7    Nonfinite clausal complementation

This chapter discusses various types of nonfinite clausal complements. We begin with subject raising, also known simply as raising. Raising is similar to subject movement in that it targets subjects, but it differs in that it moves subjects not within a single IP, but out of a lower IP (the embedded or complement clause) into a higher one (the matrix clause). In order to show that subject raising exists, we rely on the existence of a class of special subjects that includes expletive there (already mentioned in connection with mediated case assignment), weather it, and so-called subject idiom chunks. As we will see, raising is possible only when the matrix clause contains a verb or adjective belonging to a special class of raising predicates. At first glance, raising predicates resemble so-called subject control predicates, but the syntactic behavior of these two classes of predicates can be shown to be distinct. Both classes of predicates contain verbs as well as adjectives, but for simplicity, the presentation that follows focuses on verbs. In addition to subject control verbs, English also has a class of object control verbs. Again, this class of verbs superficially resembles another verb class—the so-called Exceptional Case-Marking (ECM) verbs. But just as in the case of raising and subject control, the two verb classes must be kept distinct. Finally, we discuss the proper analysis of so-called small clauses, complements with clausal semantics, but without an overt head.

Special subjects

Expletive there

As was observed in Chapter 6, expletive there is a possible subject in English, but not a possible complement ((1) = (19) of Chapter 6).

(1) a. ok There is a fly in my soup.
b. * I dislike there in my soup.

There turns out to be an additional restriction on the distribution of expletive there. As is evident from the contrast between (2) and (3), expletive there is only grammatical when the main verb of its clause belongs to a certain class of verbs (roughly speaking, verbs of coming into existence and being in existence).

(2) a. ok After their military defeat, there arose among the Plains tribes a powerful spiritual movement.
b. ok There is a problem.
c. ok There began a reign of terror.
d. ok In the end, there emerged a new caudillo by the name of Batista.
e. ok There { ensued, followed } a period of unrest and lawlessness.
f. ok There exists an antidote.
g. ok There has occurred an unfortunate incident.
h. ok There remains a single course of action.
(3) a. * There continued the same problem.
b. * There failed an available time slot.
c. * There proved toxins in the soap.
d. * There rang the mail carrier.
e. * There seems no solution.
f. * There turns out an antidote.

The verbs in (2), in contrast to others, like those in (3), are said to license expletive there—that is, to provide a necessary condition for its grammatical occurrence.

Weather it

The second type of special subject, weather it, is the subject of verbs of precipitation. As the contrast between (4) and (5) shows, other semantically conceivable subjects for these verbs are unacceptable in English.

(4)   ok It is { hailing, pouring, raining, sleeting, snowing } .
(5)   * The { air, environment, precipitation, sky, weather } is { hailing, pouring, raining, sleeting, snowing } .

We can think of weather it as standing in a mutual licensing relationship with the verbs of precipitation.

Subject idiom chunks

The third and final type of special subjects is illustrated in (6), which lists several sentential idioms. The subjects of such idioms are known as subject idiom chunks.

(6) a.   The cat has got X's tongue.
b.   The cat is out of the bag.
c. The fur will fly.
d. The jig is up.
e. The pot is calling the kettle black.
f. The shit hit the fan.

What is crucial for our purposes is that the subjects in (6) have whatever idiomatic meaning they do in connection with the rest of the idiom, but not otherwise. For instance, neither cat in (6b) nor pot in (6e) have a general metaphorical sense of secret or hypocrite, respectively. Thus, although the sentences in (7) are not ungrammatical (unlike the ones in (3) and (5)), they have only literal interpretations, which are apt to be semantically or pragmatically anomalous, as indicated by the pound sign in (7b).

(7) a.   The cat { seems, proved } safe with her.
b. # Fortunately, the pot retracted that piece of duplicitous slander.

To summarize, all three types of special subjects are special because they must be licensed by the rest of their sentence, or by some part of it. Expletive there and weather it must be licensed by verbs of (coming into) existence and verbs of precipitation, respectively, in order to be grammatical, whereas subject idiom chunks must occur with the rest of the idiom in order to receive an idiomatic interpretation.

