Prepositional NPs becoming passive subjects

We discussed verbs that must take an inherent preposition before an NP. This NP has many of the characteristics of a direct object, and can almost always passivize. Quite a range of prepositions are involved and the verbs come from a fair array of types, e.g. deal with, rely on, decide on, think about, listen to, look at, wonder at, dawdle over, search for. The object of the inherent preposition verb confess to does not readily become a passive subject but this is because confess to focuses semantically on the subject, similar to LIKING verbs, and want and possess and it is not appropriate to demote this subject. (Similar remarks apply to boast of/ about and brag of/about, but to a lesser extent—a passive here, although not common, is possible.)

Phrasal verbs, are combinations of verb plus preposition(s) which have an independent meaning, not predictable from the meanings of the component elements. An NP which immediately follows the verbal element has direct object status and may always passivize, e.g.

Np type: They put the wedding off until next week, The wedding was put off until next week

NpN type: They took me for a preacher, I was taken for a preacher

NppN type: They tied the conference in with the school holidays, The conference was tied in with the school holidays

An NP which follows the preposition(s) of a phrasal verb (with no other NP preceding it) has weak object status, and shows limited passivization possibilities, e.g.

pN type: The teachers are always picking on John, John is always being picked on (by the teachers)

ppN type: They did away with the position of Assistant Secretary, The position of Assistant Secretary was done away with

If there is a direct object NP, coming immediately after the verbal component, and a second NP following the preposition(s), as in NpN and NppN types of phrasal verb, then the later NP can never be passivized. For example, a preacher and the school holidays from the examples above could not become passive subject.

Leaving aside the special cases of inherent preposition verbs and phrasal verbs, we can now consider the question of passivization of an NP that is straightforwardly governed by a preposition. This is plainly possible in English—as in Oh dear, my new hat has been sat on and This bed was slept in by Queen Elizabeth—but in quite limited circumstances. What are these circumstances?

Recall that a direct object NP may only become passive subject when it is sufficiently special—in that semantic-syntactic context—to merit being focused on, perhaps in contrast to something else. At a party, someone could wander into another room and exclaim: Hey, mint juice is being drunk in here; one would infer that something less strong was being consumed in the room the speaker has just left. A peripheral NP, marked by a preposition, may also be passivized when it refers to something that is the most significant element in an activity. If you were handed a glass and noticed someone else’s dregs at the bottom, or lipstick on the rim, you could use a construction with an adjectival predicate—This glass is dirty; or a construction with the underlying direct object passivized—This glass has been used; or a sentence that involves passivization on a peripheral NP to the verb drink (which is the activity being alluded to)—This glass has been drunk out of.

There are two syntactic restrictions on a peripheral NP becoming passive subject:

(a) There must be no direct object present. If there is, it is only the direct object that can be passivized, not the prepositional object. Consider Someone has drunk mint juice out of this glass; it is quite unacceptable to say *This glass has been drunk mint juice out of. It is only when the object NP is omitted that passivisation on a prepositional NP becomes possible. As further examples we can quote This spoon has been eaten with but not *This spoon has been eaten beans with, and My new cushion has been sat on by a dog but not *My new cushion has been sat a dog on (corresponding to the causative Someone has sat a dog on my new cushion).

This restriction relates to our earlier observation that an NP which follows the preposition(s) of a phrasal verb may sometimes be passivized but only if there is no preceding direct object NP which comes between verbal and prepositional components, i.e. N2 may sometimes be passivized from pN2 and ppN2 but never from N1pN2 or N1ppN2.

It is not the case that a passive verb may not be followed by a direct object. Where a prepositional NP with a transitive verb is promoted into object slot, the original object remains, e.g. give Mary a book, tell Mary a story. Either of the two objects may be passivized, with the other remaining after the verb in the passive clause, e.g. Mary was told a story, A story was told Mary. The rule appears to be that if there are two objects then either may passivize, but if there is an object and a prepositional NP then passivization is restricted to the object. Only when there is no object present may the prepositional NP become passive subject.

(Quasi-exceptions to this rule involve idioms such as shake hands with, find fault with, make an honest woman of. It is possible to say things like Have you been shaken hands with yet?)

(b) The prepositional NP must not be alternatively codable as direct object. We often have available two possible syntactic frames for a given concatenation of verb and semantic roles; a certain role may be mapped onto O relation in one frame and marked by a preposition in another frame. In such a case, it will generally be passivizable only from the frame in which it is object, and not from that in which it appears with a preposition. There are several kinds of instance of this restriction:

(i) We discussed a group of transitive verbs that may insert a preposition before a direct object NP, e.g. He kicked the box, He kicked at the box. This NP may be passivized from simple direct object position, but not as a rule when there is a preceding preposition. We could say That box was kicked (and dented), This sheet has been torn but scarcely *That box has been kicked at (and dented ), *This sheet has been torn at. Inserting the preposition indicates that the activity did not relate to some specific object, or did not achieve a specific result, and in such semantic circumstances we would not expect passivization to be possible.

