You are on page 1of 16

Clinical Linguistics & Phonetics, November 2010; 24(11): 890–905

Revisiting ‘The influence of literacy in paraphasias


of aphasic speakers’

DORA COLAÇO1, ANA MINEIRO1, GABRIELA LEAL2, &


ALEXANDRE CASTRO-CALDAS1
1
Institute of Health Sciences, Catholic University of Portugal, Lisbon, Portugal, 2Department of
Neurology, Santa Maria Hospital, Lisbon, Portugal

(Received 15 March 2010; Accepted 24 June 2010)

Abstract
Literature suggests that illiterate subjects are unaware of the phonological structure of language.
This fact may influence the characteristics of aphasic speech, namely the structure of paraphasias.
A battery of tests was developed for this study to be used with aphasic subjects (literate and
illiterate), in order to explore this topic in more detail. This article aims to present the
experimental design and the results of this test battery composed of two sub-tests: (i) a naming
test with words that belong to three distinct groups: high frequency simple words (HFSW), low
frequency simple words (LFSW), and low frequency complex words (LFCW); and (ii) a word
repetition test. The variables of literacy, frequency and word morphology, and their effect on the
performance of aphasic groups, were correlated in this study. Morphology was the variable that
exercised the greatest influence on the verbal production of the participants.

Keywords: aphasia, literacy, word frequency, morphology

Introduction
Different types of analysis of the errors produced by aphasic patients are presented in this
study. The experiment was designed to allow us, on the one hand, to analyse the differences
between the errors produced by literate aphasic patients and illiterate aphasic patients, while
on the other, to analyse the differences relative to the results of word frequency and their
morphological construction.
This study is intended to be an extension of a previous study started in 2009 (Colaço,
Mineiro, Leal, & Castro-Caldas, 2009). The results of this pilot study steer the hypothesis
that there are differences in error production between literate and illiterate aphasic subjects.
This study intends to continue with this analysis of the types of errors produced by them, as
well as how the lexical variables of word frequency and morphology influence the perfor-
mance of the different groups in these types of task.

Correspondence: Alexandre Castro-Caldas, Institute of Health Sciences, Catholic University of Portugal, Lisbon, Portugal. Tel:
00351 217 214 147. Fax: 00351 217 263 980. E-mail: acastrocaldas@ics.lisboa.ucp.pt
ISSN 0269-9206 print/ISSN 1464-5076 online © 2010 Informa UK Ltd.
DOI: 10.3109/02699206.2010.511406
Influence of literacy in paraphasias 891

For the present study, the sample was widened from 6 to 25 participants, and two groups of
aphasic subjects were evaluated – fluent and non-fluent – in relation to the three distinct
variables of literacy, frequency and morphology of words.
Comparisons between the groups presented were established with the objective of under-
standing the process of lexical access in these subjects, by considering the types of error
produced by them in relation to the different types of word presented. It also intends to
analyse variables which have not yet been intensively researched, such as the effect of
morphology on paraphasias, with the objective of improving the evaluation and rehabilitation
of patients in the area of speech therapy.
Some studies demonstrate that illiterate individuals have difficulty in accomplishing tasks
that require a more explicit perception of language, that is, a certain level of phonological
consciousness (Morais, 1993). This situation has suggested the existence of differences in
phonological processing between literate and illiterate individuals, as was demonstrated by
Ardila, Ostrosky, and Mendoza (2000), and Kosmidis, Tsapkini, and Folia (2006), amongst
other authors.
Returning to this idea, in the study conducted by Castro-Caldas, Petersson, Reis, Stone-
Elander, and Ingvar (1998), the group of illiterate subjects showed slightly inferior results in
the word repetition test, and considerably inferior results in the repetition of non-words, once
again indicating differing competence in certain aspects of phonological processing of lan-
guage. It can therefore be concluded that the use of a direct or phonological route is not taken
due to the absence of visual-graphic reinforcement of the phonemes of oral language.
Through the use of neuroimaging techniques, Petersson, Reis, Askelof, Castro-Caldas,
and Ingvar (2000) observed that the acquisition of a phonological-orthographic correspon-
dence system exerts an influence on the audio-verbal process of spoken language. It appears
that the illiterate subjects lack a visual-graphic system for language, as well as explicit knowl-
edge of sub-lexical segmentation.
These differences highlight the question of cerebral organization referring to these abilities.
Previous studies have demonstrated that literacy is a factor responsible for the typical cortical
representation of language, emphasizing the cerebral dominance in speech (Cameron,
Currier, & Haerer, 1971; Damásio, Damásio, Castro-Caldas, Hamsher, 1979). Cameron
et al. (1971) also introduced the idea that language appears to be ‘bilateralised’ in illiterate
subjects.
These and other studies have contributed to the idea that the bio-functional organisation of
the brain is altered by schooling. However, these functional differences should be analysed
from a qualitative point of view, aware of the necessity of adequation of the materials of
evaluation and intervention for illiterate subjects.
As well as analysing the importance of literacy on the organization of the mental lexicon, we
also intend to study the importance of word frequency in this process. This factor has become
increasingly evident due to the work of various authors (Bormann, Kulke, & Blanken, 2007;
Kittredge, Dell, Verkuilen, & Scwartz, 2008; Morrisset & Gierut, 2002).
The results of these studies have demonstrated that frequent words are less susceptible to
error production than less frequent words, which may lead to the conclusion that the
structure of frequent words tends to remain more intact in the mental lexicon. It can also
be said that word frequency corresponds to a level of activation during lexical access,
influencing verbal production.
Bormann et al. (2007) studied the influence of word frequency by considering the semantic
paraphasias produced by aphasic individuals in a picture naming task. This study found a
significant positive correlation between the frequency of the target word and the number of
892 D. Colaço et al.

