MUST and SHOULD figured out

Those of us trying to figure out the meaning of deontic modals, especially the distinction between weak and strong necessity, should just pack in and go on vacation. There’s an official RFC1 that settles the issue.

Scott Bradner, all around internet wizard at Harvard, wrote RFC 2119 “Key words for use in RFCs to Indicate Requirement Levels” in 1997. It has this definitive pronouncement on the difference between strong and weak necessity expressions:

  • MUST: This word, or the terms “REQUIRED” or “SHALL”, mean that the definition is an absolute requirement of the specification.
  • SHOULD: This word, or the adjective “RECOMMENDED”, mean that there may exist valid reasons in particular circumstances to ignore a particular item, but the full implications must be understood and carefully weighed before choosing a different course.

Exercise for the reader: do any current semantic proposals mesh with this official pronouncement on how must and should differ?


  1. “A Request for Comments (RFC) is a publication of the Internet Engineering Task Force (IETF) and the Internet Society, the principal technical development and standards-setting bodies for the Internet.” (Wikipedia

The art and craft of semantics

The Art and Craft of Semantics is a Festschrift for Irene Heim that students, colleagues, and friends presented to her yesterday, October 30, 2014, on the occasion of her 60th birthday. I have two co-authored contributions in the collection:

The definiteness paper is a brief report on a rumored account that’s been floating around for a while. I hope there’ll be follow-up work. The modal comparison paper is a very short plea for Irene’s help with some puzzles that are simply too hard for Angelika and me.

Happy birthday, Irene!

“Only”

In today’s Boston Globe baseball preview section: “Jake Stahl (1912), Ed Barrow (1918), Terry Francona (2004), and John Farrell (2013) are the only Red Sox managers to win the World Series in their first season.”

Hmm. Yeah, no. The Red Sox have had 44 managers in their history, so when 4 of them win the World Series in their first season, does that really merit an “only”? I mean, what is the expected frequency of a manager winning a world series in their first season? Almost certainly a lot less than 4 in 44.

But the more astonishing fact is that of the 8 World Series wins the Red Sox have had, half of them were by rookie managers. (And one of those, Francona, won another one. So, 5 trophies belong to managers who won one in their first season.) That’s definitely not an “only”.

 

Up-goer five semantics

A couple of months ago, Randall Munroe’s xkcd web comic explained the design of the Saturn V rocket using only the thousand most common words of English: “the Up Goer Five explained using only the ten hundred words people use the most often”.

Explaining hard things in simple language has now become an internet meme. Just this morning, I found Walton Jones explaining his lab’s work on the genetics and neuroscience of olfaction in Drosophila: “We are interested in how little animals with six legs smell things”. There is a tumblr blog with many of these summaries.

The Up-Goer Five Text Editor makes it easy to experiment with writing down your research in the ten hundred most used words. Here’s an attempt at an up-goer five abstract for my upcoming colloquium talk at McGill (“Hedging your ifs and vice versa”, joint work with Thony Gillies):

How does the word “if” help things we say mean what they mean? It can work together with other words like “maybe” and “probably” to make things we say less strong. But how does it do that?

Many people have tried to find out how this works, but we will show that they face a big problem when one looks at people talking to each other and pointing to things the other said.

Can we do better?

There are some obstacles for a linguist. You often need to mention linguistic expressions that you work on. I was lucky that if, maybe, and probably are licit. On the other hand, “sentence” is not allowed.

Related: George Boolos’ classic exploit “Gödel’s Second Incompleteness Theorem Explained in Words of One Syllable”.

The politburo

A subset of the MIT syntax/semantics politburo, some of us even smiling:

(Taken during Rick Nouwen’s colloquium when Irene Heim was introducing the speaker, which is why she’s not in the row with us.) (Thanks to Danny Fox for the pointer at the picture and thanks to mitcho for taking the picture!) (Can you put three parentheticals in brackets in a row?) (Sure, you can.)

Modal Semantics in the News

Wired reports on the brouhaha surrounding the security of Dropbox, a geek favorite (I am including myself). Dropbox used to say that

Dropbox employees aren’t able to access user files, and when troubleshooting an account, they only have access to file metadata (filenames, file sizes, etc. not the file contents).

Now they say

Dropbox employees are prohibited from viewing the content of files you store in your Dropbox account, and are only permitted to view file metadata (e.g., file names and locations).

Any normal natural language user would interpret these two statements as distinct, I would think. (not) able to is a modal typically understood as talking about what is practically possible, while prohibited is all about things that are practically possible but circumscribed by rules and regulations.

