Vagueness vs. Ambiguity

Prof. Solum discusses vagueness vs. ambiguity, and the relationship between this distinction and the interpretation/construction decision, here.

I would characterize the difference in terms of linguistic architecture, as follows:

Ambiguity is what occurs when the sound or textual output matches more than one possible linguistic expression, where a linguistic expression includes a sound output, a meaning output, and the procedure that generates the pair of outputs. Ambiguity is the result of the fact that sounds and meanings do not perfectly correspond to one another.

Vagueness is what occurs when the meaning output of a linguistic expression, or a part thereof, corresponds to a concept (in our case, a legal concept) without clear boundaries. Vagueness is the result of the fact that our conceptual system is incomplete, so that it is not always known whether a certain object, for example, falls in a particular category. It is a necessary result in light of the fact that our conceptual system is discrete, while the world that it represents is often continuous.

The role of the interpreter/constructor in the case of an ambiguity is to select the appropriate meaning output in light of the sound output, using evidence such as the context in which the expression was generated, the likelihood that the meaning output was the one intended by the person or group articulating the linguistic expression, and the naturalness of the sound-meaning correspondence (a particular meaning can be possible but improbable, for example, if it presents great difficulty in linguistic processing).

The role of the interpreter/constructor in the case of vagueness is to modify the conceptual legal scheme so that it optimally covers the case at hand.

[edited for typos 1/17/2009]

Eat 'em up

Not law-related but diplomacy-related. Reuters reports that France's Foreign Minister Bernard Kouchner, speaking in English, caused a bit of a row when he was understood to say to an Israeli interviewer, "I honestly don't believe that it will give any immunity to Iran ... because you will eat them before."

He later explained that he was not contemplating Israeli consumption of Iran, only an illegal aerial bombing. Apparently he had meant to say "hit", but what came out, because French phonology has a highly ranked constraint against initial [h], and lacks the lax high front unrounded vowel that English uses in the word "hit", sounded like "eat."

The ministry clarified that Kouchner "regrets the unfortunate misunderstanding this phonetic confusion has caused." The confusion was of course phonological, not phonetic.

In a possible world consistent with consistency in behavior from major propaganda agencies, we will soon witness a propaganda offensive from the MEMRI hole, saying France calls for Iran to be wiped off the map.

Status update

I'm in a little over my head at law school, but I am making forward progress on my Second Amendment note. So I'll post a little bit about that.

It looks like my focus will be on one particular controversy in the DC v. Heller opinion, concerning the phrase "the right of the people to keep and bear arms". Namely, the issue of whether the fact that it's not "to keep and to bear arms" makes a meaningful difference. This is in the context of a dispute over whether the clause contemplates a single right or two.

Justice Stevens argues that the absence of "to" suggests a single right. Scalia dismisses this position as bizarre, arguing that nobody has proposed a rule that the absence of "to" in such a situation makes a difference to the meaning, and using a sentence with the noun "promise" to show that the presence or absence of "to" in such contexts makes no difference.

My paper, as I currently conceive it, takes the position that Scalia's argumentation is poor, and that the absence of "to" is in fact probative of a difference in meaning. I argue that it's not the semantics of the word "to" that makes a difference, but the syntactic structure required to accommodate the "to". And I explore the different semantics of "promise" and "right", suggesting an explanation for why the different syntactic structures end up with the same semantics in the case of "promise" but different ones in the case of "right". (I think it's the quantificational force of the modality: universal in the case of "promise", existential in the case of "right".)

[edited for typos 1/17/2009]

Update

I'm back at Case Law School for one more year. It looks like it's going to be a rough semester: I'm taking three substantive classes, the health law clinic, a bar review course and a supervised research project. The latter will be a linguistic analysis of the Second Amendment. I'm also searching for a job after graduation and taking the Multistate Professional Responsibility Exam in November. In between, I'm staying involved in the National Lawyers Guild, and working with a couple of friends to form a nonprofit to educate and advocate for greater accessibility to higher education.

I'm still hoping to post here, but it will probably not be too frequently.

A paper I wish I'd written

Shai Cohen brought this law review article to my attention. It's got so many things you want to see in a law article: linguistics, employment discrimination, eight-way ambiguities, upbraiding of judges for being insufficiently attentive to the nuances of language... well, it would have had upbraiding had I written it. Evidently the article's author, Jill C. Anderson, is less petty and more generous toward judges than I.

If I may brag, in my grad school days I wrote a paper in which I argued that the domain of events is, like the domain of entities is sometimes taken to be, divided along an individual/group axis as well as a singular/plural axis, with the upshot that sentences like Adin hit three boys five times is something like 13-ways ambiguous [aside to semanticists: I used a Landmanian analysis and took the theta-role functions to apply to group events, identifying individual events as the phases found in the literature on pluractionality]. I also managed to get an audience of linguists to see each of the thirteen or so distinct meanings. So: my contrived sentence is more ways ambiguous than Congress's carefully crafted one. Nyah!

