April 17, 2004

The magazine of anarchy on the high seas

I'm afraid that no such magazine exists, but for a brief, bright moment yesterday I thought differently. I was sitting in a coffee shop, reading the May Atlantic, and came to the blurb for William Langewiesche's (gripping, horrifying) article on the 1994 sinking of the ferry Estonia:

One of the worst maritime disasters in European history took place a decade ago. On a stormy night on the Baltic Sea, more than 850 people lost their lives when a luxurious ferry sank below the waves. From survivor testimony and other sources our correspondent has pieced together the Estonia's last moments—part of his continuing coverage for the magazine of anarchy on the high seas.

It's not surprising that I made this mistake. The trigram "the magazine of" gets 394,000 ghits, and in 28 of a sample of 30, the of-phrase expressed the magazine's theme, e.g.:

the Magazine of Christian Unrest, the Magazine of Dog Powered Sports, the Magazine of the Military-Industrial Complex, the Magazine of Type and Typography, the Magazine of Speculative Transformation, the Magazine of the Eparchy of Newton, the Magazine of Roll Your Own Cigarettes, the magazine of The Swedish Homebrewers Association

The remaining two cases were an instance of the idioms "X of choice" and "X of influence":

the magazine of choice for plastics professionals, the magazine of influence for glass industry leaders

So my vision of something like "PIRATE! the magazine of anarchy on the high seas" was linguistically probable, if sociologically unlikely. And I realized long ago that my estimates of sociological likelihood in today's world are not very sharp ones.

In the Atlantic's blurb, the two prepositional phrases don't really seem right when placed in the other order: "...part of his continuing coverage of anarchy on the high seas for the magazine." I think that this is a matter of the relative "heaviness" of the constituents -- "the magazine" is semi-anaphoric as well as short.

Here is a paper by Jennifer Arnold and Tom Wasow that discusses these heaviness effects in the case of two other constructions. Arnold and Wasow support the view that "these phenomena stem from constraints on production and planning" rather than syntactic structure or parsing effects. But I don't see how to make a production-difficulty theory explain the difficulty with the order "...part of his continuing coverage of anarchy on the high seas for the magazine" in this example. The writer can take all the time that (s)he wants to craft it -- the problem is that this reader, at least, still finds it odd.

The blurb writer has walked the sentence into a trap. The word order "coverage for the magazine of anarchy on the high seas" doesn't work because it's likely to be misparsed, and the order "coverage of anarchy on the high seas for the magazine" doesn't work because of heaviness problems. This seems to be one of those times when you just have to back up and try it from a different direction.

[Update: Tom Wasow emails to say:

I wouldn't want to say that weight effects are due entirely to the contingencies of planning and production. That would make it hard to explain their occurrence in edited written text. But past attempts to give explanations for weight effects have almost all been entirely in terms of parsing efficiency (notably, Hawkins's 1994 book, but a lot of other stuff, too). I'm sure that postponing long things helps both in both production and comprehension. Jennifer Arnold and I have tried to tease apart the predictions of the two motivations for weight effects in several studies. To the extent that we've been successful, the predictions of the production model have been better. [...].

That said, I agree that there's nothing that could save the sentence you quoted from the Atlantic.

Uh, it was the Baltic, Tom :-)...]

Posted by Mark Liberman at April 17, 2004 10:57 AM