Friends of Semantic Compositions

July 2008

Sun Mon Tue Wed Thu Fri Sat
    1 2 3 4 5
6 7 8 9 10 11 12
13 14 15 16 17 18 19
20 21 22 23 24 25 26
27 28 29 30 31    

Site Statistics

Blog powered by TypePad

November 12, 2005

Misinterpretation can be expensive if you're a genie

SC has lately become a fan of a reality TV show on NBC called Three Wishes. Whereas typical reality shows are focused on pushing people to do their worst (think Jackass, Fear Factor, or in a slightly different vein, American Idol), Three Wishes is an uplifting show where ordinary people with hard luck have their lives changed through the efforts of the show's crew. Examples of this have included the gift of a device that helps calm stuttering to a college student who dreamed of running for office, and helping a child with traumatic head injuries be able to play outside again. For the most part, the things they help out with are a matter of lacking money or influence, and by coming through with a decent production budget (and the generosity of various donors), there's a lot they can do to make people's lives better.

However, Friday night's episode featured a wish that might have been a costly mistake for the show's producers. It featured a high school student born with a leg condition that left him unable to play any sort of sports. In lieu of being an athlete, he had spent his high school years as the trainer for the school's football team. No operation or prosthetic could fix this one, but the team's players had approached the show's producers hoping that they could do something nice for him anyway.

Rather than just asking the student what he might wish for, cast member Diane Mizota took him out on the school's football field after a game, and had him whisper her his wish during a ceremony to honor his efforts. At this point in the show, the exact nature of the wish was not revealed to the viewer.

Later, it turned out that his wish had been:

I want to be an NFL trainer and do something special for my team.

The first half of this wish was fulfilled by arranging for him to spend next summer as an intern with the Cleveland Browns. But now, we finally get to the crucial linguistic problem.

There are two ways in which that sentence's referring expressions (aside from the "I", which we know is the student) might be interpreted (pay attention to the subscripts):

1) I want to be an [NFL (team)]i trainer and do something special for [my team]i.
2) I want to be an [NFL (team)]i trainer and do something special for [my team]j.

The first sentence expresses the idea that he wants to be an NFL trainer and do something special for the NFL team employing him, which presumably would be accomplished by acting in his capacity as trainer. The second sentence really expresses two wishes: that he wants to be an NFL trainer, and that he wants to do something for his high school team. The show's producers assumed that #2 was the case, and arranged for each member of the team to receive an LCD HDTV set from Dell. Whereas the first parse cost them no more than whatever the Browns might have charged for their participation, the second one added minimally $600 times whatever the number of players on the team was. Figuring on a 40-man roster, that's $24,000, and that assumes they were just monitor-sized, which did not look to be the case.

Truthfully, though, from watching this kid talk, there's no way he didn't mean #2. Your guess is as good as SC's about what else he might have meant by "do something special", but while TVs might not have an obvious connection to football ([we disagree -- the NFL, NCAA, and sports bar owners everywhere]), it was perfectly clear that his wish was being fulfilled for the right group of people. And just as clear that sometimes there really isn't any ambiguity where it would seem to be in language.

September 16, 2005

I knew I hated meetings

Everyone who has ever worked in an office has at least a passing acquaintance with motivational products like those produced by Successories. Their posters include pretty pictures with slogans like Rule #1: "If we don't take care of the customer...somebody else will."

Recently, your host became aware of a company with a much more Dilbert-like approach to the same business, Despair, Inc. Despair mocks the relentless rah-rah attitude of motivational speakers, and has a poster near and dear to SC's heart for reasons which should be readily apparent (remember that his employer is a consulting firm).

One of their posters is causing SC considerable cognitive dissonance, though. Captioned "Meetings", the slogan reads: "None of us is as dumb as all of us." The problem with this two-quantifier sentence is that while only one reading is meaningful, your host cannot get himself to stop thinking about the other one.

The intended reading might be paraphrased as "while any single person in this organization might be pretty dumb, they are less dumb than the group as a whole". Makes sense in a too-many-cooks-spoil-the-broth sort of way. But the other meaning, which SC knows doesn't really work with the quantifier none, runs something like "the absence of all employees is equally as dumb as the collection of all employees". If you replace that none with an expression that actually refers to someone, like "A few of us are as dumb as all of us", both readings are intelligible -- although of course the line loses considerably in the wittiness department.

If contemplating the nonsensical meaning of that "none of us" line bothers you as much as it does SC, you can always take refuge in the knowledge that there are worse things to worry about .

September 01, 2005

30 days hath a 2% CD

A long time ago, your host wrote about a woman he saw in a grocery store, who could say with a straight face "I only have 15 items" in her cart, when such was plainly false. Although he'd never met her before, and hasn't seen her since, he's pretty sure he's selling Chez SC to her sister, or maybe a cousin.

Let's clear up a legal point first. If you are a legal adult (the technical term is that you have reached the "age of majority"), and have not been declared incompetent by a court, then if you sign your name to a contract, it means you agree to the contract. In the case of a real estate transaction, you don't just sign -- you initial, and initial, and initial, and sign, and sign some more. If this seems a little overly emphatic, all will become clear momentarily.

In the case at hand, the buyer signed a contract with a 30-day escrow period (mathematically inclined readers may do the arithmetic with Friday as the 30th day). This means that she is obligated to have the sum total of the purchase price deposited in escrow by the 30th day, and not, say, the day after. We never met the woman to hold a gun to her head or otherwise coerce her, and she contacted our realtor 3 days before making her offer and having it accepted. So all in all, she will have been involved in this transaction for 33 days at the time of closing.

All this emphasis on 30 days ought to be raising a red flag by now for most readers. So here's the payoff. 10 days ago, SC received a call from his realtor, where she relayed 2 propositions with contradictory content. Somehow, she managed to miss that last issue. The propositions:

1) Before she knew she would be entering into a contract to purchase Chez SC, the buyer placed her down payment funds into a 30-day CD,
2) Said CD supposedly would not expire until the day after escrow was set to close.

So would we mind extending the close of escrow until the next business day, after Labor Day?

Most readers will have read those statements and performed a bit of temporal reasoning like so:

If t1 < t2, then (t1 + 30) < (t2 + 30). However, we have been told that in fact (t1 + 30) >= (t2 + 30). So since buyer says the consequent is false, but that the antecedent is true, then courtesy of the truth table for material implication, we know that the buyer's claim is that the whole implication is false. Unfortunately for the buyer, we assume an axiomization of arithmetic that says that the statement is true, which only leaves us the line in the table that says the antecedent is false, and therefore we must conclude that the buyer is full of it.

What, you didn't think about your preferred axiomization of arithmetic?

In any event, recognizing that the buyer was attempting to have a situation where she collected more interest and got SC and wife to pay for the privilege of enabling her to do so (courtesy of three days of prorated interest, association dues, and property taxes), your host responded by offering to allow the buyer to extend her escrow period, so long as she paid all costs associated with a later closing. Needless to say, she was angry.

We then received a call from our realtor a day later, informing us that by a miracle, the buyer had managed to convince her bank (which happens to be the same one that SC uses) to "let her out early" without any penalties, and by prorating the interest on the CD to account for a shorter period. Unfortunately, that's also not true, because banks can't let you out of CDs early, by law, without a penalty. Perhaps the truth is that she was hoping to roll her money into a shorter CD that nevertheless would have delayed the close, and backed out of that when it became obvious that we would not pay her for the privilege.

While your host started this blog 20 months ago hoping to demonstrate the use of "ideas...out of theoretical and applied linguistics", it really doesn't take a linguist, a philosopher or a mathematician to see through baloney like this. But it never ceases to amaze him to see how many people haven't figured that out for themselves.

