ORIGINAL ARTICLE

Alexia With and Without Agraphia: An
Assessment of Two Classical Syndromes
Claire A. Sheldon, George L. Malcolm, Jason J.S. Barton

ABSTRACT: Background: Current cognitive models propose that multiple processes are involved in reading and writing. Objective:
Our goal was to use linguistic analyses to clarify the cognitive dysfunction behind two classic alexic syndromes. Methods: We report
four experiments on two patients, one with alexia without agraphia following occipitotemporal lesions, and one with alexia with
agraphia from a left angular gyral lesion. Results: The patient with occipital lesions had trouble discriminating real letters from foils
and his reading varied with word-length but not with linguistic variables such as part of speech, word frequency or imageability. He
read pseudo-words and words with regular spelling better, indicating preserved use of grapheme-to-phoneme pronunciation rules. His
writing showed errors that reflected reliance on ‘phoneme-to-grapheme’ spelling rules. In contrast, the patient with a left angular gyral
lesion showed better recognition of letters, words and their meanings. His reading was better for words with high imageability but
displayed semantic errors and an inability to use ‘grapheme-to-phoneme’ rules, features consistent with deep dyslexia. His agraphia
showed impaired access to both an internal lexicon and ‘phoneme-to-grapheme’ rules. Conclusion: Some cases of pure alexia may be
a perceptual word-form agnosia, with loss of internal representations of letters and words, while the angular gyral syndrome of alexia
with agraphia is a linguistic deep dyslexia. The presence or absence of agraphia does not always distinguish between the two; rather,
writing can mirror the reading deficits, being more obvious and profound in the case of an angular gyral syndrome.
RÉSUMÉ: Alexie avec et sans agraphie : revue de deux syndromes classiques. Contexte : Selon les modèles cognitifs actuels, plusieurs processus
sont impliqués dans la lecture et l’écriture. Objectif : Notre but était d’utiliser l’analyse linguistique pour clarifier la dysfonction cognitive sous-jacente
à deux syndromes alexiques classiques. Méthodes : Nous rapportons quatre expériences effectuées chez deux patients, dont l’un présentait une alexie
sans agraphie suite à des lésions occipitotemporales et l’autre une alexie avec agraphie suite à une lésion gyrale angulaire gauche. Résultats : Le patient
porteur de lésions occipitales avait de la difficulté à distinguer de vraies lettres de simulacres de lettres et la lecture variait selon la longueur des mots,
indépendamment de variables linguistiques comme la partie du discours, la fréquence des mots ou l’imageabilité. Il lisait des pseudo mots et il lisait
mieux les mots dont l’épellation était régulière, ce qui indique que l’utilisation des règles de prononciation de graphème à phonème était préservée. À
l’écriture on notait des erreurs qui témoignaient qu’il se fiait aux règles d’épellation de phonème à graphème. Par contre, le patient porteur l’une lésion
gyrale angulaire gauche avait une meilleure reconnaissance des lettres et des mots ainsi que de leur signification. Sa lecture des mots dont l’imageabilité
est élevée était meilleure, mais il faisait des erreurs de sémantique et était incapable d’utiliser les règles de graphème à phonème, témoignant d’une
dylexie profonde. Son agraphie témoignait d’une altération de l’accès à un lexique interne et aux règles de phonème à graphème. Conclusion : Certains
cas d’alexie pure peuvent être une agnosie de perception de la forme des mots avec perte de la représentation interne des lettres et des mots alors que
le syndrome gyral angulaire d’alexie avec agraphie est une dyslexie profonde linguistique. La présence ou l’absence d’agraphie ne permet pas toujours
de faire la distinction entre les deux. L’écriture peut refléter les déficits de lecture qui sont plus évidents et plus sévères chez les patients atteints du
syndrome gyral angulaire.

Can. J. Neurol. Sci. 2008; 35: 616-624

Alexia is traditionally classified according to the site of
anatomic damage and to the presence or absence of deficits in
writing and oral language.1 Alexia without agraphia (‘pure
alexia’) is associated with left occipital damage1-3 while alexia
with agraphia is associated with lesions of the left angular gyrus,
and often accompanied by other left parietal signs, such as
apraxia, anomia, and Gerstmann’s syndrome.4
More recent studies have used linguistic and cognitive
approaches to categorize alexia within a hierarchy of processing
steps.1,5,6 Peripheral alexias refer to ‘sub-lexical’ impairments in
perceptual processing of letters and word-forms: reduced
616
https://doi.org/10.1017/S0317167100009410 Published online by Cambridge University Press

perceptual efficiency results in a letter-by-letter strategy, with
more difficulty for longer words, which have more perceptual
elements to be processed. Central alexias are impairments in

From the Departments of Medicine (Neurology), Ophthalmology and Visual Science,
and Psychology, University of British Columbia, Vancouver, BC, Canada.
RECEIVED JUNE 7, 2007. FINAL REVISIONS SUBMITTED MAY 5, 2008.
Correspondence to: Jason J S Barton, Neuro-ophthalmology Section D, VGH Eye
Care Center, 2550 Willow Street, Vancouver, BC, Canada, V5Z 3N9.

