I
Brain (1969) 92, 885-896.
I
I
I
I
THE SELECTIVE IMPAIRMENT OF AUDITORY VERBAL
SHORT-TERM MEMORY
BY
ELIZABETH K. WARRINGTON AND T. SHALLICE
(From The National Hospital, Queen Square, London, W.C.I)
INTRODUCTION
THE inability to repeat verbal stimuli has been recorded in the neurological
literature as a relatively isolated disability. Wernicke (1874) regarded impairment
of repetition as the cardinal feature of conduction aphasia and more recently Goldstein
(1948) has discussed repetition defects in the context of central aphasia. Both
authors discuss it as a variety of dysphasia. There have been a number of descriptions
of the symptomatology of conduction aphasia, the more recent including Dubois
et al. (1964), Konorski et al. (1961), and Geschwind (1965). All these authors
emphasize that there is no impairment of the comprehension of speech although the
ability to repeat verbal stimuli is markedly impaired. In addition to the repetition
defect these patients are reported to have great difficulty with spelling and
spontaneous writing (copying writing is intact), and much less difficulty with reading.
Expressive and nominal difficulties are present but not necessarily to any marked
extent.
Adequate repetition depends on a number of factors, the most important of which
are acoustic perception and motor speech capacity. The question arises as to
whether defects of repetition occur which cannot be regarded as secondary to other
disabilities within the language system and if so what functional systems are impaired.
Few analytical studies of tasks of repetition have been reported. Patients with left
hemisphere lesions are known to have a poor digit span (McFie, 1960), a task which
demands accurate repetition, but the relationship of this defect to other aspects of
language and memory was not examined. Dubois et al. (1964) undertook a linguistic
analysis of the speech (spontaneous and repeated) of 3 patients with conduction
aphasia, and stress that in repetition the difficulty of the task is related to the amount
of information in the message rather than the length of the message. Luria et al.
(1967) report 2 patients in whom the main symptom was an inability to reproduce
a series of phonemes, words and digits presented orally, though there was relatively
little impairment with visual presentation of the same verbal stimuli. Though this
modality-specific defect of verbal repetition is described as a disturbance of
atUniversityofNottinghamonOctober2,2011brain.oxfordjournals.orgDownloadedfrom
886 ELIZABETH K. WARRINGTON AND T. SHALLICE
"memory" no attempt was made to assess other factors, and the widely accepted
differentiation of memory functions into a long-term and short-term component
was not considered.
The digit span is a task which many experimental psychologists have used to
study short-term memory in contrast to long-term memory; it is thought to
be one measure of the capacity to recall from the short-term store (Hebb,
1961). There seems to be a prima facie case for considering whether a defect of
short-term memory could account for an inability to repeat verbal material. In
this paper a single patient is reported in whom there was a profound repetition
defect. The proposition that this repetition defect may be regarded as an impairment
of verbal short-term memory will be considered.
The question as to whether the organization of memory is a unitary process or a
two-stage process has received much attention in recent years. The strongest
evidence that there are separate short- and long-term memory systems is provided
by the specific and isolated impairment of long-term memory in amnesic subjects.
It has been clearly shown that patients with certain localized brain lesions may
have a very profound defect of long-term memory, with intact short-term memory
and with normal intellectual functions (Drachman and Arbit, 1966). Therefore if
it can be established that the short-term memory system can be impaired the
interaction of this defect with the functioning of the long-term memory processes
including trace formation and retrieval would be of theoretical interest.
CASE REPORT
K. F., a man aged 28, had a left parieto-occipital fracture in a motor-bicycle accident eleven
years before, when a left parietal subdural haematoma was evacuated. He was unconscious for ten
weeks. At first he was very dysphasic, but his speech gradually improved over the next few years.
At the age of 19 in 1959 he started having epilepsy and in 1965 when aged 25 he was admitted
under the care of Dr. Denis Williams for the investigation of this. There was a bone defect in the
left parieto-occipital region. Power in all limbs was normal, but fine movements were impaired
in the right hand. There was no plantar response on the right. Sensation was normal. Bilateral
carotid arteriograms were normal. An air encephalogram showed prominent localized dilatation
of the left trigone and occipital hom, with some dilatation of the left temporal horn.
