Wild and Crazy WTF

About this time of year back in 1991, I was reading Henry Beard’s Latin for All Occasions, a collection of sentences such as “I didn’t expect you home so soon” and “The waitress drew a smiley face on my check” into Latin. It wasn’t funny enough to buy for myself, but that was OK; I had bought it to give to my cousin for Christmas. For that reason I was being extra careful as I turned the pages, so my cousin would never know I had turned his gift into a secondhand one. For the same reason, I made Glen promise he wouldn’t tell what I had done when I shared a few of the translations with him.

Two weeks later, I unwrapped Glen’s Christmas present to me: a copy of Henry Beard’s Latin for All Occasions. There was a card tucked inside the cover. It read:

Where, When, and How Many?

I saved a section of the newspaper a year ago because there was a sentence in it I wanted to write about. But I didn’t get around to it right away, and it was of topical interest, so I ended up never doing it. It’s topical again, though, so I picked up the article off my desktop where it’s been sitting all that time, and looked at the sentence again:

They’re all violating ordinances that regulate where, when and how many campaign signs can dot the local landscape.

The coordination of where, when, and how many is like a couple of coordinations I wrote about back in 2005. What’s interesting is that campaign signs is doing two jobs.

What and How, More and Faster

Two posts back, I talked about how coordinations like when and what could force verbs that are transitive or intransitive to be both at once, and that coordinations such as more and more often could do the same thing. There’s a similar pair of syntactic structures that will force a double parsing on certain kinds of nouns. The first one is another coordination of wh-words, but this time, instead of a wh-noun and a wh-adverb, it’s a wh-determiner and a wh-adverb.

First, some background. In English, plain old nouns can’t usually function as subjects or objects in a sentence: *Cat came to the door or *I saw dentist today are ungrammatical. You have to put the noun with a determiner, such as a, the, every or your, to form a noun phrase: My cat came to the door; I saw a dentist. Some nouns, however, can act as plain nouns, combining with a determiner to form a noun phrase; or they can go without a determiner and act as noun phrases all by themselves. These are teh mass singular nouns and the plural nouns: (The) slime covered the floor; (some) squirrels keep robbing the bird feeder. With that out of the way, here’s the example I used in a January 2005 post:

Every company has its own idea of what and how information should appear….

What is the wh-determiner; put together with the noun information, it would make the noun phrase what information. But since information is a mass noun, it can also serve as an entire noun phrase without a determiner, which is exactly what it does in the phrase ideas of how information should appear. When what and when are coordinated, then (in an indirect question that doesn’t have subject-auxiliary inversion, with a mass or plural noun functioning as the subject), the noun that follows can be both a plain old noun and a full noun phrase. Here’s another example that I found more recently:

…Tropical Rainfall Measuring Mission (TRMM) satellite, the only space-based instrument that can measure where and how much rain forms deep inside a hurricane.
Laura Allen, “Endangered Robots,” Popular Science, August 2007, p. 65

In this one, the order of wh-adverb and wh-determiner are reversed, and the determiner isn’t just a simple what; it’s the more complex how much.

It turns out that for noun-vs.-noun-phrase ambiguity, just as for transitive-vs.-intransitive ambiguity, both wh-coordinations and the kind of more and… coordinations discussed earlier can flush it out into the open. It happens when the first more is a determiner (instead of a noun phrase like in the other post), and the other word is an adjective (instead of an adverb like in the other post). An example I used in a later comment sa more and faster, with more as a noun phrase and faster as an adverb. But here’s an example with more as a determiner and faster as an adjective:

We need more and faster processors.

Processors is an ordinary noun when it combines with more, but it’s a full noun phrase when it combines with the adjective faster (under the standard analysis in which adjectives form phrases of the same category as the thing they modify).

Those Three Little Words, in 16 Languages

Faithful readers of this blog (and sporadic ones who happened to read just the right posts) know that I’ve been interested in questions that contain more than one wh-word. In the course of doing some research on them, I’ve been having a set of 24 questions translated into as many languages as I can find willing native speakers for. As a result, I can now say “Who read what?” in 16 languages. In fact, I think I will:

1. Chinese (Mandarin): shei du le shenme?
2. Czech: kdo co cetl?
3. English: who read what?
4. Estonian: kes luges mida?
5. German: wer hat was gelesen?
6. Greek: pjos djavase ti
7. Hebrew: mi kara ma
8. Hindi: kisne kyaa paRhaa?
9. Hungarian: ki mit olvasott?
10. Japanese: dare ga nani o yomimasita ka?
11. Korean: nwuka mwuesul īlkessnī?
12. Macedonian: koj čto čital?
13. Russian: kto čto čital?
14. Spanish: quién leyó qué?
15. Tagalog: sino ang nagbása ng anó
16. Vietnamese: ai (đã) đọc gì

