November 2006
Seven Types of Ambiguity . . . More or Less
by Robert C. Cumbow
When I was a grad student in English Lit, many lifetimes ago, one of the most important books in literary criticism was William Empson’s Seven Types of Ambiguity. This was back in the now-lost days when “literary criticism” meant looking at good writing and saying something intelligent and useful about what made it good. Today’s English majors, to their loss, know nothing of this.
I could not now tell you what the seven types of ambiguity were, or even why there were precisely seven rather than eight, or six, or thirteen. But I mention Professor Empson’s estimable title here as a way of noting that, in literary writing, ambiguity is often a good thing. That’s because one of the techniques and purposes of literature is to expand meaning rather than limit it.
This is emphatically not the case in our profession. In law, language needs to be as precise as possible. Ambiguity invites interpretation; and while interpretation creates hours of amusement and enlightenment for English majors, it can create months of dispute, years of litigation, and hundreds of thousands of dollars in legal costs for ordinary business folk. What we want in law — at least in statutes, rules, contracts, and judicial opinions — is clarity, and preferably absolute clarity.
Anyone who’s been to law school knows that a court won’t waste its time construing a statute or interpreting a contractual provision if its meaning is plain on its face. It is only the ambiguous clause that requires judicial construction — and ends up costing lots of money for a result that will disappoint at least half of the people involved. How does ambiguity arise in legal and legislative writing, what do we do when it arises, and how do we learn to avoid it?
The deliberate ambiguity sought by the artful literary writer arises from careful selection of words and phrases that are most likely to suggest broader, deeper meanings and possibilities. The flip side of that coin is the unintended ambiguity of the legal writer or legislative draftsman who — with good intentions, but out of laziness, sloppiness, or inattention — selects a word, phrase, or structure that is capable of more meanings than the one intended. I had occasion to refer to one such kind of ambiguity a couple of months ago, when I briefly discussed the serial comma and the problems created by its absence. I wrote that “Arthur left his estate in equal parts to Gavin, Tristram and Percival” could be read to mean that Arthur’s estate is divided into three parts, one going to each of the three named individuals; or that the estate is divided into two parts, one going to Gavin, the other to Tristram and Percival together. I concluded that, for purposes of the precise clarity needed in legal writing, it’s longer but better to say: “Arthur left his estate in two equal parts, one part to Gavin, the other to Tristram and Percival.”
Back in 1999, the judges of the U.S. Court of Appeals for the First Circuit faced a similar problem when they had to interpret a compact that read, in part: “the Commission is hereby empowered to establish the minimum price for milk to be paid by pool plants, partially regulated plants and all other handlers receiving milk from producers located in a regulated area.” Appellants in the case had argued that the lack of a serial comma after “partially regulated plants” and before “and all other handlers” meant that the final phrase “receiving milk from producers located in a regulated area” modified both “partially regulated plants” and “all other handlers” — in other words, the Commission could establish minimum prices for milk handled by a partially regulated plant only if that plant received milk from producers located in a regulated area. The First Circuit rejected that argument, on grounds that such an interpretation would exempt from regulation entirely those plants that met the definition of “partially regulated plants” but did not receive milk from producers located in a regulated area.1
Similarly, the Massachusetts Appeals Court for Essex County rejected an argument that serial commas in a statute rendered its meaning ambiguous because they could be read as meaning “and” or “or.” The court held that a series of terms separated by serial commas and culminating in a final term preceded by the word “or” was to be read in the disjunctive, not the conjunctive; that is, an expression such as “a, b, c, or d” means any one of those four elements, not all of the first three or just the last one. (The disjunctive is opposed to the conjunctive. The term “or” is disjunctive: it separates the elements in a series, such that any one of them applies. The word “and” is conjunctive: it conjoins the elements in a series, such that all of them must apply.) “The obvious sense,” the court held, “dictates reading each of the commas as standing in place of an ‘or’ … the ‘or’ is to be given its ordinary disjunctive meaning.”2 The court reasoned that “The use of the disjunctive ‘or’ between each segment of the statute indicates again that a showing of any one of the activities will constitute” the violation at issue in the case. (Unfortunately, the court’s colorful display of literacy and linguistic analysis paled when it chose the phrase “between each segment.” A comma can’t appear “between” a single segment, but only between any two segments. The court would have done better to stick to its established practice of referring to the commas as “separating” the segments, rather than somehow appearing “between” each segment, in violation of the laws of both grammar and physics.)
