Lawyers should learn linguistics, part infinity
« previous post | next post »
Ken Adams, "Courtesy of the High Court of England and Wales, A Reminder that Ambiguity Is Best Left to Experts", Adams on Contract Drafting 12/11/2016:
Here’s the language at issue (emphasis added):
Without prejudice to the provisions of paragraph 3 if all of the Conditions have not been discharged in accordance with this Schedule by the Longstop Date, then either Asda or Dooba may rescind this Agreement by giving to the other not less than ten working days written notice to that effect.
So can the Agreement be rescinded if "all of the Conditions have not been discharged" with broad scope of negation, meaning that "it's not the case that all of the Conditions have been discharged", i.e. at least one Condition has not been discharged? Or should we construe the clause with narrow scope of negation, requiring that every one of the Conditions must have failed? The deputy judge reasoned as follows:
The subject of the clause is “all of the Conditions”; the characteristic which the subject is required to have is “have not been discharged”. As a matter of strict Boolean logic, the relevant characteristic is a negative one, which must affect all of the Conditions in order to fall within the clause. […]
I agree with Asda that the formula “if all … have not …” is sometimes used to mean “if not all … have”, but I do not accept that this has become its primary meaning.
Ken suggests that
Instead of assuming that it was incumbent on him to rely on his own instincts in interpreting the language at issue, the deputy judge should have considered that perhaps an entire field of study is devoted to analyzing this sort of confusion.
He cites CGEL chapter 5, §5.1 to the effect that the scope of negation is systematically ambiguous in sentences like "All of the meat wasn't fresh", and goes on to his primary point, which is to offer guidance in how to draft contracts without planting this particular sort of semantic time bomb in them.
Half a century ago, there was a theory that such differences in scope-of-negation interpretation were a matter of dialect (or idiolect) variation — see e.g. Guy Carden, "A note on conflicting idiolects", Linguistic Inquiry 1970 — but it was soon shown that the judgments in question are variable, contextual, and unstable over time even for individual subjects.
The judge's overconfidence in the power of naive "logic" and in his own interpretive judgment reminds me of a similar issue in a non-legal context, discussed in "James J. Kilpatrick, Grammarian", 7/29/2006 — where a self-appointed expert proclaimed that only narrow-scope interpretations are legitimate, so that the phrase "mass transit is not an option for all drivers" necessarily means that it's not an option for any drivers.
In that post, I quoted examples of broad-scope interpretations from William Shakespeare, Anthony Trollope, Herman Melville, and T.S. Eliot. Today I'll cite a few legal texts where a broad-scope interpretation of "if all X not Y" is clearly assumed:
From the "The Companies Act 1989 Part II (Consequential Amendments) Regulations 1995" (U.K.):
One of the effects of Part II of the Companies Act 1989 is that a Scottish firm may be eligible for appointment as a company auditor even if all the partners in the firm are not individually eligible.
A U.S. Supreme Court opinion from 1869:
If it be argued that Foster and others were not parties to the decree and not bound by it, we answer, that if all the necessary parties were not before the court when the case was originally heard, it was the duty of the court to require the plaintiffs to bring them in, and a failure to do so was ground of error in that cause.
And one from 1882:
We understand that in this case all the parties interested on one side of the controversy, and all being citizens of other states than Texas, have applied for the removal as against the parties on the other side, who are all citizens of Texas. If this is so, then this case was properly removed under the first clause of section 2 of the act of 1875. If all the parties on one side have not applied for the removal, then the case was properly removed under the second clause of the second section of said act.
Code of Federal Regulations 31 CFR § 358.5:
If all of the callable coupons associated with the corpus are not submitted with the corpus, the corpus will be converted to a non-transferable BECCS security. Any remaining callable coupons submitted with the corpus will be converted to individual non-transferable CUBES securities.
If all of the information required under this part is not provided within the specified time limit, PBGC may assess a separate penalty under ERISA section 4071 against the filer and each member of the filer's controlled group (other than an exempt entity) of up to $1,100 a day for each day that the failure continues. PBGC may also pursue other equitable or legal remedies available to it under the law.
Any participant may file a motion requesting the Commission to issue a final decision without any initial decision. If all participants join in the motion, the motion is granted, unless the Commission denies the motion within 10 days after the date of filing of the motion or, in the case of an oral motion under paragraph (c)(2) of this section, within 10 days after the motion is transmitted to the Commission. If all participants do not join in the motion, the motion is denied unless the Commission grants the motion within 30 days of filing of the motion or, in the case of an oral motion under paragraph (c)(2) of this section, within 30 days after the motion is transmitted to the Commission.
