|AUTHOR: Susan Edwards
TITLE: Fluent Aphasia
SERIES: Cambridge Studies in Linguistics, 107
PUBLISHER: Cambridge University Press
Michael P. Dean, Regional Neurological Rehabilitation Unit, Homerton
University Hospital NHS Trust, London, UK
Susan Edwards aims to cast light on the complex patterns of language
impairment that can result from brain injury, particularly on the
relationships between damage to the grammar and the lexicon and between
fluent and non-fluent forms of aphasia. She reviews empirical studies,
emphasising the value of analyses of continuous speech data, with new
examples, alongside tests that tap or elicit specific aspects of language
processing. A variety of difficulties with morphology, verbs and sentences
are seen in fluent aphasia that cannot be attributed to lexical impairment,
and variation is seen between individuals within the traditional diagnostic
The first chapter reviews attempts, dating back to the 19th Century, to
categorise various fluent aphasias. These include the influential
neoclassical or Boston system with diagnostic categories of Wernicke’s,
conduction, transcortical sensory and anomic fluent aphasia to contrast
with non-fluent categories. The features of language with which these
subgroups are characterised, and the terminology, have not been
consistently used by different authors and over time. There is also an
imperfect relationship between particular aphasia syndromes and sites of
Chapter 2 is entitled ‘Descriptions of fluent aphasia’ and is dense with
ideas and evidence. It introduces Levelt’s (1989) model of sentence
processing, which contains a number of interacting components responsible
for processing different aspects of language. This model has ‘minimal
overlap’ with the framework Edwards will use, which she describes as a dual
language system of lexical and grammatical domains. Both domains are
impaired in fluent aphasia. Edwards next describes evidence that there is
preservation of rate of speech and prosody in fluent aphasia. We then
return to the dual system and deficits to the lexical domain. Phonological
errors occur in fluent aphasia, as do whole word substitutions, often the
substitution of a semantically related word. The latter can be attributed
to a disruption of feature selection with the equal, or superior,
activation of words neighbouring the target. In connected speech it is
often difficult to identify which word is incorrectly produced, as the
speaker’s target is not known. Edwards gives examples, noting that
grammatical category but not always argument structure tends to be
preserved across substitutions. Substitutions of closed class words also
occur, which can less readily be attributed to problems with semantic features.
Next considered is a study by Butterworth and Howard (1989). They propose
a non-linguistic control system to account for the ‘constructional errors’
that occur in fluent aphasia. They argue that the same kinds of errors
occur in fluent aphasia as in unimpaired subjects but at a higher rate, due
to reduced monitoring, and that these errors have a lexical basis. The
most common form of error occurring in this study was of inflection.
Edwards argues that these and another type of error (pronouns illegally in
head positions of relative clauses) are better explained by a grammatical
deficit. She next reviews further evidence for the occurrence of
grammatical errors, including incorrect use of verb arguments.
The third chapter tackles assessment. It begins by highlighting the
importance of differential diagnosis of aphasia and disorders of language
from aetiologies such as dementia. Edwards describes sensory and physical
problems that may accompany aphasia resulting from a stroke (but not the
possibility of concomitant cognitive deficits, or changes in non-verbal
behaviour that are not diagnostic of dementia).
A number of aphasia assessment batteries that result in the examinee being
assigned to one of the (neo-) classical aphasia syndromes are described.
These syndromes are therefore based on correlations between performance on
subtests, and the features contributing to fluent vs. non-fluent
distinctions differ between tests. Example Boston Diagnostic Aphasia
Examination (BDAE) profiles are provided for Edwards’s case MG, assessed
over a number of years. The Psycholinguistic Assessments of Language
Processing in Aphasia (PALPA) do not classify subjects according to
syndrome, but rather reveal patterns of preserved and impaired skill within
a cognitive neuropsychological model of single word processing. The
battery also contains a subtest of sentence comprehension and MG’s scores
are given for this. The Sentence Processing Resource Pack contains a
comprehension test, targeting reversible sentences, and an ‘event
processing’ test, where depicted actions are matched on the basis of shared
features. Edwards argues that the latter might be performed through verb
retrieval rather than non-verbal event processing. (The difference between
fluent and other types of aphasia on these assessments is not discussed.)
Next introduced is the Verb and Sentence Test (VAST). Edwards has worked on
the English version of this. It contains comprehension tests for single
words and sentences, semantic anomaly judgements, sentence anagrams, and
has production tasks. Data are given for a group of people with fluent
aphasia (Wernicke’s and anomic). Both subgroups make reversal errors in
the sentence comprehension task (selecting distractor items that depict the
same main event with swapped thematic roles for participants) rather than
lexical errors. In production tasks, infinitive forms of verbs in
sentences are produced more readily than finite forms. These results
reinforce the view that grammatical processing is impaired in fluent aphasia.