correct responses. In other words, more semantic paraphasias were produced when naming
less frequent words. The authors demonstrated that the word frequency has an influence on
the lexical selection of fluent aphasic subjects. The conclusion of this study isn’t compatible
with Levelt’s (1989) and Levelt, Roelofs, and Meyer’s (1999) lexical access model, which
assumes that lexical selection is made at the lemma level, and is not influenced by word
frequency. However, the results of this study support Caramazza’s (1997) cascading model,
as well as the interactive model of Dell (1986).
We also intend to test the effect of the word morphology on aphasic error production. It is
known, often through the sharing of information between clinics, that word morphology
influences the performance of aphasic subjects in verbal production tasks. There is also the
idea that aphasic patients have more difficulty producing morphemes during the production
of sentences than during the production of individual words.
Many of these studies into the morphology of aphasic defects have almost exclusively been
from the perspective of flexional morphology (Jarema, 1998). Information about the influ-
ence of derivational morphology on verbal production in aphasiology is very rare.
In relation to morphology, this study dedicates particular attention to the dichotomy
between simple and complex words (derived and compound), allowing us to better under-
stand linguistic defects due to morphology.

Method
This study presents a non-random sample of aphasic individuals with the objective of
obtaining its results from two formal evaluation tests of verbal oral production: Object
naming and word repetition. The results were described and compared in accordance with
the following variables:
! Level of literacy of the participants;
! Word frequency and
! Word morphology.

Participants
Twenty-five participants of between 47–87 years of age were evaluated, of which 13 were
female and 12 male. These were divided into four sub-groups.
For each of these groups of aphasic individuals – fluent and non-fluent – literate and
illiterate subjects were included. The general groups are: (a) illiterate fluent aphasic subjects;
(b) literate fluent aphasic subjects; (c) illiterate non-fluent aphasic subjects and (d) literate
non-fluent aphasic subjects.
In reference to the group of fluent aphasic subjects, participants with Wernicke aphasia and
conduction aphasia were evaluated. Wernicke aphasia is distinguishable from conduction
aphasia as being characterized by the altered comprehension of verbal material. Difficulty in
verbal repetition is characteristic of both types of aphasia.
The types of non-fluent aphasic subjects included in the study presented identical char-
acteristics. Individuals with Broca aphasia and Motor Transcortial aphasia were included.
These two types of aphasia are characterized by non-fluent speech and good comprehension
of verbal material, only differing as the subjects with Motor Transcortical aphasia have no
difficulty with repetition tasks, contrary to the individuals with Broca aphasia. In the analysis
Influence of literacy in paraphasias 893

Table I. Sample characterisation.

Groups Participants Age Gender Aphasia

Illiterate fluent 1 73 F Conduction


2 72 M Conduction
3 73 M Conduction
4 70 F Wernicke
5 81 F Wernicke
6 72 M Wernicke
Literate fluent 7 69 F Conduction
8 51 M Conduction
9 70 M Conduction
10 74 F Conduction
11 68 F Wernicke
12 66 F Wernicke
Illiterate non-fluent 13 82 F Broca
14 51 M Broca
15 75 F Motor Transcortical
16 73 M Motor Transcortical
Literate non-fluent 17 47 F Broca
18 60 M Broca
19 81 M Broca
20 87 F Broca
21 67 M Motor Transcortical
22 49 M Motor Transcortical
23 82 M Motor Transcortical
24 56 F Motor Transcortical
25 50 F Motor Transcortical