LE JOURNAL CANADIEN DES SCIENCES NEUROLOGIQUES

translating perceptual data into meaning and speech, with
phonological, surface, and deep subtypes. These subtypes reflect
hypotheses that reading can occur via two routes: the ‘indirect’
route, in which pronunciation rules translate print to sound
(grapheme-to-phoneme conversion), and the ‘direct’ route, in
which the meaning and pronunciation of words is looked up in
an internal lexicon. Phonological alexia is characterized by
impaired grapheme-to-phoneme conversion while surface
dyslexia is characterized by difficulty accessing the internal
semantic lexicon, a process critical when reading words with
irregular spelling. Deep dyslexia presents as a combination of
both phonological and semantic errors.
In this report we describe two patients, one with pure alexia
from occipitotemporal lesions, and one with alexia with agraphia
from an angular gyrus lesion. Our goal was to apply perceptual
and linguistic tests of reading and writing to clarify our
understanding of the cognitive pathophysiology underlying these
syndromes.
GENERAL METHODS

Patient JM(Occ) is a 69-year-old right-handed, Englishspeaking man with a Masters degree who is a retired
mathematics teacher. Five months prior he had had a left
occipital hemorrhage following anticoagulation for a deep

Figure 1: A. CT scan of JM(Occ)’s lesion showing left occipitotemporal
damage. B. Goldmann perimetry of JM(Occ) showing a right
homonymous paracentral scotoma (shaded region).

Volume 35, No. 5 – November 2008
https://doi.org/10.1017/S0317167100009410 Published online by Cambridge University Press

venous thrombosis, followed by a right occipital infarct during
the same hospital stay. He complained of dyschromatopsia,
impaired reading, and difficulty recognizing people. He hadn’t
noted problems with writing or recognizing other objects. On
exam he had normal auditory comprehension, naming and
spontaneous speech. Visual acuities were 20/20 in both eyes.
Goldmann perimetry showed a small right homonymous
paracentral scotoma. The Farnsworth-Munsell 100-hue test
showed gross sorting errors for all hues. Reading was slow and
laborious, but he wrote words and short sentences well to
dictation and his own composition. Short-term memory for
words on the Warrington Recognition Memory Test was normal
(43/50). Recognition of line drawings was mildly impaired (he
did not recognize a chair or a hammock). A test of famous face
recognition confirmed prosopagnosia, and the Benton Face
Recognition Test showed a severe impairment with facial
perception (29/54). Neuro-imaging revealed a right posterior
cerebral arterial infarction and residual left occipital hemorrhage,
extending into medial temporal and parietal regions (Figures 1
and 3). In summary, he had central dyschromatopisa, alexia
without agraphia, prosopagnosia, and a mild degree of object
agnosia, a constellation that has been reported with bilateral
medial occipital lesions.7,8
Patient MP(Par) is a 55-year-old right-handed, Englishspeaking high-school graduate who was a manager in a heavy
equipment company. He presented with several weeks of daily

Figure 2: A. CT scan of MP(Par)’s lesion showing left temporoparietal
damage. B. Goldmann perimetry of MP(Par) showing a complete right
homonymous hemianopsia.

617

THE CANADIAN JOURNAL OF NEUROLOGICAL SCIENCES

Figure 3: A. Comparison of JM(Occ)’s lesion with a corresponding template showing Brodmann’s anatomically defined areas of the cerebral cortex.
Areas #18 and #19 indicate the location of the visual association areas. B. Comparison of MP(Par)’s lesion with a corresponding template showing
Brodmann’s anatomically defined areas of the cerebral cortex. Area #39 and #40 indicates the locations of the angular and supramarginal gyri,
respectively.

headaches. Neuro-imaging revealed a contrast-enhancing mass
with vasogenic edema in the left angular gyrus (Figures 2 and 3).
He had subtotal resection of the tumour, with pathology showing
glioblastoma, and subsequently completed a course of radiation
therapy and chemotherapy, following which he complained of
trouble reading and writing. On examination visual acuities were
20/30 in the right eye and 20/25 in the left. Goldmann perimetry
showed a dense macular-splitting right hemianopia (Figure 2).
He had normal auditory comprehension and spontaneous speech,
but occasional difficulty naming low-frequency items. Reading
was severely impaired and writing showed numerous spelling
and grammatical errors. He had trouble with simple arithmetical
calculations, but did not show right/left disorientation or finger
agnosia. Face and colour recognition were intact. Additional
visual tests showed normal perception of curvature, angle, and
line orientation using the Benton test, and normal recognition of
famous faces. In summary, he presented with alexia with
agraphia, acalculia, and a mild anomia, associated with a left
lateral occipito-parietal lesion.
Apparatus and analysis

Experiments were run on a Macintosh G4 computer using
Superlab 1.71 (Cedrus Inc, www.superlab.com) software in
sessions over a period of five months. Subjects were seated 57
cm in front of the computer monitor. Stimuli sizes, ranging from
2 to 5° in height for letters and pictures, were easily visible for
both patients.
The subjects performed a series of experiments, whose
methods and results are detailed below. In each experiment
statistical comparisons were made using χ2 or ANOVA analysis,
where appropriate, with level of significance set at α = 0.05.