I bring this up today because

My Five Minutes on Andrle’s Open Line

Yesterday morning I was listening to Fred Andrle, a local radio talk-show host, who had as his guest Michael Agnes, editor of the Webster’s New World Dictionary. They were taking calls about language, so I decided I’d call in and see what Mr. Agnes’s take was on some of the language questions that interest me. The topic I chose was coordinated wh words, and I spent about five minutes on the air discussing it with Mr. Agnes. You can download the podcast of yesterday’s show by going here and clicking the iTunes button (which is how I did it), or you can “copy and paste this URL into a podcasting tool” (which I don’t know how to do, but maybe you do). There are two podcasts for June 28; my call is in the second one, 28:24 in. Get it while you can–they keep only today’s and yesterday’s shows available for download.

Where Else and What Else?

The teaser for the 11:00 news last night talked about a suspect arrested in a local homicide case. It said:

We’ve got information on where else and what else he’s wanted for.

What else he’s wanted for, fine. Where else he’s wanted for, no. Looking past the non-parallel surface structure, what they must mean is, “Where else he’s wanted, and what else he’s wanted for.”

In more general terms, the coordination goes

[A and B] C D,

but actually means,

[A C] and [B C D],

instead of the strictly parallel

[A C D] and [B C D].

In this case, A = where else, B = what else, C = he’s wanted, and D = for. I’ll add this one to the list.

What and When?

The latest story about an ongoing political scandal here in Ohio led off with this sentence in today’s paper:

For the second time in two weeks, a memo has surfaced that contradicts [claims of] what and when Gov. Bob Taft’s office learned about investment losses at the Ohio Bureau of Workers’ Compensation.
(”Staff: Taft wasn’t told of losses,” Mark Niquette and Joe Hallett, The Columbus Dispatch, 9 Jun 2005, A1.)

To explain why this sentence caught my eye, I need to back up a bit. It has to do with the fact that the verb learn can be used it several different ways, two of which are relevant here. First, learn can take a direct object followed by an about prepositional phrase, as in:

I learned [something] [about the investigation].

If you wanted to make the direct object the topic of a question, you could say:

[What] did you learn [about the investigation]?

Even though learn isn’t followed by its direct object anymore, the what still fills that position, and learn is still being used with its “learn X about Y” frame. It’s as if there is something missing between learn and about, which shows up as the what at the front of the sentence.

Next, learn can be followed by just an about prepositional phrase, without a direct object, as in:

I learned [about the investigation].

If you wanted to, you could modify this sentence with adverbs:

I learned [about the investigation] {today, from the newspaper, …}

These adverbs could also show up as wh-words, like this:

{When, How, …} did you learn [about the investigation]?

So now, even though learn is next to about, like before, there is no missing direct object we need to imagine there. Learn is still being used with its “learn about Y” frame.

These two versions of learn have slightly different meanings. Both involve new knowledge entering someone’s mind, but the first learn refers to both the general field and the specific thing learned, while the second one refers only to the general field and lets the context provide the specific thing learned. Now we can go back to the opening quotation. Expanded out, the coordinated what and when question would be:

1. [what] Taft’s office learned [about investment losses], and

2. when Taft’s office learned [about investment losses]

What I find interesting, then, is that when the what and the when are coordinated in the newspaper quotation, the word learn is uttered only once, and is therefore used with both its “learn X about Y” and its “learn about Y” frames, with their different meanings, at the same time. You can’t decide (in a non-arbitrary way) whether to color the sentence green or blue. Isn’t that weird?

And it’s not just that there’s some rule saying that “Any wh words can be coordinated”; it has to be with a verb that can go with or without a direct object, such as learn. Put in a strictly transitive verb or a strictly intransitive one, or a transitive verb along with its direct object, and the coordination doesn’t work. Even if you didn’t like the what and when coordination with learn, you probably like these even less:

*What and when did get?

*What and when did you sleep?

*What and when did you get the assignment?

It doesn’t work with just any transitive/intransitive verb, either. No matter how hard I’ve tried, I just can’t make Who and when did you shave? mean, “Who did you shave, and when did you shave yourself?”

The What and How of Self-Promotion

I know, I know: You’ve been wanting to read my dissertation, but didn’t want to download it and have to print it out or view it on a computer screen. Well, the wait is over. It has been published by Routledge, and a handsomely bound, hardcover edition of my contribution to the literature can now be yours. It’s available here (discounted), here, or from the publisher itself. It’s tastefully done, too–the f-word appears on only one page (215), and even there it’s in a quoted title.