of the errors made during the repetition tests, subjects with Motor Transcortical aphasia were
excluded, as these subjects have no difficulty in completing this type of task.
The characterisation of each of the participants can be found in Table I. The criteria for the
inclusion of the participants in this study were the following:
! Portuguese was the first language (European Portuguese), excluding bilingual indivi-
duals or those fluent in other languages (e.g. immigrants);
! Absence of sensory defects (audio and/or visual) that weren’t corrected;
! Absence of psychiatric or psychological disorders;
! Absence of neurodegenerative disorders (e.g. Dementia, Parkinson Disease) and
! Absence of alcohol and tobacco dependency, toxicodependency, etc.
Criteria relevant to the performance of the participants in a naming test were also established:
! " 75% correct responses; and
! " 75% no response.
All the subjects were evaluated through a formal evaluation using the Lisbon Evaluation
Battery of Aphasia (Bateria de Avaliação de Afasias de Lisboa–BAAL),1 proving the aphasic
diagnostic and respective type. In relation to aetiology, all the subjects have vascular
aetiology.
In the group of literate subjects, only those who had learnt to read at school age were
selected, as other authors (Coppens, Parente, & Lecours, 1998) have already registered
894 D. Colaço et al.

differences in the linguistic performance of individuals who learnt to read as adults and those
who learnt at school age. The criterion for literacy was evaluated by the capacity to recognise
and write letters, not just for the capacity to write their name.
With relevance to schooling, subjects with more than 6 years of schooling were not
included, as their inclusion would introduce variables that could interfere with the literacy
analysis. The economic and cultural level of subjects should be taken into account in this type
of analysis, as it produces differences among the groups; for example, in reading habits,
abstract thought, and potentially other areas (Coppens et al., 1998). The literate subjects were
only selected from lower social–cultural backgrounds, although they had been educated, to
attempt the isolation of the literacy variable within the groups.
The details of the tasks to be completed were explained to each individual with simple
instructions. All the participants accepted agreed to take part in the study.

Instruments for data collection


The naming test included 33 objects of common use. Objects were chosen as stimuli for
verbal production since, according to a study from Reis, Guerreiro, and Castro-Caldas
(1994) of naming drawings and photographed objects, illiterate individuals with no neurolo-
gical lesions present results that are significantly inferior to the results from literate indivi-
duals. In the same study illiterate individuals, as well as literate individuals, correctly named
all of the real objects.
A pre-test with illiterate individuals was conducted before the pilot study, with the goal of
adapting the chosen stimuli for the naming test. After the test had been applied the number of
stimuli was reduced, excluding all those which weren’t named correctly.
Words were selected, pertaining to three separate groups:
(a) High frequency simple words (HFSW) (average frequency: 146);
(b) Low frequency simple words (LFSW) (average frequency: 20) and
(c) Low frequency complex words (LFCW) (average frequency: 12).
As a limit for frequency we chose 50 occurrences – according to the limit established in
Fundamental Portuguese (Bacelar do Nascimento, Garcia Marques, & Segura da Cruz, 1987) –
and as a reference the Contemporary Portuguese Multifunctional Computerized Lexicon (Bacelar
do Nascimento, Casteleiro, Garcia Marques, Barreto, & Amaro, 1997).
The Contemporary Portuguese Multifunctional Computerized Lexicon was extracted from the
sub-corpus of the Contemporary Portuguese Reference Corpus and, although containing oral
performances corresponding to only 5% of the total corpus, it is regularly used by aphasiol-
ogists in Portugal who work in this area of research.
This corpus covers a great variety of linguistic diversity, covering a large proportion of the
journalistic corpus, with the objective of establishing a common language and covering an
enormous diversity of themes. As well as the frequency of occurrences, it also contains the
phonetic transcription and morphosyntactic classification of all the words.
The objects used in the naming test according to the parameters of frequency and mor-
phology are presented in Table II.
The order established for presenting stimuli was random, that is, the items in the three word
groups were mixed-up. However, the order was kept the same for every participant, thus allowing
us to match the conditions regarding possible verbal perseverations and priming effects.
The word repetition task corresponds to the word repetition test used in the Aachen Aphasia
Test (Portuguese version). Nevertheless, only the simple word repetition test and complex
Influence of literacy in paraphasias 895

Table II. Objects used in the naming test.

High frequency simple words Low frequency simple words Low frequency complex words

! garrafa (bottle) ! banana (banana) ! açucareiro (sugar bowl)


! pêra (pear) ! gilete (razor) ! secador (hair drier)
! gravata (tie) ! pilha (battery) ! chupa-chupa (lolly-pop)
! toalha (towel) ! espelho (mirror) ! porta-chaves (keychain)
! tesoura (scissors) ! cachecol (scarf) ! despertador (alarm clock)
! fósforo (match) ! cadeado (padlock) ! sabonete (soap)
! pente (comb) ! babete (bib) ! fita-cola (sellotape)
! cinto (belt) ! rebuçado (sweet) ! esfregão (scourer)
! frasco (prescription bottle) ! alfinete (pin) ! borrifador (sprinkler)
! garfo (fork) ! alicate (pliers) ! saleiro (salt-cellar)
! anel (ring) ! régua (ruler) ! esfregona (mop)

Table III. Words used in the repetition test.