1. Letter Recognition
The goal of the three tests in this first experiment was to
determine whether the perceptual processing of letter-forms was
normal. Difficulty processing letters would be more consistent

618
https://doi.org/10.1017/S0317167100009410 Published online by Cambridge University Press

with a perceptual (‘sub-lexical’) dysfunction than a linguistic
failure.

Methods: First, a letter/non-letter discrimination test presented
subjects with 18 trials that showed two symbols sequentially for
2000ms each. One symbol was a capital letter and the other a
mirror image of the same letter, and subjects were asked to
indicate which was the real letter. Second, a vowel/consonant
test presented five trials (one for each vowel), in which capital
letters were shown, one a vowel and one a consonant, and
subjects were asked to indicate which one was the vowel. Third,
the embedded symbols test showed subjects ten words and ten
non-words (words that contain one to three non-letters; e.g.
ΔPPLE, Z∞, H3@LTH7), one at a time in random order, and
asked them to state whether the symbol string was a real word.
Subjects indicated their response with a keypress, and both
accuracy and reaction time were recorded.

Results: JM(Occ) had difficulty discriminating non-letters from
real letters (13/18 correct), but had no difficulty identifying the
vowel (5/5 correct). He incorrectly identified the non-word as a
real word in 2/10 trials: ‘F1SH’ as ‘FISH’ and ‘Z∞’ as ‘ZOO’).
In contrast, MP(Par) correctly identified the properly oriented
letter in 18/18 trials, the vowel in 5/5 trials, and all 10 non-words
as non-words. The contrasts between JM(Occ) and MP(Par)
showed a difference for letter identification (χ2 (1, N = 36) = 3.9,
p < 0.05). While the difference between the two subjects in
accuracy for discriminating words from non-words did not reach
statistical significance, (χ2 = (1, N = 20) = 2.2, p = 0.13),
JM(Occ) took far longer than MP(Par) to complete this task: 27.1
s (s.d. 14.7) for JM(Occ) and 3.2 s (s.d. 1.4) for MP(Par)
(p<0.0001). Thus JM(Occ) but not MP(Par) was impaired in
identifying stimuli as real letters or words.
2. Recognition and reading of real words and pseudo-words
The two tests of word reading in this section had two goals.

LE JOURNAL CANADIEN DES SCIENCES NEUROLOGIQUES

First, we determined whether reading varied with either
perceptual or linguistic parameters. A hallmark of peripheral
alexia from perceptual dysfunction is a word-length effect, in
which the accuracy and/or reading time is proportional to the
number of letters in the word.3,6 On the other hand, central
alexias show modulation by linguistic variables such as word
frequency, word imageability (i.e. the ability of the word to
conjure a visual image; ‘concrete’ vs. ‘abstract’), parts of speech
(i.e. nouns, modifiers and functors) and orthographic regularity
of words (whether the word can be ‘sounded out’).9 Second, we
contrasted the reading of real words with that of pseudo-words:
while peripheral alexias fail to show differences with respect to
their abilities to read real vs. pseudo-words, phonological and
deep, but not surface, dyslexics have an impaired ability to read
pseudo-words (i.e. an impaired access to phonological
knowledge).

Methods: In the first test subjects were shown 115 real words
chosen from a published list of words of known word frequency

and imageability (Table 1).5,10,11 Words varied in length from 2 to
11 letters and fell into one of ten categories: concrete high
frequency, concrete low frequency, abstract high frequency,
abstract low frequency, nouns, modifiers, functors, verbs,
irregular and regular words.11 In the second test pseudo-words
and their corresponding real words (ten trials of each type; Table
1) were presented. In both tests words were shown one at a time
with unlimited viewing time, in random order, and subjects were
asked first to indicate with a keypress if the word was a real
word, and second, to read it aloud. Immediately after the verbal
response the experimenter documented the accuracy of the
response by a second keypress. Results were analyzed for
accuracy and reading time, as indicated by the total duration to
the second key press.
Results: JM(Occ) recognized 40% (46/115) and pronounced
correctly 51% (59/115) of the words (thus to direct questioning
he could pronounce some words he did not recognize).
Increasing word length reduced JM(Occ)’s reading accuracy (χ2

Table 1: Word list employed in reading and recognition of real words and pseudo-words*
Concrete - High Frequency
action
brother
chair
farm
notice
glass
fruit
village
whistle
wound
Nouns
cabbage
kite
hammer
antelope
boy
plant
lantern
toilet
spider
lighthouse
eagle
thigh

Concrete - Low Frequency
ambulance
bomb
cabbage
clay
dad
estate
fringe
iodine
movies
plum
squeak
Modifiers
beautiful
early
quickly
friendly
green
soft
striped
round
carefully
vigorously
big
sad

Regular words
administer
hand
abdominal
spill
rug
splash
enter
dose
limp
orchestra
deputy