July 13, 2005

Oh, yes he is

SC's parents passed along a copy of the schedule for the upcoming season of a local playhouse, the Poway Center for the Performing Arts. Turns out that a production has a title of the sort that makes for SC posting material.

In a few months, they'll be putting on a stage adaptation of a book by children's author Judith Viorst, commissioned originally by the Kennedy Center. The book/production's title, "Alexander, Who’s Not Not Not Not Not Not Going to Move", has a semantic interpretation which is quite at odds with its pragmatic interpretation.

From the pragmatic standpoint, the repetition of "not" makes it particularly emphatic. SC isn't aware of a scale of child peevishness, but it probably would look something like so:

Mild: I'm not going to move.
Moderate: I am not not not going to move.
Serious: I am not not not not not  going to move.
Severe: Mooooooooooooooooooom!

Repetition certainly does serve as an intensifier. But from the semantic standpoint? Well, your host counts six "nots". Which means they all cancel out, assuming they take the same scope. Which means that the play might equally well be titled, "Alexander, Who Is Going to Move". Which, incidentally, is what he ends up doing anyway.

An SC No-Prize goes to the first reader who comes up with a screwball analysis that renders the six "noes" as a semantic negative.

June 18, 2005

Trivially true

ESPN has a blurb up for this week's baseball "power rankings" (a subjective evaluation of the teams in MLB):

For the first time in more than 30 years, Washington is on top of our baseball power rankings.

It's true -- but although the intended reading is that there hasn't been baseball in Washington, D.C. in all that time, the truth conditions of the sentence are nevertheless pretty vacuously satisfied. Until late 1994, there was no ESPN.com (and then it was called ESPNet.sportszone.com), and the power rankings didn't follow for another 5 years after that. So the suggestion that something about their rankings hadn't changed in 30 years is no different from saying that it hadn't changed for 10,000 years, or since the beginning of the universe.

This does allow for some amusing consequences, though. Despite the Yankees having been on top for much of the time since the rankings began, one can minimize their pre-2005 dominance with statements like, "The Yankees have only led the ESPN power rankings for 200 weeks in the last 15 centuries". This makes SC feel better, even if 14.9 of those centuries are frankly irrelevant.

June 08, 2005

Save at least pi, or maybe e percent

Your host just picked up his mail for the day, and inside was a most unusual offer from an insurance company. Emblazoned on the envelope, and repeated several times in the text of the enclosed letter, was the phrase:

save up to $327.96 or more

It's an odd statement for at least two reasons, although thankfully it's not false in an offensive way, like bogus past due notices:

1) Your host interprets "up to X" as meaning that the dollar amount in question is an upper bound. A charitable interpretation might be that "or more" merely cancels the upper bound. After all, it might be that the company's risk pool could tolerate the addition of a Ferrari 612 Scaglietti, and so they might be able to knock $5,000 off the premiums of such a car's owner. A more cynical ([read: likely -- ed.]) interpretation might be that they want to render the entire statement false, knowing that many (most?) of their offers will be far less attractive.

2) $327.96? That's a very random number for an advertisement. One wonders if it isn't the largest amount they've ever saved someone, based merely on the bizarre choice of digits. One rarely sees signs saying things like "This weekend only -- $43.71 off any purchase of $268.44!". It's usually more like "$50 off any purchase of $300!" or something else with nice round numbers. In puzzling over the communicative intent here, SC can't decide if it's actually a real example, or somebody's weird idea that such a specific number will be attention-grabbing ([it ain't that weird; you read the letter, didn't you? -- ed.]).

June 06, 2005

It depends on the meaning of "balloons"

In a sort of "now it can be told" moment, the first bit of formal semantics to pop up on these pages in some time.

Last week, your host was engaged in making preparations for the celebration of Mrs. SC's medical school graduation. Reservations for the restaurant had been made a month in advance, but for reasons unkown, the proprietors prefer not to associate with a florist. Therefore, it fell to your host to oversee the acquisition of flowers for the table. Wishing to preserve the maximum possible element of surprise, none of these plans were disclosed to Mrs. SC before the dinner.

However, as Mrs. SC prefers her celebrations quiet, she wanted the environment to be as balloon-free as possible. Since she also has an astonishingly good handle on your host's modus operandi, days before the actual event, she confronted him with the question:

You're not ordering balloons, are you?

Now, here are the facts of the matter. In order to fill the table with flowers, your host ordered several identical bouquets. But in order to distinguish one as particularly a gift for Mrs. SC, he ordered an additional bouquet which was both larger than the others, and decorated with a single balloon. Therefore, he felt confident replying:

No, I'm not ordering balloons.

while knowing full well that he had in fact ordered balloon.

So here is the question. Granting that the most probable interpretation of Mrs. SC's question is a demand to know whether the following logical formula is true (omitting details about time):

    AxAy( (x = SC) & Order(x,y) & (y = balloons) )

was it dishonest of your host to further parse out the plural morpheme and rewrite the formula as:

    AxAy( (x = SC) & Order(x,y) & (y = balloon) & (Cardinality(y) >= 2) )

I say no. The balloon said "Congratulations".

 

May 24, 2005

Cooperation is a two-way process

SC has been very impressed with some recent comments on an old favorite post, The Totemic Power of Names.A high-school student named Sara, presently employed at Dairy Queen, has been ably making the case that customers who actually get what they ask for sometimes get very upset over this fact. Cashiers aren't mind-readers, and if they have to throw away product because the customer doesn't want to be held accountable, it's them -- not the customer -- who will have to answer to the insidious Darth Manager.

This is a fair point, and so it's time to take another look at the situation, using an old favorite analytical tool, Grice's maxims. Two of the four strike your host as relevant:

First, the maxim of quantity, which says:

1. Make your contribution to the conversation as informative as necessary.
2. Do not make your contribution to the conversation more informative than necessary.

Second, the maxim of manner, which adds:

1. Avoid obscurity of expression.
2. Avoid ambiguity.
3. Be brief (avoid unnecessary wordiness).
4. Be orderly.

In the original post, there were a couple of assumptions which were unstated, but not necessarily valid. First, there was an assumption that cashiers wouldn't try to comply with the literal text of a customer's request if it meant making something off-menu. Second, there was an assumption that customers would think along such lines themselves, and omit only words which seemed gratuitously unnecessary to deciphering which of the possible menu items could actually be meant (like the "love" in "Choco-Cherry Love Blizzard").

But these aren't justified assumptions. Under the maxim of quantity, your host assumed that "chocolate-cherry blizzard" was adequately informative to uniquely specify an item on the menu, and that the full statement of "chocolate-covered cherry" was therefore unnecessary, on the grounds that the first was informative enough and the second was excessive. Sara points out that two different recipes are involved here, and only the latter is actually on the menu. At first, SC objected to this on the grounds that he's never seen the other one, and might have added that even if the "Choco-Cherry Love Blizzard" involved the other recipe, he's never seen them both on the menu together to create a potentially ambiguous group of menu items. However, from the cashier's perspective, "chocolate-cherry" isn't informative enough for them to be sure that the customer means that they will be happy with getting the item on the menu, as-is. Perhaps a special order is really what's meant, and so what appears to be informative enough from the customer side isn't actually so from the cashier's.

This brings us to the other issue, the question of manner. Again, the omission of words like "covered" and "love" appears to a customer who actually expects to get what's on the menu to satisfy the condition of being brief, without creating undue ambiguity or obscurity. However, a cashier can't count on the rationality of customers -- while some customers might expect that a brief expression will simply be assimilated to the nearest matching pattern, others will expect to get something different. Thus, Sara brings up the example of a customer who orders a "strawberry Blizzard" when they really want a "strawberry cheesecake Blizzard" (if we really want to be pedantic about it, DQ calls it a "strawberry cheeseQUake Blizzard"). From SC's perspective as a customer, if the only Blizzard on the menu with a name including the word "strawberry" is the one which also includes "cheesecake", then he would have expected the cashier to understand that that's what he wanted. After all, that's what's on the menu. But from the cashier's perspective, assuming that customers are reasonable enough to understand the distinction between what's on the menu and what they said isn't at all safe. After all, it's what they said -- wasn't the cashier listening?

