The Corpus in the Court . . . again.
« previous post | next post »
A note from Stephen Mouritsen:
I wanted to give you a heads up about a second judicial opinion (again by Justice Tom Lee on the Utah Supreme Court) that overtly relies on data from Mark Davies' COCA. The opinion is here, and the discussion of corpus data is found in paragraphs 36 through 40.
Here's what the case is about:
We must decide whether the juvenile court erred in concluding that a minor’s alleged solicitation of a stranger to punch her and terminate her pregnancy qualifies as an abortion, as the term is defined in the Utah Code. The juvenile court held that an assault of a woman by punching her stomach was a “procedure” intended to terminate her pregnancy and therefore fell within the statutory definition of “abortion.” Because a woman cannot be held criminally liable for seeking an abortion, the court dismissed the State’s delinquency petition against the minor.
We hold that the solicited assault of a woman to terminate her pregnancy is not a “procedure,” as contemplated by statute, and therefore does not constitute an abortion. Accordingly, we reverse the juvenile court’s order dismissing the State’s petition and remand for further proceedings.
And here is the relevant part paragraph 40 of Justice Lee's concurring opinion:
There are means at our disposal to answer the question that the majority raises in Part I.A—whether the “ordinary and accepted meaning of the term ‘procedure,’“ supra ¶ 21, is limited to medical methods. I have employed such means (corpus linguistics data) before, explaining that an empirical measure of ordinary usage may be appropriate to check our less-than-perfect judicial intuition. In this case, the majority’s confident assertion about the “ordinary and accepted meaning” of the term “procedure” in an abortion setting ultimately is based on its intuition—that although “procedure” sometimes signifies any means of accomplishing a result, an “abortion procedure” has reference to a medical procedure. I do not doubt that intuition. (In fact, empirical corpus analysis confirms it.3)
Footnote 3 explains:
In contemporary usage, “abortion procedure” references the termination of a pregnancy under medical conditions, such as in a clinic and under the supervision of a physician. This conclusion is based on a review of every instance in which the words “abortion” and “procedure” co-occur in the Corpus of Contemporary American Usage. See, The Corpus of Contemporary American English (COCA), http://corpus.byu.edu/coca/ (last visited Dec. 1, 2011). Enter “[abortion].[n*]” in the “word(s)” field and “[procedure].[n*]” in the “collocates” field, and select “LIST,”then click “search.”
This search revealed 223 co-occurrences of “abortion” and “procedure.” Of those, 106 referred to specific medical procedures such as dilation and extraction or vacuum aspiration. An additional 75 refer to circumstances in which an abortion is performed in a “clinic” or with a “doctor” or under “surgical” conditions. The remaining 27 use the terms “abortion” and “procedure” interchangeably. In 5 instances, the term “procedure” and “abortion” were not related. Not once were the terms used to connote an ad hoc, violent, nonmedical effort to terminate a fetus (as by striking the mother’s abdomen).
Stephen Mourtitsen gives some personal background:
Both Justice Lee and Professor Davies are in one respect my former bosses. I was a research assistant for Professor Tom Lee, who was a law professor before being elevated to the bench a year-and-a-half ago. I then served as his law clerk (essentially a research assistant) for a year before moving to New York, where I now work as an attorney. As a graduate student in linguistics prior to law school, I had Mark Davies as my thesis chair.
Justice Lee's conversion to the corpus cause is significant. Neal Goldfarb's brief showed the legal community that corpus data could be highly persuasive to non-linguist judges. As a consequence, I think the brief will have a dramatic effect on how advocates present arguments to the courts. Justice Lee's appeal to corpus data shows the judicial community that corpus data is not just specialized esoterica–it has a practical application to the one job that judges are supposed to be good at (but aren't), i.e., resolving textual ambiguities in a principled manner.