(Subject) raising

Data

Given what you know about expletive there, consider the contrast in (8) and (9).

(8) a. * There seems [ to continue the same problem ] .
b. * There seemed [ to ring the mail carrier ] .
(9) a.   There seems [ to be a problem ] .
b.   There seems [ to exist an antidote ] .

The ungrammaticality of (8) is not surprising, since (3e), repeated here as (10), shows that seem does not belong to the class of verbs that license there.

(10)   * There seems no solution.

But by the same token, the grammaticality of (9) is unexpected. What might the contrast between (8) and (9) be due to?

A first clue to the answer to this puzzle lies in the fact that the sentences in (8) and (9) contain not one verb, but two. Specifically, notice that although neither seem (the matrix verb in all four sentences) nor either of the complement verbs in (8) licenses there, the complement verbs in (9) do. The non–there licensing verb in the matrix clause thus turns out to be a red herring. That is, although all four sentences in (8) and (9) appear at first glance to be parallel to (10), the actual parallels are as indicated in (11) and (12).

(11) a. i. * There seems [ to continue the same problem ] . (= (8a))
ii. * There continued the same problem. (= (3a))
b. i. * There seemed [ to ring the mail carrier ] . (= (8b))
ii. * There rang the mail carrier. (= (3d))
(12) a. i. ok There seems [ to be a problem ] . (= (9a))
ii. ok There is a problem. (= (2b))
b. i. ok There seems [ to exist an antidote ] . (= (9b))
ii. ok There exists an antidote. (= (2f))

Analogous examples can be constructed for the other two types of special subjects. For instance, the contrast between (13) and (14) is parallel to the one between (4) and (5).

(13) a. It { continued, failed } [ to { hail, pour, rain, sleet, snow } ] .
b. It { seems, proved, turned out } [ to be { hailing, pouring, raining, sleeting, snowing } ] .
(14) a. * The { air, environment, precipitation, sky, weather } { continued, failed } [ to { hail, pour, rain, sleet, snow } ] .
b. * The { air, environment, precipitation, sky, weather } { seems, proved, turned out } [ to be { hailing, pouring, raining, sleeting, snowing } ] .

And the sentences in (15) have an idiomatic interpretation, even though the subject idiom chunk does not belong to the same clause as the rest of the idiom.

(15) a.   The cat seems [ to have got X's tongue ] .
b.   The cat proved [ to be out of the bag ] .
c. The fur began [ to fly ] .
d. The jig proved [ to be up ] .
e. The pot seems [ to be calling the kettle black ] .
f. The shit failed [ to hit the fan ] .

Analysis

What the facts in (8)–(15) suggest is that the subject of verbs like continue, fail, prove, seem, and turn out are not, in some sense, their own. Rather, the subject of a matrix clause containing one of these verbs originates as the subject of the nonfinite complement clause. We illustrate this by examining the derivation of (9a) in detail. We begin with the structure of the complement clause, which is given in (16). Note that the eventual matrix subject, expletive there, is licensed by being a clausemate of be (the prototypical verb of existence).

(16)    

Substituting this structure into the matrix clause yields (17).

(17)    

(17) is not well-formed as it stands, however, because it violates the principle in (18), which we have been assuming implicitly.

(18)     Exactly one member in a DP chain must occupy a position to which case is assigned.

The problematic DP is there, which cannot receive case in the position it occupies in (17). It cannot receive case from be, because even granting be the ability to assign case, be doesn't govern there. Nor can there receive case from the complement I (to). This is because nonfinite I does not bear case features; recall the ungrammaticality of (13b) of Chapter 6). Finally, as you will be asked to demonstrate in Assignment 7, there cannot be assigned case by seems in Spec(IP) of the complement clause under mediated case assignment. Unable to receive case within its own IP, there is forced to move to Spec(IP) of the matrix clause, as shown in (19), where it is assigned nominative case by finite I under spec-head agreement.