None of this is meant to say that cashiers who really don't speak English should be able to get away with blaming customers for their own failures to understand (except, of course, in non-English-speaking countries). And SC isn't backing down at all from the claim that reciting long strings of marketing-speak is gratuitous and unnecessary. But that shouldn't prevent us from having a little empathy for people who know from painful experience that not everyone says what they mean.

November 14, 2004

It depends on the meaning of "lot"

A few minutes ago on the phone with a friend, currently a resident at a prestigious Texas hospital:

I'm getting three to four hours of sleep a night. That's a lot.

Mrs. SC:

No, it's not. Residents at my school get more.

One can easily imagine constructing a scale for "lot" as a modifier for "sleep". For most of us, "a lot" of sleep is probably 7-8 hours a night, "enough" maybe being 6-7, "not much" as 4-5, etc. This is a good reminder that meaning can be context dependent -- when Mrs. SC says that her colleagues get "more" sleep, your host feels comfortable guessing that it's in the range of 0.5 - 1 hour more. Just thinking about getting that little sleep is making SC tired.

October 23, 2004

One sore loser, coming up!

While browsing ESPN's preseason NBA coverage ([don't you have something more important to do, like getting the new site ready? -- ed.]), your host saw a very difficult to interpret sentence. At least, it better be tougher than it looks like, or this guy really is nuts.

Orlando Magic GM John Weisbrod has been widely criticized for being a hockey manager out of his league -- literally -- in trying to assemble a basketball team. In this ESPN article, he comments on his philosophy:

Our objective was not to bring in a single player that wasn't a sore loser.

The double negative makes this a bit tough to unpack; let's try cancelling them out:

Our objective was to bring in a player/players that are sore losers.

Whoa there! Does he actually mean he wants a team full of headcases? Just a few sentences later, he's quoted as saying:

The NBA has been in the business of prima donna development as a league and I wanted to get out of that.

This leads SC to think that Weisbrod was trying to say that he wanted a team full of players who wouldn't tolerate losing, not that he wanted a team filled with people named Keyshawn, Terrell, and Latrell. But as far smarter linguists have written, negation is just plain hard.

October 21, 2004

How can you tell a realtor's lying?

As SC has noted many times before, he lives in an area with a very overheated real-estate market. Because of this, every day brings at least one new flyer left on the door for some realtor or other. Many of these agents write things which need to be very carefully parsed in order to determine their truth.

Thus, one particularly obnoxious agent has just left a tag on the door showing pictures of a house and boasting that it's now in escrow. In other words, it hasn't actually sold yet -- and something like 4/5ths of deals in SC's neighborhood have fallen apart in escrow in the last 6 months, a statistic being pushed heavily by one of this clown's competitors.

But here's the kicker: the ad boasts that the house "listed at $XXX,XXX" (where the X's stand for a larger number than SC cares to acknowledge, in keeping with certain thoughts of Radagast's on privacy). It occurs to your host that this is very different from saying "sold at $XXX,XXX" -- and that it has been written this way for the specific purpose of sidestepping the actual, notably lower price that the contract was signed at. Caveat venditor.

September 03, 2004

More isn't always better

Lamentably, your host didn't get a chance to watch any convention coverage today. Heard a bit on the radio while driving, but without a computer immediately at hand, nothing to write about.

However, while watching TV late this evening, something bloggable came up. As often happens during election years, some advertisers like to run election-themed advertising (here's a notably un-clever one). In particular, Miller has been running a series of commercials mocking the "King of Beers" (Budweiser) for not being democratically elected, and suggesting that their flagship beer, Miller Genuine Draft is running for "President of Beers".

In the commercial in question, Miller's advertisers assert that one fact supporting their "candidacy" is a study purporting to show that "2 out of 3 beer drinkers say that Miller Genuine Draft has more flavor than Budweiser".

SC isn't much of a beer drinker, but this is a linguistically interesting claim. Intuitively, "flavor" doesn't seem to be a scalar predicate -- either something has a particular flavor or it doesn't -- but it's very easy to find examples of specific flavors being treated as scalar predicates. And this makes sense, since the intensity of a flavor is obviously scalar. Searching for a few flavor descriptions with Google, we find some 3,000 examples of "more chocolatey", some 670 tokens of "more minty", and even some 750 examples of "more gamey". It should be pointed out that not all of these collocations actually are scalar descriptions of flavors -- one of the first hits for "minty" is "More Minty Lyrics" -- but short of using the Linguist's Search Engine, it's too hard to pick only the syntactically relevant examples.

The scalar nature of flavor descriptions might not be entirely symmetrical, though. Below, we present a table of Google counts for "more" versus "less":

more __ less __ Totals
chocolatey 3,060 (97.3%) 84 (2.7%) 3,144
minty 671 (87.8%) 93 (12.2%) 764
gamey 745 (73.3%) 272 (26.7%) 1,017


Although this is hardly an exhaustive list of flavor types, it's interesting to notice the asymmetry between positive and negative scalar comparisons. It's hard to say whether or not this is merely a frequency effect relating positive vs. negative scalar uses more generally, or if this is something specific about flavor-related language.

Assuming the latter, where might it come from? "Flavor" by itself is hard to define a scale for -- does Miller taste more like hops than Budweiser does like barley? Questions like that don't even make sense to ask, which makes it hard to make sense of a claim that one beer has "more flavor" than another. Even if we interpret that to mean that Miller's characteristic flavor is more intense on its scale than Budweiser's flavor on some other scale, it's not clear that the drinkers polled also meant that Miller's flavor was more pleasant than Budweiser's. Of course, in the absence of some solid statistics about the relative frequencies of "more" and "less", this speculation might merely rank high on the crackpot scale.

August 05, 2004

Chomsky, alone for a change

Mark Liberman discusses one of the obituaries for philosopher Sidney Morgenbesser provided by Arts & Letters Daily today. Since your host must confess that the first time he hear of Prof. Morgenbesser was when he opened up AL Daily this morning, he didn't expect that this would be the cause of any postings. It's clear enough from the obituaries that Prof. Morgenbesser was something of a wiseass, which is actually a compliment when it comes from this source.

In reading the New York Times' obituary, though, your host was caught by a sentence with amusing semantics:

In an interview yesterday, Noam Chomsky, the linguist at the Massachusetts Institute of Technology who agreed with Dr. Morgenbesser about some things and not others, called him "one of the most knowledgeable and in many ways profound thinkers of the modern period."

It's not Prof. Chomsky's quote that's interesting, but rather the definite description assigned to him: "the linguist at the Massachusetts Institute of Technology who agreed with Dr. Morgenbesser about some things and not others". Unpacking the ambiguity here is a great example of why all journalists should have to sit through a lecture on quantifier-raising before going off to spout their nonsense.

I can get four readings out of this description. Since notating them logically is a bit of a challenge without a decent equation editor, I'll put them in (hopefully) unambiguous prose:

1) There is only one linguist at MIT. That linguist is named Noam Chomsky. Noam Chomsky agrees with Prof. M. about some things, and disagrees with him about others.

2) There are many linguists at MIT. One of those linguists is named Noam Chomsky. For all linguists who work at MIT, if their name is not Noam Chomsky, then they either agree with Prof. M. about everything, or disagree with him about everything. Only Noam Chomsky agrees with Prof. M. on some things while disagreeing with him about others.

