Andrew Radford - Linguistics An Introduction [Second Edition]

He enjoyed syntax is tagged by the past tense auxiliary did in (288): (288)

He enjoyed syntax, did he?

If we assume, as in (286b) above, that (285b) is a TP headed by a T constituent containing a past tense affix, and that T in tags contains a matching affix, we can account for sentences like (288) by supposing that the auxiliary DO is used in the tag in (288) in order to support the past tense affix in the tag (i.e. in order to provide it with a suitable auxiliary stem to attach to). For obvious reasons, this phenomenon is known as Do-support.

A direct consequence of the TP analysis of clauses is that auxiliaries and verbs occupy different positions within the clause: auxiliaries occupy the head T position of TP, whereas verbs occupy the head V position of VP. An interesting way of testing whether this is correct is in relation to the behaviour of items which have the status of auxiliaries in some uses, but of verbs in others. One such word is have. In the kind of use illustrated in (289a) below, have is a perfect auxiliary (since it takes a complement headed by a verb in the perfect participle -n form), whereas in the kind of use illustrated in (289b), it functions as a causative verb (because it has a meaning akin to that of the verb cause):

(289) a.

He had gone to Paris

b.

He had a specialist examine the patient

By standard tests of auxiliarihood (cf. section 9), perfect have is an auxiliary, and causative have is a verb: e.g. perfect have can undergo inversion (Had he gone to Paris?), whereas causative have cannot (*Had he a specialist examine the

patient?). In terms of the assumptions we are making here, this means that have occupies the head T position of TP in its perfect use, but the head V position of VP

in its causative use.

Evidence in support of this claim comes from facts about cliticisation, a process by which one word attaches itself in a leech-like fashion to another (see section

10). The word had can cliticise onto the pronoun he in (289a) (forming he’d), but not in (289b), as we see from (290a, b): (290) a.

He’d gone to Paris

b.

*He’d a specialist examine the patient

How can we account for this contrast? If we assume that perfect had in (289a) is an auxiliary which occupies the head T position of TP, but that causative had in

(289b) is a verb occupying the head V position of VP, then prior to cliticisation the two clauses will have the respective (simplified) structures indicated by the labelled bracketings in (291a, b) below (the T constituent being empty in (291b)

once the past tense affix -d attaches to the verb had):

Empty categories

275

(291) a.

[TP He [T had] [VP [V gone] to Paris]]

b.

[

φ

TP He [T

] [VP [V had] a specialist examine the patient]]

If we assume that have-cliticisation is possible only when have immediately

follows the expression to which it cliticises and is blocked by the presence of an intervening constituent, it should be obvious why had can cliticise onto he in (290a) but not in (290b): had is immediately adjacent to he in (291a) but is separated from he by a null T constituent in (291b). A crucial premise of this account is that have is positioned in the head T of TP in its perfect use, but in the head V of VP in its causative use. So, have-cliticisation facts lend support to the claim that all clauses are TPs of the form subject + T + complement, and that clauses which have no overt T constituent contain a covert T which can block cliticisation.

In much the same way, we can argue that so-called bare infinitive clauses (i.e.

clauses which contain a verb in its uninflected infinitive form, but which lack the overt infinitive particle to) contain a covert counterpart of to. In this regard, consider the syntax of the bracketed infinitive clauses in (292a, b):

(292) a.

I have never known [you to lie]

b.

I have never known [you lie]

The two bracketed clauses in (292) are infinitive clauses (since in both cases the verb lie is in the infinitive form), and each serves as the complement of the verb known (so that each of the bracketed clauses is a complement clause). The bracketed complement clause in (292a) is a TP headed by the infinitival T constituent to and has the structure (293a) below. In order to maximise the symmetry between to infinitives and bare infinitives, we can analyse the bracketed bare infinitive complement clause in (292b) as a TP headed by a covert infinitive particle (symbolised below as φ) as in (293b):

(293)

a. TP

b.

TP

PRN T'

PRN

T'

you you

T V

T

V

to lie

ϕ

lie

Evidence in support of positing a covert infinitive particle in bare infinitive clauses comes from the fact that have cannot cliticise onto you in the bracketed bare infinitive clause in (294) below:

(294) a.

I wouldn’t let [you have my password]

b.

*I wouldn’t let [you’ve my password]

Why should cliticisation be blocked here? The answer is that bare infinitive clauses are TPs headed by a null infinitival T constituent φ, as shown in simplified form in (295):

276

senten ces

(295)

I wouldn’t let [

φ

TPyou [T ] have my password]

The presence of the intervening null infinitive particle is sufficient to prevent have from cliticising onto you.

The overall conclusion to which our discussion leads us is that all clauses are TPs of the form subject + T + complement (with T being overt or covert, finite or infinitival). However, this assumption proves potentially problematic in respect of certain types of infinitive clause which appear at first sight to be subjectless, and consideration of such cases leads us to another type of empty category.

PRO: the empty subject of infinitive clauses

Compare the structure of the bracketed infinitive clauses in (296a, b): (296) a.

We would like [you to stay]

b.

We would like [to stay]

Each of the bracketed infinitive clauses in (296) is a TP headed by the infinitival T

constituent to, and each bracketed TP serves as the complement of the verb like and so is a complement clause. An apparent difference between the two is that the bracketed infinitive clause in (296a) has an overt subject you, whereas its counterpart in (296b) appears to be subjectless. However, we shall argue that apparently subjectless infinitive clauses contain an understood null subject. Since the null subject found in infinitive clauses has much the same grammatical properties as pronouns, it is conventionally designated as PRO.

Given this assumption, sentence pairs such as (296a, b) have essentially the same structure, except that the bracketed TP has an overt pronoun you as its subject in (296a), but a covert PRO as its subject in (296b). These structures appear as (297a, b)s below:

(297)

TP

T'

VP

TP

T'

PRN

T V

PRN

T

V

a.

We

would like

you to

stay

b.

We

would like

PRO

to

stay

Introducing the relevant technical terminology, we can say that the null subject PRO in (297b) is controlled by (i.e. refers back to) the subject we of the would clause, or equivalently that we (i.e. the expression which PRO refers back to) is the

Empty categories

277

controller or antecedent of PRO. Verbs such as like which allow an infinitive complement with a PRO subject are said to function (in the relevant use) as

control verbs, and the clause containing the PRO subject is said to be a control clause.

An obvious question to ask at this juncture is why we should posit that

apparently subjectless infinitive complements like that bracketed in (296b) have a null PRO subject. Part of the motivation for positing PRO is semantic in