Brain (1991), 114, 1197-1225

WORD DEAFNESS AND AUDITORY CORTICAL
FUNCTION
A CASE HISTORY AND HYPOTHESIS
and

(From the 'Institute of Neurology, University Hospital Nijmegen, Nijmegen, 2Max-Planck-Institut fur
Psycholinguistik, Nijmegen, and ^Institute of Otorhinolaryngology, University Hospital Nijmegen,
The Netherlands)

SUMMARY
A patient who already had Wernicke's aphasia due to a left temporal lobe lesion suffered a severe deterioration
specifically of auditory language comprehension, subsequent to right temporal lobe infarction. A detailed
comparison of his new condition with his language status before the second stroke revealed that the newly
acquired deficit was limited to tasks related to auditory input. Further investigations demonstrated a speech
perceptual disorder, which we analysed as due to deficits both at the level of general auditory processes
and at the level of phonetic analysis. We discuss some arguments related to hemisphere specialization of
phonetic processing and to the disconnection explanation of word deafness that support the hypothesis of
word deafness being generally caused by mixed deficits.

INTRODUCTION

Pure word deafness, or auditory verbal agnosia, is a rare disorder of auditory language
comprehension. It is characterized by a selective impairment of auditory verbal
comprehension with preservation of other language functions. However, both the
selectivity for verbal material and the preservation of other language functions are matters
of dispute (Goldstein, 1974; Buchman etal., 1986). Given the rare occurrence of pure
cases, Buchman et al. dropped the adjective 'pure' and adopted 'word deafness' to
designate 'a disparity between auditory verbal comprehension and other linguistic
functions'. Thus defined, it still remains a distinctive syndrome which has drawn the
attention of many neurologists and neuropsychologists.
Opinions diverge as to the functional locus of the auditory verbal recognition disorder
in word deafness. Agnosias were originally conceived of as central recognition
disturbances, although a distinction between apperceptive and associative types was made
in its history by Lissauer (1890) and later by others (Geschwind, 1965; Vignolo, 1969;
Bauer and Rubens, 1985). In the last two decades attempts to characterize impairments
in individual patients have focused on speech perceptual deficits. Most authors have
emphasized impairments of low level general auditory processes (Jerger et al., 1969;
Albert and Bear, 1974; Chocholle et al., 1975; Auerbach et al, 1982; Miceli, 1982;
Correspondence to: Dr P. Praamstra, Institute of Neurology, University Hospital Nijmegen, PO Box 9101, 6500
HB Nijmegen, The Netherlands.
© Oxford University Press 1991

Downloaded from http://brain.oxfordjournals.org at Max-Planck-Gesellschaft on September 1, 2010

by PETER PRAAMSTRA,1 PETER HAGOORT,2 BEN MAASSEN1
THOM CRUL3

1198

P. PRAAMSTRA AND OTHERS

We report a case of word deafness, which was studied using neurolinguistic and speech
perception investigations similar to those used by earlier authors. These investigations
helped to delimit the nature of the patient's disturbance and to generate hypotheses
regarding the contribution of auditory and phonetic deficits. In the case we studied,
an unusual circumstance provided an additional source of evidence bearing on the role
of auditory versus phonetic deficits. As a Wernicke's aphasic, the patient had already
participated in various neurolinguistic experiments when he sustained a second cerebral
infarct, which was located temporoparietally in the right hemisphere and made him word
deaf. This sequence of events provided the opportunity to repeat earlier tests and to
perform new experiments in order to chart in detail the differences between the old
and the new patterns of deficits. The newly acquired deficit appeared to be limited to
a disorder of speech perception and was difficult to reduce to either an acoustic or a
phonetic type of deficit (Auerbach et al., 1982).
CASE D E S C R I P T I O N
The patient is a 57-yr-old right-handed man. He was first admitted in December 1984, when he had
suddenly experienced difficulties in speaking. His past medical history included myocardial infarct at age
44, from which he fully recovered. He is university educated and held a position as professor. His native
language is Dutch. Genera) physical examination was unremarkable. Neurological examination revealed
mild right-sided facial weakness. There was no visual field loss on confrontation. A very slight right
hemiparesis and mild right hyperreflexia were noticed. The right plantar response was extensor. There

Downloaded from http://brain.oxfordjournals.org at Max-Planck-Gesellschaft on September 1, 2010

Coslett etal., 1984; Motomura etal, 1986; Tanaka etal., 1987; Yaqub etal, 1988;
Buchtel and Stewart, 1989). In contrast, Saffran etal. (1976), Caramazza etal. (1983)
and Metz-Lutz and Dahl (1984) implicated phonetic processes, that is, a stage of auditory
processing specific to speech.
In theories of speech perception, the distinction between phonetic and auditory processes
indicates that some aspects of speech perception are specifically linguistic (phonetic
processes), while others are not (general auditory processes) (Liberman et al., 1967;
Repp, 1982; Kuhl, 1986; Pisoni and Luce, 1987). Clearly, the extraction of phonetic
structure from the physical speech signal requires the operation of basic processes that
register, for example, intensity, frequency, and duration. But how these general auditory
processes relate to specialized phonetic processes is unresolved. It is, for instance, not
known what kind of auditory processes are crucial for the perception of speech. What
is known, though, is that phonetic analysis is probably a left hemisphere-based competence
(Studdert-Kennedy and Shankweiler, 1970; Studdert-Kennedy et al., 1972; Tartter, 1988).
Hence left hemisphere lesions, especially of the auditory cortex, might have a disruptive
effect on speech perception. This expectation is, however, only partially met. Recent
studies have confirmed that phonetic analysis may be deficient in aphasia, but these
phonetic impairments are not strongly correlated with auditory language comprehension
impairments (for a review, see Blumstein, 1987). Also, studies comparing left and right
hemisphere-damaged patients have not contributed much to the identification of the
auditory processes underlying speech perception (see, e.g., Divenyi and Robinson, 1989).
Given these unresolved questions it is surprising that so many authors of recent case
reports have assumed deficits of general auditory processes as the cause of word deafness
(Albert and Bear, 1974; Auerbach etal., 1982; Miceli, 1982; Coslettetal., 1984; Tanaka
et al., 1987; Yaqub et al., 1988; Buchtel and Stewart, 1989).

WORD DEAFNESS

1199

was no hemi-inattention or apraxia. A cerebral CT scan (fig. 1) demonstrated a left middle cerebral artery
infarction. Left carotid angiography was normal. Electrocardiography showed signs of old anteroseptal
infarction. Echocardiography revealed no evidence for a cardiac source of emboli.

The patient had fluent speech, mainly consisting of jargon and empty phrases. Repetition was severely
disturbed, although individual sounds could be repeated to some extent. Auditory and written language
comprehension were both affected. He was tested formally using the Dutch version of the Aachen Aphasia
Test (AAT) (Huber et al., 1984) and classified as a severe Wernicke's aphasic. As such he participated
in a number of neurolinguistic experiments. In the course of 16 months he improved to the extent that
retesting with the ATT demonstrated a shift, on all subtests, from severely disturbed to a disturbance of
average degree.
In January 1988 the patient consulted his neurologist because of a hearing disturbance, which had been
preceded by mild complaints of headache and nausea. On admission he reported that sounds seemed louder
and noisy. It had occurred to him that he did not recognize the sound of an airplane passing and that music
sounded strange. Before the onset of this disturbance he often played the piano, but he had recently stopped.
His speech seemed unchanged, but he did not understand any spoken language. The interview with the
patient as well as his communication with relatives and nurses proceeded with written messages.
General physical examination revealed normal heart rate, rhythm, and blood pressure. There were no
bruits. Neurological examination showed no abnormalities of cranial nerve function, no visual field defect,
and normal motor and sensory functions. The tendon reflexes were symmetric, but the left plantar response
was extensor.
A CT scan revealed a hypodense area in the distribution of the right middle cerebral artery having the
appearance of a fresh ischaemic infarct. Digital subtraction angiography of the extracranial cerebral vessels
showed no abnormalities. There was right temporoparietal theta activity in an E E C Electrocardiography
and echocardiography contributed no new information.
A provisional diagnosis of auditory agnosia caused by bilateral destruction of auditory cortical structures
due to ischaemic infarcts was made. Investigations over the next one-and-a-half years confirmed this diagnosis
and showed the auditory agnosia to be of a predominantly verbal type. The patient experienced no
improvement of auditory language comprehension and continued to be dependent on written communication.
INVESTIGATIONS
The following investigations are reported: neuroradiological, audiological, and neurophysiological;
language status and nonverbal sound recognition; neurolinguistic and speech perceptual investigations.

Downloaded from http://brain.oxfordjournals.org at Max-Planck-Gesellschaft on September 1, 2010

FIG. 1. Unenhanced CT scan performed on admission after first stroke.

12W

P. PRAAMSTRA AND OTHERS

Comment. Given the nature and extent of the left and right hemisphere lesions, the patient's remaining
auditory function is probably subserved for the most part by the right hemisphere auditory cortex. Note
that this dependence on the right hemisphere already existed before the second stroke.

FIG. 2. Unenhanced CT scans performed 2 yrs after second stroke.

Downloaded from http://brain.oxfordjournals.org at Max-Planck-Gesellschaft on September 1, 2010

Neuroradiological investigation
CT scans of the left and right hemisphere infarcts (fig. 2) were classified by comparing the individual
slices with standard sections from an atlas for computerized tomography of the brain (Matsui and Hirano,
1978). Both the left and the right hemisphere lesions were middle cerebral artery infarcts, albeit with
considerable differences in size and localization. The left hemisphere lesion involved the superior temporal
gyrus and extended superiorly and anteriorly into the parietal lobe and the frontal lobe. The anterior extension
overlay the central sulcus and the precentral and postcentral gyri. Posteriorly the lesion involved the
supramarginal gyrus of the inferior parietal lobule and extended just onto the angular gyrus. The lesion
included both the superior and middle temporal gyri. Involvement of the supratemporal plane with Heschl's
gyri could be inferred reliably. The temporal isthmus containing geniculotemporal radiations was spared.
The parietal part of the lesion did not extend deep enough to involve the area lateral to the posterior part
of the lateral ventricle where the interhemisphere auditory pathways traverse (Rubens et al., 1978; Damasio
and Damasio, 1979).
The right hemisphere lesion was largely contained within the temporal lobe. The superior temporal gyrus
carried the brunt of the lesion. Like the left hemisphere lesion it extended posteriorly onto the angular
gyrus, but it was less well demarcated. Therefore, a massive destruction of the supratemporal plane including
Heschl's gyri was uncertain. The geniculotemporal radiation through the temporal isthmus was not affected.

WORD D E A F N E S S
Audiological

1201

investigation

Standard pure tone audiometry demonstrated a minor bilateral sensorineural hearing loss, which exceeded
20 dB only at frequencies above 4000 Hz (Table 1).