A few other LLOG posts that underline the potential value to lawyers and judges of relevant training in linguistic analysis:
"Rarely better than de re", 10/23/2005
"Lawyers in need of linguistic training", 8/7/2006
"Adverbial modification at the Supreme Court today", 2/25/2009
"Law as applied linguistics", 7/25/2009
"Grammatical justice is served", 5/7/2009
"The indubitable equivalent of such claims", 12/16/2009
"Corpus linguistics in a legal opinion", 7/20/2011
"An empirical path to plain legal meaning", 3/3/2012
"Corpus linguistics in statutory interpretation", 3/3/2012
"It depends on what 'the' means …", 5/14/2012
"What 'the' means", 1/28/2013
"Misreading like a lawyer", 9/27/2013
"Legal scope again", 10/30/2013
"Syntax and semantics at the Supreme Court", 1/13/2014
"SCOTUS: A fish is not a 'tangible object'", 2/25/2015
"Legal lexicography", 1/4/2015
"Anaphoric definiteness in the ACA", 3/4/2015
"We'll just have to pass it…", 6/25/2015
"'Linguists have a name for this kind of analysis'", 8/15/2015
"Fun with commas", 8/17/2015
"The billion-dollar conjunction", 12/30/2015
"Annals of conjunction", 3/1/2016
"Corpus-based judicial opinions", 7/2/2016
Ari Corcoran said,
December 13, 2016 @ 9:19 am
But one judge, at least, is prepared to take a linguistically sensible approach, see Catchwords below.
R v Qaumi & Ors (No 54) [2016] NSWSC 1067 (8 August 2016)
Last Updated: 30 November 2016
Supreme Court
New South Wales
Case Name: R v Qaumi & Ors (No 54)
Medium Neutral Citation: [2016] NSWSC 1067
Hearing Date(s): 2 August 2016
Date of Orders: 2 August 2016
Decision Date: 8 August 2016
Jurisdiction: Common Law – Criminal
Before: Hamill J
Decision: Evidentiary rulings made – see paragraphs [8], [14], [21], [22] and [27]
Catchwords: CRIMINAL LAW – admissibility of evidence – listening device recording of two accused in police cells after their arrest – lengthy recording – parties agreed on all but four passages of transcript – recorded conversation similar to evidence previously excluded – evidence excluded because police failed to record admissions – whether unfair to use evidence – relevance of evidence to the issue of whether accused had access to firearms – whether danger of unfair prejudice – assessment by accused of strength of police case – “I think we’re fucked” – not an admission or demonstrative of a consciousness of guilt – merely an assessment of the strength of the evidence
Legislation Cited: Criminal Procedure Act 1986 (NSW)
Evidence Act 1995 (NSW)
Cases Cited: Em v Queen [2007] HCA 46; 232 CLR 67
R v Qaumi & Ors (No 3) (Severance and separate trial) [2016] NSWSC 15
R v Qaumi & Ors (No 48) [2016] NSWSC 1008
Andrew Bay said,
December 13, 2016 @ 9:44 am
I feel that lawyers should pick up a programming language and structure law more like that.
Especially braces, parenthesis, and booleans.
IF (Condition A is false AND Condition B is false AND Condition C is false) then "may rescind with 10 day notice".
is different from
IF (Condition A is false OR Condition B is false OR Condition C is false) then "may rescind with 10 day notice".
Then again, lawyers like having ambiguity to argue over. That's how they make their money.
Cervantes said,
December 13, 2016 @ 10:01 am
It gets better, or worse, depending on what you're looking at.
Quoted above is Paragraph 2.3 of (Schedule 4 of) the contract:
That's Paragraph 2.3 in full; here is Paragraph 2.2 in full:
And here's the (deputy) judge on the difference:
The judge went on to find that "Sub-paragraph 2.3 does not sit happily with the alternative regime for rescission under sub-paragraph 2.2 and paragraph 3. Neither party is able to provide a satisfactory explanation for this. Accordingly the rival arguments cancel one another out" — at which point (more or less), he sided with Dooba, threw out the summary judgment, and allowed the appeal to go forward.