I write about law and corpus linguistics and I think that Justice Lee's use of the corpus (even now when he no longer has a corpus-nerd clerk at his disposal) is significant. Against the backdrop of the so-called "culturomics" movement, and the continued development of larger and better corpora, I think that corpus linguistics has the potential to reshape the way judges and attorneys do their jobs.
Stephen's reference to "Neal Goldfarb's brief" is explained and discussed in these earlier LL posts, and (especially) the links therein: "Q: Do corporations have 'personal privacy'? A: No.", 3/1/2011; "Now on The Atlantic: The corpus in the court", 3/4/2011; "Corpus linguistics in a legal opinion", 7/20/2011.
You may also be interested in Stephen's 2010 BYU Law Review article, "The Dictionary Is Not a Fortress: Definitional Fallacies and a Corpus-Based Approach to Plain Meaning".
Carl said,
December 13, 2011 @ 9:19 pm
You are a brave man to leave the comments open on this one.
Eric P Smith said,
December 13, 2011 @ 10:08 pm
I think the court’s conclusion that “the solicited assault of a woman to terminate her pregnancy is not a ‘procedure,’ as contemplated by statute, and therefore does not constitute an abortion” is very sensible.
It reminds me of an even more macabre court case in English tax law: Bourne v Norwich Crematorium Ltd, 44TC164 (1967). Norwich Crematorium claimed Industrial Buildings Allowance on the grounds that their trade of cremation consisted in “the subjection of goods or materials to a process”. Stamp, J, in the English High Court, conceded that cremation might be a process, but held that “it would be a distortion of the English language to describe the living or the dead as goods or materials”. In a famous passage he declared:
The words of a true linguist. And he reached his conclusion without the aid of corpus linguistics.
Ray Girvan said,
December 13, 2011 @ 11:21 pm
Much as I generally applaud the use of corpus data in legal contexts, I don't follow the logic in this case. It seems to be letting the letter of the definition over-ride the spirit of the law.
It was a physical action solicited as an attempt to procure an abortion. If the intention of the law was to cover such situations, it seems to be introducing an artificial distinction between different actions with exactly the same purpose, depending on whether the action is normally called a "procedure" or not.
Joyce Melton said,
December 13, 2011 @ 11:57 pm
To Ray Girvan: No, I think the intent of the statute was to make it non-criminal to seek medical help to terminate a pregnancy just for the reason that seeking to terminate a pregnancy by non-medical means should still be a criminal action. The idea is to prevent young women from seeking to abort by using coat hangers, thugs, car accidents or poisons and so steer them to safe and non-criminal medical means.
The judge was right and instead of using only his common sense, he backed up his opinion with corpus data to help insure against being reversed by some higher court.
The decriminalizing of abortions via medical procedure was the intent of the law, not simply to facilitate abortion by any means available such as jumping out a third story window — a procedure, yes, but not one envisioned by the framers of the law.
Kenny said,
December 14, 2011 @ 1:56 am
@Ray Girvan
I'm not sure if you read Justice Lee's entire opinion. I'll say up front that I'm not a linguist or a lawyer, but what he says makes good sense, though it is not completely clear from the quoted portion in this post. In paragraph 40, Judge Lee goes on to say "[His] quarrel is just that this conclusion [that procedure means medical procedure] follows from our understanding of the word “procedure” in the context of the term “abortion,” not from the fact that the statute most often refers to “medical procedures” and not just “procedures."
The definition of abortion at the time of the punching was as follows: "the intentional termination or attempted termination of human pregnancy after implantation of a fertilized ovum, and includes any and all procedures undertaken to kill a live unborn child and includes all procedures undertaken to produce a miscarriage." The definition was amended in 2010, and you can find that version of the Utah code here.
Justice Lee contends that the main opinion's linguistic reasoning was incorrect. He implies that "procedure" used in this context is not clearly limited to medical procedures. I would agree because if the phrase "any and all" is not enough to lift a word out of a specific context's definition to a more general definition, I don't know what is. Justice Lee also observes that past court rulings have tended to interpret internal terminology differences (between compounds [such as "abortion procedure"]and individual words [such as "procedure"]) to be intentional, which the main opinion glossed over by saying that the words were used interchangeably. It is also interesting to note that Justice Lee actually did not need the linguistic evidence to decide her view of the case, but she did need it to correct the arguments in the main opinion.