Psychological Assessment
K. F. was tested on the W.A.I.S. and obtained a verbal I.Q. of 79 and a performance I.Q. of
113. On the Progressive Matrices he scored 40 out of 60, which is just below the 50th percentile
for his age. His relatively poor language functions were reflected in his verbal I.Q. His ability
to express himself was halting, and some word-finding difficulty and circumlocutions were noted.
Paraphasic errors were rare, and on no occasion were neologisms used. There was a very mild
degree of nominal dysphasia, which could only be detected with uncommon names. Receptive
speech functions were well preserved. Single instructions were carried out, though there was some
difficulty with "longer" messages. Using a task similar to the token test devised by De Renzi
(1962), the conceptual difficulty of the message did not impair his performance. Reading a simple
text was slow, but reasonably accurate. On Schonell's graded word list he scored just below the
9-year level. Both oral and written spelling were very impaired (at the 6-year level). Memory
for day-to-day happenings was good and he had an adequate knowledge of recent and past events.
I
I
I
I
i
atUniversityofNottinghamonOctober2,2011brain.oxfordjournals.orgDownloadedfrom
I
I
I
I
VERBAL SHORT-TERM MEMORY 887
Immediate memory for the Binet figures was accurate. More detailed tests of long-term memory
will be given below. The most striking feature of his performance was his almost total inability
to repeat verbal stimuli. His digit span was two, and on repeated attempts at repeating two digits
his performance would deteriorate, so that on some trials his digit span was one, or even none.
His repetition difficulty was not restricted to digits; he had a similar difficulty in repeating letters,
disconnected words and sentences. Single verbal items would be repeated correctly with the
exception of polysyllabic words which were on occasion mispronounced. An analytical investigation
of this repetition defect is here presented.
EXPERIMENTAL INVESTIGATION
In this study a number of systematic procedures were devised to determine the
basis of the disorder of repetition. In each experiment a standard presentation rate
of one item per second spoken by the experimenter was used unless otherwise
stated. In each experiment 2 blocks of 10 trials were given for each condition, with
the exception of those tasks which involved continuous serial presentation of stimuli.
(1) Repetition of Numbers, Letters and Words
Repetition of strings increasing in length of numbers, letters and words was examined. Strings
of 1, 2, 3 and 4 items were presented (in this order) for each of the 3 types of verbal material.
The numbers and letters were chosen randomly, the only constraint being that no individual item
in a string was duplicated. The words were chosen at random from 4 and 5 letter words of high
frequency (AA) in the Thorndike-Lorge count. The results expressed both as the number of
correct items and number of correct strings is given in Table I.
TABLE I.—REPETITION OF NUMBERS, LETTERS AND WORD STRINGS
OF INCREASING LENGTH
String length
Numbers
Letters
Words
Performance on this task of repetition is directly related to the number of items in each string.
K. F. can only reliably repeat one item, and the proportion of items correct decreases with
increasing string length. In terms of number of strings correct his performance is already faulty
for 2-item strings and markedly poor for 3 and 4 item strings. His performance is to some extent
determined by the type of verbal material, repetition for numbers being better than for letters.
(2) Effect of Presentation Rate on Repetition of Numbers, Letters and Words
Three presentation rates were used, one item per half-second, one item per second and one
item per two seconds, for pairs of numbers, letters and high frequency words (4 or 5 letters in
length). Three blocks of 10 pairs of each type of verbal stimuli for each presentation rate were
used. The order of type of verbal stimuli and presentation rate were randomized using 3x3x3
String length
No. of items correct
No. of strings correct
No. of items correct
No. of strings correct
No. of items correct
No. of strinas correct
1 Item
20/20
20
19/20
19
20/20
20
2 Items
28/40
12
21/40
7
29/40
13
3 Items
37/60
6
26/60
2
32/60
4
4 Items
37/80
1
22/80
0
33/80
1
atUniversityofNottinghamonOctober2,2011brain.oxfordjournals.orgDownloadedfrom
ELIZABETH K. WARRINGTON AND T. SHALLICE
Latin square design. The results expressed as the total numbers of items correct and the number
of pairs correct are given in Table II.
