Raising: subject-to-subject

Linguistics 322

The term raising is often used to selectively refer to the raising of NPs to the subject position of a higher clause or to another position in a higher clause. See raising. Here, we examine raising out of tenseless clauses (infinitives). Consider the following pair of sentences.

(1)   It seems that John is happy.
(2)   John seems to be happy.

Note that this pair of sentences is nearly synonymous. (1) contains two lexical propositions (in addition to the grammatical propositions):

(3)   SEEM <X>
(4)   X= HAPPY <exp: JOHN>

Technically X is a complete proposition. Either proposition can be asserted to be false, or both propositions can be asserted to be false:

(5)    It doesn't seem that John is happy.
(6)     It seems that John is not happy.
(7)     It doesn't seem that John is not happy.

Propositional arguments are not normally assigned a theta-role. This is no real problem as it can be shown in advanced work that each theta role is nothing more than the name of an argument of a specific kind of eventuality. We could not bother or we could call it 'event':

(8)   SEEM <event: X>.

It is a moot point whether be is part of the propositional structure. In our view it is not; it is inserted as a dummy verb to function as a hosts for tense.

Both the main lexical proposition and the embedded lexical proposition are modified by TENSE--[-Past]:

(7)   [-Irreal] < [-Past] < [-Pass] < SEEM <X>]]
(8)   X= [-Irreal]] < [-Past] < [-Pass] < HAPPY <exp: JOHN>.

(7) and (8) are merged into a single string (Chomsky calls this a generalized transformation):

 (9)   [-Irreal] < [-Past] < [-Pass] < SEEM <X> [-Irreal]] < [-Past] < [-Pass] < HAPPY <exp: JOHN>>>>>>.

Now note that the basic propositions of (2) are the same as (1) even though John is the subject of seems. It is John who is happy. The difference between (1) and (2) lies in the tense of the embedded clause. In (1) it is modified by [-Past], but in (2) there is no formal tense differentiation. This construction is often called an infinitive clause, the tense form of the verb the infinitive. It is marked with the form to in English:

(8)   [TP [-Past] [seem X]]
(9)   X=[TP [-Tense] [John (be) happy]].

(8) and (9) are merged into a single string:

 (10)   [TP [-Past] [seem [TP [-Tense] [John (be) happy]]]]

Syntactically, the subject of the embedded clause raises to the subject position of the matrix clause:


[-tense] is spelled out as the particle to in English. There is some debate to what category this particle is. We won't get into this issue here. To is a free standing form--it requires no host. Hence there is no targeted movement. [-Past] targets seem and lowers. John raises first from the verb and adjoins to TP, then it raises again and adjoins to TP in the matrix clause. The rational for raising is discussed in case theory, but see also .theta roles.

 form:  Propositional Feature:
 to  [-Tense]

Note that [-Irreal] is spelled out as [empty] or as that when the clause is embedded:

(13) It seems that John is happy.

(14) It seems John is happy.

 [empty]  root and embedded  [-Irreal]
 that  embedded

This is not the complete story of the complementizer that, but it will do for now.

What is the propositional structure of the infinitive? Infinitives are very complex. Here, it seems to be simple: the complement of raising verbs seems to be lacking the tense-proposition. That is, in the case of seem and appear, the speaker has a choice of selecting a complement marking tense or one not marking tense. Chomsky has long proposed [-Tense] for this. If correct, the propositional structure of (2) is:

(16)     [-Irreal] < [-Past] < [-Pass] < SEEM <[-Irreal < [-Tense] < [-Pass] < HAPPY <experiencer: JOHN> > > > > >.

Note that TENSE now has a more complex feature structure:

(17)     TENSE -> [±Tense]

(18)     [+Tense] -> [±Past].

Raising verbs include seem, appear, and several passives formed with pleonastic pronouns:

(19)    It is believed that Mary is sad.
(20)    Mary is believed to be sad.

Other such passive verbs include understood, thought, said, presumed, assumed, and so forth.

There are raising adjectives:

(18)      It is likely that Mary will go.
(19)      Mary is likely to go.
(20)     It is certain that Eloise will faint.
(21)     Eloise is certain to faint.

Some adjectives are raising only in that they do not subcategorize a tensed complement:

(22)     *It is apt that Mary will go.
(23)     Mary is apt to go.

The subject of apt receives its theta-role from the embedded predicate. In (23) it isn't Mary who is apt, but that she might go that is apt.


Mood of the infinitive and subcategorization.