There is one notable exception: one can say Mary was shot (implying that the bullet or arrow hit her) or Mary was shot at (suggesting that the projectile missed her). In the second sentence the Target is not physically affected but she is likely to be mentally disturbed by the incident—she will have had a scare, and may realize that she is at risk of a further attack. It is here possible to passivize on an NP governed by a preposition (even though there is an alternative construction in which it is direct object); but this appears only to be possible when the NP has animate reference since only animates are affected by being shot at.

(ii) We discussed verbs that may promote a peripheral NP— normally marked by a preposition—into direct object slot, if it is particularly significant in this instance of the activity. It was remarked that the salience of such an NP is such that it may readily be passivized (especially if the transitive subject is non-specific), e.g. The English Channel was first swum in 1875, That peak has never been climbed, French is spoken throughout my chain of boutiques. Such an NP—which can be promoted to object and passivized from that slot—may never be passivized from its original post-prepositional position. That is, we could not say *The English Channel was first swum across in 1875, *That peak has never been climbed up, or *French is spoken in throughout my chain of boutiques.

Promotion to object is also possible in transitive sentences, e.g. John gave an apple to Mary, John gave Mary an apple; Granny knitted a jumper for Mary, Granny knitted Mary a jumper. The NP Mary may only be passivized from the second sentence of each pair—Mary was given an apple (by John), Mary was knitted a jumper (by Granny). Passivization of Mary from the first sentence of each pair, where this NP is marked by a preposition, is blocked on two grounds: (i) it is preceded by a direct object, and (ii) there is an alternative construction available in which it is direct object. Thus, we do not get *Mary was given an apple to, or *Mary was knitted a jumper for.

We described, for AFFECT verbs, frame I, which has Target role as O and Manip role marked by a preposition, e.g. John hit the vase with that stick, and frame II, which has Manip as O and Target marked by a preposition, e.g. John hit that stick on the vase. The Target may only be passivized from frame I, in which it is object (The vase was hit with that stick), and the Manip may only be passivized from frame II, in which it is object (That stick was hit on the vase). There are again two reasons why we cannot passivize the Manip when it is preceded by with, in frame I, or the Target when it is preceded by a preposition like on, in frame II: (i) they are preceded by a direct object NP; and (ii) there is an alternative construction in which this role is in direct object relation. This explains the unacceptability of *That stick was hit the vase with and *The vase was hit that stick on.

In summary, English may allow a prepositional NP to become passive subject (the preposition remaining after the verb) only if there is no direct object and if there is no alternative construction type in which the NP could be coded as direct object.

Passivization of a peripheral NP seldom happens in English. We can mention three of the kinds of circumstance in which it is permitted:

(a) When the referent of the NP has been (or may be) affected, to its detriment, by the activity; a by phrase, marking passive agent, is unlikely to be included, e.g. That window shouldn’t be leaned against (it might break); My new hat has been sat on.

(b) Where it is a particular feature of some object that it was involved in a certain activity with a particularly important subject. Here the significant factor is concatenation of object and verb with that subject, which must be included, in a by phrase. Imagine an auctioneer describing a four-poster: The next item is an antique bed which is reputed to have been slept in by Queen Elizabeth I.

(c) Where the NP is discourse topic and is brought into passive subject slot for this reason, e.g. Those stairs were built by John, then they were run up by Mary and promptly collapsed (cf. Bolinger 1977b: 69). (On its own Those stairs were run up by Mary is much less acceptable.)

Note the contrast between (a) and (b). If you are shown to an hotel room and notice that the sheets on the bed are all mussed up you might exclaim in annoyance: This bed has been slept in! (It doesn’t matter who by; the fact you are drawing attention to is that it’s not clean.) But in a higher-quality establishment the manager might conduct you to his finest suite and announce with pride: This bed was slept in by Winston Churchill. In the first utterance a by phrase would not normally be included; in the second it must be.

Circumstance (a) invariably deals with something that is regarded as undesirable. It does in fact shed interesting light on our cultural norms— the acceptability of This plate has been eaten off, This spoon has been eaten with, This cup has been drunk out of, This bed has been slept in illustrates the phobia English-speakers have about certain kinds of cleanliness. This knife has been cut with should be just as natural a sentence on syntactic grounds, but in fact we don’t usually care so much whether a knife has been used previously. For that reason—and that reason alone—this sentence sounds less felicitous than the others.

A wide range of prepositions may be involved in this kind of passivization, similar to those which occur as ‘inherent prepositions’—in, on, against, with, to (I’ve never been written to by Maria), at (That hat is always getting laughed at), over (The bewitched diamond is being fought over again), for (John was sent for), about (The accident is being talked about a lot), and from (Aikhenvald’s book on classifiers is always being quoted from).