Chapter 4 is on connected speech. The pros and cons of connected speech
samples versus elicited experimental data are discussed. Some studies
report differences and others find converging evidence from different task
contexts. Edwards provides a detailed analysis of four sentences elicited
from one man in a picture description task, discussed in terms of lexical
errors and with reference to generative grammar. Convergence with patterns
observed in test data from the VAST is seen. Inflectional errors also
occur and their patterning is consistent with the split INFL hypothesis
where Tense and Agreement separate. Another explanation considered is that
‘checking’ of the aspectual –ing inflection works for lexical but not
auxiliary verbs, with resultant ungrammatical sentences being produced
rather than ‘crashing’.
The Northwestern and Reading systems for coding samples of spontaneous
speech are introduced, compared and contrasted, with examples. People with
fluent aphasia produce reduced levels of complex structures, as well as
having problems with tense marking and with lexical retrieval.
Linguistic accounts of non-fluent aphasia are described in Chapter 5.
Using Government and Binding Theory, two hypotheses have been advanced.
‘Trace deletion’ is an account for features of agrammatic comprehension,
specifically difficulties understanding reversible sentences and passives.
Reviewing the literature, not all non-fluent aphasics show the patterns of
performance to support this hypothesis, while some people with fluent
aphasia do. Priming studies are described which purport to reveal the
presence or absence of the trace of moved sentence components, with mixed
results. The ‘tree pruning’ hypothesis of agrammatic production stands
Chapter 6 considers comprehension and processing problems in fluent
aphasia. Comprehension difficulties are often seen in the context of good
social interaction skills, therefore an aphasic person’s behaviour may be
interpreted as unwillingness to listen or cooperate. With reduced insight,
they may attribute breakdown in communication to the behaviour of their
communication partners. Edwards gives examples of such maladaptive beliefs
developing. There is a further review of Wernicke’s account of language
comprehension. The idea that comprehension difficulties may result from
impairment to working memory, and thereby hold a relation with short-term
memory difficulties, is next considered. Edwards does acknowledge that the
difference between language-dedicated processes and grammatical operations
is not clear; working memory is regarded here as a general rather than
language-specific process. She then further considers the comprehension of
sentences with canonical and non-canonical word order and the processing of
thematic roles. Again, similarities between people with fluent and
non-fluent aphasias are seen.
Data collected by Edwards and Bastiaanse show that most of their fluent
aphasic subjects have more difficulty comprehending sentences than single
verbs (which points to grammatical and not just lexical impairment) and
again similar performance on different sentence types is seen in fluent and
non-fluent aphasia. Difficulties with thematic role assignment are again
considered, in comprehending these sentence types and in performing a
semantic anomaly judgement task.
Chapter 7 examines more closely the language skills of MG, a man who has
had fluent aphasia for over 7 years. He has consistently been classified
with Wernicke’s aphasia by the BDAE, although there is variability in his
scores over time. He shows semantic and phonemic substitutions and reduced
semantic specificity in tests that elicit nouns and verbs and in continuous
speech. The nature of an underlying semantic impairment is considered.
Other features of MG’s performance point to grammatical problems: in
substituting words he tends to maintain word class (open vs. closed) and he
has difficulty repeating inflected words (compared to words with irregular
past tense and plural forms). In his spontaneous speech he has a reduced
level of embedded sentences and subordinate clauses.
Edwards ends her book in Chapter 8 by addressing the future of aphasia
research. She suggests that functional brain imaging studies must be
mindful of the difficulties in establishing homogeneous groups of subjects,
and that a limited range of language tasks can be investigated by such
measures. Aphasia therapy is another developing area, in terms of
identifying theoretical bases for what works for whom. The final section
is a useful summary of the data reviewed in earlier chapters.
Edwards reviews the literature and presents new data that illustrate an
important point. Fluent forms of aphasia cannot be identified with lexical
impairment alone; there is evidence for concurrent difficulties with a
variety of aspects of grammatical processing. Indeed, some of these
difficulties are not distinct from those exhibited by people with
non-fluent forms of aphasia. Moreover, within groups of people with fluent
aphasia, there is much variability in performance on language tasks; the
diagnostic category ‘fluent aphasia’ as currently used is therefore of
limited value in further characterising the individuals it subsumes.
