Ask the Cashier

I’ve been teaching academic writing at Ohio State University’s ESL Composition Program this quarter (hence the sparse blogging). After class one day last week, I stopped at a coffee shop that was right inside the building to get a Coke to go with my lunch. As I handed my money to the cashier, I noticed the sign on the register:

Ask him if I’d like a receipt? What was I supposed to say, something like “Uh, would I like a receipt?”

What kind of weird question was that? Then, to use a phrase I’ve used before, like a Necker cube flipping inside out, the phrase shifted to match its meaning. I’d been parsing it like in the diagram on the left, when really it was intended to be read like the one on the right:

In the diagram on the left, the subordinate clause if you want a receipt is a complement to the verb, just like the cashier. The role the cashier plays is the person who gets asked something, and the subordinate clause has the role of whatever question is to be asked. You can parse it this way because if is something like an honorary wh word, so subordinate clauses it heads up can go with verbs like ask or wonder: I asked {what he was doing / where they were going / whether there was any pizza left / if we were free to go}.

In the diagram on the right, on the other hand, the verb ask only has one complement: the cashier. The question that gets asked goes unspoken, and you have to get it from the context, the same as you would in sentences like Ask mom. The if-clause, meanwhile, modifies the whole thing, saying under which conditions you should ask the cashier whatever question you have. We can parse it this way because if can also be used in its regular old “if” conditional sense.

So the intended meaning was this: If the circumstance arise in which you want a receipt, ask the cashier something. From context, the most obvious question is, “May I have a receipt?”

Meanwhile, the food court in the new student union has it right:

Keep Your Promises to Yourself

I got an email from my gym, advertising some online system they have for tracking calories, planning your exercises, and who knows what else. In the message, they said using this system would help you with “keeping your promises to yourself.” That’s a great idea for what to do with my promises. If I keep them to myself, no one else will know about them, and there will be fewer embarrassing questions about whether I’ve accomplished them.

We Do What We Do

And now for a semantics-focused post on National Grammar Day. Actually, my syntax-focused post was drifting into semantics territory, when I talked about the likely and unlikely intended meanings for the sentence I was talking about. That’s OK: Although this post is mostly about an ambiguity in a sign, I’m going to use some syntactic diagrams. It’s hard to separate the two at times.

So, the sign I have in mind is one I saw for a mortgage broker. It had these encouraging words:

We do what we do to help people realize their dreams of home ownership.

The first time I read it, my reaction was, “Well, no kidding!” The sign seemed to be saying to me, “You know those things we do to help people realize their dreams of home ownership? Well, we do them.” Then the re-parse came through, and I arrived at the intended meaning: “You know those things we do? Well, we do them to help people realize their dreams of home ownership.”

The sentence was in perfectly good Standard English grammar, but there were two possible ways to structure it (i.e. two parses). The way that I happened upon first was the one diagrammed below. The tents above the words show how they clump together into phrases. Down at the bottom, the purpose infinitival phrase to help people… modifies the verb phrase (VP) do 1. (The 1 is a stand-in for the what that appears at the front of the fused relative clause.) The diagram shows this modifying relationship by having the to help people tent and the do 1 tent coming together under a bigger VP tent for do to help people. Then whole phrase what we do to help people… is the direct object of the do in the VP higher up in the diagram.

The purpose infinitival attaches to the smaller VP

The more sensible parse is this next one. Here, the infinitival phrase modifies not the little VP do 1, but the big VP do what we do. The diagram shows this by having the tent for the do what we do VP and the tent for to help people coming together under a tent for a big VP that houses them both: do what we do to help people.

The infinitival phrase attaches up high.

Enjoy the rest of National Grammar Day, and come back soon! If you have questions about grammar, send them to me at nealwhitman@yahoo.com, or address them to @LiteralMinded on Twitter. Some of them may find their way into future posts.

Before It Starts

Regular reader Adrian Morgan (you know him as the Flesh-Eating Dragon of The Outer Hoard) wrote to me about playing a game called MindTrap with some members of his family. He said:

During this game one player read me a puzzle in which the protagonist can tell people the score of any football game before it starts. The puzzle was to figure out how.

I asked whether the answer involves an ambiguity in the English language. The other player replied that it did not. I remarked that this rules out the answer I was thinking of, that the score of any football game before it starts is always 0-0 because the game hasn’t started yet. The other player replied that this is the correct answer, but that he would not have said it involves an ambiguity in the language.