(19)    

The subject's movement from the complement Spec(VP) to the matrix Spec(IP) is called (subject) raising, and matrix verbs that license such movement are called raising verbs (a less misleading term might be 'raising triggers,' since it is not the verbs themselves that undergo movement).

In (19), the subject is shown as moving through each available specifier position before coming to rest in a position where it can be assigned case. This type of movement is often referred to as cyclic. In principle, a non-cyclic alternative is available: the subject might move from the complement Spec(VP) to the matrix Spec(IP) in one fell swoop. Empirical arguments in favor of the cyclic derivation in (19) have been proposed, but unfortunately, they are not conclusive. In what follows, we will assume a cyclic derivation of subject raising as in (19) on conceptual grounds. Specifically, we will present conclusive evidence in Chapter 10 for the cyclic character of another type of movement—so-called wh- movement.

Further evidence for subject raising

Some raising verbs (though by no means all of them) can take finite that complements, as shown in (20).

(20) a.   It { appears, seems, turns out } [ that there is a problem ] .
b.   It { appears, seems, turns out } [ that it is raining ] .
c.   It { appears, seems, turns out } [ that the cat is out of the bag ] .

Sentences like those in (20) have two properties that support the analysis of subject raising just presented. First, according to the representation in (17), raising verbs have no thematic (= semantically contentful) subjects of their own; that is, the matrix Spec(VP) starts out empty. This is consistent with the fact that the matrix subject in the sentences in (20) is expletive it. Second, when the complement of raising verbs is finite, the complement subject need not undergo raising, since the I of the complement clause, being finite, is able to assign it nominative case.

Subject control

Data

Subject raising verbs are not the only verbs in English that take nonfinite complements. A second class, the so-called subject control verbs, is illustrated in (21).

(21)     Larry { attempted, decided, expected, hoped, tried } [ to travel to Bhutan ] .

Subject control verbs superficially resemble subject raising verbs because they both take nonfinite complement clauses containing to, but the two verb classes differ in their syntactic behavior in two respects. First, no subject control verb ever allows expletive it as a matrix subject.

(22)   * It { attempted, decided, expected, hoped, tried } [ that Larry { will, would } travel to Bhutan ] .

Intuitively, the reason that (22) is ungrammatical is that both the matrix verb and the complement verb require thematic subjects of their own. That is, in contrast to sentences containing subject raising verbs, these sentences require not one, but two thematic subjects—one for each verb. This is confirmed by the behavior of those subject control verbs that allow finite complements. In this case, the matrix and the complement clause each have their own thematic subject, which need not refer to the same discourse entity, as illustrated in (23).

(23) a.   Larry { decided, expected, hoped } [ that he would travel to Bhutan ] .
b.   Larry { decided, expected, hoped } [ that { we, she } would travel to Bhutan ] .

It is true, though, that when the complement of a subject control verb is nonfinite, the interpretation of the complement subject is constrained in a way that it isn't when the complement is finite. Thus, (21) can only have the interpretation in (23a), never that in (23b). When the complement of a subject control verb is nonfinite, the matrix subject can be thought of as fixing or 'controlling' the reference of the complement subject (hence the name of the verb class).

There is a second difference between raising verbs and subject control verbs. In contrast to sentences containing raising verbs, the special subjects discussed earlier are either downright ungrammatical or allow no idiomatic interpretations in sentences with subject control verbs in the matrix clause. This is illustrated in (24).

(24) a. * There { attempted, decided, expected, hoped, tried } [ to be a problem ] .
b. *? It { attempted, decided, expected, hoped, tried } [ to { rain, snow } ] .
c. The cat { attempted, decided, expected, hoped, tried } [ to get X's tongue ] .
(only literal interpretation possible)
d. # The jig { attempted, decided, expected, hoped, tried } [ to be up ] .

