Brain and Language 97 (2006) 189–199
www.elsevier.com/locate/b&l

New evidence for morphological errors in deep dyslexia 夽
Kathleen Rastle a,b,¤, Lorraine K. Tyler b, William Marslen-Wilson c
a

Department of Psychology, Royal Holloway, University of London, Egham, Surrey TW20 0EX, UK
b
Centre for Speech and Language, University of Cambridge, UK
c
MRC Cognition and Brain Sciences Unit, Cambridge, UK
Accepted 3 October 2005
Available online 8 November 2005

Abstract
Morphological errors in reading aloud (e.g., sexist ! sexy) are a central feature of the symptom-complex known as deep dyslexia, and
have historically been viewed as evidence that representations at some level of the reading system are morphologically structured. However, it has been proposed (Funnell, 1987) that morphological errors in deep dyslexia are not morphological in nature but are actually a
type of visual error that arises when a target word that cannot be read aloud (by virtue of its low imageability and/or frequency) is modiWed to form a visually similar word that can be read aloud (by virtue of its higher imageability and/or frequency). In the work reported
here, the deep dyslexic patient DE read aloud lists of genuinely suYxed words (e.g., killer), pseudosuYxed words (e.g., corner), and words
with non-morphological embeddings (e.g., cornea). Results revealed that the morphological status of a word had a signiWcant inXuence on
the production of stem errors (i.e., errors that include the stem or pseudostem of the target): genuinely suYxed words yielded more stem
errors than pseudosuYxed words or words with non-morphological embeddings. This eVect of morphological status could not be attributed to the relative levels of target and stem imageability and/or frequency. We argue that this pattern of data indicates that apparent
morphological errors in deep dyslexic reading are genuinely morphological, and discuss the implications of these errors for theories of
deep dyslexia.
© 2005 Elsevier Inc. All rights reserved.
Keywords: Morphology; Morphological errors; Deep dyslexia; Reading aloud; Visual word recognition; Acquired dyslexia; Visual errors

1. Introduction
Behavioural evidence from adults without language
impairment suggests strongly that words comprising more
than one morpheme (e.g., preWxed words, suYxed words, and
compound words) are analyzed in a decomposed manner
(i.e., in terms of their morphemic constituents) in the language system. This evidence has been gathered from across
many of the world’s languages (e.g., Arabic, Dutch, English,
French, Finnish, German, Hebrew, Italian, Spanish, Serbian;
夽
This research was supported by an MRC programme grant awarded to
L.K. Tyler. We are grateful to Lianne Older for assistance in the design of
stimuli, and to Max Coltheart and Elaine Funnell for extensive and valuable discussion concerning our results.
*
Corresponding author. Fax: +44 (0)1784 434347.
E-mail address: Kathy.Rastle@rhul.ac.uk (K. Rastle).

0093-934X/$ - see front matter © 2005 Elsevier Inc. All rights reserved.
doi:10.1016/j.bandl.2005.10.003

see e.g., Frost & Grainger, 2000; Frost, Grainger, & Rastle,
2005), and pertains to the perception of both printed and
spoken words (e.g., Frost, Forster, & Deutsch, 1997; Marslen-Wilson, Tyler, Waksler, & Older, 1994; Rastle, Davis,
Marslen-Wilson, & Tyler, 2000; Rastle, Davis, & New, 2004;
Taft & Forster, 1975) as well as to the production of spoken
words (e.g., Levelt, Roelofs, & Meyer, 1999; Melinger, 2003;
Roelofs & Baayen, 2002). We are concerned in this article
with morphologically structured representations in the reading system, and speciWcally with the manner in which such
representations may be revealed in individuals with acquired
disorders of reading.
Our investigation focuses in particular on the symptomcomplex known as deep dyslexia (Marshall & Newcombe,
1966, 1973, see also Coltheart, Patterson, & Marshall, 1980/
1987), within which morphological errors in reading aloud
are a primary feature. Deep dyslexia is an acquired disorder

