Andrew Radford - Linguistics An Introduction [Second Edition]

g. a PP such as up the mountain, or down the hill, or along the path, etc.). Conversely, however, we can maintain that the string up the electricity company in (267) is not a constituent of the phrase ring up the electricity company, and so cannot be

co-ordinated with another similar string (up is associated with ring in such constructions, and the expression ring up forms a complex verb which carries the sense of ‘to telephone’). On the basis of contrasts such as these, we can suggest that the following constraint (i.e. grammatical restriction) is part of an adequate grammar of English: (268)

Only like constituents can be conjoined; non-constituent strings cannot be

conjoined

(a non-constituent string being ‘a string of words which are not a constituent’).

264

senten ces

We are thus supposing that processes for combining words and phrases in

native speakers’ grammars are constrained by (268), and that (268) constitutes part of English native speakers’ competence.

Having established (268), we can now make use of it as a way of testing the tree diagram in (264) above. A crucial claim made in (264) is that the strings reduce taxes, to reduce taxes, try to reduce taxes and will try to reduce taxes are all constituents (of various different types). Evidence for the correctness of this claim comes from co-ordination facts in relation to sentences such as those in (269):

(269) a.

They will try to [reduce taxes] and [increase pensions]

b.

They will try [to reduce taxes] and [to cut bureaucracy]

c.

They will [try to reduce taxes] and [attempt to eliminate poverty]

d.

They [will try to reduce taxes] but [may not succeed]

Given the crucial premise (268) that only strings of like constituents can be conjoined, example (269a) provides evidence for analysing reduce taxes as a VP

since it can be conjoined with another VP such as increase pensions. Likewise,

(269b) indicates the correctness of analysing to reduce taxes as an infinitival TP, since it can be co-ordinated with another infinitival TP like to cut bureaucracy. Similarly,

(269c) shows us that try to reduce taxes is a VP since it can be conjoined with another VP such as attempt to eliminate poverty. And in much the same way, (269d) tells us that will try to reduce taxes is a finite T0 because it can be co-ordinated with another finite T0 such as may not succeed. Overall, then, we see that the assumptions about the structure of clauses embodied in tree diagrams such as (264) receive independent support from tests such as the co-ordination test (exercise 1).

Agreement, case assignment and selection

Although we’ve suggested that all phrases and sentences are formed

by successive applications of a simple binary merger operation, it’s clear that we can’t randomly combine any pair of categories by merger, as examples like the following illustrate:

(270) a.

He has/*have resigned

b.

We have/*has resigned

Given the analysis we are assuming here, sentences like (270a) and (270b) will have the respective structures shown in (271):

(271)

TP

T'

PRN T

V

a.

He has resigned

b.

We have

resigned

Sentence structure

265

In terms of the structure (271), what the contrast between (270a) and (270b) shows is that a finite T-auxiliary such as has/have must agree in person and number with its specifier/subject: since the specifier of T in (270a) is the third person singular pronoun he, the present tense T-auxiliary HAVE is marked as third person singular via agreement with its specifier he and so is ultimately realised as the third person singular form has. And since the specifier of T is the first person plural pronoun we in (270b), the auxiliary HAVE in T is marked as first person plural via agreement with its specifier and so surfaces in the first person plural form have.

This suggests that the derivation of sentence structures (i.e. the way in which they are built up) involves not only merger operations but also agreement operations.

One such operation is specifier-head agreement, which we can sketch in simplified form as follows:

(272)

Specifier-head agreement

A finite T constituent agrees in person and number with its specifier/subject A further type of operation involved in sentence formation can be illustrated by contrasts such as the following:

(273) a.

He has resigned

b.

*Him has resigned

In the grammatical sentence (273a), the subject of the clause is the nominative pronoun he, whereas in the ungrammatical sentence (273b), the subject is the accusative pronoun him. Why should it be possible to have a nominative subject in sentences like (273) but not an accusative subject? Note that we can’t simply say that this is because all clauses have nominative subjects, since this is untrue of clauses like those bracketed below, which have (italicised) accusative subjects: (274) a.

She’s keen [for him to be there]

b.

She wants [him to be there]

What’s the difference between nominative-subject clauses like (273a) and accusative-subject clauses like those bracketed in (274)? The obvious difference is that nominative-subject clauses are finite (by virtue of containing a finite T

constituent), whereas accusative-subject clauses are not. Hence, an adequate grammar of English needs to incorporate a case assignment operation to the effect that the subject of a finite clause (i.e. one containing a finite T constituent) is assigned nominative case.

More generally, let us suppose that the grammar of English incorporates a set of case assignment conditions along the lines of those given in a simplified form below:

(275)

Case assignment conditions in English

A noun or pronoun expression is assigned

a.

nominative case if the specifier of a finite T (i.e. the subject of a finite clause) b.

genitive case if a possessor (i.e. an entity possessing something)

c.

accusative case otherwise (by default, if ineligible for nominative or genitive case)

266

senten ces

It then follows that the subject pronouns in (270) will be assigned nominative case in accordance with the case assignment condition in (275a), since he/we is the specifier of the finite T-auxiliary has/have. By contrast, the subjects of the bracketed infinitive clauses in (274) cannot receive nominative case (because they are not subjects of a finite T constituent) or genitive case (because they are not possessors), and so receive accusative case by default (i.e. as a last resort/

fall-back).

We can illustrate how (275) works more generally in terms of the italicised pronouns in the examples below (A and B represent different speakers in the c, e and f examples):

(276) a.

He has lost his tax return

b.

Remember me to them̩!

c.

A: You’ve been lying to me. B: What̩! Me lie to you? Never̩!

d.

I have never understood syntax, me

e.

A: Who has finished the assignment? B: Me.

f.

A: Who is it? B It’s me.

In (276a), he is nominative because it is the subject of the finite T-auxiliary has, and his is genitive by virtue of its possessive function. In (276b), me and them receive accusative case by default – i.e. by virtue of the fact that neither is used as a finite clause subject or as a possessor: hence, accusative case is said to be the default case in English. In (276c), me is the subject of the non-finite lie clause (lie here is a non-finite infinitive form) and so receives accusative case by default. In (276d), I is nominative by virtue of being the subject of the finite auxiliary have, and the topic pronoun me at the end of the sentence receives accusative case by default. In (276e), me is used as a sentence fragment and hence carries default accusative case. And in (276f), me is used as the complement of the verb be and again carries default accusative case.