TABLE 1. PURE TONE AUDIOMETRY
Pure lone threshold (dB)
Right ear
Left ear
20
10
15
10
10
5
10
5
20
20
45
40

Comment. The patient's mild hearing loss for higher frequencies was obviously not the cause of his
auditory comprehension deficit.
Brainstem auditory evoked potentials
BAEPs were recorded on a Nicolet Pathfinder II. Stimuli were 0.1 ms clicks of 80 dB (SPL) presented
monaurally at a rate of 11/s. The contralateral ear was masked with white noise. The filter setting allowed
a bandpass of 150—3000 Hz. The potentials were recorded ipsilateral to the ear of stimulation from bipolar
derivations Cz-Al and Cz-A2.
Results. Left and right ear stimulation produced waveforms with identifiable peaks I - V , all latencies
being within normal limits.
Comment. The results of tone audiometry and BAEPs indicate that the auditory afferent pathways up
to the inferior colliculi were functioning normally.
Middle and long-latency auditory evoked potentials
Recordings of MLAEPs and LLAEPs were aimed at an evaluation of midbrain and telencephalic auditory
pathways and structures. They were obtained with the same apparatus and stimuli and from the same recording
sites as the BAEPs. The stimulation rate was 4.5/s and the bandpass 5-1000 Hz for MLAEPs and 0.25/s
and 1 -250 Hz for the LLAEPs.
Results. The Na component of the MLAEP was clearly present in the derivations contralateral and
ipsilateral to the separately stimulated ears. The latencies were normal. This component was of smaller
amplitude with right ear stimulation. The Pa component was also present in both derivations, but of smaller
amplitude in the Cz-Al derivation. Components PI, Nl and P2 of the LLAEP had normal amplitudes
and latencies in all recordings. Ipsilateral and contralateral derivations showed nearly identical waveforms.
The N2 and later components could not be identified reliably.
Comment. The generator sites of middle and long-latency auditory evoked potentials are still controversial.
The study of patients with unilateral or bilateral temporal lesions has contributed little to resolve this issue.
Bitemporal lesions have been found compatible with normal or nearly normal MLAEPs (Parving et al.,
1980; Woods et al., 1984), while they can also be absent (Ozdamar et al., 1982). There is a similar lack
of consistency regarding the LLAEPs, which have been reported normal (Parving et al., 1980; Woods
et al., 1984; Tanaka et al., 1987), delayed and of smaller amplitude (Albert et al., 1972; Miceli, 1982),
absent unilaterally (Auerbach et al., 1982), and absent bilaterally (Michel and Peronnet, 1980) in patients
with bitemporal lesions.
In an extensive review of reports of AEPs in patients with temporal lobe lesions, Woods et al. (1987)
concluded that cortical lesions only affect the Pa generator if they extend beyond the classical auditory

Downloaded from http://brain.oxfordjournals.org at Max-Planck-Gesellschaft on September 1, 2010

Frequency (Hz)
250
500
1000
2000
4000
8000

1202

P. PRAAMSTRA AND OTHERS

L A N G U A G E STATUS AND NONVERBAL SOUND R E C O G N I T I O N
Aachen Aphasia Test
The patient was tested using the Dutch version of the Aachen Aphasia Test (AAT) (Huber et al., 1980,
1984; Weniger et al., 1981; Willmes et al., 1980, 1983). In addition to eliciting and recording a sample
of spontaneous speech, the AAT has subtests for word and sentence repetition, written language, object
and situation naming, and word and sentence comprehension. In the subtest on comprehension, patients
have to select 1 picture out of 4 on the basis of auditory or visual input. The AAT further includes the
shortened version of the Token Test (TT) (De Renzi and Vignolo, 1962; Orgass, 1976). The TT is usually
presented auditorily. The TT was also presented visually using a test procedure introduced by Poeck and
Hartje (1979). This procedure involves the test materials being screened off from view during the reading
of each instruction. After 10 s (parts I-III) or 15 s (parts IV and V) the instruction was covered and the
test materials were uncovered.
On several occasions before the second cerebrovascular lesion the patient had shown a stable pattern
of performance on the AAT. The patient was tested again 3 months after his second stroke. In Table 2
these results are shown together with the last AAT before the second stroke. The interval between the
tests was 23 months. There was an interval of 10 months between the auditory and the visual administrations
of the TT after the second stroke.
Comment. The preliminary diagnosis of an auditory agnosia, in addition to the preexisting Wernicke's
aphasia, was supported by the AAT results. The qualitative assessment of spontaneous speech, which takes
into account semantic, syntactic and phonemic structure, articulation and prosody, automatized speech
and communicative behaviour, found no changes compared with his status before the second stroke.
Repetition changed dramatically. The performance deteriorated with growing complexity of the presented
items (see Table 2). On written language the patient also performed worse than before the second stroke.
But, as is shown in Table 2, this was almost entirely due to his inability to spell and to write to dictation,
both mediated by the auditory modality. The comprehension part of the AAT demonstrated a dissociation
between an almost unchanged written language comprehension and a considerably lower score on oral
language comprehension. The performance on the auditory sentence comprehension part did not exceed
chance level.
The pattern of results on the AAT obtained after the second stroke, in comparison with that before it,
shows a selective disturbance of tasks that are dependent on the auditory modality. This modality bias
also emerges from a discrepancy between the visual and auditory presentations of the Token Test (Table 3).
The results obtained in the visual version show an error score that, according to the interpretation of the
standard auditory test, is nonaphasic. Although the TT, like other subtests of the AAT, demonstrates a
modality bias, it must be interpreted with caution. The performance on the visual TT was not only much
better than the auditory TT results after the second stroke, but it was also superior compared with the
auditory TT results before the second stroke. Since the visual TT is not a standardized test we consider
the unambiguous change in performance on the comprehension part of the AAT the more reliable evidence
of a modality specific change.

Downloaded from http://brain.oxfordjournals.org at Max-Planck-Gesellschaft on September 1, 2010

areas. They further observed that the Nl component was usually not abolished with lesions restricted to
the superior temporal plane. The authors suggested that this component probably also has generators in
the inferior parietal lobe. There is, however, evidence from neuromagnetic recordings (Hari etal., 1980;
Sams et al., 1985) and from EP dipole source analysis (Scherg and von Cramon, 1986, 1990) that the
exogenous middle and long-latency AEPs are generated in the auditory cortex.
The remarkably preserved AEPs in our case may be due to the presence of remnants of intact auditory
cortex. Such remnants can produce normal evoked potentials in experimentally lesioned animals (Diamond
and Neff, 1957). In addition, the posterior parts of the inferior parietal lobes, which according to Woods
et al. (1987) have to be destroyed to annihilate the AEPs, may have contributed to their generation.
Alternatively, it may be the case that the scalp-recorded potentials, although symmetric in distribution,
are generated in the right hemisphere auditory cortex, which is obviously the least affected side. In summary,
only the modest conclusion can be drawn that the presence of middle and late AEPs attests to the integrity
of geniculotemporal radiations and at least some auditory cortical structures. Thus the evidence obtained
from AEP investigations is compatible with the CT evidence.

WORD DEAFNESS

1203

TABLE 2. AACHEN APHASIA TEST
March 1988
30
18
9
3
0
0

Written language
Oral reading
Spelling to dictation
Words/sentences to dictation

44

24

Naming
Objects (singular nouns)
Colours
Objects (composit nouns)
Situations (descriptions)

85

Comprehension
Auditory (word)
Auditory (sentence)
Written (word)
Written (sentence)

91

•

Max. score
150
30
30
30
30
30
90
30
30
30

24
0
0

25
8
11

120

91

30
30
30
30

30
30
19
12

23
29
16
17

120

73

30
30
30
30

20
10
26
17

23
24
25
19

TABLE 3. TOKEN TEST
Error score (max. 50)
Date
April 1986
March 1988
January 1989

Auditory

Visual

45
42

In sum, the AAT showed changes in comparison with the test results before the second stroke that can
be attributed to a disturbance in the auditory reception of language. The patient's Wernicke's aphasia was
essentially unchanged but compounded by an auditory deficit. Given its apparent central origin, this deficit
was characterized as an auditory agnosia.
Nonverbal sound recognition
To test the patient's ability to recognize nonverbal sounds we applied a sound recognition test modelled
after Spinnler and Vignolo (1966). It consisted of 20 environmental sounds recorded on tape (e.g., ambulance
siren, electric razor, canary, trumpet, etc.). Each sound was presented in conjunction with 4 pictures
representing the response alternatives. The correct response alternative pictured the actual source of the
sound. The other 3 pictures represented the sources of an acoustically similar sound, an object or event
semantically related to the source of the presented sound, or something unrelated. Failures to identify the
correct picture were accordingly characterized as discriminative, semantic or odd failures. The mean time
of presentation of each sound was 15 (range 8—28) s. Ten healthy control subjects (mean age 56, range
43—65 yrs) made between 1 and 5 incorrect responses (mean 3.2/20; 91 % discriminative, and 9% semantic
errors).
The original test of Spinnler and Vignolo was designed to investigate the nature of sound recognition
deficits in aphasia. It consisted of 10 different sounds and was administered twice, the maximum score
being 20. An abnormal score was established to be lower than 16. Spinnler and Vignolo detected sound

Downloaded from http://brain.oxfordjournals.org at Max-Planck-Gesellschaft on September 1, 2010

April 1986
87
26
25
21
6
9

Repetition
Phonemes
One-syllable words
Foreign words
Composits
Sentences

1204

P. PRAAMSTRA AND OTHERS

recognition deficits mainly in severe Wernicke's aphasia. Semantic errors were far more frequent than
discriminative errors. They concluded that sound recognition deficits in aphasia are associated with disorders
of auditory verbal comprehension and are not perceptual discriminative, but associative in nature.
Results. Our patient was tested 7 wks after the occurrence of the second stroke. He performed 14/20
correct. Of the incorrect responses 4 were discriminative and 2 odd.

NEUROLINGUISTIC

INVESTIGATIONS

The specific changes in the AAT test profile demonstrated a clear dissociation between language
comprehension in the auditory and visual modalities. In the presence of relatively preserved nonverbal
sound recognition, this warranted the diagnosis of auditory verbal agnosia or word deafness. Further
investigations were directed at the characterization of the underlying processing deficits.
Although the pure form of word deafness may be a fictitious entity (Buchman et al., 1986), some of
the most important studies on word deafness were carried out on relatively pure cases (Saffran et al., 1976;
Auerbach et al., 1982) or on cases in which aphasic disturbances were temporary (Albert and Bear, 1974;
Miceli, 1982; Tanaka et al., 1987). In our case, however, Wernicke's aphasia already existed before the
signs of word deafness developed. Is it also possible to attribute these later signs to the aphasia? The AAT
results suggest that such is not the case. We contend that 'central' language mechanisms did not suffer
further damage to the second cerebral infarct. This is supported by the results of a lexical decision task
performed before and after the second stroke. These results are reported below, together with those of
a syllable matching task and a task testing the ability to distinguish between sequences of Dutch and English
words.
Lexical decision
Both before and after his second stroke the patient participated in a lexical decision experiment (Hagoort,
1989, 1990). The lexical decision task requires subjects to decide whether strings of letters or sounds are
words of their language or not. In the original experiment words and nonwords were presented auditorily.
The word targets consisted of Dutch words ranging from 1 to 4 syllables. Nonwords were matched in
length and consisted of randomly selected Dutch syllables. The targets were preceded by primes, which
were always words. A subset of the primes was semantically related to the word targets, while the remainder
was unrelated. For our case, however, the primes are irrelevant. We will focus on the patient's number
of correct and incorrect responses to the word and nonword targets. After the patient's second stroke the
experiment was repeated. In addition, a visual version of the experiment was administered using the same
target words and nonwords presented in written form, without primes.
Method. The targets consisted of 128 Dutch words and 128 nonwords, which were pronounceable according
to the rules of the sound system in Dutch. Primes and targets were spoken by a female voice, tape recorded,
and presented via headphones. The test session began with 20 starter trials to acquaint the subject with
the procedure. For each target the patient had to decide whether it was a word or a nonword by pushing
one of two buttons, a 'yes' button for words and a 'no' button for nonwords. For the visual version of
the experiment the same targets were presented as a list, written in capitals, and the patient was asked
to mark each target as a word or nonword.
Results. Before his second stroke the patient made errors on 4% of the auditorily presented targets. For
the word targets, he always gave the correct response. For 7% of the nonwords, he incorrectly decided
that they were words. After the second stroke the overall error percentage increased to 49% (30% for
the words and 69% for the nonwords). In contrast, when the same targets were presented visually he
responded without hesitation and made no errors.

Downloaded from http://brain.oxfordjournals.org at Max-Planck-Gesellschaft on September 1, 2010

Comment. Our patient's score was slightly below normal performance. As the incorrect responses were
mainly discriminative, we conjecture that the disturbed performance was due to the deficit acquired after
the second stroke and not due to the pre-existing Wernicke's aphasia which, according to Spinnler and
Vignolo, generally causes semantic errors.

