Language Log

I'm spending three days in Tampa at the kick-off meeting for  DARPA's new BOLT program. Today was Language Sciences Day, and among many other events, there was a "Semantics Panel", in which a half a dozen luminaries discussed ways that the analysis of meaning might play a role again in machine translation. The "again" part comes up because, as Kevin Knight observed in starting the panel off, natural language processing and artificial intelligence went through a bitter divorce 20 years ago. ("And", Gene Charniak added, "I haven't spoken to myself since.")

The various panelists had somewhat different ideas about what to do, and the question period uncovered a substantially larger range of opinions represented in the audience. But it occurred to me that there's a simple and fairly superficial kind of semantic analysis that is not used in any of the MT systems that I'm familiar with, to their considerable detriment — despite the fact that algorithms with decent performance on this task have been around for many years.

Psycholinguist Craig Chambers sent me this photo that he snapped recently inside a large pharmacy chain store (you know the kind, where you can avail yourself of all your better-living-through-chemicals products under one roof, whether it's anti-depressant, cough syrup, your favorite crunchy snack of Olestra and yellow dye #6, jet printer ink, or the entire range of household plastics.)

Along with the photo, Craig wrote:

If you ever find yourself rubbing shoulders with an executive from Shoppers Drug Mart, you might tell them that they could use your expertise in

(a) language for in-store advertising
(b) scalar adjectives
(c) both of the above

Larry Horn posted this to the American Dialect Society's mailing list a couple of days ago:

One of the bêtes noires of the prescriptivists is "miss not Xing" in the sense of 'miss Xing'.  Here, for example, is Lederer:

Let's look at a number of familiar English words and phrases that turn out to mean the opposite or something very different from what we think they mean: […]

I really miss not seeing you. Whenever people say this to me, I feel like responding, “All right, I'll leave!” Here speakers throw in a gratuitous negative, not, even though I really miss seeing you is what they want to say.

After almost a month, I'm finally following up on the results of the single-question surveys that I asked Language Log readers to participate in. Each survey received an overwhelming 1500+ responses, and I didn't realize that I needed a "pro" (= "paid") account on SurveyMonkey in order to view more than the first 100. I owe special thanks to Mohammad Mehdi Etedali, to whom I transfered the surveys and who kindly sent me the overall percentages.

We've had several posts recently (here and here) showcasing the humorous consequences of interpreting quantifiers overly literally, with a blind eye to the usual contextual limits on their domain of interpretation. The following comic illustrates another possible pragmatic failure when it comes to quantifiers:

A posthumous collection of Shel Silverstein's poems and drawings has just been published, with the title Every Thing On It. That's also the title of a poem contained in the collection, and Buzzfeed reproduced it in a post today. The verse displays the kind of lightly subversive wordplay that Silverstein is famous for.

EVERY THING ON IT
I asked for a hot dog
With everything on it
And that was my big mistake,
'Cause it came with a parrot,
A bee in a bonnet,
A wristwatch, a wrench, and a rake.
It came with a goldfish,
A flag, and a fiddle,
A frog, and a front porch swing,
And a mouse in a mask—
That's the last time I ask
For a hot dog with everything.

That's the title of a recent paper by Caitlin Fausey and Teenie Matlock that appeared in the journal Political Psychology. It's a heartwarming title, one that permits me to dare to dream of that better day when political parties will divert rivers of cash to linguistics departments, when a grad student will be able to defend a thesis on applicative constructions in East Asian languages one day and take up a lucrative job as Washington policy wonk the next, and when volumes by Noam Chomsky and Richard Montague will be pressed into the hands of military personnel charged with the task of winning the hearts and minds of residents in troublesome, volatile nations.

The paper stems from recent interest in the persuasion sciences about the fact that how a message is expressed often has a startling impact on the choices and behaviors of its audience. Most of the attention has been lavished on questions of lexical choice, or on whether a message is framed as involving gains rather than losses. But these are happy days, and persuasion research seems to be taking a more adventurous turn, with investigators beginning to tackle questions involving finer points of semantics and their grammatical correlates.

On a recent commute via Calgary's light rail train, the following recorded announcement caught my ear: "Please stand clear of the doors as this train is trying to depart." Beyond my initial bemusement, I thought little of it. I imagined, perhaps, a harried public transit employee playing a bit fast and loose with selectional restrictions in much the same way that a certain child I know puts together jigsaw puzzles: by pounding together pieces that approximately fit together and hoping for the best. But later in the week my husband fielded a call that made me wonder whether the train announcer's overextension of animacy features wasn't in fact a crafty linguistic maneuver to increase rush-hour compliance.

The big deal in a new paper "Motivating voter turnout by invoking the self" (see also the official PNAS site, or e.g. this Discover magazine article "The power of nouns….") is that people can be manipulated into voting simply by clever use of nouns instead of verbs in a questionnaire. In each of several studies, potential voters were split into two groups and given (amongst other questions which didn't vary by group) one of two questions to answer:

Group 1 question: How important is it to you to be a voter in the upcoming election?

Group 2 question: How important is it to you to vote in the upcoming election?

Turned out that Group 1 turned out. Really. In one of the studies an amazing 95.5% of them actually turned out to vote, whereas only 81.8% of Group 2 voted. That's obviously a huge effect on voting behavior. And it appears to be caused by the use of a construction with the nominal "voter" instead of the verb "vote".

Thanks to Graeme Forbes for alerting me to this! He has given me permission to post his note to his pro-compositionality friends. [For readers for whom compositionality is a new concept: it's a central tenet of formal semantics, usually credited to Gottlob Frege (but not without some controversy): The meaning of the whole is a function of the meaning of the parts and of the way they are syntactically combined. See, for instance: this introductory handout or the entry on Compositionality in the Stanford Encyclopedia of Philosophy.]

From Graeme Forbes:

You may have already seen this, but in case not, here's an excerpt from an article in the current New Yorker, "Easy, Tiger", by David Sedaris (July 11/18 2011, p.40). It's an entertaining piece about how he "mastered" Mandarin, Japanese and German with the aid of tourist-courses on his iPod, including one from a company called Pimsleur. The "Easy, Tiger" alludes to a phrase in the section on romance in the Mandarin course. Or was it the German course? Surely not!

Jane Perlez and Eric Schmitt, "Pakistan's spies tied to slaying of a journalist", NYT 7/4/2011:

ISLAMABAD, Pakistan — Obama administration officials believe that Pakistan’s powerful spy agency ordered the killing of a Pakistani journalist who had written scathing reports about the infiltration of militants in the country’s military, according to American officials. […]

The disclosure of the intelligence was made in answer to questions about the possibility of its existence, and was reluctantly confirmed by the two officials. “There is a lot of high-level concern about the murder; no one is too busy not to look at this,” said one.

I did the stupidest thing in Bulgaria. I bought a new wallet. Never do that on a foreign trip. See if you can figure out why before you read on. (And yes, of course there's a linguistic angle.)

Luke Yeomans (pictured) had a king cobra sanctuary in Nottingham, England, and planned to open it to the public this weekend, but instead one of his cobras killed him on Wednesday with a single bite, a hefty injection of neurotoxic and cardiotoxic venom that gave him a heart attack. Sadly, the linguistic signs that he would be killed this way were already present in the record, quite clear in something he had said. I wish someone could have warned him.

The Daily Telegraph quotes Yeomans as saying this about why the snakes offered no threat to him:

These king cobras know I provide them with food and fresh water so they're not going to go out of their way to do harm to me when I do no harm to them whatsoever. People say I’m mad but it’s better than saying that you’re bad and everything I do is good. My life is about the conservation of the king cobra.