose a purpose that is barely represented in text, or is otherwise quite abstract in relation to text. Indeed, at the highest level of generality, every statute could be said to pursue “justice and security” (see Radin). But choosing that purpose could distort the means used by the statute chosen to achieve its ends by “enabling…crabbed interpretations to limiting provisions and unrealistically expansive interpretations to narrow provisions” (Scalia & Garner, at 20). This particular problem also has resonance in administrative interpretations of law, where an expansive purposive interpretation of enabling provisions could actually result in more deference to decision-makers than what the text itself allows.

Second, textualists are concerned with the realities of the legislative process and the fact that legislatures are imperfect. The takeaway from the Legal Process school, which influences purposivism, is that legislatures pursue reasonable purposes reasonably. But textualists understand that legislation, especially in the US, is a result of legislative compromise. While purposes may be clear, text pursues purposes in different ways. In this way, textualists are more concerned with the implementational rather than the ulterior purposes of legislation. Legislation can implement purposes in text in various ways.  A generalized example here is instructive:

For example, a statute providing a specific protection and a discrete remedy for purchasers of goods can be said to have as its purpose “protecting the consumer.” That would not justify expansive consumer-friendly interpretations of provisions that are narrowly drawn (Scalia & Garner, at 57).

What does this dispute between textualists and purposivists have to do with Canada? From a descriptive perspective, it describes perfectly what is happening in Canadian courts right now with regards to purpose. Normatively, Scalia & Garner’s text explains why a textualist-purposive approach is well-justified.

On the descriptive account, the Supreme Court in the past has fallen victim to the “level of generality” problem. West Fraser is a classic example. There, the dispute was whether a British Columbia statute permitted fines to be levied for workplace safety violations against owners of land on which accidents occurred. The relevant provision under which West Fraser was fined was, by its text, only applicable to “employers.” But Chief Justice McLachlin, for the majority, held that the ultimate purpose of the statute was to “promote workplace safety in the broadest sense” (see West Fraser, at para 17). This allowed her to conclude that the particular text of the section under interpretation should be interpreted to cover off West Fraser’s conduct. But here is a classic example of the purposive approach: purpose was used to interpret the text under consideration, rather than the other way around.

Justice Côté in dissent, in my view, had much better of the argument. Her view was that the relevant provision had chosen the means by which to pursue the purpose of workplace safety. The text had chosen “limited means” to pursue that purpose—by limiting fines to employers (see West Fraser, at para 107). This is a classic dispute between ulterior and implementational purposes.

Justice Côté’s view has recently been picked up in recent Supreme Court cases and in cases in the Federal Court of Appeal. I cite two examples here. First is Telus v Wellman, which I wrote about here. There, the dispute was what purpose should be chosen: for the majority, the purpose of the Arbitration Act, as directly reflected in the relevant statutory provisions, was that the Act ensures that parties abide by their agreements. But in dissent, Abella and Karakatsanis JJ would have pitched the purpose of the statute at the level of “access to justice.” Moldaver J in majority rejected the dissent’s characterization, holding that this purpose could “distort the actual words of the statute” (Telus, at para 79). The access to justice purpose was not rooted in statute. Moldaver J, then, could be said to adopt a position closer to Cote J in West Fraser, and closer to the textualist position identified by Scalia & Garner.

Similarly, in Hillier, Justice Stratas rejected the Attorney General’s attempt to cast a statute at the high level of abstraction of “administrative efficiency.” Rather, he concluded that not “every section in the Act is aimed at furthering efficiency” (Hillier, at para 35). Rather, the relevant provision under interpretation “pursues a different, more limited purpose” (Hillier, at para 35). That limited purpose governed, not the abstract purpose chosen by the Attorney General.

In these cases, the Supreme Court and the Federal Court of Appeal corrects the error in West Fraser. And here is a good point to say why it is that the textualist approach adopted by Moldaver J and Stratas JA is preferable. First, as noted above, a liberal application of the purposive approach could lead to high error costs. By prioritizing ulterior motive over implementational purpose (abstract versus specific purposes), the court could fail to understand how and why a statute achieves a particular goal. In other words, reasoning backwards from purpose (as McLachlin CJ did in West Fraser) could lead to ignoring what the text actually says, and how the text decides to pursue a particular goal. For McLachlin CJ in West Fraser, it was of no moment that the relevant provision only applied to employers. But this was the interpretive dispute at hand. The interpretive approach in West Fraser, in this sense, ignores the import of the text.

Secondly, and pragmatically, choosing more abstract purposes of statutes over more implementational ones does not actually help the interpretive task. To say that the purpose of a statute is “access to justice” will rarely do anything to determine how the text is actually supposed to be interpreted. This is because there are many different ways that a statute can methodologically choose to pursue access to justice. More likely, abstract, ulterior purposes can be used to distort text to achieve policy outcomes the interpreter likes. This is profoundly violative of the Rule of Law.

And finally, as Scalia & Garner note, perhaps the most important interpretive canon is that one which says that “[t]he words of a governing text are of paramount concern, and what they convey, in their context is what the text means” (Scalia & Garner, at 56). This sentiment has been expressed by the Supreme Court of Canada, particularly where text is “clear” (see Celgene, at para 21). It is as old as Justinian’s Digests (“A verbis legis non est recedendum”). A powerful principle of democracy justifies the canon. It is, after all, text which is enacted by our democratic institutions. Purpose should revolve around text, such that the purpose with the most reflection in text should govern. Sourcing text from purpose risks prioritizing an ideal with little democratic pedigree over the specific and finely-wrought means by which the text enacts that purpose.

Overall, and while no Canadian court will probably ever describe itself as textualist, courts in Canada are increasingly looking to text to discern purpose. In my view, this is a salutary development.