3) There is only one linguist at MIT. That linguist is named Noam Chomsky. At each university containing linguists, there is exactly one linguist who agrees with Prof. M. about some things and disagrees with him about others. If there are other linguists at other universities, they must either agree with Prof. M. about everything, or disagree with him about everything.

4) There are many linguists at MIT. One of those linguists is named Noam Chomsky. For all linguists who work at MIT, if their name is not Noam Chomsky, then they either agree with Prof. M. about everything, or disagree with him about everything. Only Noam Chomsky agrees with Prof. M. on some things while disagreeing with him about others. If there are other linguists at other universities, only one at each university may agree with Prof. M. on some things but not others; all remaining linguists at each university must agree about everything or disagree about everything.

Now, none of these strikes me as being precisely what the author actually meant. That would be:

5) Of all the people I could have called for this story, Noam Chomsky is about the only name that any of my readers will recognize. Chomsky is a linguist at MIT. Chomsky agrees with Prof. M. about some things and not others. No implications ought to be drawn about the number of other MIT linguists, or their views on Prof. M. Ditto for linguists at other universities.

The problem, of course, is that (5) is not actually something that can be extracted from the original sentence as written. Had the author written "Noam Chomsky, a linguist at MIT, who agreed with Dr. M. about some things and not others", it would have been quite clear from the indefinite article that there could be more than one linguist, and from the separation of descriptions provided by the comma that there are no implicational relations between Chomsky's status and other people's views about Prof. Morgenbesser.

It so happens that this isn't hard to figure out because the average linguist or philosopher knows better than to assume that ordinary language is really being used in the highly technical senses meant in their professions. In fact, most people probably won't ever be disturbed by readings 1-4, because they run so counter to ordinary usage. But you don't have to be a semanticist to be bothered by the sloppiness of that "the", with its unnecessary implication that some aspect of what follows is unique to Chomsky. Even if one doesn't take the time to work through all the possible readings, a description like that looks wrong on an intuitive basis. I wouldn't have thought twice about it if I hadn't read it and thought: "did this guy call everyone at MIT?". Of course he didn't -- but it wouldn't have taken much more effort for him to write clearly enough to avoid raising such questions in the first place.

July 31, 2004

That's fine

SC and Mrs. SC went to dinner this evening at a trendy pan-Asian restaurant. To be frank, it was just OK. But as your host devilishly remarked as he signed the bill, "I may not have gotten a good meal out of this, but I'm getting a post!". What caused this comment?

Mrs. SC only ate about half of her meal, and requested a box for the remainder. When she did so, the waiter replied, "I can box it for you". Mrs. SC then told him, "That's fine". Upon the waiter's return to the table, he handed her a box and left. Mrs. SC was surprised.

"Wasn't he going to box it for me?", she asked. "No, you told him not to", replied SC.

The confusion stems from the ill-defined polarity of the phrase "that's fine"; we've addressed this before with regard to the question, "do you mind?". There are two possible interpretations, which we might expand like so:

1) That's fine, I would appreciate it.
2) That's fine, I can do it myself.

To your host, the second one is the more natural interpretation of the isolated "that's fine", but that intuition is certainly open to debate. Similarly, it seems to me that a falling intonation indicates (2), but again, introspection is not necessarily a reliable guide. A flat or rising intonation doesn't strike me as conveying (1) particularly strongly.

Pursuing my own intuitions further, it seems likely that if I wanted to say (1), I would actually say "that would be fine", and that this latter phrasing would never mean (2). But (1) is not an unreasonable interpretation of Mrs. SC's original utterance; had she said "that's fine with me", (1) would be clearly understood, and it's easy to imagine that she could have produced "that's fine" through ellipsis.

Of course, there's an easy way to avoid problems of ambiguity, whether one word is at issue, or the syntax of the entire sentence: just say something that disambiguates it. But if we all spent enough time trying to compute the potential interpretations of what we want to say before we say it, I'd be writing about the waiter who ran off in frustration rather than the ambiguity of words without explicit yes/no polarity.

July 08, 2004

What's in a name? Brand equity

In today's Wall Street Journal, there's an article on the annual Harris Interactive poll of "best brands". Since the article requires a subscription, we'll just reproduce the list, and then continue:

1. Sony 2. Coca-Cola 3. Dell 4. Kraft 5. Toyota 6. Ford 7. Honda 8. Procter & Gamble 9. General Electric 10. General Motors

The question that generated this list is simply: ""We would like you to think about brands or names of products and services you know. Considering everything, which three brands do you consider the best?". For completeness, here's the article's description of the polling technique:

Methodology: This poll was conducted online in the U.S. between June 10 and 16, 2004, among a nationwide cross section of 2,136 adults. Figures for age, sex, race, education, region and household income were weighted where necessary to bring them into line with their actual proportions in the population. Propensity score weighting was also used to adjust for respondents' propensity to be online. In theory, with probability samples of this size, one could say with 95% certainty that the results have a sampling error of ±2 percentage points of what they would be if the entire adult population had been polled with complete accuracy.

Although SC is often critical of specific facts about advertising, this is ultimately what it's all about. Every name on that list conjures up a whole host of properties, a few of which might even accurately be ascribed to them. Having taken a number of marketing surveys as part of a regular panel of volunteers, it's not hard for your host to imagine some of the word associations thatmight come up: "trustworthy", "innovative", "reliable", "quality", "a good value". Every one of these associations is worth billions, and has been hard-won.

In many ways, it would be quite erroneous to confuse this list with representing companies that are the best at what they do. While Kraft is one of the world's largest producers of processed cheeses, even the most casual cheese eater would likely admit that the equally mass-production Tillamook or Land-O-Lakes cheddars taste better than Kraft 2% milk singles. For the price, though, you could easily do worse. And so it goes through a variety of categories: Dell is simply the lowest-cost assembler of mostly generic parts (the obvious exceptions being the processor and video card), Coca-Cola produces a drink that most people think is worth no more than 60 cents a serving, but is good enough that they'll hand over $2.00 in the right contexts, and as for General Electric's presence on this list, suffice it to say that the poll considers GE an electronics company, and they haven't even made any of the clocks, radios or TVs sold under their own name in years. It wouldn't be wrong to say that this list represents the world leaders in "good enough".

That's not to say that each of these companies hasn't demonstrated a capability to produce products which are world-class in other respects (nor should the manufacturing capabilities that allow such cheap mass production be dismissed as trivial accomplishments). One of the terms of art frequently used by marketers is the notion of a "halo" product, something incredible that gives the brand status on the assumption that everything else in the product line is somehow related to it. This is especially an issue in the automotive industry, which is why Ford resurrected the Thunderbird, and then decided to bring back the GT-40 when the new T-bird didn't quite live up to that role in the press. It's why Toyota needed to create Lexus, and why Honda created Acura. It's why Sony originally created their "ES" electronics, even though they then quickly brought them right back down to the manufacturing standards of their regular products. Not everyone needs a "halo"; Dell gets by just fine without one, but that's as much because nobody would have any idea what the difference was between a luxury computer and a regular one as anything about the company.

Another study which focuses more on specific products, and which also is considered to be a benchmark for positive name associations, is the annual J.D. Power vehicle dependability survey, the 2004 edition of which just came out a week and a half ago. In some ways, the methodology is absolute garbage: although the company collects more detailed data from consumers, the reported statistic is "problems per 100 vehicles", which aggregates transmission leaks, buttons that fall off the controls, and comsumer disappointment that they're not getting quite as high gas mileage as they expected into a single nebulous category (a fact which explains Hummer's perennial ranking at the bottom of the initial quality survey). The results demonstrate that "dependability" is only really a factor in reputation for brands which have little else to go on. Lexus tops the most recent list, BMW is only slightly above average, and Mercedes-Benz is far below average, and yet the evidence from annual sales indicate that consumers are willing to in fact tolerate a lot more of the same problems out of the German brands listed here than they will out of domestic Lincolns or Cadillacs. The associations of other positive properties with the BMW propeller or the Mercedes star -- better performance, more luxurious interiors -- don't even have to be true at present because of the enormous contributions of history and advertising to the meanings of the names.

