Log in

No account? Create an account
24 September 2005 @ 21:12
Time served  

“Only a judge can sentence you,” said Milo, who remembered reading that in one of his schoolbooks.

“Good point,” replied the policeman, taking off his cap and putting on a long black robe. “I am also the judge. Now would you like a long or a short sentence?”

“A short one, if you please,” said Milo.

“Good,” said the judge, rapping his gavel three times. “I always have trouble remembering the long ones. How about ‘I am.’? That's the shortest sentence I know.”

Everyone agreed that it was a very fair sentence, and the judge continued: “There will also be a small additional penalty of six million years in prison. Case closed,” he pronounced, rapping his gavel again.

—Norton Juster, The Phantom Tollbooth

The search for the shortest published sentence of the year was declared over before I was even aware that it had begun. In the aforelinked Language Log post, Geoff Pullum awards the prize to the final sentence of an article by David Grann in the Septembet 19 issue of The New Yorker. It goes like this:


Like most sentences, it requires some context in order to be fully understood. Z, in this case, happens to be the name of a lost Amazonian city whose existence was hypothesized by the explorer Percy Harrison Fawcett, and the penultimate sentence of the article runs as follows:

For a moment, I could see this vanished world as if it were right in front of me.

Pullum informs us that "Others equally short may be published, but there won't be any shorter, because we've hit the lower bound"—viz., a sentence of one letter plus an obligatory terminating punctuation mark. Michael Kaplan, at Sorting It All Out, tries to push that lower bound even lower, proposing a sentence with no letters at all. (Thanks to Mark Liberman at Language Log for supplying the link to Kaplan's post.) Again, context is crucial:

The next sentence intentionally left blank. .

I'm not sure a blank sentence is a sentence at all, but I, too, would like to challenge Pullum's assertion that "Z." represents the lower bound. Unlike Kaplan, I'll concede that a one-letter sentence is indeed the orthographically shortest possible sentence. My question is, why should we make orthography the deciding factor? We're linguists, aren't we? What we are interested in is language as a phenomenon of human cognition, and it is spoken (and signed) language that demonstrates this phenomenon in its most natural state; written language is encased in a distracting layer of explicitly constructed and agreed-upon conventions. To define the search for the shortest sentence orthographically is to construct an arbitrary challenge that focuses on a merely incidental property of the linguistic expressions under consideration.

So let's have a look at Grann's sentence in terms of sounds, not letters. There are two ways of pronouncing "Z.": a Commonwealth one and an American one. If we use the Commonwealth pronunciation, the sentence is three phonemes long:


If we use the American pronunciation, it's only two phonemes long:


The New Yorker is an American magazine, so if the sentence were about the letter Z, then we should assume the two-phoneme pronunciation. But it's not; it's about the City of Z, which was named by Col. Fawcett, who was an Englishman. So the three-phoneme pronunciation is the one we want in this context.

This means that it should be possible to come up with a sentence that is phonologically only a third as long as Grann's. A one-phoneme sentence would probably have to consist of a vowel (unless you'd like to call ruminative /m̩/ a sentence), and among English vowels, only the long ones are capable of standing alone. So here are some candidates:


The English long vowels tend to be diphthongal, so you might prefer to transcribe these as [ow] (or [əʊ], as in [həʊzə]), [uw], and [ej] (unless you're Scottish). Phonetically, then, they consist of more than just 'one sound' (though it's tricky to pin down exactly what 'one sound' is when one starts looking at phonetic details). For that matter, the sentence "Ooh!" is quite likely to have a longer phonetic duration than the sentence "Z.". But since we're looking for the shortest sentence, not the shortest utterance, I think we need to consider sounds at a somewhat abstract level: phonology, not phonetics; phonemes, not milliseconds.

Now, if we want to break the tie among the one-phoneme sentences, perhaps we could consider bringing orthography back into the picture, and look for a one-letter, one-phoneme sentence. To keep things interesting, we might want to set aside those cases in which a letter is used to refer to itself (e.g., the sentence "A." as a response to the question "What's the first letter of the Roman alphabet?"). So here is my proposal for (one of) the phonologically and orthographically shortest sentences:


That, of course, is the answer to the question "What is the base of the natural logarithm?" Here, e does not refer to the letter e, but to the number e (2.718281828459...). Since e, being irrational, has an infinite number of decimal places, this is a maximally brief sentence that refers to maximally long number, which in my opinion gives it a bit of extra charm. And because the symbol that represents the number e is specifically a lowercase letter e, this is one of the few contexts in which it is at least arguably appropriate orthographically to begin a sentence with a lowercase letter—and that means that "e." is a shorter sentence in yet another sense, for on the page it only comes up to the waist of "Z.".

Q. Pheevrq_pheevr on 25. September, 2005 09:25 (UTC)
Re: michkap@microsoft.com

Well, a blank sentence is pretty hard to beat. I had thought to respond to Geoff Pullum's post earlier, then forgot about it until I saw yours. Since I had been concentrating on short non-blank sentences, I didn't really have anything to say about ".", but I thought I should at least mention it. I think the best I can do is refer you to the following article:

Fiengo, Robert, and Howard Lasnik. 1972. On nonrecoverable deletion in syntax. Linguistic Inquiry 3: 528.

I hope that Fiengo, Lasnik, and the folks at LI will forgive me for quoting the full text of the article below; I can hardly help it.

Merlemerle_ on 25. September, 2005 13:42 (UTC)
If you follow Government-Binding (the syntactial approach my prof took), aren't there trace elements that really only exist as the byproduct of a transformation?

Such a trace would, in effect, be a lexically empty piece of prose.

I'm not sure that an entire sentence could be a trace, though. Usually they are references and the like. (but my GB training is about fifteen years rusty by now)
Q. Pheevrq_pheevr on 25. September, 2005 14:08 (UTC)

A trace can only be created by movement, and so whatever you moved will still be around somewhere higher up in the tree. You can move things from an embedded clause to a higher clause, but you can't move them all the way out of the whole sentence you're dealing with—and even if you could, then the trace you leave behind wouldn't be happy about that, because it needs to be c-commanded by something coindexed with it.

There are other silent linguistic elements, though. I don't think it's generally possible to make an entire sentence out of them, but perhaps we could imagine Michael Kaplan's sentence as consisting of a null pronoun plus an elided verb phrase....

Merlemerle_ on 25. September, 2005 17:52 (UTC)
I was thinking that a trace might be moved into a subclause position which would (effectively) be like a sentence -- in which case one could claim it as a sentence of its own.

It did seem unlikely. Aaah... "c-commanded". It's starting to come back to me. It has been a while since I have done formal linguistics.

The problem is, can you actually identify the sentence after "this page intentionally left blank" as a null sentence? Wouldn't it (by semantics, not syntax) occur on the very next page? Which is probably not blank (since those pages are usually odd-pages left as filler so chapters can start on the right-hand page).