190

K. Rastle et al. / Brain and Language 97 (2006) 189–199

of reading characterized by a number of co-occurring
impairments, including (a) semantic errors in reading aloud
(e.g., kitchen ! cooking); (b) visual errors in reading aloud
(e.g., brothel ! brother); (c) poor reading aloud of abstract
words relative to concrete words; (d) poor reading aloud of
function words (e.g., so, as, the); (e) the complete inability to
read aloud non-words; and (f) poor reading aloud of morphologically complex words. Morphological impairment in
these individuals has been evidenced both by poorer reading aloud of morphologically complex words than phonologically matched simple words (e.g., handy versus dandy;
e.g., Job & Sartori, 1984) and by the existence of morphological errors in reading aloud formed by the deletion, addition, or substitution of an aYx (e.g., swimmer ! swim;
initiate ! initiative; thickly ! thicken; e.g., Patterson, 1980).
The existence of morphological errors in deep dyslexia has
historically been viewed as evidence that representations at
some level of the reading system are morphologically structured—and has therefore required theoretical accounts of
deep dyslexia to postulate speciWc mechanisms, which when
damaged give rise to morphological errors (see e.g., Morton
& Patterson, 1980).
Research in more recent years has, however, challenged
the inference that morphological errors in deep dyslexia
necessarily reXect a level of the reading system at which
words are represented in terms of their morphemic constituents. SpeciWcally, researchers have questioned whether
morphological errors might be a consequence of damage to
the very same procedures that give rise to the visual and
semantic errors with which they co-occur (i.e., that ‘morphological’ errors are actually visual or semantic errors; see
e.g., Badecker & Caramazza, 1987; Castles, Coltheart, Savage, Bates, & Reid, 1996; Funnell, 1987; Plaut & Shallice,
1993, for discussion). One particularly inXuential claim
(Funnell, 1987, see also Funnell, 2000) is that ostensibly
morphological errors are actually visual errors, which arise
when a target word is too low in imageability and/or frequency to be read aloud. In these circumstances, a visually
similar word that can be read aloud (by virtue of its higher
imageability and/or frequency) is formed by the addition,
subtraction, or substitution of letters (see also Shallice &
Warrington, 1975). The result can be an error that appears
morphological in nature (e.g., soloist ! solo). Identifying
the true nature of these errors is, of course, crucially important for developing a coherent theoretical account of deep
dyslexia. The aim of this article is to evaluate Funnell’s
(1987) claim concerning the mechanism that gives rise to
errors that appear morphological in nature.
2. Funnell (1987): A closer look
Funnell (1987) reasoned that if “stem errors” (i.e.,
errors that preserve the stem of morphologically complex
words; e.g., soloist ! solo, swimmer ! swimming) reXect a
morphologically structured level of representation, then
they should occur only for target words that are genuinely aYxed (i.e., those words that could be lexically rep-

resented in a decomposed manner; e.g., soloist as
[solo] + [ist]). Conversely, if stem errors are a type of
visual error that arises when a target word is too low in
imageability and/or frequency to be read aloud, then they
may occur whenever a target word contains an embedding that is higher in imageability and/or frequency than
the target, irrespective of the morphological status of the
target word (e.g., they may occur in corner or billow,
which are not morphologically complex). Funnell (1987)
pointed out that this possibility could not have been
examined in previous investigations since phonological
controls for morphologically complex words used in
those investigations did not contain embedded words
(e.g., handy versus dandy; Kay, Lesser, & Coltheart, 1996).
Funnell (1987) tested her predictions in two patients previously shown to exhibit morphological errors in reading
aloud (JG and CJ), only one of which Wtted the proWle of
a deep dyslexic (JG). We focus on JG’s performance in
particular.
In her Wrst experiment, Funnell (1987) examined JG’s
reading aloud performance on sets of genuinely suYxed
(e.g., shadowy) and pseudosuYxed (e.g., irony) words
matched closely on word imageability, word frequency,
stem imageability, and stem frequency. JG produced stem
errors for each type of item (e.g., shadowy ! shadow;
irony ! iron). Further, although he produced over twice as
many stem errors for genuinely suYxed words as for
pseudosuYxed words (30 versus 13), this diVerence was
non-signiWcant. Posthoc analyses revealed that the occurrence of stem errors in JG’s reading aloud was particularly
related to the relative levels of target and stem imageability:
stem errors occurred primarily when stems were more
imageable than targets. This posthoc analysis motivated
Funnell’s (1987) third experiment. She selected a single set
of 85 words, 33 of which were genuinely suYxed words
(e.g., shadowy), 33 of which were pseudosuYxed words (e.g.,
corner), and 19 of which had other embeddings (e.g., cowl).
She classiWed these 85 words into three groups based only
on the relative levels of target and stem imageability (i.e.,
she did not consider word type as an additional variable).
Funnell (1987) found that when stems were more imageable
than targets, JG tended to produce stem errors; when targets were more imageable than stems, JG tended to produce correct responses; and when targets and stems were
both low in imageabilty, JG tended to produce omissions
and other visual errors. On the basis of these data, Funnell
(1987, p. 525) argued, “There is no evidence to suggest that
these errors reXect damage to a stage of processing in which
root morphemes and suYxes are represented as independent orthographic entities.”
Funnell (1987, Experiment 3) did not report whether
word type (i.e., genuinely suYxed, pseudosuYxed, or
embedded) had an inXuence on stem errors over and above
the inXuence of relative levels of target and stem imageability. However, her raw data indicate that this is a possibility:
JG produced stem errors for 18 of the 33 genuinely suYxed
words (54.5%), 9 of the 33 pseudosuYxed words (27.3%),

