December 13, 2007

Antarctic WTF

Said Mr Brian Lee of Telford, Shropshire, speaking to reporters after his aborted Antarctic cruise ended in a night in a drifting lifeboat a couple of weeks ago:

I am a bit disappointed we didn't get to finish our trip, but how many people have been on a ship that's hit an iceberg in the middle of the night, sunk, and lived to tell the tale?

Language Log knows exactly how many. The answer can be deduced from the syntax and semantics of Mr Lee's rhetorical question.

It all hangs on the syntax and concomitant semantics of the coordination of verb phrases in Mr Lee's utterance. Let me explain.

One possible syntactic analysis would be that the coordination begins after the word have. That would mean that the verbs that have head function in the three coordinates are been, sunk, and lived (all of them are in the past-participial form that is required for verbs governed by the auxiliary verb have marking the perfect tense). Under that analysis, the relevant sentence has this structure:

How many people have
[ been on a ship that's hit an iceberg in the middle of the night ],
[ sunk ],
[ and [ lived to tell the tale ] ]?

On that understanding, we are talking about people who sank. People who sink in iceberg-strewn Antarctic waters do not tell tales. So the number of people meeting the three specified conditions is zero.

But there is another syntactically permissible analysis. It has a coordination of three past-participial VPs, but the form of the auxiliary have that precedes them is has, in its reduced form "’s" on the end of the orthgraphic word that’s. The verb form hit is the head of the first of the three coordinates. (The verb hit is unusual in that its past participle is identical in form with its plain form, so hit can serve as a past participle like been as well as a plain form like be.) In that case this would be the structure:

How many people have been on a ship that's
[ hit an iceberg in the middle of the night ],
[ sunk ],
[ and [ lived to tell the tale ] ]?

Here the VP coordination is the predicate of an integrated relative clause attached to the noun ship. So on this understanding we are looking for a ship with three properties, not a person. And the third of the three properties is having lived to tell a tale. Ships don't tell tales. Therefore the number of such ships is zero.

On both syntactically possible analyses, then, the answer to Mr Lee's question is clear: absolutely none. No one has ever (i) been on a ship that's hit an iceberg in the middle of the night, (ii) sunk, and (iii) lived to tell the tale, including Mr Lee. And no one has ever been on a ship that has (i) hit an iceberg in the middle of the night, (ii) sunk, and (iii) lived to tell the tale, including the ship that Mr Lee was on.

I'm not dinging him for grammar faults, by the way. For one thing, there is always the possibility that the reporter transcribed his utterance incorrectly (the first rule of attributional abduction). Brian might actually have said ...have been on a ship that's hit an iceberg in the middle of the night and sunk, and lived to tell the tale, which would be fully grammatical, and then been incorrected (as we say here at Language Log) by some sub-editor who, as Martin Hardcastle put it in an email to me today, "removed the first 'and' because of the rule about multiple 'and's in a list, not realising that in this case it completely breaks the grammar." With the extra and, everything would be fine, because we would have two coordinations. In the following I number the outermost brackets of the two coordinations, and show the smaller one (which is inside the larger one) in blue:

How many people have
[1been on a ship that's
[2 [ hit an iceberg in the middle of the night ] [ and [ sunk ] ] 2]
[ and [ lived to tell the tale ] ] 1]?

But in addition to the fact that he may not have misspoken at all, I admire Brian's pluck and fortitude. This dude is cool. He and his wife were cruising Antarctica and looking at wildlife on a ship called the Explorer. It hit an iceberg at night, Titanic-style, ripping the side open, and water started filling the cabins. Having seen Titanic and learned from it, the captain gave the "Abandon ship" order early enough, and they got the lifeboats launched before the ship rolled over on its side and eventually sank.

(By the way, why is the noun phrase ship in Abandon ship anarthrous? Because in an emergency at sea there is no time to say the? Surely not. It's just another of those mystery constructions where the usual principle — that singular count nouns in English require a determiner constituent of some kind — is waived.)

Brian wasn't woken initially by the alarm. His wife Gillian woke him up. "I thought she wanted a cuddle," he told the reporters. (These people who talk about British men being uninterested in sex! Ha! They don't know what they're talking about! Brian was ready for a nice bit of cuddle time even in the small hours of the morning on a doomed vessel in the Southern Ocean as it began to fill with icy salt water.) He and Gillian spent a harrowing night in an open lifeboat among ice floes, and the following night in a hut on King George Island, and he must have been a little bit weary when he finally talked stoically to the press (he had no complaints about the wonderful holiday trip). And (if indeed his utterance was correctly reported) he got the plan for his sentence a bit messed up and produced what Language Log readers will know is called a WTF coordination. You probably would too, if you had hit an iceberg, sunk, and lived to tell the tale.

[Update: Kris Rhodes at UC Irvine pointed out that Mr Lee could have intended sunk as what The Cambridge Grammar of the English Language (chapter 15) calls a supplement, a parenthetical interpolation embellishing the point about what the ship had done, in which case the reporter would have done better to transcribe the utterance thus:

... how many people have been on a ship that's hit an iceberg in the middle of the night — sunk — and lived to tell the tale?

That would be fully grammatical, and it is certainly a possible analysis.]

Posted by Geoffrey K. Pullum at December 13, 2007 04:51 AM