Simple words Complex words

! Luz (light) ! Autocarro (bus)


! Alma (soul) ! Decreto-lei (ordinance law)
! Pasta (folder) ! Passatempo (hobby)
! Folha (sheet) ! Conta-quilómetros (speedometer)
! Sonho (dream) ! Banda desenhada (comic)
! Jardim (garden) ! Ferro de engomar (iron)
! Mestre (master) ! Desenvolvimento (development)
! Quarto (room) ! Décimo terceiro mês (thirteenth month)
! Carvão (coal) ! Irresponsabilidade (irresponsibility)
! Comboio (train) ! Marinheiro de água doce (fresh water sailor)

word repetition test were used, and the stimuli of these two tests appeared in an alternating
way during its application (simple word–complex word–simple word, etc.).
The words used in the repetition test, separated into simple and complex words, are
presented in Table III.

Procedure and analysis


Each participant was evaluated individually. In the naming task the patient was asked to name
each object. The answers were recorded on paper and in audio format. In the case of phono-
logical paraphasias, phonetic transcription was utilised using the International Phonetic Alphabet
(IPA).
The errors were classified according to the description presented in Table IV.

Results
In the naming test, of the 825 responses expected (25 participants # 33 items), a total of 711
responses were given, of which 464 were wrong answers. In the repetition test, in which 360
responses were expected (18 participants # 20 items), 329 answers were given, of which 223
were classified as errors.
896 D. Colaço et al.

The analysis of the results was centred on the comparison of the effects of the lexical
variables (word frequency and morphology) on the different groups. The results are pre-
sented as the number of correct responses produced by the different groups in the naming test
(Tables V and VI) and the repetition test (Tables VII and VIII). The descriptions of the types
of errors produced by these groups are represented schematically, where the average values of
each group can be found.

Table IV. Types and examples of errors.

Error types Error subtypes Definition Examples

Morpho- Phonological paraphasia When a switch, omission or addition of /ɐli’tεkɨ/ for pliers
phonological phonemes has occurred. (alicate /ɐli’katɨ/ in
paraphasia Portuguese)
Morphological When a switch, omission or addition of ‘responsibility’ for
paraphasia affixes has occurred. irresponsibility
Lexical- Same semantic level When the target-word was replaced by ‘comb’ for mirror
semantic another of the same category.
paraphasia Function description Description of the object by its function. ‘to shave’ for razer
Attribute description Description of the object by an attribute. ‘it is made of wool, using
needles’ for scarf
Superordinates A semantic category which encompasses all ‘sweet’ for lolly-pop
the elements of the determined category.
Paraphasias with part- When the word production corresponds to ‘sugar’ for sugar bowl
whole relationship a part of the object or something which can
be added to this object.
Unclassifiable Paraphasias with When the given response has no clear ‘lift’ for prescription bottle
errors doubtful relationship relationship with the target word.
Neologisms When the error is big enough that the given /kuRi’dɐ̃w/ for tie
response is unintelligible and without any (gravata /grɐ’vatɐ/ in
resemblance to the target word. Portuguese)
Circumlocutions Comments made about the object using ‘It is pretty, I like it’ for
expressions indirectly related to the target bib
word.
Verbal perseveration When the participant repeats the previous
verbal production
Non answers When the participants
did not answer or said
that they could not
answer.

Table V. Fluent aphasic subjects: correct responses in the naming test.

Illiterate fluent aphasic Literate fluent aphasic


Word types subjects (n ¼ 6) subjects (n ¼ 6)

High frequency simple words 4.33 % 4.08* (0–9)** 3.67 % 2.94 (0–8)
Low frequency simple words 3.33 % 3.61 (0–9) 2.17 % 1.72 (0–5)
Low frequency complex words 1.77 % 2.25 (0–6) 1.00 % 1.26 (0–3)

*M % SD;
**Minimum–Maximum.
Influence of literacy in paraphasias 897

Table VI. Non-fluent aphasic subjects: correct responses in the naming test.

Illiterate non-fluent Literate non-fluent


Word types aphasic subjects (n ¼ 4) aphasic subjects (n ¼ 9)

High frequency simple words 6.50 % 1.29* (5–8)** 6.78 % 3.03 (2–11)
Low frequency simple words 3.00 % 0.82 (2–4) 5.11 % 2.80 (1–9)
Low frequency complex words 2.25 % 1.26 (1–4) 2.00 % 2.45 (0–7)

*M % SD;
**Minimum–Maximum.

Table VII. Fluent aphasic subjects: correct responses in the repetition test.