Ken writes:
If they did, in fact, do so, the judge may have missed it because he was distracted:
RP said,
December 13, 2016 @ 11:13 am
The bit you quote from "The Companies Act 1989 Part II (Consequential Amendments) Regulations 1995" is from the Explanatory Note, which doesn't form part of the Regulations themselves. So if by legal text we mean laws or legal regulations (rather than texts explaining or describing the law) then the search is still on for a UK legal text that contains this usage.
Bloix said,
December 13, 2016 @ 12:24 pm
Andrew Bay – contracts are made and construed by non-lawyers all the time. Unlike programs, it is supposed to be possible to read and apply a contract without being a specialist. Would you want your apartment lease or your car rental agreement written in a way that required you to pay a lawyer to understand it?
As for this agreement, as a lawyer I found it perfectly intelligible and it's clear to me that the court is wrong – not because of any linguistic analysis, but because of what a "condition" is. In a contract, a condition is a requirement that Party A must fulfill before Party B's obligation to perform is triggered. A condition might be to pay money, or to provide information, or to arrange for entry into a property, or to obtain a permit, or any other thing like that.
Here, the parties have set out a list of things that are conditions, have established a date by which they must be met, and have provided a method for rescission in the event that they are not met. It is far more reasonable to interpret this to mean that the contract can be rescinded if any one of them is not met than to say that it can be rescinded only if all of them are not met. Why? Because if any one of them is not met, then under the common law rules, the counter-party need not perform at all. So you would have two rules: (1) if all conditions are not met, either party may rescind the contract; (2) if any condition is not met, the counter-party need not perform That would be a bizarre result. Why would you put in a provision that governs rescission in the event of failure to meet conditions if it didn't replace the usual common law rule?
[(myl) I agree that common sense leads to an appropriate answer in this case: but with luck, some sound linguistic background might have prevented the judge from being led away from common sense by a bizarre theory about "strict Boolean logic".]
J.W. Brewer said,
December 13, 2016 @ 12:55 pm
Just as a note on dialect variation in legalese, I am pretty sure I have never previously seen or heard the NP "Longstop Date," but it was immediately clear to me from context that it must be what American contract lawyers colloquially call a "drop-dead date," and sometimes call other things if they feel that's not the right register for the formal contract language.
Tom O. said,
December 13, 2016 @ 12:57 pm
There's a legal linguistics issue here in Michigan that I've been puzzling over. On January 10, the Michigan Supreme Court is set to hear a case involving a supposed attachment ambiguity. The language at issue is in Michigan Compiled Laws 750.520n(1):
The Michigan Court of Appeals held in People v Brantley, 296 Mich App 546 (2012) (pdf) that the long prepositional phrase starting with "for criminal sexual conduct …" modifies only "520c". In other words, someone convicted under section 520b must be sentenced to lifetime monitoring without regard to the age of the defendant or the victim. The Court of Appeals relied on the last antecedent rule. (It's at Part IV of the opinion.)
That strikes me as not just wrong, but ungrammatical. Everyone I've talked to interprets the statute to apply the age restrictions to both sections 520b and 520c, and even after reading Brantley I can't force myself to see the other interpretation. But as a lawyer, not a linguist, I don't know how to explain it. (I think the answer is that the "for" phrase has to complement "convicted," but I'm not sure.)
J.W. Brewer said,
December 13, 2016 @ 1:07 pm
Going back to the prior post linked above on "Corpus-based judicial opinions," which generally celebrated the notion of judges learning to think about linguistic issues in a sensible and empirically-driven way, I note that that post praises a decision by the Michigan Supreme Court and references positively an earlier opinion by Justice Lee of the Utah Supreme Court.
By an interesting coincidence, Mr. Trump's "official" list of potential nominees to the U.S. Supreme Court includes both Justice Lee and two members of the Michigan Supreme Court (although perhaps not ones who actually wrote in that case). I tend to doubt that it was their favorable view of corpus-based approaches to interpreting statutory language that got them short-listed, but if a Trump Administration leads to better use of corpus-based techniques at the SCOTUS level we may perhaps appreciate that the Lord works in mysterious ways.
DWalker07 said,
December 13, 2016 @ 1:47 pm
"… and all being citizens of other states than Texas, …"
Texas. It's a whole 'nother country.