Analysis

How can we capture the syntactic differences between sentences containing raising verbs and ones containing subject control verbs? The following analysis of subject control verbs is based on the idea that English has a silent pronoun PRO. In certain languages (for instance, in Italian or Spanish), silent pronouns are available in finite clauses, but in English, PRO is restricted to nonfinite clauses (for reasons that are not yet completely understood). It is this silent pronoun that occupies Spec(VP) of the subject control verb's nonfinite complement, as shown in (25), much as an overt pronoun occupies that position in a finite complement (recall (23)).

(25)    

Substituting the complement clause into the matrix clause yields (26a), and subject movement of the matrix and complement subjects yields the structure in (26b). (We assume subject movement in the complement clause purely for reasons of uniformity.)

(26) a.       b.  

Why is the complement of decide assigned the category CP rather than IP? The reason is that decide, along with certain other subject control verbs, allows indirect questions as complements, as in (27).

(27)     Larry must decide [ whether to travel to Bhutan ] .

Since the complementizer whether requires a slot, the structure for (27) must be as in (28).

(28)    

For uniformity, even subject control verbs that do not allow indirect question complements, such as promise, are assumed to take CP complements.

A further pair of verb classes

Exceptional Case Marking

Recall from Chapter 6 that expletive there is barred from complement position. We argued on the basis of this restriction that the noun phrase following it in sentences like (29) is not a matrix object, but rather the subject of an IP complement.

(29)     I expected there to be a problem.

As we would expect, the remaining two types of special subjects pattern with expletive there in analogous sentences. This is illustrated by the grammaticality of (30) and the availability of idiomatic readings in (31).

(30)     I expected it to rain.
(31) a.   I expected the cat to be out of the bag.
b. I expected the fur to fly.
c. I expected the shit to hit the fan.

For uniformity, we assume that expect takes an IP complement even when the noun phrase following it is not one of the special subjects, as in (32).

(32)     I expected her to take the job.

The elementary tree for expect for the examples in (29)–(32) is then uniformly as given in (33a), and the syntactic structure for (32) is given in (33b) (cf. (18b) of Chapter 6).

(33) a.       b.  

Because the noun phrase immediately following expect in (29)–(32) is assigned case by exceptional case marking (= our mediated case assignment), expect and verbs that pattern with it syntactically are called Exceptional Case Marking (ECM) verbs.

Object control

Certain verbs in English superficially resemble ECM verbs, but can be shown to be syntactically distinct from them. Based solely on sentences like (32) and (34), for instance, we might be tempted to classify persuade as an ECM verb.

(34)     I persuaded her to take the job.

But unlike expect, persuade does not allow the noun phrase following it to be one of the special subjects. In other words, the sentences in (35)–(37) contrast with those in (29)–(31).

(35)   * I persuaded there to be a problem.
(36) * I persuaded it to rain.
(37) a. * I persuaded the cat to be out of the bag. (* on idiomatic interpretation)
b. * I persuaded the fur to fly.
c. * I persuaded the shit to hit the fan.

The ungrammaticality of (35)–(37) follows immediately if the noun phrase following persuade is not a complement subject, but rather a matrix object. If we represent the remainder of the sentence as a CP complement in a manner reminiscent of the complements of subject control verbs, we obtain the elementary tree in (38a). As in the case of subject control verbs, we assume that the subject of the CP complement is PRO. But since the reference of the silent complement subject can be thought of as being 'controlled' by the matrix object, persuade is said to belong to the class of object control verbs. The structure that results for (34) is shown in (38b).

(38) a.       b.  

Almost without exception, control verbs with DP objects in addition to the clausal complement are object control verbs. However, promise, which takes an optional DP complement, is a subject control verb, as shown in (39).

(39) a.   They promised him to do the laundry.
b. They1 promised him2 PRO1,*2 to do the laundry.

Interestingly enough, it has been found that children have quite a difficult time acquiring the correct syntax of promise. Until they are about eight or nine years of age, they treat promise as an object control verb (C. Chomsky 1969), interpreting (39a) not as in (39b), but as in (40)!

