Wild and Crazy WTF
Posted by Neal on December 18, 2007
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:
I was going to return this after you told me you’d already read it, but then I decided that I would go ahead and give it to you anyway, with this note telling you how appalled am at your practice of buying gifts for others just so that you can enjoy them yourself first.
This year I’m giving that same cousin a copy of Steve Martin’s autobiography Born Standing Up. (I don’t want to spoil the surprise, so don’t tell him.) It’s quick and interesting reading, judging from what I read while I … while I … glanced through a copy I saw at the grocery store. On the very first page, though, I found this strange passage:
…the mouth speaking the line, the body delivering the gesture, while the mind looks back, observing, analyzing, judging, worrying, and then deciding when and what to say next.
How could a sentence that sounds so wrong have survived numerous rounds of revision and editing to end up in the final product? I’d have had no problem if Martin had coordinated two entire wh-questions, like this:
…deciding what to say next, and when to say it.
I’d also have had no problem if he’d used an option that syntacticians rather unintuitively refer to as sluicing, a radically abbreviated wh-question consisting of just the wh-item. Coordinating a full wh-question with a sluiced one would look like this:
…deciding what to say next, and when.
What he actually wrote — deciding when and what to say next — I just don’t get. Do you? I don’t like it because when I try to expand it out into two full wh-questions, one of them is *when to say next. The other one, what to say next, is OK, because the what is filling in for the missing direct object after say, just like nominal wh-words are supposed to do. But when can’t fill in for a missing direct object in *when to say next. It needs to go with a sentence (or in this case an infinitival phrase) that’s not missing a direct object (or indirect object, object of a preposition, or subject, for that matter). For example, to say something next. In my grammar, the only way you can get away with coordinating a nominal wh-word (such as what) with an adverbial one (such as when) is to have body of the question be parsable in two ways: (1) as lacking an object that would ordinarily be there, in order to go with the what; and (2) as not lacking one, in order to go with the when. Even then, it’s not entirely respectable. Here’s an example:
When and what did Gern Blanston perform?
If there is a rule in Martin’s grammar that generates this kind of wh-coordination — in other words, if he really did mean to phrase the question as he phrased it — what might that rule be like? One possibility is that in Martin’s grammar, not only can you coordinate a full wh-question with an abbreviated (sluiced) with the sluiced one coming second (as in what to say next, and when); you can also coordinate them with the sluiced question first (when and what to say next. However, saying just that much doesn’t solve everything. After all, most English speakers’ grammars already allow sluiced questions to be coordinated with full ones, so that they can say things like:
You saw that movie? When, and what did you think of it?
Here, the sluiced question when is asking about the seeing of the movie, not about the liking. In Martin’s example, the sluiced question is dependent on the full one: They both ask about saying something. (Notice also the intonational differences, as indicated by the commas.) The rule in Martin’s grammar would need to cover this point.
Another possible problem with this kind of “reverse sluicing” analysis arises if it turns out Martin can also say things like this:
The decision-making process about whether and who to scan is complex and was influenced by a range of factors….(link)
You can’t just call this a kind of sluicing, unless you want to claim that sentences like
*The decision-making process about who to scan and whether is complex.
are OK, too. Even if Steve Martin’s personal grammar doesn’t allow whether and who, there’s still the question of what kind of rule is in the grammars (of other people) that do.
The rule might be that any pair of wh-words can be coordinated, as long as at least one of them can be parsed successfully with the rest of the question. This analysis would predict that the wh-words could occur in any order, and that Martin would also find this grammatical:
…deciding what and when to say next.
In my own research, I’ve found that this kind of question, with the what separated from the phrase with the gap it’s supposed to fill, is much less acceptable than the kind Martin wrote, with the gap-filling wh-item adjacent to the gap-containing body of the question. (And those, in turn, are less acceptable than examples like when and what to give.) So probably this isn’t the rule that is governing Martin’s coordination of wh-items.
Of course, we could also just conclude that coordinations like the one in when and what to say next are just mistakes. They’re production errors made as people construct sentences on the fly. Or cut-and-paste errors made when a writer puts an extra wh-item into an ordinary wh-question and doesn’t give sufficient thought to whether it parses. Personally, that’s what I think, but I want to be careful before setting down this path, because it’s the same path taken by people who (to take a well-known example) reject singular they. They hear something like Someone thinks they’re smart and think, “Wait a minute! My rule says that they is plural, so this must be an error,” and never look close enough to discover that rules other than they one they know are at work.
This entry was posted on December 18, 2007 at 12:13 am and is filed under Christmas-related, Coordinated WH words. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.