WORD DEAFNESS

1205

Comment. The results indicate that the second stroke caused a specific deficit in accessing the auditory
input lexicon, while access to the orthographic input lexicon was normal.
Distinguishing between languages
To investigate whether he was able to discriminate Dutch from a nonnative language on the basis of
the auditory word form information, the patient, who had an active knowledge of English, was subjected
to a test of randomly presented sequences of Dutch and English words.

Results and comment. The results are presented in Table 4. The overall performance was above chance,
the response being accurate in 17 out of 23 word sequences. However, closer inspection revealed that
the patient was only accurate in identifying the English sequences (11/12), but performed at chance level
for the Dutch ones (6/11). One explanation of this dissociation is that he used a response strategy in which
uncertainty was resolved by choosing the response alternative 'English'. Alternatively, the patient might
still have been able to extract certain salient phonetic characteristics of English that are not part of Dutch.
Whatever the correct explanation for the dissociation between his performance for Dutch and English,
the results suggest that he is no longer able to derive adequate phonetic representations from verbal acoustic
input.

TABLE 4. ENGLISH-DUTCH LANGUAGE DISCRIMINATION
Language of sequence
English

Dutch

Total

Response
English

11

5

16

Dutch

1

6

7

Total

12

11

Syllable matching
The patient's performance on the auditory lexical decision task established that he is no longer able to
access the mental lexicon via the auditory route. This prevents him from recognizing spoken words. In
the process of spoken word recognition the listener has to contact stored word forms in the auditory input
lexicon. For this purpose, the listener has to extract contact representations from the speech signal
(Frauenfelder and Tyler, 1987). These contact representations allow the activation of stored word forms
that are consistent with the input. A number of different proposals have been made concerning the nature
of these contact representations (e.g., Klatt, 1979; Mehler, 1981; Pisoni and Luce, 1987). Most likely
they include information about the consonants and vowels composing a word and about the number of
syllables and the stress pattern (Pisoni and Luce, 1987). To investigate whether the patient was still able
to extract such a representation from the acoustic input, we applied a syllable matching task (P. Hagoort
and A. Witjes, unpublished). This task requires a subject to construct some sort of phonological representation
of presented word pairs and match the specified segments.
Method. In this task, pairs of bisyllabic words were presented. The list of word pairs consisted of two
blocks, each including 60 target and 40 filler pairs. The first block contained 20 target word pairs with
an identical first syllable (e.g., fortuin-formaat) (fortune-format) and 40 target pairs with no syllables

Downloaded from http://brain.oxfordjournals.org at Max-Planck-Gesellschaft on September 1, 2010

Method. Twenty-four sequences of 15 bisyllabic and trisyllabic words each were presented. Twelve
sequences consisted of Dutch words, and 12 of English words. The first sequence, consisting of Dutch
words, was not scored. The stimuli were spoken by a female voice, tape recorded, and presented via
headphones. After every sequence the tape was stopped by the experimenter, and the patient was asked
to point to one of two cards, saying 'Dutch' and 'English'.

1206

P. PRAAMSTRA AND OTHERS

Results. The results for the two blocks were analysed separately. Only the responses on the 60 target
items were analysed. To separate the patient's response bias from his sensitivity in this two-alternative
forced-choice task, the nonparametric index of sensitivity A' was computed. This measure is derived from
signal detection theory (Green and Swets, 1966; Grier, 1971). The A' value (e.g., 0.90) can be interpreted
as the expected percentage correct (e.g., 90%) on a forced two-choice procedure. For the first block the
patient's index of sensitivity (A') was 0.86 (15/20 hits and 8/40 false alarms). For the second block his
A' was 0.87 (12/20 hits and 2/40 false alarms).
Comment. The patient turned out to be fairly accurate in matching the specified syllables of the auditorily
presented word pairs. This implies that he is still able to extract at least a partial phonological representation
from verbal acoustic input. Adequate performance in the syllable matching task probably only requires
the presence of a very global phonological representation without detailed phonetic specifications (e.g.,
whether the onset consonant is a /b/ or a /p/). Given his severely impaired performance on the auditory
lexical decision task the representation derived from the input, however, seems insufficiently specified
to zoom in successfully on the appropriate word forms in the auditory lexicon.
In conclusion, the patient is clearly deficient in the ability to extract from the speech waves phonological
representations specified well enough to allow him to recognize the words of his language. A deficit at
the acoustic-phonetic level could be responsible for the impairment in constructing a fully specified
phonological representation. The subsequent investigations thus focused on the patient's auditory verbal
discrimination and identification abilities.
SPEECH PERCEPTUAL

INVESTIGATIONS

The perception of speech sounds is usually tested using the 6 stop consonants /b/, /d/, /g/, /p/, /t/ and
IVJ paired with a vowel (Saffran et al., 1976; Auerbach et al., 1982; Miceli, 1982; Tanaka et al., 1987).
These speech sounds have also been extensively used in dichotic listening research on hemisphere
specialization of speech perception (Studdert-Kennedy and Shankweiler, 1970; Studdert-Kennedy et al.,
1972). The stop consonants tax the auditory-phonetic analysis in a particular way. Whereas vowels have
a long steady state that makes them relatively easy to discriminate, the stop consonants are differentiated
from one another by rapid formant transitions within about 50 ms from onset of the speech signal (Lieberman
and Blumstein, 1988). This could make the perception of stop consonants particularly vulnerable to
impairments of auditory temporal resolution. It is precisely this mechanism by which some authors have
sought to connect impairments of general auditory processes to speech perception deficits (Tallal and
Newcombe, 1978; Auerbach et al., 1982; Miceli, 1982; Coslett et al., 1984).
The contrasts between the stop consonants can be defined by two phonetic features, as illustrated in
Table 5. Normal subjects discriminate more easily between speech sounds distinguished by two phonetic
features (e.g., /b/ vs III) than between sounds distinguished by one feature (e.g., /b/ vs Id/) (Lieberman
and Blumstein, 1988). If the discrimination of speech sounds by a patient demonstrates such a feature
effect this might be interpreted as evidence that speech sound discrimination is still supported by phonetic
processing capacities (Auerbach et al., 1982).
We performed speech perceptual investigations very similar to those used in earlier case studies (Saffran
et al., 1976; Auerbach et al., 1982; Tanaka et al., 1987; Yaqub et al., 1988). The results should provide
a measure of the severity of the speech perceptual deficit and might also provide clues to its nature.

Downloaded from http://brain.oxfordjournals.org at Max-Planck-Gesellschaft on September 1, 2010

overlapping (e.g., laken-zebra) (sheet-zebra). In the second block, 20 target word pairs were identical
in their second syllable (e.g., boeket-raket) (bouquet-rocket). In addition, each block contained 40 filler
pairs, 20 with identical first or second syllable and 20 without identical syllables. These pairs were constructed
by combining the words from the target pairs in a different way. In this way, response strategies on the
basis of the list structure were excluded (Hagoort, 1990). Each block started with 8 practice items. For
the first block the patient was required to decide for each pair whether the words had the same first syllable.
The pairs in the second block had to be judged for identity of the second syllable. The words were spoken
by a female voice and played over closed headphones. There was an interval of 500 ms between the two
words of a pair. The patient was given instructions in written form. He was asked to listen to each word
pair and to indicate whether or not the words matched with respect to the specified syllable, by pointing
to one of two cards saying 'same' and 'different'.

WORD D E A F N E S S

1207

TABLE 5. ARTICULATORY FEATURES OF THE STOP CONSONANTS
Front
(bilabial)

Place-of-articulation
Middle
(alveolar)

Back
(velar)

Voiced

Ibl

HI

'$'

Unvoiced

Ipl

III

IkJ

Voicing

Results. The test was performed at 3 and at 18 months after the second stroke with scores of 25/50 and
23/50 incorrect responses. Conversion into a nonparametric index of sensitivity gave A' values of 0.67
and 0.58, respectively. The contingency tables (Table 6) illustrate an especially poor performance on words
differing in vowel.

TABLE 6. DISCRIMINATION OF MONOSYLLABIC W O R D S CONTINGENCY TABLES*
Stimulus

Stimulus
Identical

Total

Different

Identical

Total

Different

13
(2 + 11)

3

16

19
(2 + 17)

7

26

Identical

22
(6+16)

12

34

16
(6+10)

8

24

Total

35

15

35

15

Responses
Different

* The left and right-hand table represent, respectively, the performance at 3 and at 18
months after the second stroke. The numbers in parentheses indicate the responses on word
pairs differing in vowel and the responses on pairs differing in consonant, respectively.

2. Vowel identification
We tested vowel identification in a similar way to Auerbach el al. (1982). The patient was presented
with a series of 100 vowels, /a/,/e/, I'll, lol and /u/, which were spoken by a male experimenter in random
order. The lower half of the experimenter's face was covered to prevent lip-reading. On each trial, the
patient marked 1 of the 5 response alternatives on an answer sheet.
Results, lal, Id, III,lol and /u/ were identified correctly, 5, 6, 8, 11 and 13 times out of 20, respectively.
The total score of 43% correct identification was above chance (20%).

Downloaded from http://brain.oxfordjournals.org at Max-Planck-Gesellschaft on September 1, 2010

1. Discrimination of monosyllabic words
As a first attempt at characterizing a possible speech perceptual deficit underlying the dissociation between
auditory and written language comprehension, an auditory discrimination test was administered (Crul and
Peters, 1976). The test consisted of 50 pairs of monosyllabic words, spoken by a female voice and recorded
on tape. All words began and ended in consonants or consonant clusters. The words constituting a pair
differed either in the word-initial or the word-final consonant (cluster) (27 pairs), or in the vowel (8 pairs).
There were 15 pairs with identical words. The patient had to mark the response alternatives 'same' or
'different' on a response sheet.

1208

P. PRAAMSTRA AND OTHERS

3. Vowel identification; binary choice
The patient was also tested with a forced binary choice procedure. A tape was prepared with 200 tokens
of the 5 vowels /a/, Id, HI, lol and /u/ in random order. The interval between successive stimuli was
4 s. The stimuli were spoken by a female voice and presented through headphones. The length of the vowels
ranged between 290 and 323 ms. The patient had an answer sheet with 200 numbered vowel pairs, one
of each pair being the presented vowel. On the answer sheet, each vowel was paired with each of the
other stimuli 40 times. The correct response alternative was randomly the first or the second member of
the pair. The patient answered by marking one of the response alternatives.

TABLE 7. CONFUSION MATRIX OF VOWEL IDENTIFICATIONBINARY CHOICE
lei

lal

Percentage responses
HI

lol

lul

2.5
5.0
2.5

0
0

Stimulus
/a/
Id
lit
lol

lul

87.5

7.5

7.5
7.5

80.0
12.5

65.0

15.0
7.5

7.5

2.5

5.0

10.0

2.5
7.5

62.5
10.0

12.5
12.5
67.5

4. Vowel discrimination
Vowel discrimination was tested employing the same stimuli as used in the vowel identification task.
A tape was prepared with 200 vowel pairs. Between pairs there was an interval of 4 s and between members
of a given pair an interval of 1 s. Half of the vowel pairs were identical. The stimuli were presented through
headphones. The patient responded by marking 'same' or 'different' on an answer sheet after each
presentation of a vowel pair.
Results. Results are presented in Table 8. The patient had a percentage of 92.0% correct responses.
An A' of 0.97 was computed from the overall proportions of hits (89/100) and false alarms (2/100).

TABLE 8. PROPORTION CORRECT DISCRIMINATION OF VOWELS
lal

Id
III
lol
lul

lal
1.00

lei

HI

lol

lul

1.00
1.00

1.00
0.30
0.95

1.00
1.00
1.00
0.95

1.00
1.00
0.90
0.70
1.00

5. Consonant identification; binary choice
Consonant identification and discrimination were evaluated using the stop consonants /b/, Id/, /g/, /p/,
l\l and Ik/, paired with the vowel /a/. For the identification task a tape was prepared of 300 tokens of
these 6 CV syllables, spoken by a female voice. There were pauses of 4 s between trials. The test procedure
was similar to the one used in the vowel identification task.

