December 16, 2007

In the car, on the bus

Continuing with our celebration of National Preposition Day, let me note the following judgments about prepositions with objects denoting vehicles. The preposed asterisks do not necessarily mean that the string of words is utterly ungrammatical, but just that it is grammatical only on some fairly bizarre understanding — it is not just an alternative to the other example in its pair. The degree of acceptability varies: (1b) is thoroughly strange and suggests that she was strapped to a roof rack, and (2a) is even more bizarre. The ‘?*’ in front of (5a) and (6a) indicates that (in my judgment) the status is not so clear.

(1)a. She left that morning in a car. b. *She left that morning on a car.
(2)a. *She left that morning in a bike. b. She left that morning on a bike.
(3)a. She left that morning in a taxi. b. *She left that morning on a taxi.
(4)a. She left that morning in a van. b. *She left that morning on a van.
(5)a. ?*She left that morning in a bus. b. She left that morning on a bus.
(6)a. ?*She left that morning in a train. b. She left that morning on a train.
(7)a. She left that morning in a boat. b. She left that morning on a boat.
(8)a. *She left that morning in a ship. b. She left that morning on a ship.
(9)a. She left that morning in a single-seater light plane. b. *She left that morning on a single-seater light plane.
(10)a. *She left that morning in a United Airlines 747. b. She left that morning on a United Airlines 747.

The point is not to get people to send me hundreds of emails about how they found a place where someone says "have you ever wanted to live in a bus" or "see a band record their album in a bus" (please, no emails). The conditions here are not absolute, and I know that. But there really are some gross differences between the ways we refer to travel in or on various modes of transportation. And it really does not look like they are fully and clearly predictable on the basis of meaning, or vehicle architecture, or a combination of both. Language is not that orderly.

One other thing. Consider (4b). Suppose there were a scheduled van service between a central location and some factory on the outskirts of the city. Now it sounds a bit better, doesn't it? My initial hunch is that (i) normal position for a person with respect to the vehicle, (ii) size of the vehicle relative to a person, and (iii) regularly scheduled nature of the transportation may all be relevant to some extent.

Posted by Geoffrey K. Pullum at December 16, 2007 04:56 PM