Irregular words
colonel
yacht
island
have
borough
stomach
ratio
glacier
thigh
leisure
lose

Abstract - High Frequency
go
caught
colour
dear
famous
marry
moment
peace
rich
success
tired
Functor
because
their
that
which
they
under
the
if
and
but
despite
without
instead
around
Pseudo-words
mub
jat
sust
gort
sild
stime
blape
nolden
tolt
zopped

Abstract - Low Frequency
applaud
fiction
eldest
jolly
fuss
indignation
maintenance
melody
swift
wicked
thirsty
Verb
ambulate
thrive
eat
sleep
make
laugh
sing
command
absorb
damage
debate
bleed

Real words
tub
hat
rust
fort
mild
slime
blaze
golden
bolt
topped

* words selected from (5, 10, 11)

Volume 35, No. 5 – November 2008
https://doi.org/10.1017/S0317167100009410 Published online by Cambridge University Press

619

THE CANADIAN JOURNAL OF NEUROLOGICAL SCIENCES

Table 2: Comparison of JM(Occ)’s and MP(Par)’s ability to recognize correctly
pseudo-words vs. real-words and to read words aloud
JM(Occ)
Pseudo-words
Words

Recognized
10/10 *
8/10

MP(Par)
Read
5/10 *
8/10 *

Recognized
5/10
10/10

Read
0/10
2/10

* indicates a significant difference when comparing the fraction correct observed between the
two subjects in the corresponding task (α < 0.05, 1)

(9, N = 115) = 19.9, p = 0.02) and lengthened reading time by
about 5 sec per letter (Figure 4A and B). For linguistic variables,
there was a trend to significance for regularity, with JM(Occ)
reading correctly 63% (7/11) of regular but 27% (3/11) of
irregular words (χ2 (1, N = 22) = 2.9, p = 0.08). JM(Occ)’s
reading was not significantly affected by imageability or word
frequency (Figure 5A). On the second test, JM(Occ) was 90%
accurate (18/20) in discriminating pseudo-words from real words
and correctly read 50% (5/10) of pseudo-words (Table 2).
MP(Par) recognized 77% (88/115) of words but could only
correctly pronounce 18% (21/115). There was no effect of word
length on MP(Par)’s accuracy or reading time (Figure 4). For
linguistic variables, MP(Par)’s reading did not differ overall with
the category of word (Figure 5A). However, while he read
regular and irregular words equally well (18% for both), he was
better at reading concrete (7/21 or 33% correct) than abstract
words (1/22, or 4% correct) (χ2 (1, N = 43) = 5.9, p = 0.015).
With the second test, MP(Par) was 75% accurate (15/20) in
discriminating pseudo-words from real words but could not read
any pseudo-words (0/10): for example, while he could read
“fort” and “hat”, he could not read “gort” or “jat”. We classified
MP(Par)’s 94 errors by an established system (5): 2% were
substitution errors, 10% were semantic errors, 11% were visual

errors, 21% were undeterminable and, in the remaining 51%,
MP(Par) did not offer a response. Substitution errors included
reading ‘sing’ as ‘song’; visual errors included reading ‘quickly’
as ‘question’ and ‘movies’ as ‘moving’; semantic errors included
‘beautiful’ as ‘pretty’, ‘but’ as ‘and’ and ‘cabbage’ as ‘vegetable’.
MP(Par) also made a combined visual/semantic error in reading
‘marry’ as ‘happy’.
The contrasts between JM(Occ) and MP(Par) showed that
JM(Occ) was worse than MP(Par) at recognizing words (χ2 (1,
N = 115) = 31.5, p <0.0001) but better than MP(Par) at reading
real and pseudo-words aloud (χ2 (1, N = 115) = 27.7, p <0.0001
and χ2 (1, N = 20) = 6.7, p = 0.01, respectively).
To summarize, JM(Occ) showed a word-length effect, a trend
to an effect of orthographic regularity but not of other linguistic
variables, and could read some pseudo-words, whereas MP(Par)
showed no word-length effect, but rather an influence of
imageability on reading, a more marked inability to pronounce
pseudo-words, and produced semantic paralexias.

3. Reading comprehension in MP(Par)
Despite MP(Par)’s severe difficulty in reading words aloud,
MP(Par) claimed to recognize many words. However, his high
77% rate of responding ‘yes’ to a question of ‘do you recognize

Figure 4: The influence of word length on word reading accuracy and reading time. A. Fraction of words read incorrectly as a function of word length
(number of characters in word) for JM(Occ) (G) and MP(Par) (G
G ). B. Average reading time as a function of word length for JM(Occ) (G) and MP(Par)
(G
G ). Data points indicate the average reading time for a given word length (n = 2 - 26); error bars are SD. Solid lines represent linear regression in
A (r2 = 0.85 and 0.001 for JM(Occ) and MP(Par), respectively) and weighted non-linear regression analysis in B to the data points obtained from
individual subjects (for JM(Occ), slope = 4.73 sec/letter, r2 = 0.82; for MP(Par), slope = 1.31 sec/letter, r2 = 0.53).

