Andrew Radford - Linguistics An Introduction [Second Edition]

singular form is. The pronoun he is assigned nominative case in

accordance with the nominative case assignment condition (275a), which specifies that the subject of a finite T is assigned nominative

case. The pronoun us is assigned accusative case via the default case

assignment condition (275c), by virtue of being neither the subject of a finite clause, nor a possessor. The verb helping is in the progressive

participle form because a selectional property of the progressive

auxiliary BE is that it selects a complement headed by a verb in the

progressive participle form (or, alternatively, via agreement in pro-

gressive aspect between verb and auxiliary).

20

Empty categories

So far, we have tacitly assumed that syntactic structures are projections of overt constituents (i.e. of words, phrases and clauses which have an overt phonetic form). However, as understanding of syntax has deepened, it has been argued that syntactic structures can also contain what are variously referred to as covert, null or empty constituents – i.e. ‘silent’ constituents which have no overt phonetic form. In this section, we will introduce a number of different types of empty category along with the arguments for supposing that such categories play a role in the grammar of English. In section 26, we shall see that evidence from a different source, psycholinguistic experiments, points to the importance of empty categories in sentence processing.

Empty T constituent

As a first illustration, consider how we might analyse the following set

of examples from African American Vernacular English (AAVE) – see section 16

for discussion of the (ing) variable in this variety in connection with forms such as playin’:

(279) a.

I’m playin’ baseball

b.

We/You/He/They playin’ baseball

Example (279a) contains an overt form of the T-auxiliary be – namely the contracted form ’m. However, the examples in (279b) contain no overt form of be, yet there are good reasons to suppose that they contain a covert/null/empty variant of are/is which we will symbolise as φ. If this is so, (279a, b) will have essentially the same structure, namely (280a, b):

(280)

TP

T'

VP

PRN

T

V N

a. I

’m

playin’

baseball

b. We

ϕ

playin’ baseball

271

272

senten ces

Since (280b) is a TP headed by a null variant φ of are, we can provide a straightforward account of why the subject we is nominative (because φ is a variant of the auxiliary are, and finite T-auxiliaries like are require a nominative subject), and of why the complement verb playin’ is in the -ing form (because φ is a variant of are, and it is a property of the auxiliary are that it selects a complement headed by a verb in the progressive participle -ing form).

Further evidence that structures like (280b) contain a null auxiliary comes from AAVE examples reported by Ralph Fasold such as the following (where

gonna = going to):

(281)

He gonna be there, I know he is

As we see from (standard English) examples like (282) below, in structures like these, the italicised auxiliary in the second clause (i.e. the clause after the comma) is generally a copy of that in the first clause:

(282) a.

You can do it, I know you can/*are/*have

b.

He is trying, I know he is/*must/*did

c.

They will come, I know they will/*were/*do

So, the fact that the auxiliary is appears in the second clause in (281) suggests that the first clause contains a null counterpart of is.

Although standard varieties of English don’t allow the use of a null auxiliary in sentences like (279), there are specific types of constructions in which auxiliaries can be null. In this connection, compare the two sentences in (283):

(283) a.

He was laughing and she was crying

b.

He was laughing and she – crying

The second sentence seems to contain a ‘gap’ in the position marked –. The

auxiliary was has been omitted in (283b) to avoid repetition, and we say that it has undergone a particular kind of ellipsis known as gapping (for the obvious reason that it leaves a gap in the middle of the sentence), so resulting in the structure in

(284) below:

(284)

TP

PRN T'

she

T V

ϕ

crying

That is, the clause is a TP headed by a null auxiliary φ. If we assume that φ here is a null variant of was, we can account for the fact that the subject is she (since a finite T

such as was requires a third person singular nominative subject like he or she, or the first person singular nominative subject I), and the verb crying is in the progressive

-ing form (since is requires a complement headed by a verb in this form).

If we extend this reasoning a little further, we can account for sentence pairs such as (285a, b) in a similar fashion (where capitals mark emphatic/contrastive stress):

Empty categories

273

(285) a.

He DID enjoy syntax

b.

He enjoyed syntax

Here (285a) is clearly a TP headed by an overt T-auxiliary did which is a past tense form. If we look at the internal morphological structure of did, we find that it carries the same past tense affix -d as we find in other past tense verb forms such as score-d, die-d, owe-d etc. so that did comprises an irregular past tense stem di- of the auxiliary DO and the past tense affix -d. In order to maximise the structural symmetry between (285a) and (285b), we can then propose that both clauses are TPs headed by a T constituent containing the past tense affix -d, and that the only difference between the two is that this affix is attached to the (past tense) auxiliary stem di- in (285a) but has no auxiliary stem attached to it in (285b). This means that (285a, b) have the structures in (286a, b):

(286)

a.

TP

b.

TP

PRN

T'

T'

PRN

He

He

T VP

T

VP

di+d

-d

V N

V

N

enjoy

syntax

enjoy

syntax

Let’s also assume that where T contains a tense affix which has no auxiliary to attach to, the affix is instead lowered onto the head verb of the verb phrase (by an operation which we can call Affix Attachment), in order to satisfy the requirement for the affix -d to attach to an auxiliary or verb. This will mean that in (286b), the past tense affix -d is eventually attached to (the end of) the verb enjoy, with the result that the verb is realised in the past tense form enjoyed. Such an analysis allows us to attain a unitary characterisation of the syntax of clauses, and to posit that all clauses are TPs which comprise a subject expression, a T head (which will contain an overt or covert tense auxiliary or tense affix in a finite clause, and infinitival to in an infinitive clause) and a verb (phrase) complement.

Evidence that auxiliariless finite clauses contain an abstract (present or past tense) T constituent comes from so-called tag questions. Examples of typical tag questions are given in (287) below (where the part of the sentence following the comma is called the tag):

(287) a.

He is working, is he?

b.

He can speak Swahili, can he?

c.

You will help us, won’t you?

d.

They might suspect him, mightn’t they?

e.

He could plead guilty, couldn’t he?

As examples like these show, the tag in such questions generally contains a

(positive or negative) auxiliary which copies grammatical features of the auxiliary that occupies the T position in the main clause (both auxiliaries are italicised

274

senten ces

in 287). So, for example, the main clause in (287a) contains the T-auxiliary is (which is a present tense form marking progressive aspect), and this is copied in the tag. If auxiliaries in tags copy grammatical features carried by the T constituent in the main clause, consider how we account for the fact that a sentence like (285b)