K. Rastle et al. / Brain and Language 97 (2006) 189–199

and 3 of the 19 words with other embeddings (15.8%).
These data should be treated with some caution, of course,
since the three conditions were not group-wise matched on
the variables that Funnell (1987) showed to be inXuential in
the production of JG’s stem errors (i.e., relative levels of
target and stem imageability). However, paired with the
fact that JG produced over twice as many stem errors for
genuinely suYxed targets as for pseudosuYxed targets (30
versus 13) in Funnell’s Wrst experiment, these data may
indicate that the morphological status of the target word
has an independent inXuence on the production of stem
errors in deep dyslexia.
We now turn to an examination of these issues in
another individual with deep dyslexia, DE. Our aim is to
investigate whether the production of stem errors in deep
dyslexia is inXuenced by the morphological status of the
target word—over and above any inXuence of the relative
levels of target and stem imageability and/or frequency. We
use a larger and better-controlled set of stimuli than was
used by Funnell (1987), and take advantage of specialized
statistical techniques designed to reveal independent eVects
of morphological status and other variables (e.g., target and
stem imageability) on the production of stem errors.
3. Case history
At the age of 16, he was involved in a motor-scooter
accident that was likely followed within 24 h by a middle
cerebral artery stroke. These injuries left him with extensive
language disabilities and moderate right hemiplegia (see

191

Tyler, Randall, & Marslen-Wilson, 2002, for further information). A CT scan in 1978, conWrmed by an MRI scan in
1996, showed extensive damage to the left hemisphere,
involving large sections of the middle and posterior parts of
the frontal lobe and most of the temporal lobe (see Fig. 1).
DE has worked as a store-keeper since his accident, and
was 45 years of age at the time of testing.
DE provides one of the earliest-reported cases of deep
dyslexia (Patterson & Marcel, 1977), and he has been studied extensively ever since (see e.g., Coltheart, 1980a/1987a;
Tables 2.1–2.6; Marslen-Wilson & Tyler, 1997; Morton &
Patterson, 1980; Patterson & Marcel, 1977; Patterson, 1978,
1979, 1980; Tyler, 1992; Tyler, Moss, & Jennings, 1995;
Tyler & de Mornay-Davies et al., 2002). These investigations have established beyond doubt that DE Wts the deep
dyslexia proWle: he produces the hallmark semantic, visual,
and morphological errors in reading aloud; he shows a concreteness eVect in reading aloud; he is extremely poor at
reading aloud function words; and he cannot read aloud
non-words at all.
4. Stimuli and procedure
One hundred and seventy Wve words were selected from
the CELEX English database (Baayen, Piepenbrock, & van
Rijn, 1993) for inclusion in three conditions. The ‘genuinely
suYxed’ condition comprised 52 morphologically complex
suYxed words (e.g., grower, meanness), which were generally semantically transparent (i.e., their meanings could be
derived from the meanings of their constituents). These

Fig. 1. (Top) Surface reconstruction of DE’s left hemisphere. (Bottom) Axial slices from a spatially normalized T1 weighted MR image showing the extent
of DE’s damage, which involves most of the left perisylvian language areas including the LSTG and LMTG and extending into the left inferior frontal
cortex. Talairach z-dimensions are given in the top left corner of each panel. Left D Left.

192

K. Rastle et al. / Brain and Language 97 (2006) 189–199

were all derivationally complex forms, and none of them
had inXectional endings. The ‘pseudosuYxed’ condition
comprised 62 morphologically simple words such as beaker
and bully. These words had the surface form of morphological complexity since they could be parsed into stems and
suYxes (e.g., [beak] + [er], [bull] + [y]), but they had no actual
etymological or semantic relation to their embedded stems.
The ‘embedded’ condition comprised 61 morphologically
simple words, all of which contained a potential embedded
stem (e.g., cornea). Since none of these potential stems were
followed by an English suYx, there was no basis for a
decomposition of the embedded words into morphemic
constituents.
Words in the three conditions were group-wise matched
as closely as possible on word frequency, stem frequency,
word imageability, and stem imageability. Frequency values were extracted from the CELEX database of English
written wordforms (Baayen et al., 1993). Imageability ratings on a 7-point scale (with 7 representing “highly imageable” and 1 representing “not imageable”) were collected
from two groups of 19 participants from the University of
Cambridge. One group of participants rated the imageability of half of the words in each condition and the stems of
the remaining words in that condition; the other group of
subjects rated the opposite half of words and stems in each
condition.
Frequency and imageability values of target words and
embedded stems are displayed in Table 1. There were no
signiWcant diVerences across condition in the levels of word
frequency, [F (2, 172) D 2.52, p D .083], stem frequency,
[F (2, 172) D .231, p D .794], whole-word imageability,
[F (2, 172) D 2.62, p D .076], or stem imageability, [F (2, 172)
D 1.67, p D .191]. Although none of these variables diVered
signiWcantly by condition, trends toward signiWcance did
exist. Thus, in addition to the matching carried out here,
we dealt with small diVerences between word types statistically in the analyses of the data. Stimuli are listed in the
Appendix A together with DE’s reading aloud responses to
them.
One hundred and twenty Wve morphologically simple Wller
words, which did not contain embedded words, were generated. These words were generally highly imageable (with
average ratings of over 500 in the MRC Psycholinguistic
Database; Coltheart, 1981), and were included both to disguise the presence of word embeddings in the stimulus list
and to ensure that DE would be able to read a signiWcant
number of words in the relatively lengthy testing session.