Kenny said,
December 14, 2011 @ 1:57 am
—-not @Ray Girvan
There were lots of other factors, linguistic and nonlinguistic, that most people would consider relevant to the case that are not addressed in the appellate decision. The girl in this case had already attempted to get an abortion at a clinic, but was told she was too far along in the pregnancy to get one legally. That's why she solicited a stranger to punch her (which she did 13 weeks after the clinic visit), though ultimately her baby lived. The state charged her with solicitation of murder. The problem was that the definitions of abortion and the definitions related to homicide were oddly broad and unspecific. Utah code specified as follows: "There shall be no cause of action for criminal homicide for the death of an unborn child caused by an abortion". There's no reference to agency or legality here, which I find crazy and inexplicable. I don't know Utah's law, but it seems to me she could have been charged with obtaining an illegal abortion. It also seems that neither she nor the stranger who punched her, could have been charged with killing (or attempting to kill) the baby. In order to charge her with soliciting a murder instead of some other crime, the state had to make her action fall outside of the abortion definition.
Kenny said,
December 14, 2011 @ 2:01 am
@Joyce Melton
Killing the fetus without the mother's consent is precluded from falling under the abortion definition now, and it probably was then, though because the law was changed after the events of the case, I can't tell. So, we don't have to worry about being able to charge people who kill a fetus that the mother intends to give birth to. The more interesting and troublesome situations are like the one in this case. I think this girl deserved to be charged with soliciting murder since she was not legally allowed to get an abortion, but what about a girl who could legally obtain an abortion but is too ashamed to seek one out at a clinic or can't afford to get one? If she decides to use a coat hanger or have someone punch her into miscarriage, should we be able to charge her with murder? Under the linguistic argument of the main opinion, the the stat could charge her. Under Judge Lee's linguistic analysis, the state could not. Under Judge Lee's legal analysis, it seems the state could.
Kenny said,
December 14, 2011 @ 2:05 am
Judge Lee is male. I accidentally slipped into using she at the end of my first comment. It should have said Justice Lee actually did not need the linguistic evidence to decide the case, but he did need it to correct the arguments in the main opinion.
eye5600 said,
December 14, 2011 @ 1:17 pm
Speaking from a statistical point of view, I think that trying to define a rare occurrence by reference to what is ordinary, usual, or conventional is an error. What is the sample size?
Further, there is bias is referencing the term "abortion procedure" because "procedure" is a medical term of art.
David said,
December 14, 2011 @ 5:28 pm
It's interesting, this. No mainland European court, as far as I know, would employ corpus linguistics to find out the meaning of a word. The reason is that the methodology employed by courts working in the civil law tradition permits a wider use of sources, both academic works and so-called preparatory works (perhaps the French term 'travaux préparatoires' is more used in English). In the case of Sweden, for instance, the latter are made up of governmental committee reports, the government bill to Parliament, the report by the committee in parliament, and theoretically (though now extremely rarely) by the protocol of the parliamentary debate. A bill sent by the government to Parliament comes with detailed comments on the proposed legislative changes, comments which the government knows will be used by the courts when interpreting the law in question. If the literature consulted on a certain law does not give any clue to the meaning of, say, the term "procedure", then maybe there are comments on other pieces of legislation that can shed some light. I cannot spontaneously recall reading a Swedish Supreme Court or Supreme Administrative Court case where a dictionary or text corpus (of which there are several good ones for Swedish) has had to be consulted. The nearest thing I can think of is when different language versions (English, German, French, Danish, Dutch) of EU legislation are consulted to find out how the Swedish version is to be interpreted.