The problem with studies like these is that what's being measured is very slippery. In the case of J.D. Power studies, "problems" clearly have very little correlation with brand prestige, and fare only somewhat better in relation to sales rankings. And the Harris question provides absolutely no basis for justifying the opinions expressed by the people being surveyed. But these facts merely help us to understand that the relationship between experience and language is complicated. Is Sony better at cranking out CD players than Coca-Cola is at making soda? SC would wager that the people who ranked them 1-2 have no idea how they would answer that question, and might not even recognize that they had made such a statement. Similarly, statistical evidence of the sort found in the J.D. Power survey appears to have no impact whatsoever to people who insist on the superior quality of imported over domestic vehicles. Advertising and the historical experience of the 1980s are adequate to maintain those associations well after the facts that supported them have changed. Somewhere within the complicated mix of experience and advertising lies an opportunity to make a name as valuable as the products themselves, and the Harris poll shows that, this year, Sony has done it better than anyone.

June 23, 2004

Better laughing through scalar inversions

Since the fastest way to ruin a joke is to explain it, we'll start off by quoting a line that opens an editorial in the Wall Street Journal today. Enjoy:

New Jersey is the butt of many jokes, and some of them are even unfair.

Your host brings this up because it reminds him very much of the unusual review of "Farenheit 9/11" penned by Christopher Hitchens recently, and discussed by Mark Liberman.

The technique Hitchens attempted to apply is illustrated by the line:

To describe this film as dishonest and demagogic would almost be to promote those terms to the level of respectability.

As Prof. Liberman noted, "The rhetorical trope in play is a routine one: 'To describe X as P is an understatement', where P is some scalar evaluative predicate." He goes on to note that P can be negative (i.e., "To describe this split as acrimonious would be an understatement.") or positive (i.e., "To describe this report as timely is an understatement.") The oddness of Hitchens' line derives from the fact that the implication in either normal version of the rhetorical device is that X exemplifies P to an extreme degree, but that clearly isn't how Hitchens has written things.

SC was prepared to dismiss it as "overnegation", a la Liberman, until he saw the Wall Street Journal line. The humor in it comes from the fact that the line, at least in a more charitable essay, would normally go, "New Jersey is the butt of many jokes, and some of them are even fair." That would also be intended as a punchline; what makes the Journal's line more biting is that it inverts the scale being used for comparison, rather than simply being negation of the scalar predicate.

Finding more examples like this isn't easy, as Googling for "some of them are even" yields plenty of irrelevant hits. But here's another one, a quote from a Guardian story regarding disgraced BBC reporter ([a redundancy -- ed.]) Andrew Gilligan:

Andrew gets great stories, and some of them are even true.

This one only works by inverting the scale; ending the line with "some of them are even false" would be merely anomalous.

The problem with Hitchens' attempt at using this inversion technique is that the syntax of the sentence makes it twice as challenging to interpret as it would be otherwise. Had he written it as something more like:

Farenheit 9/11 is dishonest and demagogic, and those descriptions even almost make it sound respectable.

it would have been a lot more clear what he was up to. SC gives him points for trying, but it's a stylistic flub from somebody whose craftsmanship is usually much better than that.

June 20, 2004

A Father's Day thought

Following up in the SC tradition of comments on holidays, particularly Mother's Day, an observation about fatherhood and possible worlds.

As with Mother's Day, SC went shopping for cards, and was reminded of how incredibly fine-grained the categorizations of mothers could be. This time around, though, he noticed one category in particular that led to some additional thinking: "father-to-be".

Ostensibly, such a category is intended to apply to men who know that their wives/girlfriends/one-night-stands-with-good-legal-counsel are pregnant and expecting delivery in the foreseeable future. Another possible use -- one which SC is very leery of -- would be to inform said men that their wives/etc. are now pregnant and expecting. But an optimistic Hallmark product planner might go further than that, and hope that it would also appeal to those women who believe that their men will be fathers at some time in the future. Going even further, they might introduce a new category: "father-in-potential", which could be justifiably sent to any man who, in some possible world not necessarily their own, could be a father. This would expand the universe of possible card recipients to be nearly all men (minus a few very unfortunate types), and because of the additional revenue potential, SC is frankly shocked that Hallmark hasn't beaten him to this one.

Of course, there's one scenario in which your host can even imagine for sending Father's Day cards to women. In the event of a single mother taking on the roles of both mother and father, perhaps the really aggressive Hallmark card writer could imagine a card saying "Thanks for all you do, on both Parents Days", maybe with an appropriately feminine color scheme. And then they could get around to doing the same thing the next year for single fathers; SC suggests they might want to try introducing some Mother's Day cards featuring power tools, and see if it works.

On an unrelated note, those of you looking for "Huntington contra Nunberg" will have to wait until tomorrow, while SC takes the rest of his Father-in-Potential Day off.

June 10, 2004

Humor, truth, and libel

Although there hasn't been much reason to mention it, SC is a big fan of the Detroit Red Wings. Those of you immediately suspecting gravy train/bandwagon-jumping tendencies should be aware that this has been true for some 14-15 years, and that your host in fact was in a fairly serious accident in 1999 as a result of driving from San Diego to Los Angeles to watch them play the Kings in the playoffs.

All this is by way of explaining how it is that SC occasionally reads the Detroit News (also an invaluable resource if you're an automotive buff), and so he was quite amused by this story regarding comedian Jimmy Kimmel. It seems that Mr. Kimmel said something which SC regards as a noncontroversial statement of fact:

“They're going to burn the city of Detroit down if the Pistons win,” Kimmel said during halftime of Game 2, using his customary snarky tone.

ABC broadcaster and Ann Arbor resident Mike Tirico objected immediately, saying, “Hey, hey, hey, be careful. That's my home state.”

Kimmel looked a bit stunned at Tirico's objections, then backtracked a bit by saying analyst Tom Tolbert's eye-popping plaid suit should instead be burned.

Needless to say, fans were outraged. The paper cites a local talk show host as follows:

Callers to Baligian's show were mad that Kimmel slammed Detroit without cause.

“Caller after caller, and people who were e-mailing me, were mad because what Kimmel said is simply untrue,” Baligian said. “It's not true. Detroit does not burn when teams win. It's a lie, and people are mad that somebody would say something so terrible about Detroit on national TV.

Unfortunately for the sports fans of Detroit, other people have long memories, and in spite of the city's relative calm after hockey championships (SC maintains that hockey fans are actually the most civil), the story was quite different after the Tigers' 1984 World Series win (which SC took particularly hard, having possessed one of the innumerable "Mow Down Motown" T-shirts then adorning San Diego). In spite of the fact that the web didn't exist even in anyone's mind for another half-decade, Google returns 260+ hits for "+1984 +"World Series" +Tigers +Detroit +riot". Among the evidence, a column from the Detroit News just two days ago which states:

On a more unfortunate note, Los Angeles picked up on the worst of Detroit traditions after the Lakers won NBA championships in 2000, 2001 and 2002. The post-game riot was originally made famous by the celebration that followed the Detroit Tigers’ World Series championship in 1984.

Legally, proving that a statement is true is a valid defense against a libel suit. Mr. Kimmel might be in some trouble here, on account of the fact that, while considerable fire damage accompanied the riots, the statement "the city of Detroit burned down" is at least arguably false, depending on how one defines "burn down" (does it have to be complete?). It's not clear that anybody's actually going to sue Jimmy Kimmel, but the libel accusation is just a legal formalization of the claim that a given proposition is a lie.