Illiterate fluent aphasic Literate fluent aphasic


Word types subjects (n ¼ 6) subjects (n ¼ 6)

Simple words 5.67 % 2.94* (1–10)** 6.00 % 2.97 (1–10)


Complex words 1.17 % 0.98 (0–2) 1.83 % 1.72 (0–4)

*M % SD;
**Minimum–Maximum.

Table VIII. Broca aphasic subjects: correct responses in the repetition test.

Illiterate subjects with Literate subjects with


Word types Broca aphasia (n ¼ 2) Broca aphasia (n ¼ 4)

Simple words 9.50 % 0.71* (9–10)** 8.75 % 0.96 (8–10)


Complex words 1.00 % 0 (1–1) 1.50 % 1.29 (0–3)

*M % SD;
**Minimum–Maximum.

In the analysis of the results of the naming test produced by the fluent aphasic subjects
(Table V), it was verified that both the literate and illiterate aphasic subjects produced a larger
number of correct responses for each type of word.
It can also be noted that both the literate and illiterate aphasic subjects produced a larger
number of correct responses for HFSW than for the LFSW, verifying the effect of frequency
on verbal production, though with little weight. In reference to the analysis of morphology, it
was also verified that LFCW caused more difficulty than LFSW, thus proving the effect of
morphology (Table V).
The same result was observed in the group of non-fluent aphasic subjects (Table VI).
However, the differences in the results between HFSW and LFSW were much larger than in
the group of fluent aphasic subjects. In reference to the influence of morphology, the results
appear to be identical between the fluent and non-fluent aphasic subjects.
In the group of non-fluent aphasic subjects, it was observed that word frequency had an
effect on the illiterate aphasic subjects, as there were differences between the results of the
HFSW and LFSW production, which wasn’t as evident in the group of non-fluent literate
aphasic subjects (Table VI).
898 D. Colaço et al.

Analysing Figure 1, it can be seen that the errors produced by the fluent illiterate aphasic
subjects are similarly distributed among each type of incorrect response (morpho-phonological
paraphasias, lexical-semantic paraphasias, unclassifiable errors and non-answers), except for the
lexical-semantic paraphasias. This type of error was observed more frequently in the less
frequent words (LFSW and LFCW) than in the high frequency words (HFSW).
Another important aspect that corresponds to the most frequent type of error in this group –
illiterate fluent aphasic subjects—is the production of unclassifiable errors (Figure 1).
Comparing this group (illiterate fluent aphasic subjects) with the group of literate aphasic
subjects (Figure 2), it was verified that the literate aphasic subjects also produced a large
number of unclassifiable errors, although less than the group with illiterate subjects, and
morpho-phonological paraphasias were most evident, principally in the less frequent words
(LFSW and LFCW). Strangely, identical numbers of morpho-phonological paraphasias and
unclassifiable errors were produced in HFSW (Figure 2).

Figure 1. Illiterate fluent aphasic subjects: types of error produced in the naming test.

Figure 2. Literate fluent aphasic subjects: types of error produced in the naming test.
Influence of literacy in paraphasias 899

In relation to the groups of non-fluent aphasic subjects, a larger variety of errors were seen
according to the types of word. The group of non-fluent aphasic subjects produced more
lexical-semantic paraphasias, in the HFSW and the LFSW. In the LFSW there were a larger
number of non-answers. Unexpectedly, this group produced more unclassifiable errors in the
simple words (HFSW and LFSW) than in the complex words (LFCW) (Figure 3).
In the group of literate non-fluent aphasic subjects (Figure 4), it became evident, once
again, that the complex words proved to be the most difficult, producing the most lexical-
semantic paraphasias and non-answers.
In the naming test, the results reveal differences between fluent aphasic subjects and Broca
aphasic subjects, verifying that Broca aphasic subjects produced more correct answers with
simple words than fluent aphasic subjects. In relation to the difference between the simple
and complex words, the influence of morphology on all the groups was observed (Tables VII
and VIII).
It was also verified that the illiterate Broca aphasic subjects produced more correct
responses than the literate Broca aphasic subjects, though with little weight (Table VIII).

Figure 3. Illiterate non-fluent aphasic subjects: types of error produced in the naming test.

Figure 4. Literate non-fluent aphasic subjects: types of error produced in the naming test.
900 D. Colaço et al.

Figure 5. Illiterate fluent aphasic subjects: types of error produced in the repetition test.

Figure 6. Literate fluent aphasic subjects: types of error produced in the repetition test.

Analysing the types of errors produced by the group of illiterate aphasic subjects (Figure 5),
a large variety of errors became evident, principally in the repetition of complex words, which
was also seen in the group of literate aphasic subjects. However, many lexical-semantic
paraphasias were produced in both groups, with a greater number in the illiterate aphasic
subjects, which wasn’t expected in a word repetition test (Figures 5 and 6).
In the groups of Broca aphasic subjects, the production of errors occurred most frequently
in the group of complex words, but once again lexical-semantic paraphasias occurred in the
two groups (Figures 7 and 8).