Downloaded from http://brain.oxfordjournals.org at Max-Planck-Gesellschaft on September 1, 2010

Results. The patient's performance on this task is represented in Table 7. The average performance was
72.5% correct, which is above chance (50%).

WORD DEAFNESS

1209

Results. Table 9 presents the patient's performance on the consonant identification task. The average
performance was 68.7% correct, which is above chance (50%).

TABLE 9. CONFUSION MATRIX OF CONSONANT IDENTIFICATIONBINARY CHOICE
Ida/

74

10
58
0
12
2
12

10

0
8
4

10

Percentage responses
/ga/
/pa/
10
4
4
72
4
8

0
8
82
0
0
8

/ta/

/ka/

4
16

2
4
8
2
6
42

6
6
84
20

6. Discrimination of stop consonants
The tape for the discrimination task was constructed using the same stimuli as used in the CV identification
task. The tape contained 420 pairs of CV syllables, half of which were identical. The 30 possible pairs
of different stimuli were presented in random order, each 7 times. The test procedure was the same as
the procedure used for the discrimination of vowels.
Results. Results are presented in Table 10. The average performance was 80.0% correct. The A' computed
was 0.87 (proportion hits 170/210, false alarms 45/210).
Comment. In speech perceptual tasks using isolated vowels or CV syllables, the perception of vowels
was better than the perception of consonants. The difference in performance between the vowel and the
CV discrimination tasks (tasks 4, 6) was significant at the 5% level, using the Mann-Whitney U test
(Z = 2.15, P<0.05; two-tailed). Here, as elsewhere in this section, t tests performed on arcsine transformed
scores yielded comparable results. There was a considerably smaller and not significant difference in
performance on the corresponding identification tasks (tasks 3, 5) (Z = 0.32; P = 0.75; two-tailed). While,
probably due to their longer duration and steady state, vowels were better discriminated than CV stimuli,
the assignment of phonetic labels to vowels was apparently not better than to CV stimuli. If the patient's
speech perceptual disorder was entirely due to an acoustic deficit, it would be expected that the advantage
of vowels in the discrimination task would be accompanied by a better identification of vowels compared
with consonants. Since this was not so, an additional phonetic deficit seems likely.

TABLE 10. PROPORTION CORRECT DISCRIMINATION OF CV
SYLLABLES
/ba/
/da/
/ga/
/pa/
/ta/
/ka/

/ba/

Ida/

/ga/

/pa/

Hal

0.91

0.86
0.71

0.93
1.00
0.77

0.36
1.00
1.00
0.71

1.00
0.36
0.93
0.86
0.83

/ka/
0.93
0.79
1.00
0.79
0.36
0.77

The assumption that the acoustic properties of isolated vowels (i.e., their longer steady-state portion)
accounted for their better discrimination is compatible with the finding that monosyllabic words differing
in vowel did not seem to be discriminated at all (task 1). The perception of vowels in consonant contexts

Downloaded from http://brain.oxfordjournals.org at Max-Planck-Gesellschaft on September 1, 2010

Stimulus
/ba/
/da/
/ga/
/pa/
/ta/
/ka/

/ba/

1210

P. PRAAMSTRA AND OTHERS

TABLE 11. SUMMARY OF CV IDENTIFICATION
DISCRIMINATION (TASKS 5 AND 6)
Nature of phonetic distinction/
no. of contrasting features
Voiced/voiceless
(1 distinctive feature)

AND

Identification
(binary choice)
56.7%

Discrimination
57.3%

Place-of-articulation
for voiceless stops
(1 distinctive feature)

61.7%

67.0%

Place-of-articulation
for voiced stops
(1 distinctive feature)

76.7%

93.0%

Voiced/voiceless
and place-of-articulation
(2 distinctive features)

74.2%

95.3%

A note is in order on the patient's poor discrimination of monosyllabic words and his fairly good detection
of identical syllables in bisyllabic words. Most likely, this difference is explained by the fact that each
of the monosyllabic word pairs differed in a single feature or sound contrast specially selected to detect
difficulties in auditory discrimination (Crul and Peters, 1976). In the syllable matching task the words
were not selected to have minimal sound contrast, but only such that they had (or had not) an identical
first or second syllable. The difference in performance indicates that phonological representations can be

Downloaded from http://brain.oxfordjournals.org at Max-Planck-Gesellschaft on September 1, 2010

depends partly on the extraction of phonetic attributes of the signal, while isolated vowels can be processed
on the basis of their acoustic attributes alone (Liberman et al., 1967). Due to the additional phonetic
specification, vowels in consonantal contexts are normally better identified than isolated vowels, in spite
of a shorter steady-state portion (Strange et al., 1976, 1979). The breakdown on vowels in a consonant
context implies that the patient was highly sensitive to the steady state duration of vowels. But this sensitivity
in turn might indicate that he was not able to use the phonetic cues adduced by the context, as in normal
subjects duration is not decisive (Dwyer et al., 1982). These results, therefore, cannot easily be reduced
either to an acoustic or to a phonetic deficit, but rather suggest an interaction.
In testing consonant perception we used a CV identification task with a forced binary choice procedure
similar to the task used by Auerbach et al. (1982), Tanaka et al. (1987), and Yaqub et al. (1988). Auerbach
et al. and Yaqub et al. observed better performance in their patients when the two choice alternatives
differed in two features than when they only differed in one. Presumably, to choose one of two response
alternatives on hearing 1 CV syllable, a phonological representation is generated and compared with the
phonological representations of the syllables that constitute the response alternatives. Feature effects may
arise from this comparison (Baker et al., 1981). If our patient's performance demonstrates a feature effect,
this might indicate that he can still draw on phonetic processing capacities and is not only dependent on
nonlinguistic auditory capacities. However, the difference between the performance on one-feature and
on two-feature contrasts (65.0% vs 74.2%) was not significant (Z = 0.86, P = 0.20; two-tailed). The
absence of a significant feature effect also suggests a deficit at the phonetic level.
In contrast to other authors who used the CV identification task with binary choice, we also used a
discrimination task. Although in this task the percentage correct responses were significantly higher for
stimulus pairs differing in two features than for the one-feature contrasts (95.3% vs 72.0%; Z = 2.2,
P < 0 . 0 5 ; two-tailed), the performance between different one-feature contrasts diverged even more than
between one and two-feature contrasts (see Table 11). Even an unambiguous feature effect, however, would
not necessarily imply that the patient is able to process speech at the phonetic level, since in discrimination
tasks this effect may also be attributed to the acoustic differences underlying the phonetic identity of the
speech signals.

WORD DEAFNESS

1211

constructed that include a syllabic breakdown of words, but that lack sufficiently distinctive phonetic
specifications.

Materials and procedure. Starting from a naturally spoken Dutch CVC word /bak/, 4 meaningful Dutch
CVC words were constructed. The procedure involved low-pass filtering (cut-off frequency 4 kHz, slope
24 dB/octave), digitization (10 kHz sampling rate), and storage in computer memory. Subsequently formants
were extracted with LPC-analysis and root-solving procedure using the ILS programming package (ILS,
1978). Next, the signal was resynthesized, applying a pitch synchronous energy correction (Maassen and
Povel, 1985). Because plosives and vocal murmur deteriorate after resynthesis from formants, /ak/ was
resynthesized from reflection coefficients, and the /b/ plus vocal murmur of the original spoken /bak/ was
spliced in initial position in the resynthesized /bak/. Total stimulus duration was 320 ms.
The VOT continuum (/bak/-/pak/) was constructed by changing VOT in 9 steps from —80 ms (i.e.,
80 ms vocal murmur preceding the plosive) to +12 ms (i.e., no vocal murmur and a silent gap of 12 ms
between plosive and vowel onset). Stimulus 7 had VOT = 0 ms. It should be noted that in Dutch the
phoneme boundary between voiced and voiceless plosive cognates is different from that in English. The
value around —20 ms that we found in normal subjects {see fig. 3) is consistent with previous reports
(Lisker and Abramson, 1964; Slis and Cohen, 1969).
Two place-of-articulation continua (/bak/-/dak/ and /pak/-/tak/) were constructed by manipulating the
second (F2) and third (F3) formant transitions. F2 started at 1000 Hz for the bilabial and at 2000 Hz for
the alveolar plosive, and ended at 1200 Hz in the /a/. F3 transition starting values were 2500 Hz and 3100 Hz,
respectively, and ending value 2000 Hz. By linear interpolation of F2 and F3,9 steps on the place continua
were constructed. Transition duration was 40 ms for both F2 and F3. Fl increased from 200 Hz to 750 Hz
in a much shorter transition duration (14 ms) in order to make the plosive character of the initial consonant
more prominent, and to emphasize the voiceless cognate. The analysed signal was subsequently resynthesized
as described above.
Identification tapes consisted of 90 stimuli for each continuum. Each stimulus was presented 10 times.
The interval between successive stimuli was 3.3 s. Discrimination tapes contained 78 pairs, 42 of which
differed by two steps on the continuum, and 36 that differed by 3 steps. In the /bak/-/pak/ and /pak/-/tak/
continua the intervals between pairs were 4.0 s and intervals between stimuli within pairs 2.3 s. In the
second place-of-articulation continuum (/bak/-/dak/) the interstimulus interval within pairs was 960 ms.
The patient was tested in a quiet room on three different occasions. The stimuli were presented through
headphones. In discrimination tasks, the patient responded by marking on an answer sheet the response
alternative 'same' or 'different'. In the identification tasks he had to mark the words constituting the response
alternatives.

Downloaded from http://brain.oxfordjournals.org at Max-Planck-Gesellschaft on September 1, 2010

7. CVC syllable identification and discrimination; synthetic speech
In addition to the previous tasks using natural speech, the patient was investigated using synthetic speech.
Identification and discrimination tests using synthetic speech stimuli have been developed to systematically
investigate auditory and phonetic processes in speech perception (Liberman et al., 1967). Similar tests
have been used to investigate speech perception deficits in aphasia (Basso et al., 1977; Blumstein et al.,
19776, 1984; Yeni-Komshian and Lafontaine, 1983). The synthetic nature of the stimuli allows for control
of the physical/acoustical dimensions that underlie the characterization of speech signals in terms of phonetic
features. Whereas the phonetic feature voice, for example, distinguishes between /b/ (voiced) and /p/
(unvoiced) in a dichotomous way, the acoustical property voice-onset-time (VOT) underlying the distinction
can be varied continuously. This principle is put to use in identification and discrimination tasks on VOT
and place of articulation continua. Such tests consist of stimuli that span the acoustical distance between,
for instance, /b/ and /p/ (VOT continuum) or /b/ and /d/ (place-of-articulation continuum) in a stepwise
fashion. These tests have demonstrated that there is a remarkably well-defined boundary between stimuli
on such a continuum labelled as one or the other speech signal by normal subjects. Normal subjects only
discriminate accurately between stimuli that they can assign to different phonetic categories. Two stimuli
from the same phonetic category but separated by a certain distance on the continuum are not discriminated
at better than chance level, while 2 stimuli separated by the same distance and from different categories
are. This phenomenon of categorical perception is central to the issue of whether or not the perception
of speech requires special mechanisms (Repp, 1984; Kuhl, 1986). Our patient was submitted to synthetic
speech tests for one VOT continuum (/b/-/p/) and two place-of-articulation continua (/p/-/t/ and Ibl-I&l).

P. PRAAMSTRA AND OTHERS

1212

Results. The results are shown in fig. 3. The discrimination conditions shown are those with 3 steps
between pair members. Except possibly for the /pak/-/tak/ continuum identification performance seems
random. Discrimination performance is relatively preserved in the VOT continuum.