TABLE II.—EFFECT OF PRESENTATION RATE ON REPETITION
OF NUMBERS, LETTERS AND WORDS
1 item per 1 item per 1 item per
Presentation rate $ second second 2 seconds
2 Numbers No. of items correct 30/60 45/60 59/60
No. of strings correct 8/30 20/30 29/30
2 Letters No. of items, correct 23/60 30/60 34/60
No. of strings correct 5/30 9/30 12/30
2 Words No. of items correct 26/60 38/60 51/60
No. of strings correct 2/30 15/30 22/30
As in the first experiment fewer errors occur with repetition of pairs of numbers than letters,
and pairs of words are of intermediate difficulty. For all three types of verbal material performance
is worst with the fast presentation rate (one item per half-second) and improves with the slower
rate of one item per second, and is best with the slowest rate of one item per two seconds.
This result emphasizes the need to consider whether or not faulty input or a failure of registration
of auditory-perceptual material could account for his defect of repetition. One possible explanation
of his improved performance with slower presentation rates is slow auditory perception of single
items, so that with strings of two or more items the perceptual processing is overloaded.
(3) Auditory Perception
A number of procedures were devised to determine whether the gross defect of repetition was
secondary to impaired auditory perception of verbal stimuli. That is, are individual items, under
conditions similar to those in use in conventional repetition tasks, perceived and identified
sufficiently accurately to permit subsequent repetition?
Continuous memory span.—This technique needs a subject to listen to a long series of digits
exceeding his normal span. He has no knowledge of how long the series is going to be and his
task is to recall in the correct order as many items at the end of this list as possible. K. F. was
asked to report the last item of each of 10 series of digits and letters ranging in length from strings
of 6-12 items. For both numbers and letters his performance was completely accurate. Superficially,
this result suggests that each digit or letter is being accurately processed, as at any point in a
long series an individual item was correctly reported. Preceding items do not appear to block or
prevent the registration of succeeding items.
The objection could be raised that the task could be done without registering each item in turn;
the subject noting the end of the series might be able to perceive the final item from the auditory
acoustic trace retrospectively. Therefore a matching task was devised in which the subject must
perceive each item in order to achieve an error-free performance.
Matching numbers, letters and words.—Number, letter and word strings of increasing length
(1 item—4 items) were used for this matching task. Pairs of strings were presented at the standard
rate, with a pause of 1-2 seconds between the two strings. Each block comprised 10 pairs of strings,
in 5 of which one item was different and in 5 they were identical strings, for example: 739-739
or, alternatively, 438-458. The subject was asked to respond "same" or "different" to each pair
of strings. The results expressed as the total number of correct responses (maximum for each
condition being 20) are given in Table ITT.
I
I
I
I
atUniversityofNottinghamonOctober2,2011brain.oxfordjournals.orgDownloadedfrom
I
I
I
I
VERBAL SHORT-TERM MEMORY 889
TABLE HI.—MATCHING OF NUMBERS, LETTERS AND WORDS
String length 1 item 2 items 3 items 4 items
Numbers 20 19 20 17
Letters 20 18 17 15
Words 20 19 15 19
His performance is accurate for matching pairs of single items for all three types of verbal
material. Though errors occur for longer string lengths, K. F.'s performance on this task is good
enough with matching single items (the one condition in which the memory component of the task
does not exceed his capacity) to suggest again that a failure of registration or faulty auditory
perception of verbal stimuli cannot account for his repetition defect. Any interpretation of his
better performance in matching than in recall must be treated with some caution. A same-difference
judgment is easier than an identification judgment and some correction for guessing is necessary.
It could be argued, however, that matching two strings of items could be achieved at a preidentification
level in the auditory-speech system. That is, the subject is matching on the basis of
sound patterns, and not item identification. A further procedure was devised to assess whether
verbal stimuli, presented at the same rate as in repetition tasks, are individually identified when the
memory component of the task is eliminated.
Identification of individual numbers and letters.—Lists of 20 items comprised one key item which
recurred 5 times and 15 items chosen-randomly from the same category (numbers or letters).