Dropbox takes the heroic stance that the statements are equivalent:

In our help article we stated “Dropbox employees aren’t able to access user files.” That means that we prevent such access via access controls on our backend as well as strict policy prohibitions.

As Kratzerians, we’re used to modals being heavily context-dependent, but I think Dropbox has almost no leg to stand on. Does anyone care to defend Dropbox’s semantic analysis of able to?

Many worlds

[This is a slightly edited and updated repost of an old blog post of mine. Since I had to scrub my site after a spam infestation, all my old posts only live on my hard drives and are not on the net. So, as the need or occasion arises, I will repost old posts and update them where appropriate. This one is relevant since I’ll be introducing possible worlds in my advanced semantics class next week.]

Ever since reading Max Tegmark’s article in Scientific American on the many worlds version of quantum mechanics, which mentioned David Lewis’ plurality of worlds, I have been wanting to learn more.

In high school, I was very good at chemistry but avoided physics at all costs, for no apparent reason, so I have a large deficit in that area. A good plan might be to read the textbook Physics for Future Presidents by Richard Muller at UC Berkeley. But that will have to wait for some vacation in the future.

Instead, when I saw Colin Bruce’s new book on “the many worlds of quantum”, I decided to make it my bedtime reading for this week [NB: that was written in May 2003 and I have made little to no progress since then!].

There is in fact a brief mention in the book of David Lewis, occasioned by his paper “How many Lives Has Schrödinger’s Cat?”. So, I actually decided to first read Lewis’ paper and the commentary by David Papineau later in the same issue of the Australasian Journal of Philosophy.

The basic issue seems to be that if there truly are many worlds branching off from each other all the time (something also called the “no-collapse” theory), this has great impact on our thinking about decisions, ethics, and our philosophy of life in general. Have you ever thought “I could kill myself” when you realize you made a carelessly wrong decision a while back? Well, even if you make the right decision “here in this world”, there will be another world (in fact many such worlds) where you made the other — wrong — decision and suffer the consequences. So, what difference does it really make if you make the right decision “here”? As long as the plurality of worlds is a harmless abstraction introduced to make it easier to think about the semantics of modal expressions in natural language, for example, such considerations may not be very disturbing. But if all those worlds are really real, we have a lot of thinking to do. Of course, some philosophers and some physicists have been thinking about this for a while, but it’s all new stuff to me.

Lewis points out a truly disturbing thought: in many worlds, we eventually die of course and maybe that’s fine since we don’t have any further experiences in those worlds (we may of course worry about the people and the world we leave behind in those worlds). But there is always a (however small) chance that we do not die but barely continue to live, albeit maimed and disabled. Since those are the only worlds where we continue to have experiences, we all can expect to live horrible lives at some point. Lewis writes:

“What you should predominantly expect, if the no-collapse hypothesis is true, is cumulative deterioration that stops just short of death. … How many lives has Schrödinger’s cat? If there are no collapses, life everlasting. But soon, life is not at all worth living. That, and not the risk of sudden death, is the real reason to pity Schrödinger’s cat.”

David Papineau’s commentary in the AJP tries to soothe these worries, but clearly there is much to figure out. Colin Bruce reports that “although [Lewis’] words are light, I am told by those who worked with him that he was terrified by this hypothesis [Fn: Sebastian Sequoia-Jones, pc to Bruce, March 2004]. By a cruel coincidence, he died suddenly and unexpectedly from diabetes within weeks of giving that lecture — at least in our version of reality. His paper is about to be published posthumously as I write. He must have died a badly frightened man, and the psychological impact on his colleagues was considerable.”

I am now halfway through Bruce’s book and somewhat overmatched by the strangeness of the quantum world, but I’ll continue working through it and cannot see how one can’t be fascinated by this area of inquiry. I wish I had paid more attention in high school. The fact that I did in some other worlds doesn’t help me much “here”.

Some other pointers: the entry in the Stanford Encyclopedia of Philosophy on “Many-Worlds Interpretation of Quantum Mechanics”, the Everett FAQ, Max Tegmark’s “Welcome to my crazy universe”, and a recent Fresh Air interview with physicist Brian Greene, who just wrote a book called The Hidden Reality: Parallel Universes and the Deep Laws of the Cosmos.

Present Indicative Counterfactuals

Commenting on my “If you’re Lance Armstrong …” post, Thony Gillies points out that this Hubie Brown-type conditional is a special case of the sportscasterese present indicative counterfactual conditional. As in (topical example from today’s Red Sox game, after Manny Ramirez made a spectacular catch running into the Green Monster in the 9th inning, protecting a slim Sox lead):

If Ramirez doesn’t catch that, it’s a double and the tying run is in scoring position.”

