Language Log

Headline wrangler arrested for violating the Direct Object Restriction

Headline tonight from AP: "Woman allegedly 'sprays' out of hospital."

That's terrible! I thought. But what's with that 'allegedly'? Spraying out of a hospital isn't a crime, surely!

But of course, she didn't spray out of the hospital. She sprayed her way out of the hospital. She actually pepper-sprayed her way out of the hospital. I can understand how that might be a crime.

See, thing is, in English you can describe motion events -- something going from point A to point B -- with almost any verb, even ones that don't in and of themselves refer to motion. But there's some peculiar constraints on doing so.

Verbs like 'walk', 'run', 'stroll', etc. are motion verbs in and of themselves. Their subject is the traveller; in John walked out of the hospital, John is the one going from point A to point B.

Now consider verbs like 'whistle', 'sing', 'rattle', 'wriggle', 'twitch', 'smirk', etc. The events denoted by these verbs by themselves don't imply travelling from anywhere to anywhere. With a sentence like Mary whistled, the crucial thing is for Mary to be doing some whistling. Her movement or lack therof is irrelevant.

However, even these verbs can describe motion events in English, with one crucial restriction. When you add the destination location to the sentence (to the house, out, there, whatever), thereby making it a motion event, you have to add some kind of direct object as well. It's not usually a real direct object, more a kind of an ersatz one, involving some pronominal element that refers back to the subject:

1. Mary whistled her way down the lane.
2. John smirked his way out of the room.
3. Bill sang himself to Carnegie Hall

So, what about 'spray'? Spray is a motion verb. You can use it to describe motion from point A to point B, as long as that motion happens in a characteristic spraying manner -- i.e. moving as droplets or particles propelled outward in a spreading fashion. So, e.g., the sentence below is fine:

4. The oil sprayed out of the hose.

That is, the motion itself has to be occurring in a spraying kind of way, when there's no direct object. The same fact obtains with the non-motion verbs above -- the direct object isn't required if the motion itself is occurring in the relevant way. So, famously, The bullet whistled through the window is ok without an object, because it's the bullet's motion itself that is making the whistling noise.

But 'spray' can also occur in a causative construction, where the subject of the verb is not actually spraying per se, but is rather the cause of spraying in others, as in (5) below. The crucial requirement for this sentence to be true is that the subject be the cause of some spraying -- nothing to do with whether they're moving or not moving.

5. Sue sprayed paint on the wall.

What about our headline? In the situation described by the news story, the woman emerged from the hospital in the normal human fashion. At least, she was corporeal enough to subsequently 'flee' the scene. Thus, her motion probably occurred in a normal human fashion, like walking or running or strolling or swaggering or similar. The motion itself didn't occur in a spraying fashion. Rather, the woman was the cause of spraying, which doesn't necessarily involve any motion.

So, the situation is one in which the woman is doing some (non-motion-entailing) spraying, and simultaneously moving out of the hospital. Her motion is itself not accomplished in a spraying manner. This is precisely the situation in which an ersatz direct object, like her way is required. He sprayed out of the wood chipper, yes. (Thank you, Coens!) She sprayed out of the hospital, hopefully, no.

There you go: The headline writer has wantonly violated the Direct Object Restriction on the English manner-of-motion construction. How rude.¹

¹The Direct Object Restriction observation is due to work by Beth Levin and Malka Rappaport Hovav, who know all that there is to know about what kinds of tricks can be played with English verbs. Update! Beth writes to remind me that ACTUALLY this generalization was first formulated by Jane Simpson, in her 1983 paper 'Resultatives' published in L. Levin, M. Rappaport, and A. Zaenan (eds), Papers in Lexical Functional Grammar. Bloomington: Indiana University Linguistics Club, 143-157. Many apologies, Jane! Beth and Malka did name it the DOR, though. Ray Jackendoff is still the go-to guy for the way-construction. They're all probably too busy to take calls about it, though, so comment here if you like.