VOT (/bak/-/pak/)
Identification

3 4 5 6 7
Stimulus value

2-5

Place (/pak/-/tak/)
Identification

2

3 4 5 6 7
Stimulus value

Place (/pak/-/tak/)
Discrimination

9 g

1-4

3

4 5 6 7
Stimulus value

2-5

3-6 4-7 5-8
Stimulus pair

6-9

Place (/bak/-/dak/)
Discrimination

Place (/bak/-/dak/)
Identification

2

3-6 4-7 5-8
Stimulus pair

9 g

1-4

2-5

3-6
4-7
Stimulus pair

5-8

6-9

FIG. 3. Performance in identification and discrimination tasks on three different speech continua. Solid lines represent
the patient's performance, dashed lines the averaged performance of 10 normal subjects.

Comment. From the severe impairment of both discrimination and identification performance, we inferred
that there is probably a phonetic as well as an auditory component to the patient's speech perceptual
disturbance. It could be argued that the impaired identification is fully determined by an inability to
discriminate the stimuli, which might be attributed to an auditory deficit. This is not an adequate account,
however, because discrimination on the VOT continuum was relatively preserved, while identification
was random. The contrasting hypothesis that the pattern of results is due to a phonetic deficit only is equally
unlikely. This possibility would be supported if the disturbance in identification performance had been

Downloaded from http://brain.oxfordjournals.org at Max-Planck-Gesellschaft on September 1, 2010

2

VOT (/bak/-/pak/)
Discrimination

WORD DEAFNESS

1213

S U M M A R Y OF MAJOR F I N D I N G S
The patient we studied suffered a left hemisphere stroke and acquired a Wernicke aphasia at the age
of 54 yrs. Three years later, a right hemisphere stroke caused severe deterioration of auditory language
comprehension while the patient's language expression remained unchanged. The recognition of
environmental sounds was only mildly impaired.
The left hemisphere infarct was located temporoparietally and involved most of the superior temporal
gyms. It probably left the geniculotemporal radiation intact. The right hemisphere infarct was smaller.
It also involved the superior temporal gyrus and extended posteriorly, like the left hemisphere infarct,
onto the angular gyrus. The auditory radiation through the temporal isthmus was spared.
Pure tone audiometry and BAEPs established intact auditory afference up to the midbrain. Middle and
long-latency auditory evoked potentials could be elicited both with left and right ear stimulation. A slight
asymmetry in the Pa component was consistent with CT evidence of greater destruction of the left hemisphere
auditory cortex. The PI, Nl and P2 deflections were essentially normal, suggesting that the auditory cortices
might contain some intact remnants.
Comparison of patient's performance on the Aachen Aphasia Test before and after his second stroke
confirmed that only auditory comprehension had suffered, while performance on tasks not dependent on
the auditory input channel was unchanged. The performance on an auditory lexical decision task was nearly
normal prior to the second stroke, but random after it. When the same targets were presented visually,

Downloaded from http://brain.oxfordjournals.org at Max-Planck-Gesellschaft on September 1, 2010

accompanied by an intact discrimination (Yeni-Komshian and Lafontaine, 1983). Our patient's overall
discrimination performance, however, was evidently impaired, which rules out this possibility.
The selectively preserved discrimination on the VOT continuum might be due to the perception of voicing
differences being more readily mediated by general auditory mechanisms, as the feature voice is more
closely tied to acoustically invariant properties than place-of-articulation (for critical discussion, see
Summerfield, 1982). This explanation implies a deficit at the phonetic level. However, while in the synthetic
speech tasks the voicing distinction was better perceived than the place contrast, the reverse was true using
natural speech. Although differences between results obtained with natural speech and results obtained
with synthetic speech are not uncommon in patients (Miceli et al., 1982; Caramazza et al., 1983), the
discrepancy in our results cautions us not to overstate this argument (see also Addendum).
The speech perceptual investigations may be summarized by the conclusion that the patient's level of
performance was comparable with that found in other cases of word deafness attributed to a speech perceptual
deficit (Saffran et al., 1976; Miceli, 1982; Auerbach et al., 1982; Yaqub et al., 1988). As to the pattern
of performance, several findings argued against a reduction of the speech perceptual disorder to an auditory
processing deficit only. (1) Discrimination of vowels was superior to the discrimination of consonants,
while identification of vowels and consonants was not significantly different, though more severely impaired.
This dissociation between discrimination and identification performance argues for a phonetic deficit. (2)
The better discrimination of isolated vowels compared with consonants was not mirrored in the discrimination
of monosyllabic words differing either in (medial) vowel or in a consonant. Since vowels in a consonantal
context have a shorter steady-state portion, this might be explained in terms of an auditory deficit. But
in view of evidence that consonantal context provides additional phonetic specification that aids vowel
recognition, a phonetic deficit has to be assumed to explain the apparent inability to use these cues and,
thereby, the strong dependence on steady-state duration of vowels. (3) The absence of a significant feature
effect in CV identification with binary choice and the lack of an unambiguous feature effect in CV
discrimination implies that recognition of these sounds is mainly dependent on nonlinguistic auditory
capacities. (4) In the presence of a relatively preserved discrimination on the VOT continuum, the otherwise
severe disturbance of identification and discrimination on synthetic speech continua indicates that both
auditory and phonetic levels of speech perception are affected.
Taken together, these results point to a combination of auditory and phonetic impairments. The strength
of this conclusion, however, is limited by the fact that auditory and phonetic deficits cannot be assessed
independently given that the auditory processes essential to speech perception are largely unknown. Also,
the results do not allow any firm statement with respect to the relative contributions of the two types of
impairment.

1214

P. PRAAMSTRA AND OTHERS

GENERAL DISCUSSION

In many cases of pure word deafness that have been reported, the syndrome was
accompanied by aphasic disturbances (Goldstein, 1974; Buchman et al., 1986). Most
case reports, however, do not explicitly take issue with the presence of aphasic symptoms
and treat them as merely contaminating the word deafness. The possibility of an
underlying deficit that contributes both to the aphasic disturbance and to the word deafness
in these patients is rarely considered {but see Caramazza et al., 1983; Franklin, 1989).
The patient we studied is a rare instance of the coexistence of aphasia with word deafness,
in that the aphasia already existed and was documented at the time the patient acquired
word deafness. The word deafness could, therefore, clearly be distinguished from the
aphasia. This does not preclude a deficit related to the aphasia also being relevant to
the word deafness. Surely the word deafness of our patient would not have occurred
if the right hemisphere stroke had not been preceded by one in the left hemisphere.
The nature of the left hemisphere's contribution to word deafness will be our main
focus in this discussion. In the analysis of our case, we chose to distinguish auditory
and phonetic stages of speech perception, since earlier case reports claimed a specific
deficit at either one or the other level (Saffran et al., 1976; Auerbach et al., 1982;
Yaqub et al., 1988). While we suggested a mixed deficit, the arguments we provided
are not incontrovertible, given the fact that there is no generally accepted account of
how auditory and phonetic processes are related, let alone a way to assess auditory and
phonetic deficits independently. However, as we will discuss below, the left hemisphere
specialization of phonetic processing capacities and the explanation of word deafness
as a disconnection syndrome favour a mixed deficit rather than a pure auditory or phonetic
deficit. We further discuss how this ties in with present knowledge on speech perceptual
deficits in aphasia.
Auditory vs speech modes of perception
The distinction between auditory and phonetic levels or modes of perception originates
partially in research on categorical perception (Liberman et al., 1967). Recent
developments in categorical perception research, such as the use of animal models and
the demonstration that nonspeech sounds can be categorically perceived, have changed
the field, but not eliminated the distinction (Repp, 1984; Kuhl, 1986). Of particular
interest to the present discussion is the support for the distinction from studies on

Downloaded from http://brain.oxfordjournals.org at Max-Planck-Gesellschaft on September 1, 2010

he made no errors. Furthermore, the patient could not distinguish auditorily presented Dutch words from
English words very well. On a syllable matching task, however, he showed a fairly good performance.
This suggested that he was still able to extract some sort of phonological representation from the
acoustic-phonetic input, including a syllabic breakdown of the words.
Further investigations focused on speech perceptual abilities, revealing an impairment of phoneme
identification and discrimination using monosyllabic words, vowels and CV syllables. The voicing contrast
was perceived less well than the place-of-articulation contrast. The performance on synthetic speech tasks
using place-of-articulation and voice-onset-time continua was also severely disturbed. Only discrimination
of stimuli on the VOT continuum showed a preserved phoneme category boundary. The overall pattern
of results provided arguments for a speech perceptual disturbance to which both general auditory and phonetic
deficits seem likely to contribute.

WORD DEAFNESS

1215

Downloaded from http://brain.oxfordjournals.org at Max-Planck-Gesellschaft on September 1, 2010

hemispheric specialization using dichotic presentation of speech stimuli (for a review,
see Tartter, 1988). Ear asymmetries obtained in dichotic studies have been interpreted
as due to hemisphere asymmetries in processing capacities. Since stop consonants display
a right ear advantage not displayed by vowels, it has been suggested that consonants
require a speech mode of perception, which is left hemisphere based, while vowels can
be processed in both hemispheres by general auditory mechanisms (Studdert-Kennedy
and Shankweiler, 1970; Studdert-Kennedy et al., 1972). Evidence from patients with
hemisphere lesions and evidence from split-brain research also indicate that phonetic
processing capacities are, indeed, unique to the left hemisphere (Oscar-Berman et al,
1975; Zaidel, 1976; Levy and Trevarthen, 1977). Finally, animal data suggesting a
left hemisphere specialization for species-specific calls provide circumstantial evidence
for the engagement of left hemisphere specialized processes in speech perception (Petersen
et al, 1978; Miller and Jusczyk, 1989).
The distinction between auditory and speech modes of perception has been invoked
by several authors to explain the dissociation between verbal and nonverbal auditory
agnosia (Albert et al, 1972; Albert and Bear, 1974; Lambert et al, 1989). Although
most patients that have been reported displayed difficulties in both verbal and nonverbal
sound recognition (Gazzaniga et al, 1973; Chocholle et al, 1975; von Stockert, 1982;
Miceli, 1982; Buchtel and Stewart, 1989), the occurrence of relatively pure cases does
support a double dissociation (Lambert et al., 1989). The distinction between auditory
and speech modes of perception might indeed be relevant to the explanation of this
dissociation. However, the way it was applied so far merely restated the observation
that the processing of speech and nonspeech sounds diverge at some level, and did not
elucidate the functional loci of verbal and nonverbal auditory agnosia.
The distinction between auditory and speech modes of perception has also been invoked
to account for different patterns of deficits in word deafness. Auerbach et al (1982)
argued that the speech perceptual deficit encountered in word deafness must be considered
prephonetic if it can be attributed to a disorder in acoustic processing. He specified
this acoustic deficit as a disorder of temporal auditory acuity, to be demonstrated by
psychoacoustic measures like click-counting and click-fusion. While this deficit is
localized at a prelinguistic level, it has its major manifestation at a linguistic level, as
the auditory temporal resolution suffices for the perception of nonverbal sounds but
is insufficient to resolve the rapid acoustic transients characteristic of certain speech
sounds. Auerbach et al formulated a number of criteria that purportedly distinguish
patients with a deficit at the acoustic (prelinguistic) level from patients with a deficit
at the phonetic (linguistic) level. While they described a case of the former type, a welldocumented case that fits the latter type has been reported by Saffran et al. (1976).
The model proposed by Auerbach et al (1982) connects the distinction between
phonetic and acoustic deficits in word deafness with the lesion type being unilateral
or bilateral. It associates the acoustic type of deficit with bitemporal lesions ('type I
word deafness') and the phonetic deficit with unilateral left hemisphere lesions ('type
II word deafness'). Thus the model offers an explanatory scheme that relates deficits
at different stages of processing to distinct lesion patterns. We want to argue in the
next section, however, that the distinction between acoustic and phonetic deficits in
word deafness must be dissociated from the distinction between unilateral and bilateral
cases.