Adrian, of course, was right. This is an attachment ambiguity involving the phrase before it starts. I’m going to follow Geoff Pullum’s analysis of subordinating conjunctions like before, and classify them as prepositions that can take either noun phrases or sentences as their objects. So before it starts will be a prepositional phrase. Under the “how is that possible?” reading, it attaches up high, to the entire verb phrase tell people the score of any football game, as in this diagram here:

The How-is-that-possible? parse

Under the “who cares?” reading, it attaches down low, modifying the nominal score of any game. (I’ve accidentally labeled score of any game as N instead of Nom, but I’m not going to redo it now.)

The Who-cares? reading

Adrian continues:

To me, it’s a mystery how someone could deny that this involves an ambiguity in the language, when to my way of thinking it involves a rather prototypical example of one. What do people think “ambiguity in the language” means? I can only suggest that the incident supports Geoff Pullum’s observation that most people think of language as a big bag of words — hence, to them, an ambiguity in the language can only mean an ambiguity in a word.

[I]t’s interesting as an example of how most people are not accustomed to thinking about language in the way that a linguist would: they think of it in terms of vocabulary rather than syntax. Later in the game there was another puzzle of which [the other player] remarked that he would say it involved an ambiguity in the language: but in that case it was a simple case of polysemy, which reinforces my interpretation of the incident.

Family Owned and Imitated

A tire shop that opened a year or two ago puts funny messages on its marquee. They’re so funny that I can’t seem to recall any of them right now, except of course for the one I’m going to tell you about now. It said:

Family Owned and Imitated

Family owned: So a family, let’s call them the Smiths, owns this business. Family imitated: A family (presumably the Smiths again) also imitates this business. The Smiths imitate their own business? How is that possible? Maybe it’s like that that Greek family I read about. They ran a chocolate shop in nearby Granville for years, but then had a falling out, so that there are now two chocolate shops, run by two branches of the same family, located within two blocks of each other in downtown Granville, each claiming to possess the truest version of the family’s recipes for chocolate confections.

But never mind that. I’m pretty sure all they’re saying is that this business is family-owned, and that it’s imitated. This reading makes sense: Lots of businesses say that they’re imitated, usually before a warning that they’re never equalled or duplicated. In this reading, the coordinated elements are family-owned and imitated, as illustrated on the left.

To get the reading that leads you to imagine a rift in the family, you have to parse it with just owned and imitated as the coordinated elements, with family applying to both, as illustrated on the right. So why did I want to parse it this way, anyway, since it gives the weird and unlikely reading?

It’s at least partly because of the common collocation that the sign is harking to: Family Owned and Operated (or sometimes, family owned and run). In those phrases, family is clearly intended to form a compound with both owned and operated, as in the diagram. After all, who’d want to say that a family owns some particular place of business, and that (get this) someone operates it? If it’s open at all, the latter claim is obvious, and stating it violates the principle of Relevance. Only if it’s taken to mean “family-operated” does the statement say something useful: The fact that some place is run by the family that owns it might not be obvious to the casual observer.

By using this recognizable phrase as their point of departure, they primed me to parse Family Owned and Imitated in the stupid way. Now that I think about it, though, family owned and operated could be useful as a deceptively ambiguous phrase, for a family that has recently contracted out the operation of its family business but doesn’t want to change the wording in their advertisements. I wonder if that’s been done. Do any of you know of businesses that advertise that they’re “family owned and operated”, and are operated by someone other than the family?

: : : : : : : : : : : :

Cute as a Button in the Eye

We went to see Coraline last week. It’s an OK movie; it does a good job at setting a creepy mood, but even by its own internal logic, it doesn’t quite make sense (unlike, for example, Monsters, Inc. or the Toy Story movies). One of the best scenes has Coraline’s “Other Father” in a parallel world improvising a song in her honor. It’s an upbeat, catchy melody, which I learned in the end credits was written and performed by They Might Be Giants (you know, the guys who did “(You’re Not the) Boss of Me“). Here, see for yourself:

If you’ve seen the movie, or even just the previews (or read the graphic novel the movie is based on), you know that one of the unsettling details of the parallel world that Coraline visits is that all its inhabitants have buttons instead of eyes. They Might Be Giants have cleverly alluded to this fact in the line that goes:

She’s cute as a button in the eyes of everyone who ever laid their eyes on Coraline.