Reaching this conclusion requires the exploration of complex data. I
enjoyed working through the many examples. The volume would benefit from a
more clearly signposted organisation. Only one level of section heading is
used throughout, and headings do not always make explicit the relationship
between the content that follows and the developing themes. For example,
Chapter 6 contains consecutive sections headed: ‘Understanding words and
sentences’, ‘Comprehension difficulties with sentences’, ‘Comprehension on
single verbs and sentences’ and ‘Comprehending sentences’.
There were a couple of points where I was left with questions. The first
concerns the treatment of some test scores. Tables 6.1 and 6.2 contain
results from 9 people with fluent aphasia on two tests. Edwards reports
that one person (CG) fell within the normal range for verb comprehension.
It is worth noting that 2 others do too (MG and MS). In the analysis of
sentence comprehension scores, chance is taken as 45-55% correct even
though there are 4 response alternatives. This is because few errors were
made to the lexical distractors and so these distractors are discounted.
However, scoring in this way means that two subjects (JoH and IM) score
below the chance range for non-canonical sentences—a result that requires
some explanation. Further, MF is described as approaching above chance for
canonical sentences, yet her score is 50% for these items. When describing
the Verb and Sentence Test, Edwards states that all control subjects
obtained at least 98% correct on all subtests (p.85), yet in Table 3.5 the
control range for the verb comprehension test is given as 38-40 out of 40.
Thirty-eight out of 40 is 95% correct. These apparent errors detract from
the arguments which the figures are used to support.
The second point concerns the comparison of scores, often expressed as
percentage correct, between different tasks. For example, Edwards reports
that subject MG scores 95% correct in comprehending verbs by selecting
action pictures from an array of 4 items, which contrasts with a ‘poor
score’ of 55% correct in naming action pictures, thus demonstrating that
his comprehension of verbs is considerably better than his ability to
produce them in test conditions (p.187). The naming test is, however,
simply more difficult than the comprehension test as it requires semantic
specification of the target to the exclusion of all other possibilities,
while selection from an array of 4 requires semantic specification only to
the exclusion of those 4 items.
Detailed exploration of the exact nature of MG’s lexical difficulties is
not necessary to Edwards’s main thesis; it is the findings of grammatical
impairment and focus on connected speech that are important. However, as
she presents specific test scores and discusses them at some length, I did
feel that this discussion could have made greater reference to accounts of
single word processing. For example, an effect of imageability on MG’s
naming is noted (p.189) but this is not considered as an explanation for
the difference seen in naming objects/nouns (high imageability) versus
verbs (low imageability) without recourse to their grammatical category.
This has been a topic of recent debate (e.g. Berndt, Haendiges, Burton &
Mitchum, 2002; Bird, Howard & Franklin, 2000). On p.210, Edwards appears
to equate the presence of semantic paraphasia with semantic impairment and
phonological paraphasia with phonological impairment. In fact, damage to
phonological output lexical representations, or impaired access to them,
may produce semantic errors, as related representations that are intact or
accessible receive activation (see Nickels, 2004).
I think this book will be of value to those linguists who are concerned to
consider how empirical data (here neurolinguistic and neuropsychological)
may constrain purely formal accounts of language. And for the clinician
who believes the fluent vs. non-fluent distinction in aphasia is equivalent
to lexical vs. grammatical impairment, or paragrammatism vs. agrammatism
for that matter, there is an important message.
Berndt, R.S., Haendiges, A.N., Burton, M.W., & Mitchum, C.C. (2002).
Grammatical class and imageability in aphasic word production: Their
effects are independent. Journal of Neurolinguistics, 15, 353–371.
Bird, H., Howard, D., Franklin, S. (2000). Why is a verb like an inanimate
object? Grammatical category and semantic category deficits. Brain &
Language, 72, 246-309.
Butterworth, B. & Howard, D. (1989). Paragrammatism. Cognition, 26, 1-37.
Nickels, L.A. (2004). How much semantics is enough? The relationship
between comprehension and production in aphasia. In Murdoch, B.E., Goozee,
J., Whelan, B.-M. & Docking, K. (Eds). Proceedings of the 26th World
Congress of the International Association of Logopedics and Phoniatrics.
| ABOUT THE REVIEWER:
ABOUT THE REVIEWER
Michael Dean Ph.D. is a Speech and Language Therapist working with brain
injured adults, in a post-acute neurological rehabilitation unit within the
National Health Service. The unit is located in a culturally and
linguistically diverse part of London. He has research interests in
aphasia and cognitive neuropsychology.