SC applauds the instinct of people to defend their home, and it would be inappropriate to expect that Detroit residents should flood the phone lines instead with self-criticism. But your host nevertheless has to question how one can simultaneously think both 1) a riot including much fire damage occurred in Detroit as a consequence of a sporting victory, and 2) anyone asserting (1) is a liar. It's a paradoxical mental state, and SC doesn't envy the people straining to maintain it.

[Edited at 1:24 p.m. on 6/10/04 to correct typo.]

April 28, 2004

An interesting epistemic problem

A tragic story from Minnesota this morning, about a group of teenagers who died from carbon monoxide poisioning while exploring a cave near St. Paul. Apparently, when people start fires in the caves -- which this group did not do -- carbon monoxide builds up in the cave's atmosphere, and leaks out slowly at best.

It would seem desirable, given the peculiarly poor ventilation of these caves, to simply seal them off. In dismissing that possibility, the local fire chief said:

"There are entrances and exits that we don't even know of."

That's an interesting statement to make about an "known unknown" (we discussed the related phenomena of "unknown unknowns' previously here and here). It's a stronger claim than "there are probably entrances and exits that we don't even know about".

Your host finds himself quite confused about how to properly analyze the truth conditions of such a statement, in either probabilistic or "absolute" form. On the one hand, the statement strikes him as empirically unverifiable -- even if one went out in search of previously unknown entrances, and found a few, both versions would still be open to further testing.

And yet, it also strikes SC as intuitively erroneous to dismiss the claim as meaningless. It's quite compatible with a possible-worlds interpretation along the lines of "in all possible worlds, there exist entrances not in our model of each world". The probabilisitc version might even be tautologically true, softening the truth conditions to "there exists a possible world with entrances not in our model of this world" (although, as connoisseurs of possible worlds know, in principle no such world must necessarily exist).

Perhaps "unknown unknowns" and their half-known cousins are really just the rest of the world catching up to the philosophical discussion of counterfactuals. A brief search along these lines turned up an abstract that points toward a solution. It might be the case that positive statements about "unknown unknowns" are really just ceteris paribus claims with the necessary operators being implicit. If the fire chief's statement is rephrased as "all other things being equal, there are entrances/exits we don't know about", then we can more readily imagine how things might not be equal -- more time spent searching for entrances, maybe a little infrared imaging, maybe a couple tons of dynamite to get rid of the caves (note that SC isn't advocating that, it's just one way that our knowledge of the caves' condition might change). On the other hand, as Alice Drewery points out in the abstract linked above, we can judge claims like "there are entrances we don't know about" without knowing the full list of things that might change our belief. Perhaps it's not so much important that we have ideas about what sorts of things we don't know, as that we know what might go wrong with the things we do.

April 26, 2004

Illocutionary force minus the force

As your host has mentioned on a number of occasions before, he relies on taking a couple of trains to commute to his office. Frequently, in the course of waiting to change trains in the morning, he stops at the Burger King in the station to get a carton of orange juice, and if he's feeling hungry, an order of BK's not-at-all-good-for-you french toast sticks.. This morning, as with many others, the same dialogue played out:

SC: One orange juice and an order of french toast sticks, please.
Cashier: For here, or to go?
SC: For here, please.
Cashier: OK, your total is (X). [exchange of money, passage of time]
Cashier: [puts order in bag] Here you go.
SC: [shrugs]

It's not that the store doesn't have any trays. It's not even the case that the cashier has a pathological aversion to handling the trays. It's just that in this particular context, the question "for here or to go?" is only part of a script (not necessarily even one mandated by the restaurant, just socially). It's more efficient for the cashier to simply put everything into bags, which lets her move faster than trying to pay attention to what the customers actually asked her to do with each order. If she didn't go through the motion of asking, though, customers might be more annoyed by the disruption of the routine than they are by the matter of whether or not their request was actually acted on.

This same thing happens in other commercial contexts. In many grocery stores, when the bagger starts handling your order, they'll just start putting things in plastic bags, and mutter "plastic OK?", hoping that the customer will just ignore the usual dialogue, again because plastic bags are much more convenient for the task at hand. Occasionally, SC is moved to enough peevishness to wait for them to put in a few items and then say, "I'll take paper, thanks", but only if the bagger is acting especially lazy.

Even in social contexts where people are speaking as equals, and not as customer/employee or some other formal relationship, there are plenty of exchanges where it's a mistake to respond to a request literally. When somebody says "how are you?", the last thing they actually want to hear about is that you put in 15 hours at the office yesterday and your goldfish just died ([these are not actually biographical details of SC, at least not recently -- ed.]). They want to hear "fine", "good" or some other short, positive response. Of course, the degree to which this is true varies by relationships -- the people SC happens to see on the train platform each morning are more likely to prefer a short, information-free response than Mrs. SC.

So the next time you feel like nobody's listening, just remember -- they're only doing it because they're trying to serve you. Or at least it's nice to tell yourself so.

April 08, 2004

Plausible but not logical

As part of SC's new life as a marketer, he has to pay careful attention to the wording of proposal solicitations. It's important to understand what your prospective customer is actually asking for, which may not always be what it sounds like they're asking for. But sometimes, it's best not to take their requests too seriously.

Consider this passage from a current DARPA solicitation:

How can plausible but not logically valid reasoning be used in a consistent and pragmatic way to get to reasonable conclusions? Can we build a real-world-scale cognitive system that has a principled foundation for its representation and reasoning capabilities?

It's hard to determine what coherent meaning ought to be assigned to this statement. Intuitively, your host wants to agree that there is such a thing as "plausible but not logically valid reasoning" -- if he says, "It's a sunny day out; therefore, I'll spend the day in front of my computer", it ought to be a reasonable inference to anyone who knows he's a geek. But it's hardly logically valid, unless we're starting from a world model including the axioms: "If it's sunny out, vampires will die"; "Spending the day in front of a computer will prevent an individual from being exposed to sunlight"; "SC is an individual"; and my personal favorite, "SC is a vampire" ([the only evidence for that is your screwy sleeping schedule, bat-boy -- ed.]).

It's not at all clear to SC, though, that "plausible but not valid" meshes well with "consistent", at least not where mechanical reasoning processes are involved (although the hardcore philosophical determinist will reply that this just means the world model isn't sophisticated enough). Humans make statements that at least look like inferences all the time, but which aren't really logically coherent. That's what the DARPA folks are trying to get at, but SC's not sure how desirable it really is to have machines emulate that sort of behavior, especially not if they're making life-and-death decisions. Taken as a whole, the statement above reads: "Make us a principled system that makes unprincipled conclusions". Brilliant!

March 25, 2004

Noam Chomsky's blog

Brian Weatherson passes along news that a blog penned by someone named Noam Chomsky has recently started up. Brian expresses surprise, though, that the author refers to himself by full name; since SC likes the way he puts it, we'll quote:

I wouldn't have guessed that Noam Chomsky calls Noam Chomsky "Noam Chomsky", but if it's good enough for Rickey Henderson I guess it's good enough for the Noam.

SC is not at all convinced, though, that the statement "Noam Chomsky has started a blog" is true. Although the signature after each post says "Posted by Noam Chomsky", your host thinks the opening post indicates something rather different:

This blog will include brief comments on diverse topics of concern in our time. They will sometimes come from the ZNet sustainer forum system where Noam interacts through a forum of his own, sometimes from direct submissions, sometimes culled from mail and other outlets -- always from Noam Chomsky.