1216

P. PRAAMSTRA AND OTHERS

Mixed deficits in word deafness
In our view, a classification of word deafness as acoustically or phonetically based
bears no relation to the lesion being unilateral or bilateral. In addition, the dichotomous
classification itself may be criticized on several grounds.
First, the acoustic deficit documented in bilateral cases might also be present in patients
with unilateral lesions. We know of only one unilateral case in which the presence of
acoustic deficits was investigated (Albert and Bear, 1974). This patient evidenced an

Downloaded from http://brain.oxfordjournals.org at Max-Planck-Gesellschaft on September 1, 2010

The explanation of pure word deafness as a disconnection syndrome
Pure word deafness has long been recognized as a sequel either to a left hemisphere
temporal lobe lesion or to bilateral temporal lobe lesions (Schuster and Taterka, 1926;
Geschwind, 1965). The explanation of pure word deafness as a disconnection syndrome
(Geschwind, 1965) unifies unilateral and bilateral cases in that it assumes that in unilateral
cases the lesion extends deep subcortically and involves the interhemispheric auditory
pathways. This explanation is strengthened by evidence from dichotic listening studies
which elucidated the functional significance of interhemispheric auditory connections.
Sparks and Geschwind (1968) established that sectioning the corpus callosum results
in an almost complete suppression of left ear dichotic verbal information. This finding
confirmed that left ear dichotic verbal information, arriving in the right hemisphere
auditory cortex, has to be transferred to the left hemisphere by commissural auditory
fibres in order to be processed as language.
The disconnection explanation of pure word deafness also derives support from
neuroradiological studies (Rubens etal., 1978; Damasio and Damasio, 1979), which
demonstrated that patients with left hemisphere lesions and a left ear extinction in verbal
dichotic listening tasks have a consistent pattern of lesions extending deep to the wall
of the lateral ventricle. It can be concluded that unilateral left hemisphere lesions may
indeed cause damage to the interhemispheric auditory pathways and deprive the left
hemisphere of auditory input via this pathway. According to Geschwind's disconnection
theory, a lesion of this type may have the same effect on language function as bilateral
temporal lobe lesions.
Most of the case reports published in the last decade, including Auerbach et al. (1982),
subscribe to a disconnection type of explanation of pure word deafness. But whereas
the disconnection hypothesis unifies unilateral and bilateral cases in a single explanation,
Auerbach et al. chose to subdivide the syndrome again into two distinct types. It is
unclear how Auerbach et al. intended their adherence to the disconnection hypothesis
to be reconciled with their subdivision of pure word deafness into different types. Since
the disconnection hypothesis implicates both auditory cortices in the causation of word
deafness (either by direct lesion involvement or by a disconnection), a classification
of word deafness into auditory and phonetic types cannot be based on the differential
specialization of left and right auditory cortex. This does not imply that unilateral and
bilteral cases must be identical in all respects. Whereas only bilateral cases seem to
go through an initial episode of cortical deafness (Mendez and Geehan, 1988), unilateral
cases may be more frequently accompanied by aphasic symptoms, purely contingent
on the considerable extension of a lesion which, according to the disconnection hypothesis,
must involve the interhemispheric auditory connections and the auditory cortex at the
same time.

WORD DEAFNESS

1217

The main thrust of this discussion is that most patients are not sufficiently documented
to allow a classification according to the dichotomous scheme proposed by Auerbach
et al. (1982). Moreover, evidence can be gathered from recent literature amounting
to counter-examples to the proposed explanatory scheme. This prepares the ground for
an alternative hypothesis that implies the involvement of auditory as well as phonetic
processes in the causation of word deafness. The findings in our own case, indicating
a mixed deficit rather than an isolated acoustic or phonetic deficit, are consistent with
the proposed hypothesis. The way word deafness evolved from Wernicke's aphasia in
our patient may further elucidate the hypothesis.
A remarkable feature of our patient is the fact that the change in his language status,
caused by the right hemisphere infarct, was limited to the auditory input modality, as
demonstrated by the A AT results before and after the second vascular accident. The
second stroke did not even cause temporary changes outside the domain of the auditory
recognition disturbance. This indicates that the patient's residual language capacities
after his first stroke were still subserved by the damaged left hemisphere, which was
only dependent on the right hemisphere for transfer of auditory information. Presumably
the severe involvement of the left hemisphere auditory structures compromised the left
hemisphere specialized processes that support phonetic analysis. This did not result in
specific speech perceptual difficulties as long as general auditory processes subserved
by the right hemisphere functioned adequately, given that intact interhemispheric
connections allowed an interplay between the two hemispheres. The subsequent

Downloaded from http://brain.oxfordjournals.org at Max-Planck-Gesellschaft on September 1, 2010

impairment of auditory temporal resolution only slightly milder than that displayed by
the bilateral case of Auerbach et al. (1982).
A second reason for questioning the proposed classification of word deafness is the
improbability of an acoustic deficit as the only cause of the syndrome in all bilateral
cases. A recent case report carefully documented a severe psychoacoustic deficit in a
patient with bilateral lesions (Tanaka et al., 1987). We suspect that even the severe
deficit of this case only caused word deafness because it occurred together with a higher
level deficit. This patient had gradually recovered from Wernicke's aphasia due to a
left temporal lobe lesion when she experienced a right hemisphere stroke that caused
word deafness. The authors documented deficits in temporal resolution (Albert and Bear,
1974) and in temporal sequencing (Efron, 1963; Swisher and Hirsh, 1972). However,
these deficits were not limited to the auditory modality, but also encompassed the visual
and somatosensory modalities. A supramodal temporal sequencing disorder seems to
be associated with left hemisphere lesions that cause aphasia, though not with left
hemisphere lesions as such (Efron, 1963). The presence of this supramodal deficit suggests
persistent effects due to the left hemisphere lesion, in spite of the patient's reported
recovery from aphasia. Furthermore, for an acoustic deficit one would predict vowel
perception to be better than consonant perception, consistent with the criteria formulated
by Auerbach et al. (1982). The patient's performance was indeed slightly better on
vowels, but at the same time the recognition of nonverbal sounds was only mildly
disturbed. We would have expected that an auditory processing disorder that severely
disrupts vowel perception would also strongly affect the recognition of environmental
sounds. In sum, these features of the case reported by Tanaka et al. (1987) provide
reasons to express doubts on an auditory processing deficit as the only cause of this
patient's word deafness.

1218

P. PRAAMSTRA AND OTHERS

Speech perceptual deficits in aphasia
In spite of the evidence we have already cited in favour of left hemisphere specialized
capacities for phonetic analysis, the disconnection explanation of word deafness could
make one more inclined to take both hemispheres for equipotential in this respect, or
even doubt the existence of special mechanisms for phonetic analysis. The question to
be dealt with in this section is therefore whether studies on speech perceptual deficits
in aphasia support left hemisphere specialized phonetic processes.
One study by Oscar-Berman et al. (1975) explicitly addressed phonetic processing
in left and right hemisphere-damaged patients. It was found that left hemisphere patients,
in contrast to right hemisphere patients, were impaired in the use of a phonetic code
in identification tasks using dichotically presented CV syllables. These results support
a distinction between auditory and phonetic levels of speech processing and strongly
suggest that it is the left hemisphere that is specialized for the conversion of auditory
parameters of speech into phonetic attributes.
The majority of studies pertinent to the issue have focused on the question to what
extent speech perceptual deficits contribute to auditory language comprehension
impairments in aphasia. These studies have in large part been motivated by the claim
by Luria (1970) that the comprehension deficit of Wemicke's aphasia should be attributed
to a specific impairment of phonetic analysis. A number of studies using phoneme
identification and discrimination tasks have found perceptual difficulties but have failed
to show a consistent relation between speech perception and auditory language
comprehension (Basso et al., 1977; Jauhiainen and Nuutila, 1977; Blumstein et al.,
1977a, b, 1984). Other studies found only a weak relation (Varney and Benton, 1979;
Varney, 1984; Miceli et al., 1980; Yeni-Komshian and Lafontaine, 1983; Riedel and
Studdert-Kennedy, 1985). It thus appears that impairments in phonetic perception that
can be demonstrated in aphasia are not a major factor in aphasic disturbances of auditory
language comprehension. However, the results do not refute the original assumption
that there are specialized left hemisphere mechanisms for phonetic analysis. This is
supported, first, by the fact that right hemisphere-damaged subjects evidenced no deficit

Downloaded from http://brain.oxfordjournals.org at Max-Planck-Gesellschaft on September 1, 2010

involvement of the right hemisphere auditory cortex, then, sufficiently degraded the
auditory input to linguistic processing stages that a specific speech perceptual deficit,
that is, word deafness, resulted. We thus suppose that, before the second stroke, the
patient was already partially dependent on nonlinguistic auditory capacities to compensate
for deficient phonetic processing.
The interpretation we propose of our case might apply more generally. We suggest
that word deafness is caused by deficits both of phonetic processes and of nonlinguistic
general auditory capacities. This hypothesis strongly relies, first, on the disconnection
hypothesis as an account of the underlying neurological events, and, secondly, on the
left hemisphere's specialization for phonetic analysis. These two elements do not in
all respects fit together, however. Whereas the disconnection hypothesis makes word
deafness conditional on bilateral lesions or a disconnection, it is unclear why, given
the existence of specialized phonetic processes for speech perception, a selective
disturbance of these processes by a left hemisphere lesion does not suffice to produce
severe speech perceptual problems. Evidence relevant to this issue can be obtained from
the study of speech perceptual deficits in aphasia.

WORD DEAFNESS

1219

CONCLUSIONS

The present conception of word deafness is far removed from the original conception
of the agnosias as central disturbances of recognition (Bauer and Rubens, 1985). Like

Downloaded from http://brain.oxfordjournals.org at Max-Planck-Gesellschaft on September 1, 2010

in phoneme perception (Goldblum and Albert, 1972; Basso etal., 1977; Blumstein et al.,
1911b; Varney and Benton, 1979). In addition, a number of studies consistently found
Wernicke's aphasics worse than other aphasia types on phoneme perception tasks
(Goldblum and Albert, 1972; Blumstein et al., 1911b; Baker et al., 1981). This might
be related to temporal lobe damage and concurrent involvement of auditory cortical
structures that presumably underlie phonetic processing capacities. Finally, in some
studies that explored categorical perception in aphasia, phoneme discrimination was
better preserved than identification performance (Blumstein et al., 1911b; Riedel and
Studdert-Kennedy, 1985), that is, a dissociation that argues for an impairment of phonetic
perception (Yeni-Komshian and Lafontaine, 1983; Riedel and Studdert-Kennedy, 1985).
In short, these studies on speech perception and auditory language comprehension in
aphasia are compatible with the existence of a specialized phonetic competence subserved
by the left hemisphere.
The puzzling fact that remains to be explained, then, is why a disturbance of phonetic
perception by a left hemisphere lesion contributes so little to impairments of auditory
language comprehension in aphasia. Part of the answer is provided by work conducted
by Blumstein and coworkers. Whereas most investigations referred to in the last paragraph
set out to isolate speech perceptual deficits from higher level language impairments,
Baker et al. (1981) and Milberg et al. (1988) also addressed the question of whether
speech perceptual deficits interacted with semantic impairments and the process of lexical
access. These studies suggested an intimate interaction between phonological and semantic
factors in auditory word processing. More specifically, the first study indicated that
perceptual deficits may become manifest with increased semantic demands, even if they
are not evident in perceptual tasks. Thus there might be a small but more pervasive
influence of speech perceptual deficits on auditory language comprehension in aphasia
than earlier investigations suggested.
Within a modular framework, Blumstein (1987) summarized the investigations referred
to above by stating that the speech perception system, that is, the phonetic/phonological
processor, is not functionally autonomous since it can be affected by higher level language
impairments. Our conjecture that word deafness might be due to an interaction of auditory
and phonetic deficits emphasizes the obvious possibility that it can also be affected by
lower level auditory impairments. This proposal is in some sense complementary to
Blumstein's view. The relatively selective disturbance of speech perception in word
deafness seems difficult to account for in purely auditory terms, given that neither auditory
temporal resolution, nor any other auditory process can be singled out as crucial to
intact speech perception. The evidence from aphasia on the other hand makes clear that,
while left hemisphere lesions can cause speech perceptual deficits, these deficits alone
do not account for major impairments of auditory language comprehension. Therefore,
the hypothesis of a mixed deficit in word deafness might be in better agreement with
our present knowledge of auditory cortical function than the hypothesis of isolated acoustic
or phonetic deficits.

