Andrew Radford - Linguistics An Introduction [Second Edition]

c. TABLE

d. APPLE

When the aphasics produced errors in this experiment, it was typically an error of type (207c), i.e. an exchange based on the semantic similarity between the test word (chair) and the name of the depicted object (table) – in this case, the similar items are co-hyponyms. Errors such as (207b), based on phonological similarity, specifically on rhyme, were much less frequent, and wild ‘paraphasias’ such as

(207d) were produced only by severely impaired subjects (note that these subjects did not actually produce these errors in their speech in this study, hence referring to them as ‘paraphasias’ is an extended use of this expression). These findings indicate that the meanings of words and their associative links in the mental lexicon are accessible to Wernicke’s aphasics, and that only in severe cases of vocabulary deficit do the associative processes themselves begin to break down.

The examples cited in the last category in (208) below indicate that the notion of

‘semantic relatedness’ which we are relying on here has to be interpreted fairly liberally if it is not going to exclude significant numbers of cases; the fact remains, however, that the overwhelming proportion of paraphasias do appear to be

explicable in terms of one semantic relation or another.

In sum, the following effects have been found in content-word paraphasias

from aphasics:

(208) I.

Frequency effects:

Low-frequency content words yield more paraphasias than high-frequency

words.

II.

Categorisation-level effects:

a. Hyponym exchanges: sparrow → owl

b. Use of superordinates: sparrow → bird

III.

Similarity effects:

a. Semantic exchanges: hair → comb

b. Pragmatic exchanges: flowers → visit (flowers and visits are often

associated in everyday life)

In general, the content-word usage of Wernicke’s aphasics is markedly poorer than in normal speakers. Thus there are more errors, but the types of errors (as set out in 208) are familiar from normal subjects. When normal speakers are under

Lexical disorders

219

stress, or are distracted or confused, their word usage too is influenced by word frequency and semantic similarity, and they produce errors with all the characteristics of paraphasias (cf. section 14). Thus, there do not seem to exist any qualitative differences in content-word usage between aphasics and normal speakers, and, apart from severe cases of jargon aphasia, the organisational principles of the mental lexicon, in terms of levels of categorisation and associative processes, are not affected by the deficit. We now turn our attention to our second major category of language disorders, SLI (exercises 3 and 4).

Dissociations in SLI subjects’ inflectional systems

There is a consensus that SLI children have problems in the area of

inflectional morphology, and at first sight, the picture we get from examining the language of such children is very similar to that of agrammatism in Broca’s

aphasia. Specifically, SLI subjects often omit grammatical function words and bound morphemes encoding case, gender, number, person etc., or they use them incorrectly. It also seems that in SLI children, the development of inflectional morphology comes to a standstill at an early stage, and that beyond that point the acquisition process cannot advance without difficulties.

Consider the following examples from a ten-year-old SLI child:

(209) a.

you got a tape recorders

b.

the four bus go in Boucherville

c.

when the cup break he get repair

d.

the Marie-Louise look at the bird

e.

the superman is say good-bye and hiding

f.

the ambulance arrive

In these examples, we see problems in number marking within noun expressions (209a, b), an inappropriate pronominal choice (209c), an inappropriate determiner choice (209d), difficulties with participle forms and auxiliary verbs (209e) and in subject–verb agreement (209c, f).

How can we explain the difficulties of SLI subjects in the area of inflection?

One interesting proposal is that SLI individuals’ ability to learn inflectional rules is impaired relative to their ability to memorise and store individual words.

Consider the two inflected verb forms in (209) which are irregular and correct (got and is). By contrast, regular verb inflections such as the third person singular -s are omitted (go, break, get, look, and arrive). These data indicate that SLI subjects can retrieve irregular verbs such as got and is from memory –

equivalently from the relevant lexical representations – but that they cannot generate the third person singular forms of verbs. Recall that we are assuming that these do not appear in lexical representations since they are derivable by regular processes. SLI subjects have problems learning regular inflectional

rules, while at the same time their ability to retrieve irregular forms, which are

220

words

stored as part of a verb’s lexical entry, remains intact. In short, whereas normal speakers appear to possess two distinct psychological mechanisms for inflection, a rule system that attaches regular affixes, e.g. the third person singular -s, to stems, and a set of irregular forms such as got and is which are stored in memory, SLI subjects’ knowledge of inflection is selectively impaired. In

support of this rule-deficit hypothesis on the nature of SLI, it has been reported that SLI subjects produce practically no overregularisations of plural or past tense affixes; such overregularisation would, of course, indicate productive use of these affixes (see section 13). This, then, is a further indication that SLI individuals have more problems with regular rules of inflection than

with accessing irregulars from memory, and it is this selective impairment

which enables us to conclude that two psychological capacities (the ability

to implement rules and the ability to retrieve forms from memory) can be

dissociated.

Results from other SLI studies have indicated that the linguistic deficit is even more selective than has been suggested above. One of these investigated SLI

children’s performance on two regular inflectional affixes, the plural -s (two book-s) and the third person singular present -s (she arrive-s). It was found that the SLI children’s usage of the third person singular present -s was only 36 per cent correct, whereas 83 per cent of their -s plurals were correct, this difference being statistically significant. Despite the fact that both affixes are regular, SLI subjects’

performance with the plural is considerably better than with the tense/agreement suffix. Notice also that the two inflectional morphemes tested in this study are homophones, displaying identical phonologically conditioned allomorphy (see

section 10); this rules out phonological explanations for the observed difference.

Taken together, these findings indicate that the different grammatical functions of the affixes is the controlling factor. SLI subjects seem to be significantly less impaired in their use of noun plurals than in their use of the affix encoding subject–verb agreement and tense. We will come back to these findings in section

26, after we have discussed the structure of the sentences SLI subjects typically produce (exercise 5).

In summary, we have seen that language disorders such as aphasia and SLI do not involve global disruptions of the mental lexicon and the grammar, but rather selective deficits to otherwise normal lexical and morphological systems. In the so-called agrammatic errors produced by Broca’s aphasics, word-structure properties, categorial features and inflectional paradigms are respected, and the notion