Four lists of numbers in which the key item was 3, 6, 2 and 8 respectively and 4 lists of letters in
which the key item was H, X, B and K were presented at a rate of one item per second. K. F.
was told the key item for each list and asked to tap each time it was presented. The number of
correct responses to key items and neutral items is given in Table IV.
TABLE IV.—IDENTIFICATION OF INDIVIDUAL NUMBERS AND LETTERS
Key items Neutral items
Numbers 20/20 59/60
Letters 19/20 60/60
Two errors were recorded, one on a number list and one on a letter list. This finding indicates
that each individual item can be perceived and identified serially at a rate of presentation
comparable to that used in repetition tasks and furthermore there is no deterioration in performance
on lists of 20 items, which is comparable to those repetition tasks requiring repetition of 10 2-item
strings. Therefore, on this task, which demands rapid acoustic perceptual identification but has
no "memory" or verbal response component, performance is not detectably impaired.
Identification of categories of words.—Three lists of 40 words were compiled, of 10 key words
belonging to a particular category and 30 neutral words. The categories were: Countries, Colours
and Animals; one category represented by 10 different words was used in each of the 3 lists. K. F.
was told the category of the key words in each list and asked to tap each time he identified a word
as belonging to that category. The 40-word lists were read at an average rate of one per second.
The number of correct responses to key words and neutral words is given in Table V.
TABLE V.—IDENTIFICATION OF WORD CATEGORIES
Category Key words Neutral words
Country 9/10 30/30
Colour 9/10 29/30
Animal 10/10 28/30 .
atUniversityofNottinghamonOctober2,2011brain.oxfordjournals.orgDownloadedfrom
890 ELIZABETH K. WARRINGTON AND T. SHALLICE
A maximum of two errors in a forty-word list was recorded. On this task K. F. was required
not merely to identify a particular number or letter but to determine whether or not each word
belonged to a particular category without prior knowledge of what the individual key words would
be. There is no possibility of succeeding on this task by matching auditory percepts at a
pre-identification level in the auditory-speech system.
(4) Articulation and Related Disorders ofLanguage
Single verbal stimuli were almost invariably correctly repeated; there was some deterioration
with two items and a very marked difficulty with three or more. Particularly as K. F. has some
expressive speech impairment the question arises as to what extent the phenomenon of gross
failure of repetition could be accounted for by articulatory difficulties or dysphasia. It might be
assumed that an articulatory defect per se would not be directly related to the number of verbal
stimuli, but to the difficulty of the individual speech sounds, though the possibility remains that
difficulty with articulation of the first item would be followed by greater difficulty with a second if
it is necessary to do it quickly. By comparing performance on recall and recognition of verbal
stimuli it is possible to assess the role of articulation processes in genesis of the repetition defect.
Recognition bypointing.—Recognition by pointing was assessed for strings of numbers increasing
in length from one to four items. Immediately after verbal presentation of each string a card printed
with the numbers 1 to 9 arranged randomly in three rows of three was displayed and K. F. was
asked to point to the items in the string. The arrangement of the numbers in the choice situation
was varied from trial to trial. The number of correct responses for each length of number strings
and the number of strings correct is given in Table VI.
TABLE VI.—RECOGNITION BY POINTING
String length 1 item 2 items
Numbers No. of items correct 20/20 34/40
No. of strings correct 20 15
Recognition of numbers by pointing to each individual item does not result in better performance
than conventional repetition of number strings {see Table I). If an articulatory defect were
contributing to the poor performance in repetition one would expect a significant improvement on
this recognition task where no verbal response is required. It may be argued that recognition tested
by pointing to written numbers is mediated by sub-vocal repetition. Therefore a situation was
devised in which two items must be perceived, but only one must be reported, i.e. part-reporting.
Part-reporting.—The subject was asked to repeat the first or second item of a pair without
knowing which item was to be recalled. Two sets of 10 pairs of letters and numbers were presented
at the standard rate.
TABLE VII.—PART-REPORTING
1st item 2nd item Total
2 numbers 5/10 3/10 8/20
2 letters 4/10 4/10 8/20
Comparing these results with the first experiment on repetition {see Table I) it can be seen that
efficiency of part-reporting is shown to be no better than "whole" reporting. As reducing the
response load in this way does not improve performance, this finding supports the view that
K. F.'s disability cannot be explained as a difficulty in articulating more than one speech sound in
rapid succession. It can be shown that in certain circumstances his ability to repeat verbal stimuli
is markedly better.