Discussion
Some of the results obtained in this study confirm the findings of previous studies; however,
others deserve to undergo a more detailed analysis as they present us with interesting and as
yet under-debated data.
Influence of literacy in paraphasias 901

Figure 7. Illiterate Broca aphasic subjects: types of errors produced in the naming test.

Figure 8. Literate Broca aphasic subjects: types of error produced in the naming test.

In general and relative to the types of error produced by the different groups it can be
concluded that the most recurrent errors of the four groups were those produced in the
complex words, indicating that morphology is the principle factor in error production.
Surprisingly, the word frequency didn’t seem to affect the performance of most of the
groups, contrary to studies described in the literature, specifically Leal (2003), Bormann et al.
(2007) and, amongst others, that found a correlation between word frequency and a better
performance in word production.
There were small differences between the average number of correct results in the LFSW
and HFSW in all the groups. In spite of demonstrating a small tendency towards correct
answers in the simple words, the difference is slight.
Perhaps this result is attributable to the fact that the linguistic corpora constructed for
contemporary European Portuguese is largely made up of large collections of written
902 D. Colaço et al.

linguistic data. Could it be that the notion of frequency is basically a notion of the frequency of
written words? What then is the impact of writing on the criteria for word frequency while
factoring in the influence on the lexical selection of aphasic subjects? This is a question to be
studied carefully in the future.
While historically the English language corpora (e.g. SEU–Survey of English Usage, Brown
Corpus and Birmingham Collection of English Texts) has a strong inclination to balance the
design of written and oral material, the Portuguese corpora doesn’t follow this tendency.
As well as insufficient oral material, the inclusion of words that reflect contexts of day-to-
day life is also under-represented in the Portuguese corpora. In this study, objects in common
use and recognisable by the majority of people were included, objects that we all use in our
daily routines, whether we are literate or illiterate. This type of lexicon is perhaps insufficiently
represented in the Portuguese corpora, which probably explains these results, in which
frequency isn’t a distinct factor.
The influence of word frequency can also be analysed by studying lexical processing and
the effect of frequency on this process. Even though various authors have defended the
influence of word frequency on lexical access, it would be of great interest to discover the
specific level of access on which word frequency has an effect.
Levelt’s (1989) and Levelt et al.’s (1999) access model, the discrete two-stage model,
assumes one level of lexical concepts and two levels of lexical representations, lemmas and
word forms. In this model, there are no feedback connections between word forms and
lemmas. We don’t know to what extent the differences between this model and
Caramazza’s (1997) cascading model and the interactive model of Dell (1986) reflect the
influence of word frequency; however, it is a factor that should be investigated further.
It is also important to distinguish word frequency from age of acquisition, as has been
attempted in previous studies (e.g. Kittredge et al., 2008), and to continue to investigate the
importance of this variable during lexical access.
Specifically, it is particularly interesting to highlight the differences between literate and
illiterate aphasic subjects. Literate aphasic subjects (fluent and non-fluent) produce a larger
number of phonological errors than illiterate aphasic subjects (fluent and non-fluent), reflect-
ing organised lexical access through a strategy of phonological programming, as was sug-
gested in Castro-Caldas (2000). This data reveals that illiterate aphasic subjects do not seem
to think in terms of word segmentation, possibly as they do not make the grapheme–phoneme
association. Their phonological errors seem to arise from a deficit in lexical–phonological
access, for example, /’f‫כ‬rf/ for ‘fósforo’ (match). In the case of literate aphasic subjects,
phonological errors seem to lie in a difficulty in phonemic planning, which is what happens
in the case of /alpi’netɨ/ for ‘alfinete’ (pin). Such a dichotomy in phonological access errors and
planning errors is described in the literature by Kohn and Smith (1994).
The illiterate aphasic subjects appear to utilise, with more frequency than the literate aphasic
subjects, non-phonological strategies for naming the presented objects, and as such commit
other types of errors, such as diverse semantic errors (e.g. the substitution of pear for ‘apple’),
neologisms (e.g. /pɐtɐ’titu/ for ‘cadeado’ (/kɐdi’adu/–Portuguese for padlock) or circumlocutions
(e.g. ‘it’s very beautiful’, ‘I would really like it’, ‘I know, I know, I knew, now it’s very difficult’).
The variable of word morphology, with a distinction between simple and complex words,
appears to have affected all the groups in the analysis, as observed in the worse performance of
the literate fluent group in naming complex words than the illiterate aphasic subjects. In the
case of the literate subjects, lexical access and lexical recovery appear to be based on morpho-
phonological planning, whereas it is based on strategies of a different nature (semantic or
other) in the case of the illiterate subjects.
Influence of literacy in paraphasias 903