1220

P. PRAAMSTRA AND OTHERS

ADDENDUM
An anonymous reviewer proposed that the preserved discrimination of /bak/-/pak/ might be due to an
artificial cue caused by the fact that the voiced and voiceless stimuli had different proportions of original
and resynthesized speech. We investigated the possibility that voiceless stimuli sounded more unnatural
than voiced stimuli by presenting 8 normal subjects with the same stimuli as used in the identification
tasks on the VOT continuum and the two place-of-articulation continua. The subjects, who were naive
with respect to synthetic speech continua, were asked to rate the 'naturalness' of the stimuli on a 5 point
scale, ranging from —2 to +2. Prior to analysis, the 5 response categories were combined into 3 categories
(unnatural, neutral, natural). The results demonstrated no difference in the percentage 'neutral' responses
on the three continua (6%, 4% and 4%). The percentage 'natural' responses for each stimulus of the three
continua is represented in fig. 4. The response patterns on the two place-of-articulation continua show,
not unexpectedly, that the stimuli at the ends of the continua were most frequently rated as natural. The
VOT continuum diverges from this pattern, but the stimuli at the voiced extreme of the continuum are
not rated more frequently as natural than the stimuli at the voiceless extreme. Therefore, the pattern that
would be expected if the suggested explanation for the preserved discrimination on the VOT continuum
were right, was not obtained.

Downloaded from http://brain.oxfordjournals.org at Max-Planck-Gesellschaft on September 1, 2010

many patients reported in recent decades, our patient had a speech perceptual disturbance
of sufficient severity to explain the word deafness. The documentation on his language
status before and after his second stroke rules out that changes occurred other than those
related to the disturbance of the auditory input channel. The very discrete change in
the language status, occasioned by the right temporal lobe infarction, strongly suggests
that, apart from auditory input via the right hemisphere, there had been no compensation
of any importance of left hemisphere language functions by the right hemisphere, since
in this case a more general change should be expected.
The nature of the speech perception impairment in our patient does not fit in well
with the hypothesis of Auerbach etal. (1982), stating that speech perceptual disturbances
in word deafness can be classified as due to either a prephonetic, acoustic deficit or
a deficit at the linguistic level of phonetic perception. In contrast to the proposal by
Auerbach et al., we conjecture that word deafness is due to an interaction between an
impairment of general auditory processes and a deficit at the level of phonetic analysis.
The presence of deficits at both these levels is coherent with the left hemisphere
specialization of phonetic processing capacities and with the necessary involvement of
both auditory cortices (by direct lesion involvement or by a disconnection) in word
deafness.
Evidence from aphasia, reviewed by Blumstein (1987), suggests that a deficit at the
level of phonetic analysis gives rise to a disturbance of auditory language comprehension
only in the presence of a higher level language deficit. Similarly, according to our
hypothesis, a deficit at the phonetic level gives rise to a manifest disorder of speech
perception only if it is compounded with a lower level deficit of general auditory
processes, that is, when a left temporal lobe infarct extends deep enough to involve
interhemispheric auditory fibres, or when it is accompanied by a right temporal lobe
infarct. This implies that speech perceptual processes, in spite of their left hemisphere
specialization, are quite robust in the face of left hemisphere damage.

WORD

DEAFNESS

1221

1UU
VI

p

C
O

80'

•s>

\

p

60'

, s%

^SV

...

3
_c

c

/ s'

40-

"'"--. /

20-

1)

u

CJ

0.

01

2

3

4

5
6
7
Stimulus value

8

9

FIG. 4. Naturalness ratings by 8 normal subjects of the stimuli used in 1 VOT continuum and 2 place-of-articulation
continua. Hatched line = bak-pak; broken line = pak-tak; continuous line = bak-dak.

ACKNOWLEDGEMENTS
We wish to thank S. Garnsey, A. Lahiri and A. S. Meyer for their comments on the manuscript. We
are grateful to R. E. Wong Chung who kindly allowed us to study the patient under his care, and to
B. P. M. Schulte for his constant support. H. O. M. Thijssen performed the CT scans and helped in their
interpretation. E. J. Colon and S. Kooijman performed the evoked potentials investigations. E. Gonera
was the first to recognize the syndrome and drew our attention to the patient. We are most indebted to
A.O. for his patient cooperation. During part of this research the first author was supported by a grant
from the Nederlandse Hartstichting (no. 88.204).

REFERENCES
ALBERT ML, SPARKS R, VON STOCKERT T, SAX D (1972) A case study of auditory agnosia: linguistic

and non-linguistic processing. Cortex, 8, 427-443.
ALBERT ML, BEAR D (1974) Time to understand: a case study of word deafness with reference to the
role of time in auditory comprehension. Brain, 97, 373—384.
AUERBACH SH, ALLARD T, NAESER M, ALEXANDER MP, ALBERT ML (1982) Pure word deafness: analysis

of a case with bilateral lesions and a defect at the prephonemic level. Brain, 105, 271-300.
BAKER E, BLUMSTEIN SE, GOODGLASS H (1981) Interaction between phonological and semantic factors

in auditory comprehension. Neuropsychologia, 19, 1 — 15.
BASSO A, CASATI G, VIGNOLO LA (1977) Phonemic identification defect in aphasia. Cortex, 13, 8 5 - 9 5 .
BAUER RM, RUBENS AB (1985) Agnosia. In: Clinical Neuropsychology. Second edition. Edited by
K. M. Heilman and E. Valenstein. New York and Oxford: Oxford University Press, pp. 187-241.
BLUMSTEIN SE (1987) Speech perception and modularity: evidence from aphasia. In: Motor and Sensory
Processes of Language. Edited by E. Keller and M. Gopnik. Hillsdale, NJ: Lawrence Erlbaum,
pp. 257-275.
BLUMSTEIN SE, BAKER E, GOODGLASS H (1977a) Phonological factors in auditory comprehension in aphasia.

Neuropsychologia, 15, 19—30.
BLUMSTEIN SE, COOPER WE, ZURIF EB, CARAMAZZA A (1977ft) The perception and production of

voice-onset time in aphasia. Neuropsychologia, 15, 371—383.

Downloaded from http://brain.oxfordjournals.org at Max-Planck-Gesellschaft on September 1, 2010

o

1222

P. PRAAMSTRA AND OTHERS

BLUMSTEIN SE, TARTTER VC, NIGRO G, STATLENDER S (1984) Acoustic cues for the perception of place

of articulation in aphasia. Brain and Language, 22, 128-149.
BUCHMAN AS, GARRON DC, TROST-CARDAMONE JE, WICHTER MD, SCHWARTZ M (1986) Word deafness:

one hundred years later. Journal of Neurology, Neurosurgery and Psychiatry, 49, 489—499.
BUCHTEL HA, STEWART JD (1989) Auditory agnosia: apperceptive or associative disorder? Brain and
Language, 37, 12—25.
CARAMAZZA A, BERNDT RS, BASILI AG (1983) The selective impairment of phonological processing:
a case study. Brain and Language, 18, 128—174.
CHOCHOLLE R, CHEDRU F, BOTTE MC, CHAIN F, LHERMITTE F (1975) Etude psychoacoustique d'un

of Neurophysiology, 20, 300-315.
DIVENYI PL, ROBINSON AJ (1989) Nonlinguistic auditory capabilities in aphasia. Brain and Language,
37, 290-326.
DWYER J, BLUMSTEIN SE, RYALLS J (1982) The role of duration and rapid temporal processing on the
lateral perception of consonants and vowels. Brain and Language, 17, 272—286.
EFRON R (1963) Temporal perception, aphasia and deja vu. Brain, 86, 403-424.
FRANKLIN S (1989) Dissociations in auditory word comprehension; evidence from nine fluent aphasic
patients. Aphasiology, 3, 189-207.
FRAUENFELDER UH, TYLER LK (1987) The process of spoken word recognition: an introduction. Cognition,
25, 1-20.
GAZZANIGA MS, VELLETRI GLASS A, SARNO MT, POSNER JB (1973) Pure word deafness and hemispheric

dynamics: a case history. Cortex, 9, 136 — 143.
GESCHWIND N (1965) Disconnexion syndromes in animals and man. Parts I and II. Brain, 88, 237-294,
585-644.
GOLDBLUM MC, ALBERT ML (1972) Phonemic discrimination in sensory aphasia. International Journal

of Mental Health, 1, 25-29.
GOLDSTEIN MN (1974) Auditory agnosia for speech ('pure word-deafness'): a historical review with current
implications. Brain and Language, 1, 195-204.
GREEN DM, SWETS JA (1966) Signal Detection Theory and Psychophysics. New York and London: John
Wiley.
GRIER JB (1971) Nonparametric indexes for sensitivity and bias: computing formulas. Psychological Bulletin,
75, 424-429.
HAGOORT P (1989) Processing of lexical ambiguities: a comment on Milberg, Blumstein, and Dworetzky
(1987). Brain and Language, 36, 335-348.
HAGOORT P (1990) Tracking the Time-course of Language Understanding in Aphasia. Unpublished doctoral
dissertation. The Netherlands: University of Nijmegen.
HARI R, AITTONIEMI K, JARVINEN M-L, KATILA T, VARPULA T (1980) Auditory evoked transient and

sustained magnetic fields of the human brain: localization of neural generators. Experimental Brain
Research, 40, 237-240.
HUBER W, WENIGER D, POECK K, WILLMES K (1980) Der Aacherier Aphasie Test. Aufbau und Uberpriifung
der Konstruktion. Nervenarzt, 51, 475—482.
HUBER W, POECK K, WILLMES K (1984) The Aachen Aphasia Test. Advances in Neurology, 42, 291 - 3 0 3 .
ILS (1978) Interactive Laboratory Systems. Santa Barbara: Signal Technology.
JAUHIAINEN T, NUUTILA A (1977) Auditory perception of speech and speech sounds in recent recovered
cases of aphasia. Brain and Language, 4, 572—579.
JERGER J, WEIKERS NJ, SHARBROUGH FW, JERGER S (1969) Bilateral lesions of the temporal lobe: a case
study. Ada Oto-Laryngologica, Supplement 258, 7—51.

Downloaded from http://brain.oxfordjournals.org at Max-Planck-Gesellschaft on September 1, 2010

cas de 'surdite' corticale'. Neuropsychologia, 13, 163 — 172.
COSLETT HB, BRASHEAR BR, HEILMAN KM (1984) Pure word deafness after bilateral primary auditory
cortex infarcts. Neurology, Cleveland, 34, 347 — 352.
CRUL TH, PETERS H (1976) Auditive Discriminative Test. Handleiding. Amsterdam: Swets and Zeitlinger.
DAMASIO H, DAMASIO A (1979) 'Paradoxic' ear extinction in dichotic listening: possible anatomic
significance. Neurology, New York, 29, 644-653.
DE RENZI E, VIGNOLO LA (1962) The Token Test: a sensitive test to detect receptive disturbances in
aphasics. Brain, 85, 665—678.
DIAMOND IT, NEFF WD (1957) Ablation of temporal cortex and discrimination of auditory patterns. Journal

WORD DEAFNESS

1223

KLATT DH (1979) Speech perception: a model of acoustic-phonetic analysis and lexical access. Journal

of Phonetics, 7, 279-312.
KUHL PK (1986) Theoretical contributions of tests on animals to the special-mechanisms debate in speech.