620
https://doi.org/10.1017/S0317167100009410 Published online by Cambridge University Press

LE JOURNAL CANADIEN DES SCIENCES NEUROLOGIQUES

Figure 5: Parameters influencing reading and writing. Fraction of words read aloud correctly (A) and written correctly to dictation (B) as a function
of word class for JM(Occ) (I) and MP(Par) (I
I); n = 11 - 14 in each group. C, Fraction of words written correctly as a function of word length (number
of characters in word) for JM(Occ) (I) and MP(Par) (I
I) CHF = concrete high frequency, CLF = concrete low frequency, AHF = abstract high
frequency, ALF = abstract low frequency, N = noun, Mod = modifier, V = verb, F = functor, IR = irregular, R = regular.

this word?’ could simply reflect a criterion bias rather than true
recognition. To verify recognition, we examined whether he
knew the meanings of written words.
Methods: We gave MP(Par) a picture/word matching test. This
consisted of 20 pictures presented four times each (80 trials
total). On each trial a word was written below the picture. On one
trial the word was the correct (index) name, on one it was a

Table 3: Word list employed in reading comprehension in
MP(Par)
Index Word

PhonologicallyRelated Word

book
star
mother
shoe
carrot
bouquet
fan
king
pianist
paper
oyster
dog
church
stomach
sword
acrobat
money
needle
square
Tree

cook
scar
doctor
hat
cannon
bodice
ram
wink
pinch
party
ocean
door
child
store
swing
action
muddy
kneel
squeak
beef

SemanticallyRelated
Word
card
moon
uncle
foot
salad
blossom
water
movies
orchestra
ink
water
armadillo
school
tickle
soldier
dance
income
prick
rug
branch

Non-Related
Word
bride
kitchen
cushion
dump
ring
garage
kiss
hug
drink
stem
fire
limp
scrub
war
iodine
jelly
butter
cut
lake
train

Volume 35, No. 5 – November 2008
https://doi.org/10.1017/S0317167100009410 Published online by Cambridge University Press

phonologically similar word, on one a semantically related word
and on the last an unrelated word (see Table 3). MP(Par) was
asked first to state if the word matched the picture and second to
read the word aloud. Words were obtained from published lists:
all were concrete and controlled for frequency effects. There
was no time limit.

Results: MP(Par) was 15/20 correct on trials when pictures were
shown with their real name: this 75% accuracy rate is nearly
identical to his 77% rate of acknowledged word recognition in
the previous experiment. In all five instances when MP(Par)
claimed incorrectly that the picture did not match its index word
(‘stomach’, ‘needle’, ‘fan’, ‘square’ and ‘paper’), he had
difficulty reading the word aloud. On 2 of the 15 trials on which
he did correctly match the picture and its index word (‘piano’ and
‘acrobat’), he read the index word incorrectly (‘pianist’ and
‘trapeze’). Thus, although correct matching was needed for
MP(Par) to read index words, on occasion this lead to semantic
paralexias.
MP(Par) was 100% accurate in recognizing that the picture
did not match unrelated words or phonologically similar words.
However, he incorrectly stated on 5 of 20 trials that a
semantically related word matched the picture. On these five
trials he did not read the word correctly, making semantic
paralexical errors. For example, he incorrectly matched the word
‘orchestra’ with a picture of a ‘pianist’ and read the word as
‘symphony’, and incorrectly matched the word ‘soldier’ with a
picture of a ‘sword’ and read the word as ‘knife’.
This experiment suggests that MP(Par) has relatively
preserved word recognition and comprehension, although
comprehension does not guarantee successful reading.12 This test
also suggests that his semantic paralexical reading errors
originate while accessing word meaning prior to activating
phonological output, consistent with hypotheses that semantic
reading errors are generated because of compromised access to
phonological output lexicon.13,14
4. Writing words
Few studies have assessed the reading and writing abilities of
individual patients systematically in parallel. Some suggest that
621

THE CANADIAN JOURNAL OF NEUROLOGICAL SCIENCES

the cognitive processes for writing parallel those of reading.15,16
Patients with deep dyslexia may show ‘deep dysgraphia’17 while
those with letter-by-letter reading may exhibit ‘surface
dysgraphia’.18 However, others assert that alexic features may be
dissociated from those of agraphia.19

Methods: Using the same real words as in the second
experiment, the two subjects were asked to write words in
response to dictation. There was no time limit and results were
analyzed for accuracy. MP(Par) completed only one portion of
this task.