In the word repetition test the effect of morphology on verbal production is well known, and
this study also shows a considerable difference in results between simple and complex words.
In relation to the type of errors produced in this test, there is some interesting new data,
specifically the production of lexical-semantic paraphasias. Independent of the group, this
type of paraphasia was unexpected in a word repetition test, which signifies that a semantic
activation to an audioverbal stimulus occurred in which the objective was to only reproduce
what was heard. Initially, this brings to mind different strategies for different types of input –
visual (real object naming) vs. audio (repetition of words produced by the researcher), which
wasn’t observed in this study.
Comparing the data of the two groups of fluent aphasic subjects (literate and illiterate) in
the word repetition test, it was verified that the illiterate aphasic subjects produced more
lexical-semantic errors than their literate opposites and also a larger number of unclassifiable
errors. This fact could be interpreted as a lexical selection strategy supported by semantic
relations and more ‘diffuse’ activations, such as circumlocutions, instead of the morpho-
phonological approximations produced by the literate aphasic subjects.
Making a general analysis between the results of the fluent aphasic subjects and the non-
fluent aphasic subjects, there is a definite difference in the number of correct responses of
each group, with a noticeable advantage in the group of non-fluent aphasic subjects. This
analysis could have taken into account the aphasia quotient (AQ) of the subjects, however it
appears as if the differences in the types of error produced by these groups reflect the
descriptions of the types of aphasia. The group of non-fluent aphasic subjects produced a
larger number of non-answers and lexical paraphasias, while in the group of fluent aphasic
subjects the quantity of non-answers is much less, and a larger number of morphological
paraphasias were observed.
As the fluency was unaffected, the fluent aphasic subjects attempted approximate produc-
tions of the target words, in many cases producing paraphasias with a doubtful relationship
with the target word, or circumlocutions (with a greater quantity in the illiterate group), but
also referred morpho-phonological paraphasias (with a greater quantity in the literate group).

Conclusions
We consider that the analysis of the types of error produced by the sample in this study is a
contribution to the linguistic knowledge of the differences between literate and illiterate
subjects, and to the importance of word frequency and morphology in the preparation of
the material for the evaluation and intervention of these types of patients.
In this study, the illiterate aphasic participants gave surprising results in the sense that they
had less difficulty, compared with the literate group, in the naming of complex words.
Could it be that the results shown by the illiterate aphasic subjects are anchored in storing
strategies, processing and lexical retrieval, different to those whose brain has been ‘reorga-
nised’ by their access to reading and writing at a precocious phase of their existence?
This work emphasises the importance of studying morphology, specifically regarding word
formation and the understanding of the real impact of these sorts of stimuli in the linguistic
defect caused by aphasia. In the future we intend to study this dimension, bearing in mind
that it has persistently been forgotten by researchers, who have been focused almost exclu-
sively on flexional morphology. The importance of studying the correlation between naming
defects and distinct morphological stimulus will possibly shed some light on the areas of
choice of stimulus and development of materials of evaluation and therapeutic intervention
for aphasia.
904 D. Colaço et al.

Another finding of this study concerns the fact that the results were contrary to the
tendency described in literature of what concerns the impact of word frequency, specifically
the occurrence of fewer paraphasias in relation to frequent lexical items. This data appears to
be particularly relevant in the sense that we should ask about the ‘type of frequency’ that is
represented in the linguistic corpora of European Portuguese.

Acknowledgements
We wish to acknowledge the collaboration of all the aphasic participants of this study. We
would also like to thank the six SLTs who collected the data from the aphasic participants,
especially Inês Rodrigues, member of the Cognitive Neurosciences Group (GNC) of
Catholic University of Portugal. Thanks also to the other members of this group for their
helpful comments and suggestions. Finally, we would like to thank Richard Esdaile for his
help revising this article in English.
Declaration of interest: The authors report no conflicts of interest. The authors alone are
responsible for the content and writing of the paper.

Note
1. BAAL (Castro-Caldas, 1979; Damásio, 1973; Ferro, 1986) evaluates the speech of adults, and is the only battery
validated for the Portuguese population. This battery follows the taxonomic criteria that make it possible to
classify the classic types of aphasia.

