Andrew Radford - Linguistics An Introduction [Second Edition]

(204)

andare

ospedale.

Non

credere

parola

to-go

hospital.

Not

to-believe

word

(205)

capucetto

rossa (capucetto rosso would be correct)

riding hood-masc.

red-fem.

(Note that both 204 and 205 would be marked with a * in standard Italian – the lack of annotation here indicates that the expressions do occur in agrammatic Italian speech.)

Furthermore, the range of errors that aphasics produce is rather restricted and narrowly constrained. Erroneous infinitive inflections occur, as in (204), but only on verbs and never on nouns, and gender mistakes, such as that in (205), are also found, but only on nouns and adjectives and never on verbs. This observation suggests that significant remnants of Italian morphology remain in place.

Studies on other languages, e.g. French, Hebrew and Russian, lead to the same outcomes as may be derived from Italian and justify a number of general conclusions. Specifically, it appears that agrammatics respect:

216

words

a.

the word-structure properties of their native language;

b.

the categorial features of bound morphemes;

c.

inflectional paradigms.

We shall now say a little more about each of (a) to (c).

The generalisation in (a) covers the fact that agrammatics never produce words, stems or roots that would violate word-structure properties of their language.

Thus, bound inflectional morphemes are dropped in English-speaking agram-

matics, but the consequence of this is the occurrence of stems which can function as words (e.g. walks → walk). However, such morphemes are not dropped, for

example, in the speech of Hebrew-speaking agrammatics. It therefore seems that the broad distinction between word-based morphology (English) and stem-based morphology (Italian, Hebrew) is retained in the grammars of agrammatics.

The generalisation in (b) describes the fact that agrammatics seem to know the categorial identity of affixes, in the sense that they retain knowledge of the categories to which specific affixes can be attached. Thus, verb inflections, e.g.

infinitive endings, are only attached to verbs, never to nouns; conversely, case suffixes are never attached to verbs but only to nouns.

The third observation in (c) is that agrammatics still have inflectional paradigms. What this acknowledges is that many of the inflectional errors agram-

matics produce are exchanges between individual cells of morphological

paradigms, e.g. feminine gender is incorrectly used instead of masculine gender, as in (205). It is important to be clear that this is a stronger generalisation than (b), which does not rule out the replacement of one type of nominal affix by another nominal affix, say, replacing a gender affix by a number affix. But, in fact, this does not occur and the contents of the inflectional paradigms are typically intact.

To take a particular case, it is as if the agrammatic knows that case affixes attach in a specific slot, but makes incorrect choices from the available set of case affixes.

Taken together, these findings indicate that agrammatism cannot be accounted for in terms of a global simplification process by which functional elements are simply deleted from the linguistic output. Rather, the linguistic impairments are more specific, and the proper understanding of agrammatism requires notions

such as word-structure properties, categorial features and morphological paradigms. Linguistic theories of agrammatism will be considered in section 26, after we have extended our discussion to include the syntactic disorders that occur in these patients.

Another myth of the clinical classification of aphasias is that Broca’s aphasia is mainly a production disorder and that comprehension is largely unimpaired in these patients. This view was mainly based on lack of knowledge, specifically on the fact that in the clinical interview, comprehension is not systematically studied.

Rather, clinicians ask patients everyday questions such as How did the stroke come about?, and agrammatic patients answer such questions appropriately. But this does not mean that comprehension is unimpaired, as the meaning of such

questions could be directly inferred from the meaning of the content words and the

Lexical disorders

217

context in which the question is posed. An important feature of English telegrams, which gave rise to the characterisation of agrammatic speech as ‘telegraphic’, was the omission of function words (e.g. ARRIVE HEATHROW TOMORROW 3PM

STOP HEAVY BAGS PLEASE MEET STOP JOHN), and such telegrams were

typically understood by their recipients, giving a clear indication that the presence of function words is not always necessary for understanding to occur. In the 1970s, aphasiologists started to carry out experimental studies on agrammatism, the results of which clearly demonstrated that agrammatics have comprehension

problems with functional elements which are similar to those they show in

production (exercises 1 and 2).

Paraphasias

Errors in the use of content words, i.e. paraphasias, are reported to be

characteristic of Wernicke’s aphasics. What kind of content words cause difficulty and how can we account for the error patterns? First of all, performance of

Wernicke’s aphasics on content words is affected by the frequency of the word in the vocabulary: infrequent words take longer to retrieve and are more often inaccurately retrieved than frequent words. Secondly, and more importantly, the typical error patterns that occur in paraphasias can be explained in terms of the structures which characterise the mental lexicon such as we have already met in the previous three sections of this part of the book. Consider the data in (206) from object-naming experiments; in such experiments, subjects are presented with a picture of an object and are simply asked to name it:

(206)

target picture: SHARK

subjects’ responses:

a. fish

b. trout

c. guitar

d. rainbow trout

Among these responses, (206a, b) represent the common types, and we can understand what is going on here by referring to section 13 where we distinguished between three levels of categorisation in taxonomies: the basic level (where we find such words as trout, shark and guitar), the superordinate level ( fish, musical instrument, fruit, etc.) and the subordinate level (rainbow trout, great white shark, bass guitar, etc.). These notions, as well as being significant in understanding the child’s acquisition of words, have also proved important in the study of how visually presented objects are categorised by normal adult subjects.

Such subjects typically categorise an object (e.g. by naming it) at the basic object level, despite the fact that logically it could be categorised at a variety of other levels. In object-naming experiments with Wernicke’s aphasics by contrast, the subject’s typical naming response to the picture of a shark is either the superordinate level term (fish) or a prototypical element from the basic set (trout) (see

218

words

section 12 for the notion of prototypicality). Wild paraphasic misnaming as in

(206c) occurs only in severely impaired subjects, and responses at the subordinate level such as that in (206d) are practically non-existent.

In another set of experiments, the role of phonetic and semantic similarity in aphasics’ perception of category names has been tested. Aphasics were asked to match a picture of an object from a set of multiple-choice pictures to a test word presented orally by the experimenter. A typical situation is schematised in (207):

(207)

test word: chair

subjects’ choices: a picture of

a. CHAIR

b. STAIR