Results: Despite denying writing problems, JM(Occ) wrote only
56.5% (65/115) of the words correctly. He was more successful
when writing shorter than longer words (χ2 (9, N = 115) = 22.4,
p = 0.01). Linguistic variables showed no effects. Writing of
regular words did not differ significantly from irregular words
(χ2 (1, N = 22) = 1.71, p = 0.19) (Figure 5B, C), and there was
no effect of word imageability or word frequency, JM(Occ)
rarely made errors of substitution, transposition or addition; the
vast majority of his errors were phonological: for example,
‘bleed’ as ‘bleade’, ‘success’ as ‘sucess’, ‘striped’ as ‘strypt’ and
‘colonel’ as ‘kernell’.
MP(Par) demonstrated a marked agraphia (20% correct rate;
6/30), independent of word class or word length (Figure 5B, C).
Common errors made by MP(Par) were addition, substitution
and transpositions: ‘eagle’ as ‘eagaef’ (addition), ‘ambulance’ as
‘amburance’ (substitution) and ‘farm’ as ‘fram’ (transposition).
He demonstrated an impaired understanding of the rules of
phonology. For example, ‘spider’ was spelled as ‘sitre’, ‘kite’ as
‘kibe’, ‘hammer’ as ‘harer’. No semantic spelling errors were
noted.
DISCUSSION

We examined the performance of two subjects with different
lesions and different traditional alexic syndromes on three
different tasks. The results illustrate distinct contrasts in the
perceptual and linguistic properties of not only their reading but
also their writing abilities.

JM(Occ): Medial occipitotemporal damage and peripheral
‘word-form agnosia’
JM(Occ)’s alexia appears to reflect a failure to generate
accurate word-forms. He had trouble perceiving letters and with
word-form decision tasks, and displayed a classic word-length
effect. His reading was not affected by linguistic variables such
as part of speech, word frequency, or imageability. These
findings are characteristic of pure alexia. Patients with this
disorder cannot access word knowledge and rely at best on a
slow letter-by-letter reading strategy.3,5,6 JM(Occ)’s neuroimaging showed a left occipital hemorrhage and a right occipitotemporal infarct: the former is consistent with the association of
pure alexia with lesions of the left medial occipital cortex.2,3,6,20
The traditional explanation of occipital alexia is a
disconnection syndrome.21,22 This postulates a failure of
information transmission to language centers in the left angular
gyrus from both the left hemifield, because of damage to callosal
fibers in the splenium, and the right hemifield, usually because
of right hemianopia, but sometimes from damage to fibers
622
https://doi.org/10.1017/S0317167100009410 Published online by Cambridge University Press

surrounding the occipital horn of the lateral ventricle.23 The
disconnection hypothesis is most strongly supported by atypical
cases, such as a combination of damage to the splenium and a
right hemianopia from a lesion of the left lateral geniculate
nucleus.24,25
However, others argue that some cases of pure alexia may
represent an agnosia for visual word forms.26,27 This ‘word-form
agnosia’ may actually be a more generalized perceptual defect
for certain complex visual stimuli, of which words and letters are
the most prominent examples.3,8,28,29 Word-form agnosia has
been postulated to result from damage to the left fusiform gyrus,
which fMRI experiments suggest contains a visual word-form
area.30,31
Two features not usually considered part of pure alexia
further support a hypothesis of impaired word-form processing
in JM(Occ). First is his trend to greater difficulty for words with
irregular spelling. Typically this is considered a feature of
surface dyslexia, attributed to an inability to access semantic
information or activate the stored sound of the word,5 so that
reading can only be accomplished through the phonological
route. Loss of word-form knowledge thus results in difficulty
with irregular words, for which phonological rules of
pronunciation are unhelpful. Second, JM(Occ) also demonstrated a similar reliance on phoneme-to-grapheme rules in his
writing. This type of agraphia has also been described with some
forms of surface dyslexia.6,15
In summary, JM(Occ) has trouble identifying letters and the
more letters in a word the more difficulty he has identifying the
word. His spelling errors as well as his difficulties pronouncing
irregular words suggest degraded knowledge of word forms. He
employs grapheme-to-phoneme rules to achieve better pronunciation of regular than irregular words and more successful
pronunciation of pseudo-words than MP(Par). JM(Occ) similarly
relies on phoneme-to-grapheme rules when writing. A perceptual
or ‘peripheral’ word-form agnosia with loss of internal
representations of letters and word forms would account for the
majority of his deficits, and is consistent with a lesion of the left
fusiform gyrus.

MP(Par): Left angular gyral lesion and central dyslexia
MP(Par) has preserved perception of letters and better explicit
recognition of words, which he demonstrates with equally good
matching of words to pictures. He shows no effect of word length
on reading. These features suggest that perceptual processing of
letters and words is relatively intact. Rather, many aspects of his
reading are consistent with deep dyslexia.5,12 These include
better reading of words with high imageability, semantic
paralexias, and a severe inability to use phonological rules to
read pseudo-words. Although MP(Par)’s reading was best for
concrete high-frequency nouns and worst for modifiers, he did
not show a statistically significant part-of-speech effect, which is
usually another feature of deep dyslexia. This may reflect the
severity of his defect with all classes of words or, as others have
suggested, an extension of the imageability effect on his reading
(i.e. nouns tend to have higher imageability than other words).5
The origins of deep dyslexia remain controversial. It may
represent either damage to multiple left hemispheric reading
modules or the loss of left hemispheric reading skill, with
residual reading reflecting the abilities of the right hemisphere.

