Three syntactic canons

[This is Part 4 of my look at Reading Law: The Interpretation of Legal Texts by Antonin Scalia and Bryan Garner. (Part 1; Part 2; Part 3.)]

Included in Reading Law’s list of 57 canons of interpretation are seven that are described as “syntactic canons.” Of these, three are the most important:

Last-Antecedent Canon. A pronoun, relative pronoun, or demonstrative adjective generally refers to the nearest reasonable antecedent.

Series-Qualifier Canon. When there is a straightforward, parallel construction that involves all nouns or verbs in a series, a prepositive or postpositive modifier normally applies to the entire series.

Nearest-Reasonable-Referent Canon. When the syntax involves something other than a parallel series of nouns or verbs, a prepositive or postpositive modifier normally applies only to the nearest reasonable referent.

In this post I’m first going to look at these canons fit in with the existing law and then at whether they hang together as a coherent set of principles. And since the post is a long one, here’s the executive summary: First, the canons as formulated in Reading Law differ from the existing law in some important ways. Second, they don’t hang together as a coherent set of principles.

We’ll start by comparing the Scalia/Garner version of these canons with their, um, antecedents.

First up is the Last-Antecedent Canon, which is Scalia and Garner’s version of what’s generally called “the rule of the last antecedent.” In an attempt to avoid confusion, I’ll refer to the Scalia/Garner version by the name they use and to the underlying rule from which their Canon derives as the rule of the last antecedent.

Both the Canon and the rule are intended to provide a default rule for resolving attachment ambiguities in which a modifier comes after the words or phrases that it’s modifying. The rule is stated in the following terms in the most recent edition of the leading treatise on statutory interpretation: “Referential and qualifying words and phrases, where no contrary intention appears, refer solely to the last antecedent.” (Singer & Singer, Statutes & Statutory Construction (7th ed. 2007)). This is in effect a preference for low attachment. As applied to Liparota v. United States (which is discussed in the previous installment of this series), it would support the defendant’s interpretation over the government’s. government’s interpretation over the defendant’s.

The defendant’s interpretation

The government’s interpretation

[On edit: The order of the diagrams was changed.]

The scope of the rule of the last antecedent has been described in cases, treatises, and law dictionaries in a variety of different ways. One formulation is quoted above (“referential and qualifying words and phrases”); here is a sampling of the others (sources are listed at the end of the post):

“relative words”

“relative or qualifying words or phrases”

“a limiting clause or phrase”

“qualifying words or phrases”

“where one phrase of a statute modifies another”

These descriptions are all rather vague, and Reading Law increases the level of precision by defining the Last-Antecedent Canon as applying only with regard to specified word classes (pronouns, relative pronouns, and demonstrative adjectives.) The result is to give the Canon a narrower scope of coverage than the rule of the last antecedent as it has come to be applied, but the expressions that are excluded from the Last-Antecedent Canon are included in the Nearest-Reasonable-Referent Canon. The latter canon covers the interpretation of modifiers occurring in expressions whose syntax “involves something other than a parallel series of nouns or verbs,” and like the Last-Antecedent Canon, it prescribes a preference for low attachment.

Scalia and Garner explain that the Nearest-Reasonable-Referent Canon has been carved out of the rule of the last antecedent in the interest of terminological accuracy:

Although this principle is often given the misnomer last-antecedent canon (see § 18), it is more accurate to consider it separately and to call it the nearest-reasonable-referent canon. Strictly speaking, only pronouns have antecedents, and the canon here under consideration also applies to adjectives, adverbs, and adverbial or adjectival phrases…

But the division of the rule of the last antecedent into two separate canons represents more than just grammatical wonkery.

The Nearest-Reasonable-Referent Canon differs from the rule of the last antecedent in that it applies not only to modifiers that follow the words or phrases that they modify but also to those that precede them. It refers to both “postpositive” and “prepositive modifiers,” and the commentary elaborating on the canon makes it clear that the canon “applies not just to words that precede the modifier, but also to words that follow it.”

This is an innovation. I say that not only because Reading Law cites no authority (and offers no explanation) for extending the Nearest-Reasonable-Referent Canon to premodification as well as postmodification, and not only because I’m unaware of any previously-recognized canon of interpretation that has such an effect, but also because of the dog that didn’t bark.

If such a canon existed, one would expect it to have been cited in, say, the dissenting opinion in Liparota. As you’ll recall, Liparota had to do with the scope of knowingly in the phrase whoever knowingly uses, transfers, acquires, alters, or possesses [food stamps] any manner not authorized by [law]. The two competing interpretations are represented visually in the tree diagrams above. The government’s interpretation would have been supported by the Nearest-Reasonable-Referent Canon, if such a canon existed. So when the Supreme Court rejected the government’s argument by a vote of 6–2, the canon would presumably have been mentioned in the dissenting opinion—if such a canon really existed. But no such canon was mentioned. The dog didn’t bark.

On to the Series-Qualifier Canon, which provides, “When there is a straightforward, parallel construction that involves all nouns or verbs in a series, a prepositive or postpositive modifier normally applies to the entire series.” Thus, this Canon states a preference for this structure

rather than this one

[Update: “XP”, for the non-linguists in the audience, means “phrase of any type.” A noun phrase is an NP, a verb phrase is a VP, and so on. “X” is a variable standing for all the possible lexical categories (aka “parts of speech”) around which a phrase can be constructed.]