I
J items
24/60
0
4 items
33/80
0
I
I
atUniversityofNottinghamonOctober2,2011brain.oxfordjournals.orgDownloadedfrom
I
No. of items correct
No. of strings correct
No. of items correct
No. of strings correct
57/60
17/20
42/60
7/20
69/80
13/20
51/80
3/20
I
I
I
VERBAL SHORT-TERM MEMORY 891
Reinforcement and repetition.—The effect of reinforcement by repeating each individual item
of a series on the subsequent repetition of the whole series was tested. Three or four numbers (or
letters) were presented individually, and each repeated individually. Immediately after repeating
the fourth digit, K. F. attempted to repeat all items in their correct order. The total number of
correct responses and the number of strings correct is given in Table VTH.
TABLE Vm.—REINFORCEMENT AND REPETITION
String length 3 items 4 items
Numbers
Letters
This result shows that in certain conditions the actual repetition of more than two verbal stimuli
is much improved, which is further evidence against articulatory difficulties being a causal factor
in this syndrome. Further examples of conditions giving rise to normal or near-normal verbal
responses will be reported in the next section of the results of long-term memory tests.
(5) Long-term Memory and Learning
Paired-associate learning.—The- paired-associate learning test from the Wechsler Memory
Scale (1945) was administered according to the standardized procedure. The test consists of 10
paired associates (6 easy and 4 hard), and the final score is based on 3 learning trials. The mean
score obtained by 50 normal subjects with an age range of 20-29 was 15-72 (S.D. 2-81).
K. F. obtained a score of 14-5 which is less than one standard deviation below the mean for his
age group. K. F. was given a fourth trial on which he recalled all 10 associations; when retested
after a delay of six hours on his first trial he recalled 9 out of 10 associations.
Learning of incomplete words and pictures.—Warrington and Weiskrantz (1968) described a
technique for measuring learning and retention from recognition of fragmented words and pictures.
The method of learning to criterion (two error-free trials) was used and the test was repeated
with a twenty-four-hour delay. Number of errors and number of trials to criterion for K. F.
and a normal control group of 40 subjects whose age range was 20-40 (mean age 27) are given in
Table IX. On this task K. F. performs within the normal range.
TABLE IX.—LEARNING OF INCOMPLETE WORDS AND PICTURES
Patient Control group
Test Retest Test Retest
Gollin pictures Initial score 14 10 19 10
Trials 2 1 3 1
Incomplete words Initial score 23 11 20 . 11
Trials 4 2 3 2
Ten-word learning.—A list of 10 high-frequency words was presented auditorily at the standard
rate. Subjects were required to recall as many words as possible from the list immediately after
presentation. This procedure was repeated until all the 10 words were recalled (not necessarily
in the correct order). K. F. needed 7 trials. Twenty normal controls took an average of 9 trials,
4 of the subjects failing on the task after 20 trials (Stevenson, 1967). After an interval of two
months he was able to recall 7 of these 10 words without relearning.
atUniversityofNottinghamonOctober2,2011brain.oxfordjournals.orgDownloadedfrom
892 ELIZABETH K. WARRINGTON AND T. SHALLICE
(6) Repetition of Visually Presented Numbers and Letters
The procedure described in the first experiment, relating repetition to length of string, was
replicated using visual rather than auditory presentation of the verbal stimuli. Strings of printed
numbers and letters werepresented visually at the standard rate of one item per second.
Immediate verbal recall was attempted. The total number of items correct and the number of
strings correct is given in Table X.
TABLE X.—VISUALLY PRESENTED NUMBERS AND LETTERS
String length
Numbers
Letters
1 item
20/20
20
20/20
20
2 items
39/40
19
37/40
17
3 items
48/60
10
48/60
11
4 items
51/80
6
46/80
2
On this task his performance is superior to his recall of auditorily presented numbers and digits
{see Table I). Only occasional errors occur with 2 item strings and he is still relatively efficient with
3 or 4 item strings. The question arises as to whether his superior performance on this task is
directly related to the modality of the input or to the effectively longer exposure of each item.