J.W. Brewer said,
December 13, 2016 @ 2:07 pm
In partial defense of the learned judge, if you draft sentences in a fairly formal document with an explicit "If P, then Q" syntactic structure, you might reasonably be taken to be inviting the reader (esp if the reader has the sort of background where he or she might plausibly have taken a course or two on formal logic in college and thus know just enough to be dangerous) to muddle up the grammar of some sort of formal logic (whether Boolean or otherwise) with the grammar of natural language. That doesn't mean the judge got it *right* even within the grammar of some plausible formal logic, of course. If you don't want your reader to muddle up natural-language grammar with formal logic, perhaps it's a good idea to avoid the sort of stylistic cues (both register and lexicon) that could encourage such muddling.
Cervantes said,
December 13, 2016 @ 2:09 pm
Tom O, compare that language to "A person convicted under section 520b or under 520c for …"
Guy said,
December 13, 2016 @ 2:38 pm
"As a matter of strict Boolean logic, the relevant characteristic is a negative one, which must affect all of the Conditions in order to fall within the clause."
I think the most offensive part was of this reasoning is the idea that "strict Boolean logic" somehow specifies that negation always has scope over quantification, when of course the relative scopes would be specified not by logical principle (because both are logically possible) but would be expressed in a formal language by syntactic rules that make the scopes unambiguous, whereas this is written as an ambiguous English sentence. It could have been made unambiguous by either "not all have" or "none have" instead of "all have not", thus avoiding the judge reaching this seemingly absurd conclusion based on linguistic misanalysis.
Guy said,
December 13, 2016 @ 2:50 pm
Or rather that quantification always has scope over negation, I should say.
Bloix said,
December 13, 2016 @ 3:45 pm
Tom O. – the Michigan statute issue is that prosecutors can always find a way to make a hideously draconian statute (a seventeen-year old who has sex with a 13-year old has to wear an ankle cuff GPS device FOR LIFE?) even more draconian.
Bloix said,
December 13, 2016 @ 4:05 pm
"common sense leads to an appropriate answer"
a canon of contractual construction is that documents should be read as a whole and provisions should be construed to carry out the parties' intention. Here the judge seems to have become enamored with his half-understood "Boolean logic" to the exclusion of traditional legal rules. This isn't a failure of the law; it's the failure of a judge to apply the law.
Cervantes said,
December 13, 2016 @ 6:43 pm
J.W. Brewer:
Noun and verb, from cricket; used by analogy (and certainly in real-estate development contracts) to mean "of last resort," etc.
Geoff said,
December 13, 2016 @ 8:58 pm
The bizarre thing about these legal attachment ambiguity and scope of negation cases is that it's so easy (providing you have a suitably primed editor's mind) to avoid the problem with a few extra words or commas. If an ambiguity gets to court, that just shouts 'incompetent drafting'.
I had a similar case in my own will quite recently. The intended meaning was 'if all [the heirs named so far] predecease me, the following alternative conditions apply.' The lawyer drafted 'If I am not survived by all [the heirs named so far]….' I couldn't convince her there was any problem. In the end I let it pass, since in context the non-intended interpretation is obviously absurd.
Brett said,
December 13, 2016 @ 9:44 pm
@Tom O.: While the reading you prefer certainly strikes me as the most natural one, I don't think anything strictly forecloses the other meaning. So there is a real ambiguity, and it seems to me that then the rule of lenity calls for the more awkward reading. So there is a balancing necessary.
Geoff said,
December 13, 2016 @ 10:09 pm
@ Tom O
'A person convicted under section 520b or 520c for criminal sexual conduct…'
So who invented this 'last antecedent rule'? The intended meaning is obvious. The claimed rule is ridiculous.
The other meaning might be reasonable if the sentence had been:
'A person convicted under section 520b, or under 520c for criminal sexual conduct…'
Graeme said,
December 13, 2016 @ 10:29 pm
As a lawyer I'm with Bloix and can't see what the linguistic fuss is. 'Condition' has a strong and settled meaning – it's a contractual promise that goes to the root of the contract. In contrast to a mere 'warranty'. So there is literal clarity on the face of the phrase if it was written by a solicitor or attorney. That's just basic reading in context.
Any ambiguity could only arise if the rest of the contract or set of conditions showed a different purpose. Without seeing the judgment/contract we can't say. But perhaps some things were labelled 'conditions' which were in substance warranties. If so the failure isn't in the stock clause discussed here, but in the substantive assessment of interests and risks in the list of 'conditions'.
John Swindle said,
December 13, 2016 @ 10:49 pm
The Oxford dictionary online says longstop was a fielding position in cricket but is no longer generally used. The longstop date would be when that change occurred.