References
Ardila, A., Ostrosky, F., & Mendoza, V. (2000). Learning to read is much more than learning to read: A
neuropsychologically-based learning to read method. Journal of the International Neuropsychological Society, 6,
789–801.
Bacelar do Nascimento, M.F., Casteleiro, J., Garcia Marques, M.L., Barreto, F., & Amaro, R. (1997). Léxico
Multifuncional Computorizado do Português Contemporâneo. Lisbon: Programa Praxis.
Bacelar do Nascimento, M.F., Garcia Marques, M.L., & Segura da Cruz, M.L. (1987). Português Fundamental –
Métodos e Documentos. Tomo 1. Lisbon: INIC e CLUL.
Bormann, T., Kulke, F., & Blanken, G. (2007). The Influence of word frequency on semantic word substitutions in
aphasic naming. Aphasiology, 22, 1313–1320.
Cameron, R.F., Currier, R., & Haerer, A.F. (1971). Aphasia and Literacy. Journal of Language & Communication
Disorders, 6, 161–163.
Caramazza, A. (1997). How many levels of processing are there in lexical access?Cognitive Neuropsychology, 14, 177–208.
Castro-Caldas, A. (1979). Diagnóstico e evolução das afasias de causa vascular. PhD Dissertation. Lisbon: Faculty of
Medicine.
Castro-Caldas, A. (2000). A herança de Franz Joseph Gall. Lisbon: McGraw-Hill.
Castro-Caldas, A., Petersson, K.M., Reis, A., Stone-Elander, S., & Ingvar, M. (1998). The illiterate brain: Learning
to read and write during childhood influences the functional organization of adult brain. Brain, 121, 1053–1063.
Colaço, D., Mineiro, A., Leal, G., & Castro-Caldas, A. (2009). The influence of literacy in paraphasias of aphasic
speakers. In V. Marrero & I. Pineda (Eds.), Linguistics: The challenge of clinical application (pp. 151–157). Madrid:
Euphonia Editions.
Coppens, P., Parente, A., & Lecours, K. (1998). Aphasia in illiterate individuals. In P. Coppens, Y. Lebrun, &
A. Basso (Eds.), Aphasia in atypical populations (pp. 175–202). London: Lawrence Erlbaum Associates.
Damásio, A.R. (1973). Perturbações neurológicas da linguagem e de outras funções simbólicas: contribuição do
estudo clínico e laboratorial das afasias, apraxias e outras disfunções da actividade superior, para o conhecimento
da semiologia e da fisiopatologia do sistema nervoso. PhD Dissertation. Lisbon: Faculty of Medicine.
Damásio, H., Damásio, A.R., Castro-Caldas, A., & Hamsher, K. (1979). Reversal of ear advantage for phonetically
similar words in illiterates. Journal of Clinical Neuropsychology, 1, 331–338.
Influence of literacy in paraphasias 905

Dell, G.S. (1986). A spreading-activation theory of retrieval in sentence production. Psychological Review, 93, 283–321.
Ferro, J.M. (1986). Neurologia do comportamento – Estudo da correlação com a tomografia axial computorizad, PhD
Dissertation. Lisbon: Faculty of Medicine.
Jarema, G. (1998). The Breakdown of morphology in aphasia. In B. Stemmer & H.A. Whitaker (Eds.), Handbook of
Neurolinguistics (pp. 221–234). San Diego: Academic Press.
Kittredge, A., Dell, G.S., Verkuilen, J., & Scwartz, M. (2008). Where is the effect of frequency in word production?
Insights from aphasic picture-naming errors. Neuropsychology, 25, 463–492.
Kohn, S.E., & Smith, K.L. (1994). Distinctions between two phonological output deficits. Applied Psycholinguistics,
15, 75–95.
Kosmidis, M.H., Tsapkini, K., & Folia, V. (2006). Lexical processing in illiteracy: Effect of literacy or education?
Cortex, 42, 1021–1027.
Leal, G. (2003). A influência da frequência de uso das palavras na capacidade de nomeação dos afásicos.
Undergraduate dissertation. Alcoitão: Escola Superior de Saúde do Alcoitão.
Levelt, W.J.M. (1989). Speaking: From intention to articulation. Cambridge, MA: MIT Press.
Levelt, W.J.M., Roelofs, A., & Meyer, A.S. (1999). A theory of lexical access in speech production. Behavioral and
Brain Sciences, 22, 1–75.
Morais, J. (1993). Phonemic awareness language and literacy. In R.M. Joshi & C.K. Leong (Eds.), Reading disabilities:
Diagnosis and component processes (pp. 175–184). Dordrecht: Kluwer Academic.
Morrisset, M., & Gierut, J. (2002). Lexical organization and phonological change in treatment. Journal of Speech,
Language, and Hering Research, 45, 143–159.
Petersson, K., Reis, A., Askelof, S., Castro-Caldas, A., & Ingvar, M. (2000). Language processing modulated by
literacy: A network analysis of verbal repetition in literate and illiterate subjects. Journal of Cognitive Neuroscience,
12, 364–382.
Reis, A., Guerreiro, M., & Castro-Caldas, A. (1994). Influence of educational level of non brain-damaged subjects on
visual naming capacities. Journal of Clinical and Experimental Neuropsychology, 16, 939–942.

You might also like