With visual presentation the stimuli are exposed for one second while in the comparable auditory
task, using a one second rate of presentation, the auditory stimulus is present for much less than one
second. Therefore his ability to repeat tachistoscopically presented verbal stimuli was assessed.
Tachistoscopic presentation of numbers and letters.—Strings of numbers and letters of increasing
length (1-4 items) were presented tachistoscopically. A 250 msec, exposure duration was used
throughout. K. F. was asked to delay his verbal report of the visual stimuli for two seconds after
presentation. The total number of correct items and the number of correct strings is given in
Table XI.
TABLE XI.—TACHISTOSCOPIC PRESENTATION OF NUMBERS AND LETTERS
String length
Numbers
Letters
1 item
20/20
20
20/20
20
2 items
40/40
20
36/40
16
3 items
51/60
15
40/60
8
4 items
52/80
4
32/80
0
His performance is accurate for 1 item strings and the total number of items correctly recalled
in the 2, 3 and 4 item strings is greater than with an auditory presentation. Again he has greater
difficulty with letters than numbers.
DISCUSSION
An attempt has been made to analyse the marked and specific impairment of
repetition in a single patient. This patient had a reliable digit span of one and his
performance deteriorated with progressively longer strings of digits. Experimental
psychologists regard digit span as involving verbal short-term memory, therefore
the possibility was considered that specific impairment of verbal short-term memory
could account for his inability to repeat verbal material. To establish this it is
essential to show that the repetition defect is not secondary to other disabilities in
language.
I
I
I
atUniversityofNottinghamonOctober2,2011brain.oxfordjournals.orgDownloadedfrom
I
I
I
I
VERBAL SHORT-TERM MEMORY 893
Two alternative explanations of. his disability have been investigated. First, that
his repetition defect is secondary to impaired auditory perception, resulting in faulty
registration of verbal stimuli and second, whether impairment of motor speech
functions could account for his difficulty in repetition. A number of procedures
were devised to determine the efficiency of his auditory perception. An explanation
of his performance on the continuous span task and on the matching task in terms
of a perceptual defect does not appear adequate. Performance on the matching
task appears better than on recall; although an exact comparison is impossible
without an elaborate theory to deal with guessing corrections, with strings of two
or more items his performance is not entirely error-free. As there is a small but
definite memory load in this matching task as well as a perceptual component, it
is clearly an unsatisfactory way of differentiating between a perceptual and memory
defect. More convincing are the results of the tapping tests, where there is no
memory component in the task, but rapid successive perception of auditory material
is required. No difficulty was encountered on these tasks. It is therefore implausible
to account for the disorder of repetition as secondary to impaired auditory perception
or faulty registration of verbal stimuli.
The task of recognition by pointing needs no verbal response, yet performance
in terms of accuracy is no better than with verbal recall. This finding, with the
"part-reporting" finding, strongly suggests that neither impairment of motor
speech functions nor faulty articulation were contributing to the repetition defect.
Furthermore, there are a number of test procedures in which the patient is able to
produce a normal number of verbal responses. His repetition is improved after
reinforcement by repeating each item individually and he is able to learn a 10-word
list as efficiently as normal control subjects.
A cardinal feature of the repetition defect is that it is directly related to the
"memory" load of the task. The original observation of a very poor verbal span,
disproportionately bad compared with his comprehensive and expressive language
functions, led to the present series of investigations. All subjects have a limited
capacity to recall a series of digits or letters, or indeed any verbal stimuli, and this
limitation is regarded as a characteristic of the "short-term" memory store {see
Hebb, 1961). In the present case it is suggested that the capacity of the S.T.M.