Michael Watts said,
December 15, 2011 @ 2:56 am
I agree with eye5600 — it doesn't seem to make sense to
a) decide we need to understand the meaning of "procedure" in the phrase "abortion procedure"
b) examine a corpus to find that nearly all instances of the word "procedure" in the immediate context of abortion can be interpreted in a narrow, medical sense, and then
c) conclude that the phrase "abortion procedure" can only be interpreted in the narrow, medical sense
This chain of reasoning utterly fails to account for the possibility that whenever it is uttered in the context of abortion, "procedure" can be reasonably interpreted to refer to the broad sense ("method") of the word, but that virtually all "method"s for abortion discussed in the corpus happen to be medical procedures. (A possibility which I find entirely plausible, for reasons ranging from the perceived greater safety and acceptability of clinics to the lack of a need to overcome strangers' natural reluctance to repeatedly punch a pregnant woman.)
Kenny said,
December 15, 2011 @ 3:36 am
@Michael Watts
Judges typically don't decide that terms can never have certain definitions. They look for what the intended meaning of terms used in a law are likely to be. I don't think that a reasonable person could look at the corpus evidence and say that the term "abortion procedure", used in a law with multiple references to doctors and medicine, means something besides a medical procedure. A reasonable person could, however, look at the word "procedure" alone and conclude that it can and probably does include things that are not medical procedures, even in the context of a law about abortion.
Justice Lee said just that: "My quarrel is just that this conclusion [that "procedure" means "medical procedure"] follows from our understanding of the word 'procedure' in the context of the term 'abortion,' not from the fact that the statute most often refers to “medical procedures” and not just 'procedures.'"
Michael Watts said,
December 15, 2011 @ 4:02 am
I'm actually ok with the idea that since the law is full of "references to doctors and medicine", its use of the word "procedure" might most likely refer to medical procedures. But as I understand it, that's the opposite of what Judge Lee is saying: he says that we should pay less attention to the general contents of the statute, and more attention to the particular words immediately surrounding the instance of "procedure" that we want to interpret (and most specifically, the surrounding word "abortion").
Here's the thing… I don't think a corpus is a useful tool for answering the question he asks (which I'm interpreting as "what does 'procedure' mean in this immediate context: broad or narrow sense?"). The problem is that even if *every* instance of the token 'procedure' in an abortion-related context within the corpus was both intended and understood in the broad sense, then, as long as most discussed abortion procedures (broad sense) are also narrow-sense medical procedures, which seems likely, the corpus can't reveal that the broad sense was intended. All it can show is that when the token 'procedure' occurs in an abortion context, it's almost always compatible with the narrow-sense meaning. But the fact that that meaning is compatible doesn't imply that it was meant, or that we would be restricted to it in other circumstances.
(In fact, where one sense of a word is a strict subset of another sense, I'd think it would be extremely difficult to show, for any utterance, that the broad sense couldn't have been meant.)
Michael Watts said,
December 15, 2011 @ 4:23 am
Furthermore, while Judge Lee conducts corpus research by looking for cooccurrences of 'procedure' and 'abortion', 'procedure' does not cooccur with 'abortion' in the defining clause which appears to be at issue ("includes any and all procedures undertaken to kill a live unborn child and includes all procedures undertaken to produce a miscarriage"). I have to admit it would be unwise to include the word 'abortion' in a clause purporting to define that word, but still… it's being defined by reference to unadorned 'procedures'. And as you point out, this instance of 'procedure' is modified by the very broadening phrase "any and all", which Judge Lee rejects.
So, as I see it, the main opinion's argument boils down to "procedure means medical procedure", whereas Lee's boils down to "procedure means medical procedure, because what's being defined is the word 'abortion'". Question: if the drafters of the law had meant to define an abortion in terms of any broad-sense procedure, how could they have done that to satisfy Lee (bearing in mind that specifying "any and all procedures" is not sufficient)?
Finally, I apologize for this comment being more relevant to the comment thread than to the post. Judge me by the one preceding this.