In the above cases, it was fairly easy for the court to find the correct, intended meaning, despite the ambiguous omission of the serial comma, by simply considering the intent of the statute and comparing it with the result that would follow from the Appellants’ proposed interpretation. But the placement of serial commas can create knottier ambiguities. The New Mexico Supreme Court in 1969 faced the challenge of interpreting the state’s kidnapping and false-imprisonment statutes, which epitomize both ambiguous and misleading legislative drafting:
Kidnaping [sic] is the unlawful taking, restraining or confining of a person, by force or deception, with intent that the victim be held for ransom, as a hostage, confined against his will, or to be held to service against the victim’s will.
False imprisonment consists of intentionally confining or restraining another person without his consent and with knowledge that he has no lawful authority to do so.
Interpreting these statutes is a daunting task, which the New Mexico Supremes undertook with aplomb.3 I commend it to those interested enough to look it up, but won’t repeat their analysis here — except to note the court’s especially useful distinction between parenthetical commas, which enclose certain phrases, and serial commas, which separate the several terms that make up a sequence. But metal more attractive is found in rewriting the offending statutes.
We know we’re in trouble when we see that the definition of kidnapping uses the term “unlawful.” There are two problems with this. First, since the definition comes from the New Mexico criminal code, it is supposed to tell us what is unlawful, so using the term “unlawful” in defining an unlawful act is spectacularly unhelpful. Second, including the term “unlawful” here suggests that there may be actions of the type defined in this statute that are lawful. If anyone can think of an example of the lawful taking, restraining or confining of a person, by force or deception, with intent that the victim be held for ransom, as a hostage, confined against his will, or be held to service against the victim’s will, please let me know. On second thought, don’t — I think the legislature probably inserted the word “unlawful” when someone noticed that, without it, a police officer arresting someone could be guilty of kidnapping. Still, there must have been a better way of making that point than to settle on language that essentially means “the following behavior is against the law — but only if it’s unlawful.”
But moving on to the substance of the statute: Kidnapping may consist of taking, restraining, or confining a person. The statute uses the disjunctive, so any one of those activities will suffice. It’s equally evident that, whichever one of these is done, it must be done by force or deception — again, either of the two will suffice. And it is also clear that there must be a specific intent. But in the intent provision the statute breaks down. As written, the statute could refer to any of three possible intents: (1) that the victim be held for ransom, as a hostage, (2) that the victim be confined against his will, or (3) that the victim be held to service against his will. However, the statute could also be read as referring to four possible intents: (1) that the victim be held for ransom, (2) that the victim be held as a hostage, (3) that the victim be confined against his will, or (4) that the victim be held to service against his will. The latter appears more likely, given the fact that not all hostages are held for ransom.
The false imprisonment definition is not as complex, but it does contain an ambiguous pronoun. Here the intent is to confine or restrain (“intentionally confining or restraining another”), but in addition to the absence of consent there is also a knowledge requirement: “with knowledge that he has no lawful authority to do so.” Who is the “he” referred to in this knowledge provision? We know that a pronoun must refer back to its antecedent, and agree with it in gender and number. But the only word in this definition that could possibly be the antecedent of the pronoun “he” is “another person” — in other words, the victim! So the law is telling us that false imprisonment consists in intentionally confining or restraining another person, without his consent, and with the knowledge that he, the victim, has “no lawful authority to do so.” To do what? To give his consent? To withhold it? No, we are really lost in the Twilight Zone now. The “do so” must refer to the confining or restraining, and the “he” in the knowledge clause must mean the perpetrator, not the victim, even though the sentence, as constructed, contains no reference to a perpetrator.
So, humbly submitted for your approval, as Rod Serling used to say, here is a proposed rewrite of the statutes at issue in the New Mexico case:
Kidnapping is the taking, restraining, or confining of a person, by force or deception, without lawful authority, with intent that the victim be (1) held for ransom, (2) held as a hostage, (3) confined against his will, or (4) held to service against the victim’s will.
False imprisonment consists of intentionally confining or restraining another person, without his consent and with knowledge that the perpetrator has no lawful authority to do so.
I resorted to a list of numbered items in the definition of kidnapping, and I make no apology for it. Numbering is not very endearing in literary or social writing, but it can be absolute salvation for the lawyer or legislator who quests after the Holy Grail of Absolute Clarity.
More recently, the Supreme Court of Wisconsin had to interpret that state’s recreational immunity statute to determine whether a property owner was liable for the injuries of an invitee on his property, or was immune because the invitee was engaged in recreational activity, of whose consequences the invitee assumed the risk. The statute defined an “owner” as a person who “owns, leases or occupies property,” and defined “property” as “real property and buildings, structures and improvements thereon, and the waters of the state.” Because the invitee was injured in a fall from a structure, the applicability of the recreational immunity statute’s definition of “property” became critical. This is an even more difficult type of ambiguity than the one explored by the New Mexico Supreme Court, because it went to the heart of the issue in the case.
