Andrew Radford - Linguistics An Introduction [Second Edition]

immediately deals with the fact that English, and any other language, has a

potentially infinite number of sentences (see the Introduction, pp. 3f.).

So far, we have restricted our discussion to the question of how phrases are formed. However, as we saw in the previous section, linguists draw a distinction

Sentence structure

261

between phrases and clauses. For example, the reply given by speaker B in (261)

below is a clause, containing the subject they and the predicate try:

(261)

speaker a: What will the government do?

speaker b: They will try to reduce taxes.

An obvious question to ask is how clauses are formed – or, in more concrete

terms, what the structure of speaker B’s reply is in (261).

As already noted, there are interesting similarities between infinitival to and auxiliaries like will/would, shall/should, can/could, may/might, etc. For example, as illustrated earlier in relation to the sentences in (257), to typically occupies the same position in a clause (between subject and verb) as an auxiliary like will.

Moreover, just as will requires after it a verb in the infinitive form (cf. will show/

*will showing/*will shown), so too does infinitival to (cf. to show/*to showing/

*to shown). Furthermore, infinitival to behaves like a typical auxiliary (e.g. will) but unlike a typical verb (e.g. want) in allowing ellipsis (i.e. omission) of its complement:

(262) a.

I don’t really want to go to the dentist’s, but I know I eventually will

b.

I know I should go to the dentist’s, but I just don’t want to

c.

*I know I should go to the dentist’s, but I just don’t want

The fact that to patterns like the auxiliary will in several respects strengthens the case for regarding them as belonging to the same category. As noted earlier, since it is a core property of auxiliaries that they mark tense, and since a clause containing infinitival to often has future time reference, it has been suggested in much recent work that the two are different exponents of the category of T(ense).

(It should be noted, however, that in work in the 1980s, auxiliaries and infinitival to were taken to belong to the category INFL/inflection, the general idea behind this label being that finite auxiliaries inflect for tense/agreement, and infinitival to serves much the same function in English as do infinitive inflections in languages like Italian: however, we will adopt the more recent T analysis here – see also

section 10, p. 134.)

Having established that auxiliaries like will are exponents of the category T, let’s now return to the question of how clauses like that produced by speaker B

in (261) are formed. The simplest assumption (and hence the most desirable theoretically) is to posit that clauses are formed by exactly the same binary (i.e.

pairwise) merger operation which leads to the formation of phrases. This being so, we can suggest that the clause They will try to reduce taxes is formed by first merging the T-auxiliary will with the verb phrase try to reduce taxes to form the expression will try to reduce taxes, and then merging this larger

expression with the pronoun they to form the complete clause They will try to reduce taxes.

At first sight, it might seem plausible to claim that the expression will try to reduce taxes is a TP (i.e. tensed auxiliary phrase), and that when combined with the pronoun they it forms a pronoun phrase. But this can’t be right, since it would

262

senten ces

provide us with no obvious way of explaining why it is ungrammatical for speaker B to reply as in (263) below:

(263)

speaker a: What will the government do?

speaker b: *Will try to reduce taxes.

If complete phrases can be used to answer questions, and if will try to reduce taxes is a complete TP, how come it can’t be used to answer A’s question in (263)?

The answer which we shall give to this question here is that will try to reduce taxes is an incomplete phrase. Why? Because auxiliaries require a subject, and the auxiliary will doesn’t have a subject in (263). More specifically, let’s assume that when we merge a T-auxiliary with a verb phrase (VP), we form an incomplete

tense phrase which is often denoted T

¯, pronounced T-bar. For typographical

convenience, we shall follow many authors in using T0 (although readers should bear in mind that this too is pronounced T-bar̩!) and that only when we merge the relevant T-auxiliary with its subject do we form a TP (i.e. a complete tense phrase).

Given these assumptions, the clause They will try to reduce taxes will have the structure in (264):

(264)

TP

PRN

T'

They

T

VP

will

V

TP

try

T

VP

to

V

N

reduce

taxes

In a structure such as (264), the position occupied by the pronoun (PRN) they which serves as the subject of will is said to be the specifier position within TP. It is important to be clear that the term specifier (like the terms subject and complement) is the label of a grammatical function and not a grammatical category; thus, in (264) the function of specifier is fulfilled by the PRN (i.e. word belonging to the PRN category of pronoun) they. A specifier precedes the head of the phrase

containing it, whereas a complement follows its head: so, the PRN they precedes will in (264) because it is the specifier (and also subject) of will, whereas the VP

try to reduce taxes follows will because it is the complement of will; likewise in a determiner phrase (DP) such as such a pity, such is the specifier of (and so precedes) the head determiner (D) a, and pity is the complement of (and so

follows) a – for discussion of DP structures, see section 20; similarly, in a prepositional phrase (PP) such as right inside it, right is the specifier of (and so precedes) the preposition (P) inside and it is the complement of (and so follows) inside.

Sentence structure

263

Tests for constituency

Tree diagrams such as (264) provide a visual representation of what we claim to be the syntactic structure of the corresponding sentence. But this raises the question of how we can test whether claims made about structure in tree diagrams are correct. One way to do this is to make use of traditional tests which are designed to determine structure. We shall look at just one of these, relating to the phenomenon of co-ordination. English and other languages have a variety of co-ordinating conjunctions like and, but and or which can be used to

co-ordinate (that is conjoin or join together) expressions such as those bracketed below (see section 9, p. 134):

(265) a.

[fond of cats] and [afraid of dogs]

b.

[slowly] but [surely]

c.

[to go] or [to stay]

In each of the phrases in (265), a co-ordinating conjunction has been used to conjoin the bracketed pairs of expressions. Clearly, any adequate grammar of English will have to provide a principled answer to the question of what kinds of strings (i.e. sequences of words) can and cannot be co-ordinated.

It turns out that we can’t just co-ordinate any random set of strings, as we see by comparing the grammatical reply produced by speaker B in (266) below with the ungrammatical reply in (267):

(266)

speaker a: What did he do?

speaker b: Run up the hill and up the mountain.

(267)

speaker a: What did he do?

speaker b: *Ring up the electricity company and up the gas company.

Why should it be possible to co-ordinate the string up the hill with the string up the mountain in (266), but not possible to co-ordinate the string up the electricity company with the string up the gas company in (267)? We can provide a principled answer to this question in terms of constituent structure. More specifically, we can maintain that the string up the hill in (266) is a constituent of the phrase run up the hil (up the hill is a PP), and so can be co-ordinated with another similar type of phrase (e.