Seem, appear, certain, sure, clear subcategorize an irrealis complement. In other words, their complements are asserted by the speaker subject to the modifying effect of the main verb. The complement can only be asserted including denial of the assertion by negation (negation):

(24)     It seems that John is not happy.

(25)     John seems not to be happy.

It is not possible to question the complement clause:

(26)     *It seems that is John happy?

Thus the predicates of this claim subcategorize a complement with the propositional structure containing irrealis and plus tense. Let us call the theta role of the complement :eventuality":

 ____ [MOOD [-Irreal]] < [TENSE [+Tense]]>

When the same verbs subcategorize an infinitive, the take [-Tense] rather than [+Tense]. Otherwise there is no difference:

 ____ [MOOD [-Irreal]] < [TENSE [-Tense]]>

On purely speculative grounds, there is the possibility that the infinitive is possible here since the complements can only be irrealis. Infinitives can never mark questions. However, infinitives may mark other moods. These infinitives have different properties. We won't go into mood much here.

The mood of infinitival complements of predicates of the apt-class is not irrealis. Note that the following two sentences are nearly synonymous:

(29)     Mary is apt to go.
(30)     Mary might go.
(31)     It is probable that Mary will go.

Modal auxiliaries mark modality. Modality deals with forms of irrealis such as possibility, obligation, possibility, predictability, promise, and so forth. The term is hard to define, as mood c-commands modality:

(31)     If Mary is apt to go, is Mary apt to come back?

We will put the question of modality aside here noting that apt takes an infinitival complement marking modality. Such predicates are not assertive and hence that cannot take a tensed complement:

(32) *It is apt that Mary will go/goes/has gone.

There is another class of adjectives that subcategorize tensed clauses but these adjectives subcategorize a subject argument:

(29)     John is certain that he is going.

(30)     Mary is sure that the earth is flat.

In (24) John is an argument of certain. He is experiencing certainty. The pronominal he is an argument of going. There is no raising here. We will identify the propositional form of certain as "CERTAIN-2:":

(31)     CERTAIN-2 <experiencer: JOHN> < MOOD [-Irreal]] <TENSE [+Tense [-Past]]] ... GO <agent: [+Human, +Gender, -Fem, -Pl, +Def]>

 he  [+Human, -Pers, +Gender, -Fem, -Pl, +Def, Nom]
 him  [+Human, -Pers, +Gender, -Fem, -Pl, +Def, Acc]

The feature [+Gender] refers to forms in which gender is distinguished like humans (except babies in some circumstances), pets, and higher animate forms in some circumstances. The feature [-Gender] refers to other animals and all inanimate objects. In the case of animals and some botanical species, it does not mean that these species do not have a gender distinction, but that we do not perceive a gender difference in this species. The pronoun it has the following features:

 it  -Pers, -Gender, -Pl, +Def, Nom/Acc

The propositional structure of 'is going' is not the progressive, but some form of the future mood. We put it aside here.

In (31) both the experiencer of the root clause and the agent of the embedded clause raise to the subject position of their respective clauses where each are assigned the nominative Case. Suppose that CERTAIN-2 took an infinitival complement. He in the embedded complement receives no tense:

(34)     *John is certain he to go.

(35)     *John is certain him to go.

In this case the underlying pronominal form cannot receive Case form to and it cannot raise to the root clause, since John raised to TP to be assigned Case by [+Tense]. Such constructions are doomed from the onset and thus they do not occur.

In cases of uncertainty whether an adjective is of the raising kind or not, sentential idioms make useful tests. A so-called sentential idiom is one where the verb and its arguments constitute a single proposition where the arguments of the verb have no semantic content:

(28)     All hell broke loose.
(29)     The shit hit the fan.

Both these idioms nearly have the same proportional content: something like " trouble occurred." These idioms occur in raising constructions, but they do not occur where the adjective assigns an argument to the surface subject:

(30)     It seems that the shit has hit the fan.

(31)     The shit seems to have hit the fan.

(32)     *The shit seems that it has hit the fan.

(33)     It is certain that the shit will hit the fan.

(34)     The shit is certain to hit the fan.

(35)     *The shit is certain that it will hit the fan.

"The shit hit the fan" is an idiom and contains no arguments. To get the idiomatic reading, "the shit" must start in the embedded clause where it gets the idiomatic reading. Note that (32) and (35) are ungrammatical because "the shit" cannot raise out of the embedded clause. The subject of the embedded it is marked for the nominative Case. There is no need for it raise, and it cannot be marked twice for Case. The idiom tests always distinguishes between raising predicates and non-raising predicates.

To return to course outline Click here.

Go to .

This page last updated 7 MR 2001