store is severely reduced so that in any task making demands on this function his
performance is impaired:
Specific impairment of long-term memory with intact short-term memory is well
recorded in amnesic subjects. It is therefore of interest to consider long-term
memory functions in the present case. The patient's verbal learning (auditory
presentation) appears to be quite satisfactory; considering that he is dysphasic, it is
remarkably good. In these learning tasks he has little or no difficulty in producing a
verbal response per se. Similarly, his learning and retention of visually presented
verbal material (incomplete words) was entirely satisfactory. Therefore, if it is
accepted that his short-term memory system is impaired, the finding, when considered
with the converse finding in amnesics, that his long-term memory processes are
atUniversityofNottinghamonOctober2,2011brain.oxfordjournals.orgDownloadedfrom
894 ELIZABETH K. WARRINGTON AND T. SHALUCE
normal, provides evidence for a double dissociation of function within the memory
system. Further experimental evidence to support this point of view and a discussion
of the relevance of this case for general theories of memory are reported by Shallice
and Warrington (1970).
Using this formulation it is possible to account for the effect of presentation rate
on his ability to repeat verbal stimuli. Glanzer and Cunitz (1966) have shown that
decrease in presentation rate improves recall from the long-term store but does
not improve recall from the short-term store. Moreover it is known that in long-term
memory the act of recall increases the strength of the memory trace (Horowitz
et al., 1966; Tulving, 1967). Therefore, in conditions favouring recall from the longterm
memory store one would predict improved performance. In fact this is the
case; our patient improves in repetition tasks when the presentation rate is decreased,
and when each item is repeated individually before recall (increasing strength of
trace) his performance is better.
There are many points of similarity between the present case and those of conduction
aphasia reported in the neurological literature, in particular intact comprehension
together with severe impairment of repetition. Geschwind (personal communication),
however, has suggested that conduction aphasics characteristically have greater
difficulty within the sphere of expressive speech functions, yet Dubois et al.
(1964) state in the conduction aphasic "repetition is as disturbed as spontaneous
language, sometimes even more so." It is probable that our patient presents the
characteristic defect of conduction aphasia in a particularly pure form, but we
cannot exclude the possibility that his is an essentially different syndrome.
Geschwind (1965) reviewed the available anatomical data in published cases of
conduction aphasia and concluded that the fasciculus arcuatus, a "tract which runs
from the posterior superior temporal region, arches round the posterior end of the
sylvian fissure and then runs forward in the lower parietal lobe, eventually to reach
the frontal lobe," is the critical site of the lesion. Dubois et al. (1964) state that
there is invariably damage in the region of the supramarginal and angular gyri of
the left hemisphere, but no autopsy findings are available in any of their cases. In
the 2 cases reported by Luria et al. (1967) the temporo-parietal cortex of the left
hemisphere was involved. In the present case the maximum injury was in the
posterior parieto-occipital region, which is not inconsistent with the anatomical
correlates reported in previous cases of conduction aphasia. There appears to be
considerable agreement that damage to the posterior dominant hemisphere is
invariably present. Shallice and Warrington (1970) have reported findings which
suggest that this short-term memory defect is one of storage rather than retrieval,
which is difficult to reconcile with Geschwind's view that conduction aphasia results
from a tract lesion.
Many instances of modality-specific cognitive disabilities have been recorded in
the neurological literature. These functional dissociations are not only of diagnostic
importance but are of considerable theoretical interest. Our patient had greater
difficulty in reporting letters and digits presented auditorily than visually (at the
I
I
I
I
atUniversityofNottinghamonOctober2,2011brain.oxfordjournals.orgDownloadedfrom
I
I
I
I
VERBAL SHORT-TERM MEMORY 895
standard rate of 1 item per second). Furthermore, his auditory span is shorter than
his visual span for letters and digits presented tachistoscopically, though normal
subjects have longer auditory than visual tachistoscopic span; this applies even if a
delay is introduced before recall to allow the raw visual trace (called by Neisser
(1967) the ikon) to decay. In both patients reported by Luria et al. (1967) recall of
visually presented verbal stimuli was considerably better than recall of auditory
stimuli. Such findings have relevance to models to visual memory (e.g. Sperling,
1967, and Neisser, 1967). A superior visual span shows that it must be possible to
recall verbal information presented visually without necessarily involving the
auditory verbal short-term memory system. Neisser (1967) may have over-emphasized
the process by which letters after being identified are retained in an auditory store
until recalled. One explanation of these results is that immediate recall of visualverbal
material is at least in part directly from a separate post-perceptual visual
short-term memory store.