LE JOURNAL CANADIEN DES SCIENCES NEUROLOGIQUES

MP(Par)’s better reading of nouns and words with high
imageability and his better access to semantic information (in the
form of accurate picture-word matching) are consistent with a
hypothesized right hemispheric ability to store semantic but not
phonological information.32,33 Neuroanatomic data for deep
dyslexia have been limited because this rare condition is most
often associated with large perisylvian lesions and almost always
accompanied by a profound aphasia.9 There are uncommon
reports of more focal temporoparietal lesions and clinical
syndromes. One case with longstanding resection of a left
temporoparietal tumour had ‘profound language disturbances’
and a hemiplegia that suggests more frontal involvement than
described.34,35 A second patient with a post-traumatic subdural
hematoma and left parietal contusion had deep dyslexia and
agraphia (whose characteristics were not examined) with a
severe anomia as the only other deficit.36 Another patient with
trauma to the left inferior-middle temporal gyrus presented with
deep dyslexia, agraphia (again, not characterized) and normal
oral comprehension.37
MP(Par) had only a mild anomia and his lesion was centered
on the angular gyrus, though, as with all tumours, the limits of
his lesion are not sharply demarcated. A recent meta-analysis of
fMRI studies concluded that the angular gyrus is activated
during semantic access and the neighboring posterior superior
temporal gyrus during grapheme-to-phoneme conversion.38
From MP(Par)’s data, we suggest that the classic syndrome of
alexia with agraphia is characterized by deep dyslexia and a
corresponding deep dysgraphia, and that the angular gyrus may
be a key substrate for this linguistic deficit.

JM(Occ) and MP(Par): agraphic disorders
Although JM(Occ)’s clinical and radiologic features are
consistent with pure alexia, more detailed assessment of his
writing showed significant spelling errors consistent with
excessive reliance on phoneme-to-grapheme rules. As with his
alexia, JM(Occ)’s writing deficits likely reflect loss of
information about visual word-forms. This supports claims that
the visual word form area also serves as a graphemic-output
lexicon (18). Whether these subtler deficits in writing are typical
of all cases of so-called pure alexia is unclear. It may be that the
presence or absence of surface dysgraphia may distinguish
agnostic forms of pure alexia from the disconnection variety,
since writing would not be affected by the latter.
Likewise, MP(Par)’s writing mirrors his reading defect. Just
as his pseudo-word reading shows a severe inability to use
grapheme-to-phoneme conversion rules, his writing errors show
no phonological relation to the correct word. While he did not
show any semantic substitution errors in writing, this does not
imply that he has only phonological agraphia: in that disorder,
preserved semantic access to the lexicon allows correct spelling
of real words,18 which he did not show. Thus, his writing
demonstrates impaired phoneme-to-grapheme and semantic
routes, consistent with a deep dysgraphia mirroring his deep
dyslexia.
In summary, our examination of these two patients with
different classic syndromes of alexia show that at least some
variants of pure (occipital) alexia are likely due to a word-form
agnosia that creates parallel deficits in reading and writing. In
contrast, alexia with agraphia (angular alexia) is associated with

Volume 35, No. 5 – November 2008
https://doi.org/10.1017/S0317167100009410 Published online by Cambridge University Press

better perceptual processing of words but more significant
linguistic dysfunction, generating parallel deep dyslexia and
deep dysgraphia. These data suggest that medial occipital and
angular gyral lesions are distinguished not so much by whether
writing is intact but by whether both reading and writing show
perceptual or linguistic-based deficits.
ACKNOWLEDGEMENTS

JB was supported by the Canada Research Chair program and
a Michael Smith Foundation for Health Research Senior
Scholarship. We thank P. Teal and G. Redekop for patient
referrals. We dedicate this paper to the memory of MP(Par),
whose patience and cooperation were invaluable.
REFERENCES

1.