I love the syntactic ambiguity here. More specifically, it’s an attachment ambiguity. In the normal reading, the prepositional phrase in the eyes of everyone who ever laid their eyes on Coraline functions as a sentential adverb, modifying the sentence She’s as cute as a button, as shown in the diagram on the right.

However, anyone who has been watching the movie up to this point is well primed to parse the prepositional phrase as modifying the noun button, as illustrated in the diagram on the left. Ordinarily this parse would be unconsciously discarded, in the same way as we’d never even think about parsing Kim disassembled the TV with a flat screen to mean that Kim used a flat screen to disassemble the TV. But in the context of the movie, both parses are salient, and both make sense (as long as you’re willing to stretch the meaning of in to include in place of, or on if you imagine the buttons to be placed on top of the eyes).

The only flaw in the exploited ambiguity is the clash between the singular a button and the plural the eyes. It’s hard, even impossible I’d say, to get a wide-scope reading of the eyes, so that we’re talking about one button for each eye. I keep thinking about a single button that is somehow in (place of) both eyes at once. That’s an extragrammatical correction you just have to grant in the name of artistic license.

: : : : : : : : : : : :

Killing His Wife for the Second Time

The Ridger wrote about this headline in a post I linked to a couple of weeks ago:

Jury Convicts New York Man of Killing Wife for the Second Time

Funny low attachment

Serious high attachment

She identified three different ways the headline could be true. First of all, there’s the low-vs.-high attachment ambiguity. The low-attachment reading is the strange one, suggesting that the man killed his wife, she came back from the dead, and he killed her again. The syntactic structure that goes with this reading is diagrammed on the left, with for the second time modifying just the gerund phrase killing [his] wife.

The high-attachment reading is the one in which the man is convicted twice for murdering his wife. (The Ridger explains: According to the news article, he was convicted once, but the first verdict was overturned.) It is diagrammed on the right, with for the second time modifying the larger VP convicts a man of killing his wife.

On top of this ambiguity is the unresolved question of whether we’re talking one wife or two. If the man remarried after murdering his first wife, then the low-attaching reading is not nonsensical after all. This same goes for the high-attaching, two-convictions reading: He could have been convicted once for murdering wife #1, and convicted a second time for murdering wife #2.

When I linked to The Ridger’s post, I called this a four-way ambiguity, but in the comments I mentioned that I didn’t think the question of how many wives was a true ambiguity. The Ridger doesn’t think so, either; her opinion is that “people with too much time on their hands … pretend it is.” So why isn’t it a true ambiguity?

I’ll focus on just the two-killings reading to illustrate. To make it easier, I’ll just go with X killed his wife for the second time. Translating it into quasi-logical notation, we can write it as

exist times T1, T2, such that

1. at T1, exists y such that wife-of(X, y) & kill(X, y)
2. at T2, exists z such that wife-of(X, z) & kill(X, z)

For this to be true, all that is necessary is that there be an individual y at T1 who is X’s wife, and an individual z at T2 who is X’s wife. y and z could be the same person, but they don’t have to be. That’s not an ambiguity; it’s different situations that would make an utterance true, just like I ate a sandwich could be true whether I ate a peanut butter and jelly sandwich or grilled chicken sandwich with lettuce, tomato, and onion. Semanticists refer to this kind of non-ambiguity as vagueness.

You know, now that I’ve explained all that, a different kind of ambiguity has occurred to me. The sentence could also be true in a world where it’s OK to kill your spouse once, but illegal to do it more than once, and the jury convicted the man for killing one wife too many. I’m not sure what to blame for that ambiguity, but I think it’s the same kind of thing that goes on in a sentence like

He got a speeding ticket for going less than 100mph.

The most easily accessible reading is the weird one, in which it is illegal to go less than 100mph. But I’ve also read a sentence like this in a news article describing a situation like this one: The speed limit was 65; the driver was doing 85; 85 is less than 100; therefore, the driver got a ticket for going less than 100 miles per hour. Further thoughts on this kind of ambiguity will have to wait until a later post.

: : : : : : : : : : : :

Like a Racehorse

DGM of Sunny Side Up has written about her encounters with an evil nurse at the ER. At one point in the story, she tells us,

I had to pee like a racehorse.

Now DGM is a bit sensitive about me putting stuff she writes under the linguistic magnifying glass, so let me say that I wasn’t planning on commenting on this sentence, any more than I’d comment on most other idioms I come across. But that was before one commenter asked: