The syntax of fear

...

7a. He feared [her to hit Peter]

7b. He feared [Peter to be hit by her]

8a. He persuaded [her] [to hit Peter]

8b. He persuaded [Peter] [to be hit by her]

In the examples above we can notice that the meaning only changes when the embedded clause in (8a) is passivized. This is because persuade is selecting one independent and nominal internal argument NP, her. Therefore its arguments not only change place but also modify its meaning when the embedded clause in (8a) is passivized, as we can see in (8b) because the verb is selecting the argument, therefore the person who is persuaded is different. ; this leads us to think that the NP and the to-infinitive are independent clauses, and that would confirm the inherent nature of persuade as a potential control predicate (that is, one whose complement (NP) is associated with the subject of the embedded clause). Nevertheless, in (7a) the verb fear is not selecting an independent-nominal internal argument as the verb persuade does, but rather it is selecting a unique clause [NP + to-infinitive], which is a piece of evidence for this version of fear to be an ECM verb. In the examples of (7); it does not matter if we change the form of the embedded clause into a passive, because it will keep the same meaning. What the passivation is useful for in this case, is to emphasize one argument over another.

Based on the data in (3) we will now analyze (2). First we start analyzing (2a) to guess if it is a subject control structure or a raising one as we have said before. It is important to say that fear is here a two-place predicate as we can see in the lineal analysis down below:

(2) a. He feared [to go out after nine]

x y

fear (x,y)

go (x, y)

So, as we saw on (3a) fear selects the external argument, in this case, NP he (x) and the internal argument which is a default embedded clause CP, to go out after nine (y). Now, in the embedded clause, the lexical predicate would be go which at the same time selects an external argument and an internal one, the external argument would be a covert NP, PRO. Out would be a lexical preposition, contentful, that it is not selected by the verb but it is sister with the verb of the embedded clause, go. In short, the internal argument is a locative PP that may be followed by a NP or not.

In order to satisfy the requirements of to go; we need the presence an external argument but as it is not actually phonetically realized; (is not present) we know that it is somehow associated with the subject of the matrix clause. We propose the existence of a covert NP with no phonetic realization called PRO.

Due to the fact that the verb fear selects its external argument as we saw in the evidences before, it is very probable that (2a) results to be a subject control structure. This would entail that the subject of the verb is syntactically represented as a covert NP PRO that is given null case by the inflection to.

As a piece of evidence for this, we are going to make a test to confirm that (2a) is a subject control structure, which is to passivize the subordinate clause in two different verbs as we did to the complex catenative construction of fear, one that is a raising structure verb and the one that we are dealing with on this exercise: fear. This will allow us to see whether the verb fear is imposing selectional restrictions on its external argument, in such a case it would be a subject control structure, otherwise, it would be a raising one, as we will see below:

9a. Peter feared [to hit her]

9b. Peter seemed [to hit her]

10a. She feared [to be hit by Peter]

10b. She seemed [to be hit by Peter]

Thanks to this evidence given below we see that the verb fear imposes selectional restrictions to its external argument. Due to the change in meaning of (9a) to (10a), now we are able to say that (2a) is a subject control structure. To corroborate this we add one last test, including “that” paraphrase with expletive “it”‘:

11a. *It feared that he went out after nine

11b. It seemed that he went out after nine

Continuing with (2b), we can presume it is a sentence with an ECM verb, based on the data analyzed for (3b). An ECM is structurally possible because the clausal complement is an exceptional clause (IP) so that the verb can assign accusative case to the subordinate subject.

(2b) They feared [the new king to be another unjust ruler]

x y

ECM verbs differ from (object) control verbs in several characteristics, such as: exceptionally case marked NPs don’t receive thematic roles from the verb in the main clause ( aren’t arguments of it). How we know this? In the following construction, the interpretation doesn’t correspond to that found in cases where the verb selects a NP without an infinitive:

12a. I fear him to be a liar

12b. I fear him

If ECM verbs assigned a thematic role to their objects, we would expect them to be incompatible with expletives (just like object control verbs), but what we find is that they are compatible with them:

(13a) They fear there to be another unjust ruler

Summing-up, the NP in the lower spec, IP receives accusative, not nominative. Recall that infinitive particle to doesn’t assign case to its specifier. This suggests that the case of the objects in ECM constructions must come from the matrix verb. This is exceptional because the verb assigns case to the specifier of its complement, in contrast where to the more normal situation where it assigns case to its complement (not to a constituent within it).

Finally, we will examine the examples in (4):

(4) a. !!It feared to rain

b. It threatened to rain

Based on the evidences given before; we can now say that in (4) we have two different predicates with two differents types of structure. In the case of (4a) we see it is semantically anomalous, this is due to the fact, as we said previously,