Is Prop 8 language negative? Maybe not, but it's downward-entailing

The media is reporting on a California judge's dismissal of a lawsuit brought by an anti-gay group to change the wording in the summary of Proposition 8, a proposal to amend the state constitution to ban gay marriage. Roger Shuy covers it in the Language Log, here.

According to news reports, the lawsuit alleged that the summary, which reads "Eliminates the Right of Same-Sex Couples to Marry," "was argumentative, misleading and prejudicial," because eliminates is "a negative, active transitive word - grammar that had rarely, if ever, been used in a state ballot title." The plaintiffs preferred the old title, "Limit on Marriage," which was changed by Attorney General Jerry Brown after the Supreme Court held in May that same-sex couples could marry.

California Superior Court Judge Timothy M. Frawley's dismissed the challenge, saying that "[p]etitioner has failed to explain why the term 'eliminates' is inherently argumentative, while the term 'limit' is not."

One can argue about the inherent argumentativity, misleadingness or prejudiciality of the choices of language. But in at least one important sense, the anti-gay crowd is right that eliminate is more negative than limit. That's because eliminate, unlike limit, is downward-entailing in its complement, which, as shown in Bill Ladusaw's brilliant dissertation, correlates with the licensing of negative polarity items.

Here's an explanation for the uninitiated. It was noticed that certain words, like ever, are licensed in negative contexts, in the sense that they can exist in negative contexts but not the corresponding affirmative contexts. For example:

Nobody ever told me there'd be days like these
*Somebody ever told me there'd be days like these
Yochanan didn't ever tell me there'd be days like these
*Yochanan ever told me there'd be days like these
(Following convention, the asterisk indicates ungrammaticality.)

This led words like ever to become known as "negative polarity items" or NPIs for short. But many other words license NPIs without being logical negators.

Yeshaayahu denied that he ever smoked pot.
*Yeshaayahu affirmed that he ever smoked pot.

Few people ever walked on the moon.
*Many people ever walked on the moon.

Shmuel supported gay marriage before he ever found out his son was gay.
*Shmuel supported gay marriage after he ever found out his son was gay.

Everyone who's ever lived in California knows it's a crazy place.
*Someone who's ever lived in California knows it's a crazy place.

Some of these licensing words seem negative and can be rephrased using negatives, or else broken down into component parts that include negatives. Deny can be recast as say that not, and few can be restated as not many. But this is not true of all NPI licensers. Before is not the same as not after (because neither covers contemporaneous events) and every doesn't mean not some. Moreover, in neither of these pairs is one member of the pair more intuitively negative than the other.

What Ladusaw showed was that the concept of downward entailingness was a better predictor of the ability to license NPIs than the concept of negativity. A context is downward-entailing if and only if in that context, replacing a set with its proper subset preserves the truth of the sentence. For example:

Yisroel didn't eat fruit -->
Yisroel didn't eat bananas

Bananas is a proper subset of fruit. These sentences exemplify the fact that the scope of negation is a downward-entailing context. If the first sentence is true, the second one is necessarily true as well. The following sentences show that this property is shared by deny, few, before and the first argument of every.

Dafna denied eating fruit -->
Dafna denied eating bananas
(on one reading of the sentence)

Few people ate fruit -->
Few people ate bananas

Ofer says a bracha before eating fruit -->
Ofer says a bracha before eating bananas

Everyone who eats fruit is gay -->
Everyone who eats bananas is gay

These entailments do not hold of the counterparts of these words.

John ate fruit --/-->
John ate bananas

Dafna affirmed eating fruit --/-->
Dafna affirmed eating bananas
(on any reading of the sentence)

Many people ate fruit --/-->
Many people ate bananas

Ofer says a bracha after eating fruit --/-->
Ofer says a bracha after eating bananas

Someone who eats fruit is gay --/-->
Someone who eats bananas is gay

Now, back to the proposition. Eliminate is a downward-entailing word, and licenses ever.

Ryvka eliminated fruit from her diet so as not to appear to be homosexual -->
Ryvka eliminated bananas from her diet so as not to appear to be homosexual

Ryvka eliminated the situations in which she ever had to feel guilty.

Limited is not downward entailing, and doesn't license ever.

Ora limited the fruit in her diet so as to appear less homosexual --/-->
Ora limited the bananas in her diet so as to appear less homosexual

*Ora limited the situations in which she ever had to feel guilty.

These judgments are somewhat subtle, but I think they're correct; or if not correct, at least onto something.

Solum on defining "strict construction" and "judicial activism"

Prof. Solum's Legal Theory Lexicon here attempts to debunk, rather than define, the terms "strict construction" and "judicial activism", which he does by showing that straightforward attempts to define the terms fail for one reason or another.

He concludes that "'strict construction' and 'judicial activism' are simply not very useful as theory terms for academic constitutional lawyers." He fails, however, to explain why their use is so widespread. The obvious answer seems to be that they are useful terms of propaganda. It's much easier to rail against judicial activists and offer strict constructionist judges as an alternative if you don't have to state your objections or proposed alternatives with any clarity or specificity.