Mark Mandel said,
December 14, 2016 @ 12:48 am
Jeez, haven't these legal beagles ever heard of "none" and "unless"? As in
"unless all of the Conditions have been discharged"
and
"if none of the Conditions have been discharged"
which are unambiguous and easily understood by the layperson, unlike "if all of the Conditions have not been discharged".
Neal Goldfarb said,
December 14, 2016 @ 1:29 am
@ Tom O:
Interesting case. I agree with you that the PP "for sexual conduct…" is most naturally ready to apply to both sections, even though that runs contrary to the tendency for local/recent (= low) attachment.
I think two factors are at work. First, the locality/recency preference is probably at its weakest when each of the two items in question is very short. Compare, e.g., men and women living in New York. To me, it's easier to understand the PP as modifying both Ns, unless a prosodic break is added before women. And note that if the locality/recency preference results (mainly) from working-memory limitations rather than syntactic preferences, this is probably what you'd expect.
Second, it seems to me that there is a strong tendency to read section 520b or 520c as a single unit, and in order to easily understand the PP as attaching only to 520c, that preference has to be overcome. If you try to force the Brantley reading by inserting a prosodic break before and 520c, the result sounds substantially stranger than adding such a break in my men and women example.
The more I think about it, the more I think that prosody is a major factor here.
Are there any real psycholinguists in the house?
Ran Ari-Gur said,
December 14, 2016 @ 2:05 am
> I agree with Asda that the formula “if all … have not …” is sometimes used to mean “if not all … have”, but I do not accept that this has become its primary meaning. [emphasis mine]
This is interesting. The judge seems to be operating under a theory of "English degenerates over time": since one reading strikes him as more logical, he assumes it must be the original one, with the other one being a recent corruption.
Breffni said,
December 14, 2016 @ 3:50 am
Replacing "all" with "any" would seem to do the trick: "if any of the conditions have not been discharged…". Although if some crashing pedant of a drafter insisted on singular agreement with "any", that would introduce a new ambiguity: "if any of the conditions has not been discharged…" could be taken to mean exactly one and no more.
Use of bulleted lists would resolve a lot of scope ambiguities. In the case Tom O raises, the oddball reading corresponds to:
The presumably intended reading would be represented as:
I agree with Brett that there's a real ambiguity here and no ungrammaticality. You can see the unintended reading better if you reverse the disjuncts:
Tom O. said,
December 14, 2016 @ 10:33 am
Thank you for all the replies. You've helped me clarify my own thoughts about this.
I meant to include a link to the Michigan Supreme Court's page for the case where they are considering this issue: People v Comer. Oral arguments will be January 10, and they will have a live stream, if anyone is that interested.
John Williams said,
December 14, 2016 @ 11:00 am
As a corpus linguist, I find this sort of thing fascinating.
By coincidence, as a private individual, I was an interested party in a (UK) legal case a few years back that hinged partly on the meaning of 'benefit solely from'. There were two sources of funding available (Fund X and Fund Y) on which various parties had a claim. The key document determining the distribution of the funds contained a sentence along the lines:
Now, this could mean:
(1) Only Person P is entitled to benefit from Fund X. or
(2) Person P has no claim on Fund Y.
As a good corpus linguist I looked up the phrase in the largest corpus I have available, enTenTen13. Of the 15 examples, a few were indeterminate (or a different meaning of benefit) as to the two interpretations. Among the others, there appeared to be a majority for interpretation (2). Crucially however, the one instance from a legal/contractual context seemed to favour interpretation (1):
So what is the nature of the evidence here? We seem to be no further forward.
In the end, I was not allowed to be expert witness in my own cause (and other corpus linguists would have been equally suspect since I know a lot of them), and the case was settled out of court.
James Wimberley said,
December 14, 2016 @ 11:11 am
Andrew Bay: You can reach clarity in complex conditions without special punctuation. Nested indents will do it.
Is there some hoary legal tradition, justifying monstrous sentences in continuous prose, that it is important to save precious parchment?
Guy said,
December 14, 2016 @ 12:33 pm
@Geoff
I'm not sure that there is any ambiguity in "haven't … all", where "all" is determiner in a complement in linear position after the negation. Doesn't negation always have scope over quantification here? A brief look at COCA seems to confirm this, though I can see how someone might confuse themself into thinking "all" is replaceable by "any" in that context, even though I'm skeptical that it ever is.