As far as I have been able to determine, there has never been a canon of interpretation called anything like “the Series-Qualifer Canon” or a canon that is limited in its application to parallel constructions involving nouns or verbs in a series. However, I suspect that this canon is derived from a “rule” that is sometimes invoked to justify a particular reading of a statute: “When several words are followed by a clause which is applicable as much to the first and other words as to the last, the natural construction of the language demands that the clause be read as applicable to all.” (For example.)

That statement isn’t much of a rule (hence the scare-quotes around rule) because it’s entirely circular. Whether in a given case the clause in question is “applicable as much to the first and other words as to the last” is precisely the question that has to be answered. This probably explains why (if I’m right about the source of the Series-Qualifier Canon) Scalia and Garner didn’t use the original wording.

By framing the Series-Qualifier Canon in terms of syntax rather than semantics, Scalia and Garner are presumably trying to introduce the same kind of precision that they did in their framing of the Last-Antecedent Canon. That impulse is a noble one, but unfortunately the attempt fails.

The problem is that the categories defined in the two canons are overlapping. The category defined by the Series-Qualifier Canon—parallel constructions consisting of nouns or verbs (more appropriately, noun phrases or verb phrases)—includes parallel constructions that are modified by relative clauses. And relative clauses, in turn, generally begin with relative pronouns and therefore come within the Last-Antecedent Canon.

In fact, two such constructions are provided as examples illustrating the Series-Qualifier Canon (pg. 148):

Institutions or societies that are charitable in nature

A wall or fence that is solid

(Contra The Cambridge Grammar of the English Language, Scalia and Garner regard that as a relative pronoun; see Reading Law, pg. 142.)

Why are these examples treated as being subject to the Series-Qualifier Canon rather than the Last-Antecedent Canon? Your guess is as good as mine. The book gives no explanation; it doesn’t even recognize the issue.

And the issue is a serious one, because the conclusion reached by applying the Series-Qualifier Canon is precisely the opposite of the conclusion reached by applying the Last-Antecedent Canon: under the Series-Qualifier Canon, the modifier applies to every item in the series, while under the Last Antecedent Canon [This paragraph was added on edit.]

The conflict between the two canons is unintentionally highlighted by the 7th edition of Black’s Law Dictionary (the first edition edited by Garner), where the entry for the rule of the last antecedent says, “an application of this rule might mean that in the phrase Texas courts, New Mexico courts, and New York courts in the federal system, the words in the federal system might be held to modify only New York courts and not Texas courts or New Mexico courts.” That example could just as well be used to illustrate the Series-Qualifier Canon—in which case in the federal system would modify all three items in the series. [This paragraph was modified on edit.]

And there are plenty of other cases in which the rule of the last antecedent is applied to a parallel series of conjoined noun phrases or verb phrases. (For example, this one, this one, and this one.) In all of those cases, the application of the rule of the last antecedent resulted in an interpretation different from what the Series-Qualifier Canon would have required. So in addition to conflicting with the Last Antecedent Canon, the Series-Qualifier Canon can’t be justified as an accurate summary of the case law. [This paragraph was modified on edit.]

Here are some more questions that Reading Law doesn’t answer: What distinguishes “parallel construction[s] that involve[] all nouns or verbs” from all other constructions such that high attachment is preferred in the former but low attachment for the latter? And for that matter, can either of these preferences be justified on purely linguistic grounds? In other words, does either of them represent a valid generalization about how expressions in English are understood? These questions are important because without such a justification, the preferences incorporated in these three canons are purely arbitrary. Which is not something that legal principles ought to be.

For my attempt to answer those questions, stay tuned. [Update: For that discussion, go here.]

Sources for the prior formulations of the categories to which the rule of the last antecedent applies:

“relative words”—Herbert Broom, Selection of Legal Maxims 190–91, *292–94 (1845) (link)

“relative or qualifying words or phrases”—Black’s Law Dictionary (rev. 4th ed. 1968); Jabez Gridley Sutherland, Statutes and Statutory Construction 349 (1891) (link)

“a limiting clause or phrase”—Barnhart v. Thomas, 540 U.S. 20, 26–27 (2003) (link)

“qualifying words or phrases”—Black’s Law Dictionary (7th ed. 1999) (Bryan A. Garner, editor in chief)

“where one phrase of a statute modifies another”—O’Kane v. Apfel, 224 F.3d 686, 690 (7th Cir. 2000) (link)

Disclosure: I have two points of involvement in Bryan Garner’s work other than as a consumer. An excerpt from a brief I wrote appears as an example in his book The Winning Brief: 100 Tips for Persuasive Briefing in Trial and Appellate Courts. In addition, as the first edition of Black’s Law Dictionary that Bryan edited was nearing completion, I was one of about two dozen people who were each given several hundred entries to proofread and to suggest revisions.

On edit: I’ve taken out a statement that the Series-Qualifier Canon states a preference for high attachment, because I realized that I was using that term in an unusual way and therefore that the statement was likely to generate confusion. I’ve also made some changes and additions to the discussion of the Series-Qualifier Canon to make it clearer that the canon conflicts with the Last-Antecedent Canon.

2 responses to “Three syntactic canons

  1. It is interesting that your examples with relative clauses are not ambiguous in languages that have agreement in number of the relative pronoun and its referent. Is English especially ambiguous a language?

  2. Reblogged this on Law&Language and commented:
    I found this blog post by Neal Goldfarb at LawnLinguistics so fascinating, I had to repost it. It is more linguistic (in a technical sense) than what normally appears on L&L, but it is very much related and overall quite fascinating. Enjoy!

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s