This seems like a good time to do a post or two about some of the topics I covered in my dissertation. One of them came up just last week in a book I was reading, specifically in this sentence:

Every company has its own idea of what and how information should appear….
Richard Curtis, How to Be Your Own Literary Agent, p. 116.

Is that sentence a little odd for you? It is for me, in the coordination of what and how. I’ll expand out the coordination so that each of these words heads up its own question:

Every company has its own idea of:

• what information should appear
• how information should appear

In the what question, information is used as a noun. You can substitute other nouns for it and still have a syntactically well-formed sentence: what person should appear, what item should appear, etc.

In the how question, however, information draws on its powers as a mass noun (see previous post) to function not as a mere noun, but as a complete noun phrase! If you try to substitute an ordinary noun into this question, it won’t work: *how person should appear, *how item should appear, etc. By contrast, if you substitute a noun phrase for information, it works just fine: how the report should appear, how Kim should appear, etc.

Therefore, in the coordinated what-and-how sentence earlier, information is used simultaneously as a noun and a noun phrase. That’s pretty weird, especially being as how nouns and noun phrases are typically viewed as having different semantic types (predicates and individuals, respectively). Conventional wisdom has been that words (or phrases) can’t be used with more than one semantic type at a time–at least, not outside of puns. So is the quotation from Richard Curtis a mistake, or is it actually generated in his (and maybe other people’s) grammar? Corpus linguistic and experimental research on this kind (and other kinds) of “mixed-wh interrogative” is presented in Chapter 3 of my dissertation. Own it today!

Blinded by the Rules

Arnold Zwicky at Language Log has put his finger on something that has bothered me when I’ve researched various linguistic phenomena that catch my ear. He writes:

Much of my recent research has to do with syntactic variation in English…sometimes on details of constructions that are for the most part uncontroversial, sometimes on phenomena that are very widespread but are condemned by some usage manuals, sometimes on relatively infrequent and largely disregarded phenomena. I seem to have specialized in variation that isn’t tied in any obvious way to the standard extralinguistic factors (geographical region, class, age, sex, race/ethnicity), although a few of the variables are associated with informal style or with speech as opposed to writing.
…
Now, I’m used to having people, especially non-linguists, respond to some of my data through the lens of rules they’ve been taught. Being Blinded By the Rules, I call it. It seems that once you’ve had a generalization about grammar, however spurious, made explicit for you, you can no longer judge language like a normal person; a little learning is a dangerous thing. You may deny that you use some variant — possessive antecedents for pronouns, split infinitives, stranded prepositions, certain types of “dangling modifiers” — when in fact you use it with some frequency. You may make tortured attempts to avoid this variant. You will certainly discredit reports that other people use a variant that you don’t — say, Isis (”The problem is is that I don’t speak that way”), GenXso (”I’m so not going to talk about this”), or themself (”Everybody should get themself a research project”). You’ll be inclined to treat these usages as errors, not as real linguistic variants, that is, parts of somebody’s grammar (maybe your own).

Here’s an example of a “relatively infrequent and largely disregarded” phenomenon that I’ve talked about earlier, and have encountered rule-induced blindness in discussing: the coordination of what and who in a question. As I mentioned in the earlier post, when what and who is the subject of a question, the verb sounds much better to my ear as a singular than a plural: I’d say what and who was found rather than what and who were found. On the other hand, there is the well-known rule that coordinated noun phrases count as plural (barring certain exceptions that I won’t get into here).

Faced with these two facts, I trust the instincts that I’ve gained from a lifetime of speaking English, and conclude that the rule for coordinated noun phrases doesn’t work for wh noun phrases the same as it does for ordinary noun phrases. Or rather, that for at least some of the population of English speakers it didn’t. But when I was describing the phenomenon to someone Blinded By The Rules, he unquestioningly, automatically, ceded authority to the rule on coordinated noun phrases. To be fair, it might have been that this rule really did apply to any and all coordinated noun phrases in BBTR’s grammar, but he didn’t put it that way. Instead, he said, “Well, as far as I’m concerned, John and Marsha is married is just plain wrong!”

Well, of course it is! I never said anything about coordinations like John and Marsha! I was talking about coordinations of wh words like what and who. But in BBTR’s mind, if the rule spoke of coordinated noun phrases, then by golly it meant all of them! Nevermind that the authors of the English grammar books probably never even thought about what happened in such a relatively infrequent situation; the rule’s been written, and there’ll be no changing it now.