In other words, the content will always be statements made by or articles written by Prof. Chomsky. However, if by "Noam Chomsky's blog", we mean some entity which is a member of a set defined as "a web publication where Noam Chomsky personally has sat down at a computer and used the Movable Type interface to post an article", then SC submits that the set is empty and the phrase "Noam Chomsky's blog" refers to an entity much like "Pegasus" and "the present King of France". While your host knows a thing or two about writing in the third person about himself, he reads the quote from the blog to mean that it's a clipping service run by a third party (apparently Z Magazine) with Chomsky's endorsement.

March 14, 2004

Flexible designators

Over the weekend, the invaluable Arts & Letters Daily publicized a story by a woman who apparently was assigned a cell-phone number previously belonging to comedian Chris Rock. Predictably, she couldn't resist the urge to commit a little mischief when famous people called; SC loves this story:

LAURA: [Curious, and ready to tackle the unknown] Hello?

CALLER: Is Chris there?

LAURA: [Inquires politely] Who's calling?

CALLER: It's Spike.

LAURA: [Mischievously inquisitive] From...?

CALLER: [Blurts out, in an annoyed tone] It's Spike Lee.

LAURA: [Speechless, stunned, mouth frozen open. Guess wasn't quite ready enough to "tackle the unknown." Takes longer than usual to respond, and when does, does so very slowly] Uh... well... actually... you have the, uh... wrong number.

Linguistically, the story raises a couple of interesting questions. Most of the vignettes feature people calling on behalf of someone else, like so:

LAURA: Hello?

CALLER: Hi, I'm calling from Jerry Seinfeld's office. Jerry would like to get two tickets for Chris' show in L.A. this weekend. Would that be possible?

Clearly, the caller wasn't actually expecting to get Mr. Rock (SC feels a bit like the New York Times, famously referring to Meat Loaf as "Mr. Loaf"), or he would have reacted with surprise to hear a woman's voice. So when the author says that she was assigned "Chris Rock's cell phone number", it's ambiguous between two interpretations: 1) a number belonging to someone in Chris Rock's entourage, or 2) Chris Rock's personal cell phone. Not being fortunate enough to be acquainted with Cell Phone Etiquette of the Rich and Famous, your host cannot reasonably speculate on how often callers expect that Mr. Rock personally answers "his" cell phone.

But there's a specific case which your host finds especially interesting:

LAURA: Hello?

CALLER: Hi, is Chris there?

LAURA: Who's calling, please?

CALLER: It's Jack Nicholson.

In the next millisecond, frozen in time, I nearly dropped dead, but then I thought, "Wait a minute." Adam Sandler totally sounded like Adam Sandler. I quickly put two and two together here, and they didn't add up. With a hefty dose of skepticism, I boldly commented, "This doesn't sound like Jack Nicholson." The caller replied, "I'm his assistant."

Previously, we've talked about rigid designators, which are explicitly referential expressions. Although for most people's names, we know that there are actually plenty of people that the name could refer to, celebrity names surely must approach rigid designator status in most people's mental models. When one says "Madonna", it's far more likely that they're referring to Louise Veronica Ciccone than to anyone else. Similarly, your host suspects that this person is the usual referent of "Jack Nicholson".

So, strictly speaking, "It's Jack Nicholson" was a false reply to the question of who's calling. One response might be that in fact, the caller's statement was merely shorthand for "it's a designated agent of Jack Nicholson". But such talk is: 1) cumbersome, and 2) reflective of a more sophisticated philosophy of language than the Hollywood crowd has any real interest in/need for. Perhaps, then, in a social circle where the names are all understood to uniquely refer to particular individuals, we can analyze names as being rigid designators which can be assigned by an act of the individuals they refer to. It would be absolutely disastrous for credit card companies to accept this sort of behavior in discourse with hundreds of thousands of clients, but when the social group is a little more intimate, flexible designation makes perfectly good sense.

March 11, 2004

Speaking the language of terror

By now, the news of the bombing in Madrid is common. As in the first day or so after September 11th, the question of responsibility is open, and subject to considerable speculation.

Geoff Pullum, writing about a discussion on the BBC World Service, quotes a Prof. Halliday of the London School of Economics as saying "For a start they speak a language that no one can learn." This comes in reference to the Basques, an ethnic group living near the French/Spanish border. The language is famous for being wholly unrelated to anything in geographic proximity (or elsewhere, really) which would account for its origins. Prof. Pullum explodes the notion that "no one" who isn't a Basque can learn the language. But Prof. Halliday ought to have been vaguely embarrassed to even have originally put it like that.

SC assumes a logical form for the statement which is something like (using A for the logical "all" operator):

AxAy( (Person(y) & Is_Basque(y) & Speaks_Basque(y)) & (Person(x) & !(Speaks_Basque(x))) )

If we wanted to be really rigorous about it, we could add terms making explicit the relationship between speaking Basque and learning it, but SC will assume that anyone who speaks Basque managed to successfully learn it. Otherwise, I've tried to capture the part that Basques speak Basque with the first group of terms, and that "no one" speaks Basque with the second group. Regardless, assuming our variables range over the set of individuals, some x and y values will be identical -- i.e., Basques -- and the statement is a contradiction. Not that one really needs to have written that out in logical terms to have seen the problem.

It's not at all clear, though, that ETA, a Basque terrorist group, is responsible for the attack. Semioticians, the sort of people who are interested in figuring out why particular language items carry symbolic meaning (as opposed to semanticists, who worry about how), will be quick to note that today is the 11th of the month. And the evidence is somewhat mixed in regard to possible suspects. Investigations of this sort of thing ought to turn on real evidence, not the speculations of linguists, but it strikes SC as likely that whoever is responsible didn't pick the date accidentally.

March 05, 2004

What are the odds of that?

Two weeks ago, negation was the hot topic of linguistics blogs. SC's still out looking for examples (not too consciously, just taking notes when he finds them) and came across an interesting one this morning.

From Rob Neyer, SC's favorite baseball writer:

"Oh, and this doesn't really fall under the purview of today's column, but what are the odds against Esteban Loaiza winning 21 games and posting a sub-3.00 ERA again? Pretty long, I would guess."

In the context of the column, it's clear that Mr. Neyer's intent is to express the opinion that Esteban Loaiza will revert to his usual unfortunate state of Esteban Loaiza-hood, and put up 10 wins and a 4.5-ish ERA ([which will be 10 more wins, a real number of runs allowed, and several million dollars more than you'll be seeing, pal -- ed.]). However, as written, Mr. Neyer has claimed that it would be unlikely for Mr. Loaiza not to repeat the performance, as the odds against him are what are being called "long".

So the SC research staff checked the relative frequency of some "odds" phrases in Google:

"odds of *": 860,000 hits
"odds for *": 765,000 hits
"odds against *": 94,700 hits

A cursory examination of the results suggests that "odds of" is pretty much used correctly, as a description of the probability of something (albeit often metaphorically, and not with actual calculations). "Odds for" is similar to "odds of". "Odds against", though, has two vaguely opposing uses:

From a recent MSN Money article: "Odds against an audit are good"
From the Houston Chronicle, June 28, 2003: "Rudy T's doctors like odds against cancer".
From CNN, September 5, 2003: "Olympic swimmer defies odds against diabetes"

In the first case, the parse is something like NP[odds against an audit] VP[are good], where the meaning of odds is exactly like the for/of cases discussed previously, and the odds in question are low. However, for the latter two, it's something more like S'[Rudy T's doctors like odds] PP[against diabetes], and the expression is meant to signify that the odds being discussed are highly likely, and favor the overcoming of whatever's against them. Here, the negation isn't carried in the lexical meaning of "odds"; it's buried in the syntax.