Apropos my little working paper from 1997 on The Presupposition of Subjunctive Conditionals, I had a brief email exchange with Larry Horn about this kind of conditional. Here are some passages from my main message to Larry and his responses:

At 4:06 PM -0400 5/22/98, Kai von Fintel wrote:

Dear Larry —

thanks for your message. As an absolutely incorrigible sports addict, I have been struck by this construction many times. [It first struck me at the same time when I realized that the NFL was using instant replay to ascertain the truth of such counterfactuals as “If Sanders hadn’t pushed Rice, he would have come down inbounds”.]

Actually, upon further review I do not agree with your assessment:

[Larry had tried to convince me that “The existence of this construction clearly forces the severance of grammatical mood from the semantico-pragmatics of counterfactuality. There’s no presupposition, conventional implicature, whatever, in these cases of indicative “if p (then) q” that p is epistemically possible.”]

1. First possible response (not one I would be attracted to): this construction is ungrammatical, these guys are confused. Hence nothing follows about grammar.

Serious question: this seems much more localized to sportscasters than for example the reporter’s simple present, which also surfaces in stage directions and other circumstances (which has led people to actually propose funky semantic analyses of this use of the simple present). How widespread is this counterfactual indicative?

[Larry responded: “That is an interesting question. I’ll raise it on ADS-L, where we had an earlier discussion of the PICFC (as I’ll abbreviate the construction), since this would be an interesting sort of isogloss. All other cases of sportscasterese (not just the “shazam historical present” that Erich Woisetschlaeger and John Goldsmith discussed a decade or two back, but also e.g. the extension of certain currency descriptions for other uses: “he’s hitting a buck fifty”; “there’s a buck ten left in the quarter”; “a cornerback weighing a buck seventy-five taking down a tank like Ironhead!”) or sportsplayerese (“my bad”) are attested elsewhere, but I haven’t come across PICFCs outside of SportsWorld.”]

2. Second possible response (somewhat more believable, but still not great): these are run-of-the-mill indicative conditionals with a presumption of epistemic possibility. They would receive an analysis along the lines of the historical present tense, whatever that may be. Something like: “present” with respect to a temporarily assumed/imagined speech time (which is actually in the past of the real speech time), “epistemically possible” with respect to a temporarily assumed/imagined epistemic state (which is actually one that the speaker knows s/he’s not in (anymore)). This is the kind of move that I report in my subjunctive paper as the move favored by Portner for why some “subjunctives” do not seem straightforwardly “counterfactual”, according to him they are counterfactual but only with respect to a temporarily assumed point of view.

3. Third possible response (perhaps the one I would spend most energy on if I were to expand my paper to include discussion of this construction): I say in my paper that indicatives do not actually carry any direct presupposition/implicature. Subjunctives do. And it is the choice of an indicative over a subjunctive that may usually be interpreted as indicating epistemic possibility. In other words, subjunctives are marked and only good for uses where (at least some) worlds outside the set of epistemically possible worlds are quantified over. Indicatives are unmarked and thus in principle usable in many more circumstances; but of course usual considerations will limit their use.

There’s possibly quite a lot more to say about this construction. For example, one might wonder whether it could be used in the Anderson-type argument:

“If he had taken arsenic, he would show exactly these symptoms.”

This is impossible in a normal indicative:

“??If he took arsenic, he shows exactly these symptoms.”

But now imagine a sportscaster who hasn’t paid much attention to what is happening peripherally to the game. Rodman totally flips out and throws the ball at some spectators. Now, one thing that would explain his behavior is that he was heckled. Can our sportscaster say:

“If Rodman is heckled by the guy, he does exactly this. So, perhaps he was heckled.”

[Larry replied: “Funny you should bring these up. On the example-laden scrap paper I was typing my message from, I had an observation to the effect that Anderson-type non-CF contexts are impossible with PICFCs, but then I started to doubt the conclusion, although my Andersonesque examples weren’t as convincing as yours. Another context would be the SportsCenter replay: Dan Patrick says Rodman should be suspended for his antics, but Kenny Mayne points out that they don’t have complete footage of what preceded the incident—after all, [your PICFC here].”]

Anyway, thanks again for your message. If I expand the paper for publication in a journal, I will try to take this construction into account.

which I never did.

I double-checked the archives of the ADS-L list, but no new insights into this constructions seem to have turned up, except the tip that David Carkeet, author of various novels with a linguist hero, mentioned the construction in a New York Times “On Language” column on July 22, 2000.