Kenny said,
December 16, 2011 @ 12:58 am
@Michael Watts
I reread Lee's opinion and realized I'd understood it backwards. You are correct: Lee did basically say that proof for the meaning of the term "abortion procedure" was proof for the meaning of procedure. Nevertheless, Lee spend a lot of space explaining that "procedure" in the abortion statute does not on its face exclude the braod reading that it includes nonmedical procedures, so how did a judge who seems to have a good linguistic head on his shoulders take such a myopic view?
From a linguistics perspective, it seems troubling that someone could be versed in the rhetoric and then so misuse the methodology. Also troubling for the prospects of linguistic evidence in court is that Lee didn't use it.He decided the case on the interplay of the homicide statute and the abortion statute (the court decided the homicide statute had no meaning if "procedure" was broad). There was no linguistic argument for why a deficiency in the homicide statute should have bearing on the meaning of the abortion statute. Even if we accept the notion that "procedure" has to mean "medical procedure", the homicide statute remains defective because it would say that someone who performed unwanted abortions using medical means could not be tried for homicide but a girl who sought an abortion through non-medical means can.
Unfortunately, I think the answer to your question is that no likely wording of the abortion statute would have been attributed with the broad meaning. The only kind of wording that would have made the problem with the homicide statute explicit would have been "any and all procedures, whether usually considered medical or not". There's no deniability there, but even the American penchant for redundant and wordy legal language would be hard pressed to get someone to write that.
Ben Hemmens said,
December 16, 2011 @ 4:36 pm
No mainland European court, as far as I know, would employ corpus linguistics to find out the meaning of a word.
I second that from Austria. Over the years, my appreciation of code-civil legislation has grown. The imperative to avoid ambiguity in drafting of statutes probably comes from the competence of the constitutional court (Verfassungsgerichtshof) to declare laws or passages of laws invalid. Which it regularly does.
Practical avoidance of ambiguity is assured by the fact that drafters of statutes are generally senior permanent officials, each of whom may oversee a particular subject area for a period that outlasts several ministers.
Issues of interpretation of existing statutes like this are mostly handled by the Oberster Gerichtshof, which would search the chain of reasoning and definitions within the act in question and then back up the hierarchy of statutes to the constitution. If it finds a definite gap, it doesn't start reading the tea leaves (and probably also not the corpora) as to what the legislators might have meant, but refers the law to the constitutional court to be tested. If there is a real gap in definition, the law itself gets thrown out along with the individual judgement.
An important figure at the beginning of the republican Austrian constitution and the constitutional court was Hans Kelsen, who spent the later part of his career at Berkeley. Anyone interested in this tradition is probably well advised to look him up.
Alex said,
December 17, 2011 @ 4:29 am
I agree with Michael Watts; I was not persuaded by the argumentation. To me it seems clear that any method intended to produce an abortion is an "abortion procedure" (whether it is successful or not, logical or downright boneheaded).
I am also a bit skeptical about the use of corpora in legal cases because it restricts our understanding of a word to those uses which happen to be represented in the (I assume, written) corpora, thus only representing a small subset of the population. Of course, corpora allow more people to contribute to and ground a definition than before (when it was the judge's intuition), which is great, but I think many people will overreach and assume corpora to be more complete and encompassing than they are. In this case, a little abstract reasoning would have done a lot of good rather than simply taking the statistical facts at face value. Furthermore, no doubt that if corpora become a thing, as it were, carefully groomed "corpora" to suit every lawyer's particular need will pop up.
Some women engage in self-administered herbal or non-medical methods of abortion (certainly not safe, but unfortunately not unheard of), and others use the services of people who, well, are not medical doctors operating in a respectable clinic. It is not ideal that women choose to end their pregnancies in these ways, but from a descriptive point of view, these are abortion procedures. Ineffective and sometimes dangerous or lethal, as with the punching, but still. A woman's right to choose must be about basic control of her reproductive health and body rather than access to a limited number of medical operations.