Here’s another example, also involving wh questions. This sentence appeared on page F1 of the Columbus Dispatch of July 12:

Digital video recorders and video-on-demand systems give a new generation of viewers greater control over what, when and how they watch television.

As it happens, I was talking about this example with the same person as I mentioned earlier. BBTR’s reaction: “How they watch television… when they watch television… what they watch television… that’s wrong, man!” He had zeroed right in on what had caught my attention: the lack of parallelism in the coordination of what, when, and how. The latter two could combine with they watch television just fine, but the what definitely could not. This kind of coordination is definitely not licensed in his or my grammar.

But BBTR went further: “That sentence makes no sense! It’d have to be what they watch on TV, and when and how they watch it.”

“But wait!” I said. “You just made sense of it right then! So why can’t we say that some people’s grammar lets them do this kind of a coordination, with the kind of interpretation that you just gave it?”

About this time, I realized that we’d driven several miles past our turn, and had to stop the conversation while I found a place to turn around and backtrack. To tell you the truth, I don’t dismiss the possibility that the what, when and how sentence really is an error, produced in the course of rewriting the sentence a few times and not carefully checking the final version. But still… do an Internet search for strings like “what and when” or “what and how” (or if you prefer, read Appendix A in my dissertation) and you’ll find enough questions like the one quoted above that dismissing them all as errors becomes a real test of how stubborn you’re willing to be.

What Do You Do with What and Who?

Last week, there was a story in the Metro section of the Columbus Dispatch about digging up and reburying remains from an old cemetery. It was a forgotten cemetery, rediscovered when construction began for an office complex, and now that the relocation is underway, the property owner “plans to use charts she’s drawn each day to map out in detail who and what were found.”

When I read that sentence, something made me stop. After a few seconds, I figured out what it was. It was the plural were, instead of a singular was. I had never realized this about my own internal grammar, but it seems there’s a difference between coordinated subjects in a declarative sentence and coordinated subjects in a wh-question, summed up below:

1. *A man’s body and a rosary was found.
2. A man’s body and a rosary were found.
3. Who and what was found?
4. *Who and what were found?

At first, I thought the problem was that who and what just about always take a singular verb, even if we know we’re talking about more than one who or more than one what. For example, even if I tell you, “A lot of people are coming to my party,” it would sound pretty strange for you to ask, “Really? Who are coming?” And even if I tell you that there were a lot of items on the breakfast buffet, you wouldn’t ask, “What were there?” You’d say “What was there?” The only exception that I can think of is the case where the verb is a form of be, and after it is a plural noun phrase, as in the examples below:

1. Who are these people?
2. Who were they?
3. What are a few of your favorite things?
4. What were the causes of the Spanish-American War?

But even though who and what can’t ordinarily take a plural verb, it seems like the and-coordination of who and what should trigger plural agreement on the verb, just like the coordination of A man’s body and a rosary does. But why does it feel so wrong to me, while plural verbs with ordinary subjects, as in (2), are no problem? My gut feeling is that Who and what was found? is to be read as Who was found and what was found?, with the singular verb agreeing with each wh word individually. But if I’m allowed to do that kind of interpretation on a wh question, why can’t I do it with the statement*A man’s body and a rosary was found ?

Time for some Google corpus linguistics, to see how others handle who and what,or for that matter what and who, as a subject. I searched for each of these strings, excluding the phrases “what’s what” and “who’s who” to filter out a lot of irrelevant hits. In addition to who and what or what and who being the subject of the sentence, the sentences were not allowed to be in the exception case mentioned above, with be as the verb, followed by a noun phrase. I wanted to be sure that it was the and-coordination that was responsible for any plural verbs, not the plurality of a following noun phrase. For each string (who and what, what and who), the first 20 hits meeting these requirements were recorded. Here are the results:

• who and what
  • singular: 70%
  • plural: 30%
• what and who
  • singular: 90%
  • plural: 10%

So it looks like I’m not alone in my feeling about the singular agreement with who and what, what and who. For the people whose grammars are like mine in this respect, the question is still why there is this difference between wh coordinations and coordinations of ordinary noun phrases. The Cambridge Grammar of the English Language is silent on this point. It does talk about coordinated-wh questions, but has only two examples (p.874). One of them coordinates where and when (no help), and the other coordinates two wh noun phrases that are used as direct objects (still no help).

For the people whose grammars require plural agreement, there is less to explain, but I still wonder about the examples I found: How many of these authors’ internal grammars truly required plural agreement, and how many of these authors started with the singular, wondered about it during copyediting, and finally let the agreement rule for ordinary noun phrase coordinations override their native instincts?