Language Log often weighs in when courts try to nail down the meaning of a statute. Laws are written in natural language—though one might long, by formalization, to end the thousand natural ambiguities that text is heir to—and thus judges are forced to play linguist.
Happily, this week's "case in the news" is one where the lawyers managed to identify several relevant considerations and bring them to the judges for weighing.
Most news outlets reported the case as being about the Oxford comma (or serial comma)—the optional comma just before the end of a list. Here, for example, is the New York Times:
- Lack of Oxford Comma Could Cost Maine Company Millions in Overtime Dispute (news article)
- For Want of a Comma (opinion piece)
However, it wasn't quite that simple. The appeals court's decision explicitly considers several competing tidbits of evidence in much the same way that natural language understanding software tries to do. Humans usually manage this subconsciously, effortlessly, and successfully: most sentences are easily understood despite being formally ambiguous. But sometimes even humans hit a tricky case—or a case with $10 million at stake.
75 truck drivers delivered perishable dairy products. Were they entitled to overtime pay? In Maine, overtime pay does not apply to
(a) The canning, processing, preserving, freezing, drying, marketing, storing, packing for shipment or distribution of: (1) Agricultural produce; (2) Meat and fish products; and (3) Perishable foods.
The appeals court essentially considered whether statute (a) should be read as (b) or (c). In other words, when listing activities in (a), were the legislators omitting (b)'s final comma or (c)'s final conjunction?
(b) The canning, processing, preserving, freezing, drying, marketing, storing, packing for shipment, or distribution of …
(c) The canning, processing, preserving, freezing, drying, marketing, storing, or packing for shipment or distribution of …
Under (b), "distribution" is exempt from overtime, so the dairy needn't pay the truck drivers extra. But under (c), it's only "packing for distribution" that is exempt, so they must pay.
Bayes' Theorem tells us that to evaluate the plausibility of (b), we should consider both whether the legislature would have wished to make rule (b) and also whether someone who wished to make rule (b) would have phrased it as (a). Any other reading, such as (c), should be similarly evaluated.
The district court observed that Maine legislators might reasonably have intended (b), and that if so, they would very likely have phrased it as (a). Indeed, "Maine statutes invariably omit the serial comma from lists"—just as the Maine Legislative Drafting Manual recommends.
The appeal lawyers argued that the legislators had actually intended (c). Here, the final category is "packing for shipment and distribution." But this interpretation hits a hurdle that the news stories don't mention. While (c) might be equally sensible public policy, it's more of a stretch to suppose that the legislators would have phrased it as (a). That would mean they omitted a final conjunction rather than just a final comma. Maine legislators mainly include that final conjunction, just like the rest of us. The appeal lawyers didn't find any Maine laws permitting citizens to "eat, drink, be merry" or calling for a flag of "red, white, blue"—a rhetorical device known as asyndeton.
Nonetheless the appeals court agreed that (c) was plausible enough to stay in the running. (I sort of agree, though I'd explain the missing "or" as more probably an editing error than a deliberate use of asyndeton.)
How did the court arrive at this reasonable doubt? To bolster their case, the lawyers for (c) questioned whether (b) was so plausible after all. If (b) had been the intent, they asked, would the legislators really have written it as (a)? Why wouldn't they have used parallel construction and written (a'), with a list of all gerunds?
(a') The canning, processing, preserving, freezing, drying, marketing, storing, packing for shipment[,] or distributing of …
Now, the lawyers for (b) argued back that (c) wasn't so plausible either because shipment = distribution and therefore "packing for shipment or distribution" would have been redundant. But the lawyers for (c) came back with evidence that shipment ≠ distribution. They also noted that this sword cut both ways, since if shipment = distribution, then someone who meant (b) wouldn't have had to change the word, but could have written
(a") The canning, processing, preserving, freezing, drying, marketing, storing, packing for shipment[,] or shipment of …
In the end, the appeals court found that neither side had a knockdown argument on the question of whether (b) or (c) would be more likely to give rise to (a):
We are reluctant to conclude from the text alone that the legislature clearly chose to deploy the nonstandard grammatical device of asyndeton. But we are also reluctant to overlook the seemingly anomalous violation of the parallel usage canon that Oakhurst's reading of the text produces.
Nor was it clear from the legislative history whether the state had meant to enact (b) or (c). With Bayes' Theorem not providing a clear winner, the judges fell back on an established tiebreaking mechanism. They simply construed the statute in favor of the truck drivers because the broad purpose of employment law is to protect employees.
This back-and-forth might give you a sense of why it's so challenging for computers to understand or translate human language, or to use it as a basis for decisions! It will be some time before we can automate judges and lawyers. In the meantime, however, technology might be able to provide some help to legislators. Grammar checkers could be extended to flag potentially consequential ambiguities when a bill is being drafted. I don't know of much work on this problem, beyond the detection of puns, but here's a 2010 paper at least.
A final remark for those of you who came to this post because you have opinions on the Oxford comma. Whichever comma convention we adopt, most sentences will be easy to interpret, and a few will be confusing. But at least this legal case provides some variety in the example sentences. When people argue that omitting the Oxford comma creates ambiguity, they usually adduce examples that are ambiguous because the conjunction "Y and Z" could be misread as an appositive. (In the 3-element list "X, Y and Z" this reading involves no serial list. In the 4-element list "W, X, Y and Z" it involves a serial list with asyndeton.) The example in this legal case instead misreads "Y and Z" as the object of a preposition ("V, W, X for Y and Z"). Or perhaps it's not a misreading, if you agree with the truck drivers.
Update: Kevin Underhill's entertaining blog "Lowering the Bar" finds it laughable that the appeals court was willing to balance asyndeton against non-parallel usage. I agree that the former story seems orders of magnitude less probable. (Again, if you want to argue for (c), you're better off claiming that the omission of "or" was a drafting error—which escaped notice because of the subsequent "or"—rather than a rare literary device. It's just that courts are averse to correcting the text of laws that it's not their job to write.)
In the comments below, Mats Rooth, Neal Goldfarb, and Guy explain why the non-parallel usage in (b) would be perfectly natural, while a_George explains why (c) would reflect a more coherent legislative intent.