Returning to Rob Neyer's sentence, it's fairly clear that what happened is a confusion between the two parses because of subject-auxiliary inversion. "What are the odds against X?" looks, on the surface, like the second parse of "odds". Had he first written "The odds against Esteban Loaiza repeating his 2003 performance are long", it would have been immediately obvious what his mistake was, and he would have instead written: "What are the odds against Esteban Loaiza repeating? Pretty good, I'd say." In case Mr. Neyer should happen to come across this post, I only point it out because I know he cares about getting these things right.

February 27, 2004

Analyzing a crazy woman

Yesterday, SC wrote about the frustration of watching someone violate the terms of using an express lane in a particularly egregious manner. This morning, he realized he has a better analysis to offer.

Recalling that the woman said "I only have ten items", when she had more like double that, SC originally wrote "it's difficult for SC to decide if her statements in fact qualified as lies". This was due to the fact that she could have meant "I only have ten items in each order". Your host has changed his mind -- she was a liar.

The basis for deciding this is not mere peevishness ([although you'll have a hard time convincing anyone who knows you, pal -- ed.]). It's the fact that she said "only". Larry Horn provides this definition: "The semantics of only says this: it asserts that no proposition from the set of relevant contrasts C other than the one expressed by its sister sentence [alpha] is true". I'm not sure if he's attributing that formulation to Kai von Fintel based on the way the abstract is posted; if so, consider the credit assigned. Regardless, we can represent her statement as a conjunction of logical statements:

Ex((Rude woman(x) & (Possesses_10_items(x)) & ~(Possesses_11_items(x) & ~(Possesses_12_items(x)...)

SC is sure the speaker who prompted this posting would disagree with the formulation's first term, but SC's not terribly interested in hearing her complaints right now.

Given that her statement was "I only have 10 items", the above formula is only true if there exists a person x who is a rude woman and possesses exactly 10 items. If she possesses 11 items, then the term Possesses_10_items remains true, but now Possesses_11_items(x) is also true, and so its negation -- the actual term in the formula -- is false. Since "and" statements are only true when all of their conjuncts are true, her statement was false, and she was a liar.

But wait! What about the pragmatic claim that she might have actually meant "I only have 10 items in each order"? Let's resolve that by recourse to Grice's maxims. Readers encountering them for the first time may think that Grice was lecturing about how people should speak. In fact, Paul Grice offered these as his formulation of the rules which appear to guide our approach to conversations. They're an approximation of our expectations about how other people communicate with us, as well as how we try to formulate our own statements (although some people disagree vehemently). SC thinks they're a good starting point.

So how might she have violated Grice's maxims? The issue of quantity seems to vindicate her, as Grice's conditions for that could be construed by a malicious speaker as "say no more than the minimum necessary to establish your position". Since she said something which was true under a very strained interpretation, SC will give her a reluctant pass here, even though he thinks she's a weasel. The issue of quality is a toss-up; on the surface, she made a clearly false statement, but we're analyzing the "in each order" interpretation here. She could truthfully claim that she doesn't believe the "in each order" statement to be false, and SC would have to agree (again, while swearing furiously). The evidence is right there in her cart.

As for relevance, well, she was definitely addressing the situation; namely, the question of whether she had too many groceries to be checking out in the express lane. So she's clean. But then we get to manner. Did she avoid obscurity? Absolutely not. Did she avoid ambiguity? Nope -- she only gets a pass on the first two maxims by assuming that she was very ambiguous. Was she brief? Well, yes. Last, but not least, orderly? Hmmm...this would-be parliamentarian rules her..."Out of line!".

So she strikes out on both truth-conditional and pragmatic grounds, and we can now conclude that she's a liar. However, due to the fact that SC has now been obssessing about this for at least 15 hours, he'll conclude with an apocryphal story from which you can draw your own conclusions about his mental health.

Two monks who have sworn never to touch a woman are walking by a river. They come across a woman who is desperate to cross, but cannot swim. One of the monks promptly picks her up on his shoulders, and carries her across. The other monk says nothing. Hours later, though, he can no longer restrain himself, and says, "How could you touch that woman? You've broken your vows!". The monk who carried the woman replies, "I put her down by the side of the river. You've been carrying her ever since."

February 25, 2004

Can a stock bid for a company?

Like most people, your host spends a lot of time worrying about a fellow named Mort. Mort Gage. Mr. Gage discovered SC's checking account some time ago, and removes a large part of it with disturbing regularity. Judging by the number of people who appear to be similarly afflicted, Mr. Gage must be a very wealthy man by now, and so SC wishes he'd decide he's satisfied and just go away.

Since, however, the shadowy Mr. Gage has not indicated plans to do so anytime soon, SC finds that he spends a lot of time attempting to compensate for the ravages of Mr. Gage's attentions. As a consequence of this, your host is well acquainted with the writings of various members of the financial and business press, including the Wall Street Journal and TheStreet.com.

This morning, writing for TheStreet.com in regard to Comcast's abortive attempt to buy Disney, hedge fund manager Doug Kass commented that:

Unlike my view, with Microsoft playing an integral role, Reif reminds us that CMCSK can call upon investments in Time Warner, Liberty Media, Time Warner Cable and multiple cable partnerships in order to raise the cash component.

Over a number of years, SC has noticed that people working in finance will frequently refer to companies by their stock symbols. This is not an unreasonable shorthand, so long as it's unambiguous. But in Comcast's case, CMCSK is not identical with stock representing ownership of the entire company! In fact, there are two different stocks held by different owners of Comcast, as well as a third stock representing additional assets owned by Comcast as a parent company. And there are other companies where the ownership is even more fine-grained, like General Motors, which has no less than 12 distinct stocks representing ownership of different parts of the company, or different classes of shareholder rights for the same part of the company.

So when Mr. Kass says that CMCSK is pursuing ownership of Disney, a reader with SC's general disposition might conclude that in fact the decision is not unanimous among all owners of Comcast, and that holders of CMCSA disagree. Of course, such a reading is not the intent of Mr. Kass, nor of anyone else SC can recall writing this way. If the sentence actually was meant to be read as referring exclusively to only some portion of the company, it would become practically impossible to read financial writing -- does part of Comcast intend to acquire all of Disney? Or does part of Comcast intend to acquire one (or more) of the 7 publicly traded parts of Disney? For any company of significant size, the number of potential interpretations of the sentence would become proportional to the number of stocks representing interest in the company.

Of course, given the substantial amounts to be made in merger-and-acquisition banking, to say nothing of the legal fees, some enterprising banker or lawyer might come across this discussion of agonizing hairsplitting and say, "What a great idea!". Rather than disabuse them of the notion, SC will simply point out that valid contact information is available on this website in order to discuss SC's cut of the resulting deals.

February 23, 2004

The joys of the semantic method

Kai von Fintel has thoughtfully provided a link to an interesting and worthwhile paper for those interested in the polarity of yes/no questions. It's a tad dense, so SC won't read it all for a day or two, but for those who aren't going to be reading the paper at all, your host would like to share the sorts of examples, found in this paper, that make doing semantics fun:

Scenario: S hates both Pat and Jane. The prospect of an excursion without them pleases S. S does not have any previous belief about whether either of them is coming or not. A: Pat is not coming. S: Great! Is Jane not coming (either)? That would be the best!!! S’: # Great! Isn’t Jane coming (either)? That would be the best!!!
Scenario: Michael has been upset at Sue since yesterday’s meeting. The speaker is wondering how this could have been avoided. The speaker has no belief about what Sue should or should not have done. A: Michael has not been happy with Sue since yesterday’s meeting. S: Should she not have talked to him (at the meeting) / (yet)? S’: # Should she not have talked to him already?

If the reader is beginning to suspect that semanticists are actually closet romance novelists -- YOU'RE RIGHT!

More seriously, this is how semantics usually pro