Guy said,
December 14, 2016 @ 1:27 pm
@James Wimberly
But you have to be careful to do ithe indents well. In United States v. Hayes, the Supreme Court interpreted the language:
"Except as provided in subparagraph (C),[2] the term “misdemeanor crime of domestic violence” means an offense that—
(i) is a misdemeanor under Federal, State, or Tribal [3] law; and
(ii) has, as an element, the use or attempted use of physical force, or the threatened use of a deadly weapon, committed by a current or former spouse, parent, or guardian of the victim, by a person with whom the victim shares a child in common, by a person who is cohabiting with or has cohabited with the victim as a spouse, parent, or guardian, or by a person similarly situated to a spouse, parent, or guardian of the victim."
This structure suggests that the "committed" clause is modifying "use or attempted use of physical force, or threatened use of a deadly weapon", but the Court concluded that the conjunction between (i) and (ii) ended at the word "weapon" and the "committed" portion was logically outside (ii) and modifying "offense". The Supreme Court's interpretation was probably the intended one, but to reach it you basically have to ignore the subsection structure and formatting and read it as if it were one long rambling passage.
If you just add a paragraph break after weapon, then the structure and text are in harmony, but not doing so made the structure and text point in different directions.
Guy said,
December 14, 2016 @ 1:34 pm
Additionally, at oral argument, Justice Scalia was lead away from the majority's interpretation because the "that-" was outside (i), so when the lawyer suggested that the "committed" part was outside (i) and (ii), he tried to read it as "offense that comitted" (deleting (i) and (ii) but not the "that" that goes with them), which doesn't really parse right. Presumably if "that" had just been placed in (i) this wouldn't have been how he had tried to read it. I think this is a pretty good example of a case where the formatting and subsection structure made a muddle of a text that would have been much clearer without it.
Alan Gunn said,
December 14, 2016 @ 2:41 pm
I agree that lawyers should learn more about language than they do. But what, specifically, should they learn? In my opinion, just enough so that they understand that it makes no sense, especially in the US, where legislation is drafted with little care, to try to resolve cases by close analysis of the particular statutory language. I was a tax lawyer. The best tax decisions, on the whole, came from the Tax Court, because its judges are tax professionals who (with exceptions) usually understand what the purposes of the stature in question are and what result makes sense. Generalist judges, like those of the Supreme Court, generate a lot of terrible tax decisions because they (again, with exceptions) can't be bothered to figure out what the statute is meant to do, and so they try resolve questions purely by reading the statute's words, too often with a wooden literalism.
Unfortunately, most of the legal literature on statutory interpretation is written by people whose field of practice or academic study is litigation. So they look for shortcuts that will let lawyers decide cases on matters that are heavily codified, like taxation and commercial law, without taking the trouble to understand what the objectives of the statutes in question are. Some of it almost seems to suggest that judges don't need to know anything except the English language and where to get a dictionary. That sort of thing doesn't help.
Neal Goldfarb said,
December 15, 2016 @ 2:14 am
Guy: Regarding Hayes, you might be interested in the amicus brief that I filed in in the case when it was at SCOTUS. The brief dealt only with the linguistic issues, and was filed on behalf of several linguistics professors (Georgia Green, Ray Jackendoff, Jeffrey Kaplan, and Roger Shuy), as well as Ted Gibson, who is a cogsci professor at MIT.
The brief's position on the linguistic issues it discussed was the opposite of what the Court ultimately held. I'm biased, of course, but I don't think the Court made any serious effort to deal with the brief's argument.
In any event, Mark's not the only person who's trying to increase the linguistic sophistication of lawyers and judges.
chris said,
December 16, 2016 @ 8:18 pm
@Guy: If you just change the part just after "weapon" to ", and (iii) is committed by…" then you much more clearly get the majority result (although I might argue for also then breaking down (iii) into (a), (b), (c), and (d)).
And it makes the text correspond to the structure of the phrase "misdemeanor crime of domestic violence"; (i) establishes that there was a misdemeanor crime, (ii) that it was violent, and (iii) that the relationship between criminal and victim was a domestic one.
Partly, I think, it hinges on the ambiguity of the term "offense" as both a category of similar offenses, and a specific offense that was actually committed; a domestic relationship is a (disturbingly common) circumstance of offense-instances, but would rarely be an element of an offense-category, so requiring that seems unlikely to be the meaning intended by the legislature. But then, Scalia was known for his disregard for legislative intent.