Experimental Biology, 45, 233-265.
LAMBERT J, EUSTACHE F, LECHEVALIER B, ROSSA Y, VIADER F (1989) Auditory aphasia with relative

sparing of speech perception. Cortex, 25, 71—82.
LEVY J, TREVARTHEN C (1977) Perceptual, semantic and phonetic aspects of elementary language processes
in split-brain patients. Brain, 100, 105—118.
LIBERMAN AM, COOPER FS, SHANKWEILER DP, STUDDERT-KENNEDY M (1967) Perception of the speech

of the Royal Society of London, B, 295, 333-352.
MENDEZ MF, GEEHAN GR (1988) Cortical auditory disorders: clinical and psychoacoustic features. Journal
of Neurology, Neurosurgery and Psychiatry, 51, 1-9.
METZ-LUTZ M-N, DAHL E (1984) Analysis of word comprehension in a case of pure word deafness.
Brain and Language, 23, 13—25.
MICELI G (1982) The processing of speech sounds in a patient with cortical auditory disorder.
Neuropsychologia, 20, 5—20.
MICELI G, GAINOTTI G, CALTAGIRONE C, MASULLO C (1980) Some aspects of phonological impairment

in aphasia. Brain and Language, 11, 159—169.
MICHEL F, PERONNET F (1980) A case of cortical deafness: clinical and electrophysiological data. Brain
and Language, 10, 367—377.
MILBERG W, BLUMSTEIN S, DWORETZKY B (1988) Phonological processing and lexical access in aphasia.

Brain and Language, 34, 279-293.
MILLER JL, JUSCZYK PW (1989) Seeking the neurobiological bases of speech perception. Cognition, 33,
111-137.
MOTOMURA N, YAMADORI A, MORI E, TAMARU F (1986) Auditory aphasia: analysis of a case with bilateral
subcortical lesions. Brain, 109, 379-391.
ORGASS B (1976) Eine Revision des Token Tests: I. Vereinfachung der Auswertung, Itemanalyse und
Einfiihrung einer Alterskorrektur. II. Validitatsnachweis: Normierung und Standardisierung.
Diagnostica, 22, 7 0 - 8 7 , 141.
OSCAR-BERMAN M, ZURIF EB, BLUMSTEIN S (1975) Effects of unilateral brain damage in the processing
of speech sounds. Brain and Language, 2, 345—355.
OZDAMAR 6 , KRAUS N, CURRY F (1982) Auditory brain stem and middle latency responses in a patient
with cortical deafness. Electroencephalography and Clinical Neurophysiology, 53, 224—230.
PARVING A, SALOMON G, ELBERLING C, LARSEN B, LASEN NA (1980) Middle component of the auditory

evoked response in bilateral temporal lobe lesions: report on a patient with auditory agnosia.
Scandinavian Audiology, 9, 161 — 167.
PETERSEN MR, BEECHER MD, ZOLOTH SR, MOODY DB, STEBBINS WC (1978) Neural lateralization of

species-specific vocalizations by Japanese macaques (Macaca fuscata). Science, 202, 324—327.
PISONI DB, LUCE PA (1987) Acoustic-phonetic representations in word recognition. Cognition, 25, 21-52.

Downloaded from http://brain.oxfordjournals.org at Max-Planck-Gesellschaft on September 1, 2010

code. Psychological Review, 74, 431—461.
LIEBERMAN P, BLUMSTEIN SE (1988) Speech Physiology, Speech Perception and Acoustic Phonetics.
Cambridge: Cambridge University Press.
LISKER L, ABRAMSON AS (1964) A cross-language study of voicing in initial stops: acoustical measurements.
Word, 20, 384-422.
LISSAUER H (1890) Ein Fall von Seelenblindheit nebst einem Beitrage zur Theorie derselben. Archivflir
Psychiatrie und Nervenkrankheiten, 21, 222—270.
LURIA AR (1970) Pathology of temporal lobe systems and the syndrome of acoustic aphasia. In: Traumatic
Aphasia: Its Syndromes, Psychology and Treatment. Edited by A. R. Luria. The Hague: Mouton,
pp. 104-141.
MAASSEN B, POVEL D-J (1985) The effect of segmental and suprasegmental corrections on the intelligibility
of deaf speech. Journal of the Acoustical Society of America, 78, 877—886.
MATSUI T, HIRANO A (1978) An Atlas of the Human Brain for Computerized Tomography. Tokyo:
Igaku-Shoin.
MEHLER J (1981) The role of syllables in speech processing: infant and adult data. Philosophical Transactions

1224

P. PRAAMSTRA AND OTHERS

Brain and Language, 3, 209-228.
SAMS M, HAMALAINEN M, ANTERVO A, KAUKORANTA E, REINIKAINEN K, HARI R (1985) Cerebral

neuromagnetic responses evoked by short auditory stimuli. Electroencephalography and Clinical
Neurophysiology, 61, 254—266.
SCHERG M, CRAMON D VON (1986) Evoked dipole source potentials of the human auditory cortex.
Electroencephalography and Clinical Neurophysiology, 65, 344—360.
SCHERG M, CRAMON D VON (1990) Dipole source potentials of the auditory cortex in normal subjects
and in patients with temporal lobe lesions. In: Auditory Evoked Magnetic Fields and Electric Potentials.
Advances in Audiology, Volume 6. Edited by F. Grandori, M. Hoke and G. L. Romani. Basel and
London: Karger, pp. 165-193.
SCHUSTER P, TATERKA H (1926) Beitrag zur Anatomie und Klinik der reinen Worttaubheit. Zeitschrift
fur die Gesamte Neurologie und Psychiatrie, 105, 494—538.
SLIS IH, COHEN A (1969) On complex regulating the voiced-voiceless distinction. II. Language and Speech,
12, 137-155.
SPARKS R, GESCHWIND N (1968) Dichotic listening in man after section of neocortical commissures. Cortex,
4, 3 - 1 6 .
SPINNLER H, VIGNOLO LA (1966) Impaired recognition of meaningful sounds in aphasia. Cortex, 2,
337-348.
STOCKERT TR VON (1982) On the structure of word deafness and mechanisms underlying the fluctuation
of disturbances of higher cortical functions. Brain and Language, 16, 133—146.
STRANGE W, VERBRUGGE RR, SHANKWEILER DP, EDMAN TR (1976) Consonant environment specifies

vowel identity. Journal of the Acoustical Society of America, 60, 213—224.
STRANGE W, EDMAN TR, JENKINS JJ (1979) Acoustic and phonological factors in vowel identification.
Journal of Experimental Psychology: Human Perception and Performance, 5, 643—656.
STUDDERT-KENNEDY M, SHANKWEILER D (1970) Hemispheric specialization for speech perception. Journal
of the Acoustical Society of America, 48, 579—594.
STUDDERT-KENNEDY M, SHANKWEILER D, PISONI D (1972) Auditory and phonetic processes in speech
perception: evidence from a dichotic study. Cognitive Psychology, 3, 455—466.
SUMMERFIELD Q (1982) Differences between spectral dependencies in auditory and phonetic temporal
processing: relevance to the perception of voicing in initial stops. Journal of the Acoustical Society

of America, 72, 51-61.
SWISHER L, HIRSH IJ (1972) Brain damage and the ordering of two temporally successive stimuli.

Neuropsychologia, 10, 137-152.
TALLAL P, NEWCOMBE F (1978) Impairment of auditory perception and language comprehension in
dysphasia. Brain and Language, 5, 13—24.
TANAKA Y, YAMADORI A, MORI E (1987) Pure word deafness following bilateral lesions: a psychophysical
analysis. Brain, 110, 3 8 1 - 4 0 3 .
TARTTER VC (1988) Acoustic and phonetic feature effects in dichotic listening. In: Handbook of Dichotic
Listening: Theory, Methods and Research. Edited by K. Hugdahl. Chichester: John Wiley,
pp. 2 8 3 - 3 2 1 .

Downloaded from http://brain.oxfordjournals.org at Max-Planck-Gesellschaft on September 1, 2010

POECK K, HARTJE W (1979) Performance of aphasic patients in visual versus auditory presentation of
the Token Test: demonstration of a supra-modal deficit. In: Auditory Comprehension: Clinical and
Experimental Studies with the Token Test. Edited by F. Boiler and M. Dennis. New York: Academic
Press, pp. 107-113.
REPP BH (1982) Phonetic trading relations and context effects: new experimental evidence for a speech
mode of perception. Psychological Bulletin, 92, 81 — 110.
REPP BH (1984) Categorical perception: issues, methods, findings. In: Speech and Language: Advances
in Basic Research and Practice, Volume 10. Edited by N. J. Lass. New York: Academic Press,
pp. 243-335.
RIEDEJL K, STUDDERT-KENNEDY M (1985) Extending formant transitions may not improve aphasics'
perception of stop consonant place of articulation. Brain and Language, 24, 223—232.
RUBENS AB, JOHNSON MG, SPEAKS C (1978) Location of lesions responsible for the 'paradoxical ipsilateral
ear effect' with dichotic listening tests in patients with aphasia due to stroke. Neurology, New York,
28, 396.
SAFFRAN EM, MARIN OSM, YENI-KOMSHIAN GH (1976) An analysis of speech perception in word deafness.

WORD DEAFNESS

1225

VARNEY NR (1984) Phonemic imperception in aphasia. Brain and Language, 21, 85—94.
VARNEY NR, BENTON AL (1979) Phonemic discrimination and aural comprehension among aphasic patients.
Journal of Clinical Neuropsychology, 1, 65 — 73.
VIGNOLO LA (1969) Auditory agnosia: a review and report of recent evidence. In: Contributions to Clinical
Neuropsychology. Edited by A. L. Benton. Chicago: Aldine, pp. 172-208.
WENIGER D, WILLMES K, HUBER W, POECK K (1981) Der Aachener Aphasie Test. Reliabilitat und

WOODS DL, CLAYWORTH CC, KNIGHT RT, SIMPSON GV, NAESER MA (1987) Generator of middle- and

long-latency auditory evoked potentials: implications from studies of patients with bitemporal lesions.
Electroencephalography and Clinical Neurophysiology, 68, 132 — 148.
YAQUB BA, GASCON GG, AL-NOSHA M, WHITAKER H (1988) Pure word deafness (acqifired verbal auditory
agnosia) in an arabic speaking patient. Brain, 111, 457—466.
YENI-KOMSHIAN GH, LAFONTAINE L (1983) Discrimination and identification of voicing and place contrasts
in aphasic patients. Canadian Journal of Psychology, 37, 107 — 131.
ZAIDEL E (1976) Auditory vocabulary of the right hemisphere following brain bisection or hemidecortication.
Cortex, 12, 191-211.
{Received January 23, 1990. Revised July 3, 1990. Accepted August 14, 1990)

Downloaded from http://brain.oxfordjournals.org at Max-Planck-Gesellschaft on September 1, 2010

Auswertungsobjektivitat. Nervenarzt, 52, 269—277.
WILLMES K, POECK K, WENIGER D, HUBER W (1980) Der Aachener Aphasie Test. Differentielle Validitat.
Nervenarzt, 51, 553-560.
WILLMES K, POECK K, WENIGER D, HUBER W (1983) Facet theory applied to the construction and validation
of the Aachen Aphasia Test. Brain and Language, 18, 259—276.
WOODS DL, KNIGHT RT, NEVILLE HJ (1984) Bitemporal lesions dissociate auditory evoked potentials
and perception. Electroencephalography and Clinical Neurophysiology, 57, 208-220.

Downloaded from http://brain.oxfordjournals.org at Max-Planck-Gesellschaft on September 1, 2010