(40)     They1 promised him2 PRO2,*1 to do the laundry.

Small clauses

At least since Aristotle, it has been assumed that sentences fall naturally into two parts: one that refers to an entity under discussion and another in which something is said about the first part. In logic, these two parts of a sentence are called the subject and the predicate, and the predicate is said to be predicated of the subject.
Note: A note of clarification is in order about the term predicate. Sometimes the term is used to refer to verbs, adjectives, or even nouns in isolation. This is the sense in which the term is used in the collocation 'raising predicates.' At other times, the term is used to refer to a potentially much larger piece of structure: namely, everything in a clause or sentence except the subject. The reason that one and the same term is used is that in predicate logic, predicates in the first sense generally share with predicates in the second sense the property that they take subject arguments and return propositions.

Ironically, an exception to this generalization are raising predicates, which map propositions to propositions without the benefit of an additional argument. In this sense, they resemble adverbial modifiers. Not surprisingly, many raising predicates allow paraphrases with sentence adverbs. Cf. Beth appears to have taken the job at Stanford, Beth has apparently taken the job at Stanford.

The predication structures that we have considered so far have all been projections of finite or of nonfinite I, as illustrated in (40) and (41). In these examples, our focus is on the embedded subject-predicate structures; the relevant subjects of predication are in boldface, and the relevant predicates are italicized.

(40) a. Finite I:   I expect that he will be off the ship immediately.
b. They found that he he was an expert.
c. They proved that the theory was false.
(41) a. Nonfinite I:   I expect him to be off the ship immediately.
b. They found him to be an expert.
c. They proved the theory to be false.

It is also possible, however, for predication structures to lack an overt I, as shown in (42).

(42) a.   I expect him off the ship immediately.
b. They found him an expert.
c. They proved the theory false.

Such minimal predication structures are known as small clauses.

What syntactic structure should we assign to small clauses? The parallel between (41) and (42) suggests that small clauses might simply have the structure of ECM complements, except that I and (in general) V are silent. But such an analysis is inconsistent with a wider array of facts. First, not all matrix verbs that subcategorize for small clauses subcategorize for ECM complements.

(43) a. Small clause:   They made him { (into) a star, sing the National Anthem } .
b. They let him sing the National Anthem.
(44) a. ECM complement: * They made him to { be (into) a star, sing the National Anthem } .
b. * They let him to sing the National Anthem.

Second, the syntactic category of the small clause predicate depends on the matrix predicate, as shown in (45)–(49).

Note: Some idiomatic exceptions to (47a) are disregarded, such as to let free, loose.

(45) a. A I consider them proud of their achievements.
b. D I consider her a friend.
c. P I consider them in the running.
d. V * I consider him enjoy the National Anthem.
(46) a. A His experience has made him considerate.
b. D His experience has made him an expert.
c. P * I'll make him off the ship immediately.
d. V I made him sing the National Anthem.
(47) a. A * I let him happy.
b. D * I let him a freedman.
c. P I let the cat out of the bag.
d. V I let him sing the National Anthem.
(48) a. A * I saw the sky cloudy, the outcome inevitable.
b. D * I saw him a burglar.
c. P I saw him in the doorway.
d. V I saw him enter the building.
(49) a. A * I expect them proud of their achievements.
b. D * I expect him an expert.
c. P I expect him off the ship immediately.
d. V * I expect him like the National Anthem.

Stowell 1983 concludes from facts like these that small clauses are uniformly maximal projections of lexical heads, which the matrix predicate subcategorizes for. The predicate of the small clause is an intermediate projection, and the subject of the small clause occupies the specifier position. The structures for (45a–c) and (46d) under this analysis are shown in (50).

(50) a.       b.  
c.       d.  

