Andrew Radford - Linguistics An Introduction [Second Edition]

(493)

quest-o

macchin-a

this-masc.

car-fem.

‘this car’

Notice, however, that in this error (and others like it), the categorial identity of the inflections in question is always respected; that is, agrammatics do not, for example, attach verbal affixes (e.g. infinitive endings) to nouns or adjectives and vice versa.

How can we account for such a selective impairment in syntactic terms?

Like all heads, functional categories are each associated with a set of properties.

A general syntactic property of the category T, for example, is that it always takes a VP as its complement. In addition, as we have seen in section 20, T is specified for abstract grammatical features such as tense ([+Past] or [−Past]), which determine the temporal value of the sentence (e.g. past or present). D, on the other hand, which requires a nominal complement, is associated with features such as number, gender and definiteness. The basic idea is that, in agrammatism, the specific values of the features associated with functional categories are lost or unspecified – in other words, although categories like T or D are present, they are underspecified (see

section 24 for a similar idea in connection with the early speech of children).

Syntactic disorders

381

Consider, for illustration, the syntactic representation of the sentence The boy kissed a girl in normal Standard English (494a) and in agrammatic English (494b): (494)

a.

CP

C

TP

ϕ

DP

T'

D+Def

N

T+Past

VP

the

boy

ϕ

V

DP

kissed

D

N

–Def

girl

a

b.

CP

C

TP

ϕ

DP

T'

D0Def

N

T0Past

VP

ϕ

boy

ϕ

V

DP

kiss

D

N

0Def

ϕ

girl

Compare the feature contents of the D- and T-heads in (494a) and (494b).

Grodzinsky argues that the crucial property in (494b) is that the internal feature specifications of these two heads have unspecified feature values, indicated by the

‘0’ (we adapt the notation to make it consistent with earlier parts of this book).

This means that the D-head and the T-head are left unspecified with respect to definiteness and tense: in contrast to unimpaired English, the head D position of DP in agrammatism is not specified for a definite [+Def] or an indefinite [−Def]

determiner, and likewise the head T position of TP is not specified as carrying a past tense feature [+Past] or a present tense feature [−Past] (this feature ultimately being realised on the main verb if T does not contain an auxiliary). As a

consequence, English-speaking agrammatics leave the functional category

heads empty, which results in ‘telegraphic’ sentences such as Boy kiss girl.

In languages such as Hebrew, Russian and Italian, in which the option of omitting inflections is not generally available, agrammatics randomly choose some inflectional element to fill the slot, and this choice typically results in inflectional errors.

Consider the gender error in (493) above. In Italian, DPs have to be specified for gender features such as [Masc-Gen] or [Fem-Gen], and an expression such as

questa macchina (‘this car’) has the structure in (495):

382

senten ces

(495)

DP

DFem-Gen

N

questa

macchina

Agrammatic patients have lost the values of syntactic features such as gender, and in their grammars, the features have no specifications; see the structure in

(496).

(496)

DP

D0-Gen

N

questo/questa

macchina

This means that agrammatics can attach any kind of gender inflection to the

determiner in the D position. This sometimes produces gender errors as in (493), although Italian agrammatics will also produce the correct agreement pattern questa macchina on occasions. But the option of omitting gender affixes entirely is not available in this case, as this would produce illicit words such as *quest- in Italian, and agrammatics do not violate word-structure contraints of their particular language.

Thus, despite performance differences, i.e. omissions of functional elements in English-speaking agrammatics and inflectional errors in Hebrew- and

Italian-speaking agrammatics, the underlying deficit is the same: the functional categories in their syntactic representations have lost their internal feature specification (exercises 1 and 2).

Paragrammatism

At first sight, the spontaneous speech of Wernicke’s aphasics appears

to be fluent, with normal prosody and syntactic structure. However, although the sentences these patients produce are quite long and complex, they are not always syntactically well formed and contain various kinds of errors, e.g. word exchanges and exchanges of whole constituents as well as blends of different constituents.

This cluster of properties is called paragrammatism in the clinical literature.

Consider as an illustration the various attempts in (497) by a Wernicke’s aphasic to name a lady’s shoe that was shown to him.

(497)

experimenter: What is this? (= a lady’s shoe)

patient:

Yes sir. Now there there I remember. I have you there what

I thought was the … a lady. one. another. with a very short.

very very clever done. do that the one two. go. but there’s the

liver. and there is the new. and so on. It is a document. late …

These utterances are spoken at a very high speed with only a few pauses,

and the sentences are not so much characterised by a reduction of syntactic

Syntactic disorders

383

complexity (as in the case of agrammatism), as by the juxtaposition of

incompatible sequences. There seems to be a consensus among aphasiolo-

gists that paragrammatic errors do not result from an independent syntactic

disorder, but that they are symptomatic of patients’ lexical problems, specifically their word-finding difficulties which we briefly examined in section 15.

It has been found that blends and syntactic errors typically occur at points at which the patient is trying to retrieve content words, particularly nouns.

They start to produce a sentence, and at points at which they experience word-finding problems, change the sentence plan or start again. Crucially, however, the syntactic structure of the various fragments including the internal structure of functional projections is the same as that of normal subjects. Thus, paragrammatism is not a genuine syntactic disorder, but rather a secondary effect of patients’ lexical disorder (exercise 3).

Specific Language Impairment (SLI)

Finally, we will look at the syntactic errors in the speech of specifically

language-impaired children. We will focus on word order, and we will also briefly comment on SLI therapy. In corpora of English-speaking SLI children who

experience great difficulty with inflection and omit functional elements such as determiners or subject–verb agreement markers (see section 15), errors in word order are hard to find. Does that mean that word order is relatively well preserved in SLI? This is not necessarily the case as the word-order system of English is rather simple, and it might well be that SLI subjects do show word-order

problems in a language which has a more complex system. Hence, the ques-

tions we are going to consider are as follows: do SLI children have genuine

word-order problems? And does the picture we get from English-speaking SLI

children hold in general, so that SLI can be said to affect inflection, but not word order?

Let us look at German-speaking SLI children in the light of these questions.

Speech therapists have noticed that the most salient syntactic error in the speech of German SLI children is that they almost always place the verb at the end of the clause, as, for example, in (498a, b):