One party maintained that the statute defined three types of property: (1) real property, (2) buildings, structures and improvements thereon, and (3) waters of the state. The other argued that only two types of property were defined: (1) real property together with any buildings, structures and improvements thereon, and (2) waters of the state. The crucial point raised by this dispute was whether a property owner is immune from liability for recreational injuries incurred on any buildings, structures and improvements that he owns; or is immune only if such injuries occur on buildings, structures and improvements situated on land owned by the owner. In other words, the question was whether the immunity provision applied if the owner owned the land but not the structure, or the structure but not the land it stood on, or if he had to own both in order to be immune.
The court reasoned that if the legislature had meant to limit immunity to injuries incurred on structures owned by the owner of the real property on which they stood, they would have said so explicitly. “We cannot rewrite [the statute] in the exercise of interpreting it,” the court wrote. But a dissenting judge, in a rich and enjoyable opinion discussing the statute comma by comma, pointed out that at least four differing definitions of “property” could equally reasonably be interpolated from the statute as written, and that resort to legislative intent was unjustified.4 Probably the judges all went to bed wishing that the injury had occurred upon the waters of the state.
One of the most common types of ambiguity arises from the use of a disjunctive or conjunctive series with a modifying clause, where it is unclear whether the modifying clause applies to all of the terms in the series or just the one nearest it. A simple example is the phrase “frozen meats, fruits, and vegetables.” Are only the meats frozen, or are all three of the things on the list frozen? There is no “rule” of English grammar or statutory construction that helps us with this. The only effective way to deal with an ambiguity of this kind is to nip it in the bud. The writer should spot it and rewrite in order to make sure she has clearly said what was intended. If the modifier applies to only one of the terms in the series, a good rewriting technique is to put that term last: “fruits, vegetables, and frozen meats” leaves no ambiguity. If, on the other hand, the modifier applies to the entire series, the only way to make that clear is to use more words. Brevity may be the soul of wit, but it is not always the clearest way to write. “Frozen meats, frozen fruits, and frozen vegetables” is soporifically repetitious, but at least it leaves no doubt. A better approach might be something like “frozen products, namely meats, fruits, and vegetables” or “meats, fruits, and vegetables, all frozen.”
Our own Appellate Division Three faced a case in 19905 that required it to interpret the language of a real estate contract that read, in part, “Seller agrees to sell to Buyer and Buyer agrees to purchase from Seller the equipment stated in Schedule 1 attached hereto and incorporated herein; the 1982 fruit crop and the following legally described real property …” The drafters’ use of a semicolon after “herein” and no punctuation at all between “fruit crop” and the description of the real property gave rise to a dispute as to what was actually being sold, and whether the language gave rise to a general warranty against encumbrance. The court wrote: “Had the series … been separated by a semicolon after crop — or a comma used after equipment and again after crop — there would be no ambiguity.” Deciding that the parties clearly intended the sale of three things — the equipment, the crop, and the real property — the majority re-punctuated the contract and decided the case accordingly. But a dissenting judge maintained that the majority had rewritten the contract, in essence reforming the agreement without sufficient justification in the record. Is it the judges’ job to interpret and enforce a contract or statute according to what its language says, or to rewrite it according to what the court believes it was meant to say, even though it ended up saying something different?
That is a perennial question, and different courts have answered it in different ways. It’s not unreasonable, though, to expect judges to be language police. By reaching opinions based on what laws and contracts actually say, rather than what their drafters may or may not have meant, judges send a crucial message to lawyers and legislators everywhere: Word choices have consequences, and those who are charged with putting the law into words should therefore be very careful about the words they choose.
Robert C. Cumbow is a shareholder with Graham & Dunn, Seattle, where he counsels clients on trademark, copyright, advertising, and technology issues. He is a past chair of the WSBA Intellectual Property Section and the Bar News Editorial Advisory Board, and is currently a board member of Washington Lawyers for the Arts. He teaches at Seattle University School of Law and writes on law, language, and film.
NOTES
1. New York State Dairy Foods, Inc. v. Northeast Dairy Compact Commission, 198 F.3d 1 (1st Cir. 1999).
2. Commonwealth v. Silva, 488 N.E.2d 34 (Mass. App. 1986).
3. State of New Mexico v. Elsie Clark Jr., 455 P.2d 844 (N.Mex. 1969).
4. Peterson v. Midwest Security Insurance Company, Inc., 636 N.W.2d 727 (Wis. 2001). Though it has limited applicability to Washington law, this case is a great read for property lawyers, personal injury lawyers, and grammarians alike.
5. Sackman Orchards v. Mountain View Orchards, 56 Wn. App. 705, 784 P.2d 1308 (Wash, App. Div. Three, 1990).