Stowell's analysis is attractive because it captures the similarities between ordinary and small clauses in a straightforward way. In both cases, the predication structure has a head, the predicate is an intermediate projection, and the subject occupies a specifier position. However, his analysis cannot be maintained for DP small clauses. Specifically, it fails to account for the minimal variant of (45b) in (51), where the DP predicate contains a possessor, which under Stowell's analysis would compete with the small clause subject for Spec(DP) (Heycock 1991).

(51)     I consider her Tanya's friend.

Given (51), then, DP small clauses must be projections of a head other than D. The semantic similarity between (51) and (52) suggests that the head in question is a silent counterpart of as.

(52)     I regard her as Tanya's friend.

What syntactic category should we assign to as (and by implication, to its silent counterpart)? The German translation equivalent of (51) in (53a) and the English example in (53b) suggest that as is a P.

(53) a.   Ich halte sie für Tanyas Freundin.
I hold her for Tanya's friend-fem
'I consider her Tanya's friend.'
b. I (mis)took her for Tanya's friend.

Unlike an ordinary P, however, as would subcategorize for AP and PP in addition to DP, though not for VP, as shown in (54).

(54) a.   I regard her as [AP friendly ] .
b. I regard her as [PP in the running ] .
c. * I regard her as [VP be { Tanya's friend, friendly, in the running } ] .

But classifying as as a P incorrectly leads us to expect the absolute with illustrated in (55) to behave parallel to (52) and (54a,b).

(55) a.   With Bush as [DP the GOP's likely presidential candidate ] …
b. * With so little plasma as [AP available ] …
c. * With all your kids as [PP in college ] …

The contrast in (55) forces us to posit at least two homonymous instances of as: one that takes DP complements and another that takes AP and PP complements. For mnemonic convenience, let us assign the two instances of as the syntactic categories Pr(edication)-D and Pr-X, respectively (the syntactic category Pr, though not the finer-grained analysis just proposed, is due to Bowers 1993). We can then say that absolute with subcategorizes for Pr-D, but not for Pr-X, whereas regard subcategorizes for both.

Let us return now to our original focus of attention, the silent counterpart of as in small clauses. The semantic similarity between (51) and (52) leads us to analyze DP small clauses as in (51) as projections of a silent Pr-D head. The contrasts between the (a) and (c) examples in (46)–(49) suggest that the Pr-X category just proposed needs to be split yet again into Pr-A and Pr-P, which take AP and PP complements, respectively. Since the overt predication head as takes no VP complements, there is no reason to assume a Pr-V head, and we can therefore maintain Stowell's analysis for VP small clauses like the (d) examples in (46)–(48). According to the analysis just proposed, the structures corresponding to those in (50) are as in (56) ((56d) = (50d)).

(56) a.       b.  
c.       d.  

A final apparent puzzle remains. As we saw in (55a), absolute with subcategorizes for Pr-D. Not surprisingly, therefore, its variant in (57a), with a silent Pr-D head replacing as, is grammatical. However, the small clauses in (57b,c) contrast with their as counterparts in (55b,c).

(57) a.   With Bush [DP the GOP's likely presidential candidate ] …
b. With so little plasma [AP available ] …
c. With all your kids [PP in college ] …

This problem is only apparent, however, since heads must be able not only to subcategorize for the syntactic category of their complements, but to specify that category's head as well. Independent evidence for this ability comes from examples as in (58), where a head selects not just a PP complement, but a PP complement headed by a particular preposition.

(58) a.   faith { in, *on } your ability
b.   rely { on, *at, *to } someone

Selection of this sort is analogous to a verb governing the morphological case of its DP complement, and we can assume that appropriate information percolates along the spine of the complement in both cases. Granting this, we can now revise our earlier statement about absolute with as follows. Contrary to what we stated earlier, absolute with is actually able to subcategorize for the maximal projections of all three Pr categories (Pr-D as well as the subcases of the erstwhile Pr-X—namely, Pr-A and Pr-P), provided that they are headed by the silent counterpart of as. However, when the complement of absolute with is headed by the overt head as, as it is in (55), the head is required to be of the category Pr-D, which in turn is restricted to taking DP complements.