Black SE, Behrmann M. Localization in Alexia. In: Kertesz A,
editor. Localization and neuroimaging in neuropsychology. San
Diego: Academic Press; 1994. p. 331-76.
2. Dalmas JF, Dansilio S. Visuographemic alexia: a new form of a
peripheral acquired dyslexia. Brain Lang. 2000;75(1):1-16.
3. Behrmann M, Nelson J, Sekuler EB. Visual complexity in letter-byletter reading: "pure" alexia is not pure. Neuropsychologia.
1998;36(11):1115-32.
4. Goethals M, Santens P. Posterior cortical atrophy. Two case reports
and a review of the literature. Clin Neurol Neurosurg. 2001;
103(2):115-19.
5. Coslett HB. Acquired dyslexia. Semin Neurol. 2000;20(4):419-26.
6. Bub D. Alexia and related reading disorders. Neurol Clin. 2003;
21(2):549-68.
7. De Renzi E, di Pellegrino G. Prosopagnosia and alexia without
object agnosia. Cortex. 1998;34(3):403-15.
8. Buxbaum LJ, Glosser G, Coslett HB. Impaired face and word
recognition without object agnosia. Neuropsychologia. 1999;
37(1):41-50.
9. Marshall JC, Newcombe F. Patterns of paralexia: a psycholinguistic
approach. J Psycholinguist Res. 1973;2(3):175-99.
10. Marchand Y, Friedman RB. Impaired oral reading in two atypical
dyslexics: a comparison with a computational lexical-analogy
model. Brain Lang. 2005(3);93:255-66.
11. Coltheart M. The MRC psycholinquistic database. Q J Exp Psychol.
1981;33A:497-505.
12. Newton PK, Barry C. Concreteness effects in word production by
not word comprehension in deep dyslexia. Cog Neuropsychol.
1997;14:481-509.
13. McLeod P, Shallice T, Plaut DC. Attractor dynamics in word
recognition: converging evidence from errors by normal
subjects, dyslexic patients and a connectionist model. Cognition.
2000;74(1):91-114.
14. Colangelo A, Buchanan L. Semantic ambiguity and the failure of
inhibition hypothesis as an explanation for reading errors in deep
dyslexia. Brain Cogn. 2005;57(1):39-42.
15. Beeson PM. Remediation of written language. Top Stroke Rehabil.
2004;11(1):37-48.
16. Ogden JA. Phonological dyslexia and phonological dysgraphia
following left and right hemispherectomy. Neuropsychologia.
1996;34(9):905-18.
17. Cipolotti L, Warrington EK. Does recognizing orally spelled words
depend on reading? An investigation into a case of better written
than oral spelling. Neuropsychologia. 1996;34(5):427-40.
18. Rapcsak SZ, Beeson PM. The role of left posterior inferior temporal
cortex in spelling. Neurology. 2004;62(12):2221-9.
19. Kim HJ, Chu K, Lee KM, Kim DW, Park SH. Phonological
agraphia after superior temporal gyrus infarction. Arch Neurol.
2002;59(8):1314-16.
20. Mao-Draayer Y, Panitch H. Alexia without agraphia in multiple
sclerosis: case report with magnetic resonance imaging
localization. Mult Scler. 2004;10(6):705-7.

623

THE CANADIAN JOURNAL OF NEUROLOGICAL SCIENCES
21. Bub DN, Arguin M, Lecours AR. Jules Dejerine and his
interpretation of pure alexia. Brain Lang. 1993;45(4):531-59.
22. Geschwind N. Disconnexion syndromes in animals and man. I.
Brain. 1965;88(2):237-94.
23. Damasio AR, Damasio H. Hemianopia, hemiachromatopsia and the
mechanisms of alexia. Cortex. 1986;22(1):161-9.
24. Stommel EW, Friedman RJ, Reeves AG. Alexia without agraphia
associated with spleniogeniculate infarction. Neurology.
1991;41(4):587-8.
25. Tamhankar MA, Coslett HB, Fisher MJ, Sutton LN, Liu GT. Alexia
without agraphia following biopsy of a left thalamic tumor.
Pediatr Neurol. 2004;30(2):140-2.
26. Warrington EK, Shallice T. Word-form dyslexia. Brain. 1980;
103(1):99-112.
27. Benito-Leon J, Sanchez-Suarez C, Diaz-Guzman J, Martinez-Salio
A. Pure alexia could not be a disconnection syndrome.
Neurology. 1997;49:305-6.
28. Rentschler I, Treutwein B, Landis T. Dissociation of local and
global processing in visual agnosia. Vision Res. 1994;34(7):
963-71.
29. Miozzo M, Caramazza A. Varieties of pure alexia: the case of
failure to access graphemic representations. Cogn Neuropsychol.
1998;15(1/2):203-38.
30. Sakurai Y. Varieties of alexia from fusiform, posterior inferior
temporal and posterior occipital gyrus lesions. Behav Neurol.
2004;15:35-50.

624
https://doi.org/10.1017/S0317167100009410 Published online by Cambridge University Press

31. Cohen L, Martinaud O, Lemer C, Lehericy S, Samson Y, Obadia M,
et al. Visual word recognition in the left and right hemispheres:
anatomical and functional correlates of peripheral alexias. Cereb
Cortex. 2003;13(12):1313-33.
32. Coslett HB, Monsul N. Reading with the right hemisphere:
evidence from transcranial magnetic stimulation. Brain Lang.
1994;46(2):198-211.
33. Coltheart M. Deep dyslexia is right-hemisphere reading. Brain
Lang. 2000;71(2):299-309.
34. Colangelo A, Stephenson K, Westbury C, Buchanan L. Word
associations in deep dyslexia. Brain Cogn. 2003;53(2):166-70.
35. McEwen S, Westbury C, Buchanan L, Libben G. Semantic
information is used by a deep dyslexic to parse compounds.
Brain Cogn. 2001;46(1-2):201-5.
36. Nolan KA, Volpe BT, Burton LA. The continuum of deep/surface
dyslexia. J Psycholinguist Res. 1997;26(4):413-24.
37. Glosser G, Friedman RB. The continuum of deep/phonological
alexia. Cortex. 1990;26(3):343-59.
38. Jobard G, Crivello F, Tzourio-Mazoyer N. Evaluation of the dual
route theory of reading: a metanalysis of 35 neuroimaging
studies. Neuroimage. 2003;20(2):693-712.