SUMMARY
A patient is reported who had a gross impairment in the repetition of auditory
verbal stimuli after a left parietal injury. His verbal span was reliable for only one
digit and a comparable difficulty with letter and word spans was recorded. There
was much less difficulty with comparable visually presented verbal stimuli. Verbal
learning and verbal long-term memory were not affected. The defect cannot be
accounted for by faulty auditory perception or motor speech defect, so a specific
impairment of verbal short-term memory is proposed to account for the inability
to repeat verbal material. The findings are discussed in relation to the interaction
of short- and long-term memory functions and attention is drawn to the modality
specificity of the defect.
ACKNOWLEDGMENTS
We are most grateful to Dr. Denis Williams for his permission to report our findings on this
case. We wish to thank Dr. R. T. C. Pratt and Dr. A. Baddeley for their comments on the
manuscript, and Dr. A. R. Jonckheere for helpful discussions.
REFERENCES
DE RENZI, E., and VIGNOLO, L. A. (1962) The token test: a sensitive test to detect receptive disturbances in
aphasics, Brain, 85, 665-678.
DRACHMAN, D. A., and ARBIT, J. (1966) Memory and the hippocampal complex. II. Is memory a multiple
process? Archs. Neurol., Chicago, 15, 52-61.
DUBOIS, J., HECAEN, H.F ANGELERGUES, A., MAKFRAS DU CHATELIER, A., and MAROE, P. (1964) Etude
neurolinguistique de l'aphasie de conduction, Neuropsychologia, 2, 9—44.
GESCHWTND, N. (1965) Disconnexion syndromes in animals and man, Part n , Brain, 88, 585-644.
GLANZER, M., and CUNTTZ, A. R. (1966) Two storage mechanisms in free recall, /. verb. Learn, verb. Behav.,
5, 351-360.
GOLDSTEIN, K. (1948) "Language and Language Disturbances." New York: Grune and Stratton.
atUniversityofNottinghamonOctober2,2011brain.oxfordjournals.orgDownloadedfrom
896 ELIZABETH K. WARRINGTON AND T. SHALLICE
HEBB, D. O. (1961) Distinctive features of learning in the higher animal. In: "Brain Mechanisms and
Learning." Edited by J. F. Delafresnage. Oxford: Blackwell, pp. 37-46.
HOROWITZ, L. M., NORMAN, S. A., and DAY, R. S. (1966) Availability and associative symmetry, Psychol.
Rev., 73, 1-15.
KONORSKI, J., KOZNHWSKA, H., and STEPIEN, L. (1961) Analysis of symptoms and cerebral localization of
audio-verbal aphasia. VLIth International Congress of Neurology, Excerpta Medica, International
Congress Series No. 38, p. 22.
LURIA, A. R., SOKOLOV, E. N., and KJJMKOWSKJ, M. (1967) Towards a neurodynamic analysis of memory
disturbances with lesions of the left temporal lobe, Neuropsychologia, 5, 1-10.
MCFDJ, J. (1960) Psychological testing in clinical neurology, J. nerv. ment. Dis., 131, 383-393.
NPTSSPR, U. (1967) "Cognitive Psychology." New York: Appleton.
SHALUCE, T., and WARJUNGTON, E. K. (1970) The independent functioning of the verbal memory stores:
a neuropsychological study. In preparation.
SPERUNO, G. (1967) Successive approximations to a model for short-term memory, Ada psychol. (Amst.),
27, 285-292.
STEVENSON, J. F. (1967) M.Sc. Thesis (unpublished). University of Cambridge.
TULVINO E. (1967) The effects of presentation and recall of material in free-recall learning, J. verb. Learn.
verb.Behav., 6, 175-184.
WARJUNGTON, E. K., and WEKKRANTZ, L. (1968) New method of testing long-term retention with special
reference to amnesic patients, Nature, Lond., 217, 972-974.
WERNICKE, C. (1874) "Der aphasische Symtomenkomplex." Breslau: Cohn & Weigert.
{Received 18 March 1969)
I
I
I
I
atUniversityofNottinghamonOctober2,2011brain.oxfordjournals.orgDownloadedfrom