DE was presented with target and Wller words in random
order for reading aloud. Words were typed onto note cards,
and the experimenter recorded responses during the session. DE was tested on these words twice at an interval of
approximately six months.
5. Results
DE’s reading aloud responses for both testing sessions
are provided in the Appendix A. He made numerous errors
in reading aloud the target words. As expected given his
deep dyslexia proWle, his errors included semantic errors
(e.g., caret ! jewel, lotion ! cream), visual errors (e.g.,
haggle ! haggis, muster ! muscles), visual then semantic
errors (e.g., pastor ! pasta, spaghetti), morphological errors
(e.g., sexist ! sexy), and morphological then semantic
errors (e.g., exactly ! ex-sing something). Errors also
included a number of novel morphological constructions
(e.g., arsonist ! arsoner; goddess ! godery; illness ! illy).
Responses from both testing sessions were added
together in order to increase the power of the analyses and
to ensure that the eVects we observed were not particular to
a speciWc testing session. For this reason, session was
included as a covariate in all analyses. Responses were
scored as correct if DE initially produced the correct
answer, or if he made an immediate self-correction following an incorrect response. There were Wve occasions in
which DE produced a successful self correction: two occasions for the suYxed items (killer ! kill, killer;
buzzer ! bell, no buzzer); one occasion for the pseudosuYxed items (billion ! million, billion); and two occasions
for the embedded word items (dingo ! australian, dingo;
billow ! bill, billow).
Incorrect responses were then examined for the presence
of stem errors. A stem error was deWned as any incorrect
response which contained the stem. Thus, a stem error
could be the result of deletion of an ending (e.g.,
irony ! iron) or the substitution of an ending—whether
that substitution be an aYx (e.g., swimmer ! swimming) or
another type of ending (e.g., pastel ! pasta). Instances in
which DE produced the entire target word with an inXectional ending (which was always –s, e.g., bunion ! bunions;
sweetie ! sweeties) were not considered to be stem errors.
Table 2 displays percentages of correct responses, stem
errors, and other kinds of error across the three conditions.
We analyzed correct responses and stem errors using
logistic regression analyses. These analyses enabled us to

Table 1
Stimulus characteristics (means and standard deviations) for words in each condition

SuYxed
PseudosuYxed
Embedded

Word frequency

Word imageability

Stem frequency

Stem imageability

Mean: 13.35
SD: 25.24
Mean: 30.90
SD: 71.08
Mean: 14.50
SD:29.48

Mean: 4.00
SD: 1.39
Mean: 4.18
SD: 1.47
Mean: 4.59
SD: 1.41

Mean: 91.94
SD: 160.16
Mean: 123.32
SD: 302.21
Mean: 120.93
SD: 305.87

Mean: 4.49
SD: 1.54
Mean: 4.79
SD: 1.69
Mean: 5.05
SD: 1.59

K. Rastle et al. / Brain and Language 97 (2006) 189–199
Table 2
Percentages of correct responses, stem errors, and other errors for words
in each condition

SuYxed
PseudosuYxed
Embedded

Correct
responses (%)

Stem
errors (%)

Other
errors (%)

35.58
43.55
51.64

51.43
26.61
30.32

12.99
29.84
18.04

assess the inXuence of condition on these binary dependent
measures while accounting for any inXuences of the continuous covariates target imageability, stem imageability, target frequency, and stem frequency. Session was also
included as a binary covariate in these analyses. For each
analysis, we tested the statistical signiWcance of these independent variables with the Wald test (hereafter, W).
Results revealed no inXuence of condition on the production of correct responses [W (2) D 1.38, n.s.]. Of the
covariates, word imageability had the strongest inXuence
on the production of correct responses [W (1) D 37.29,
p < .001]: words of high imageability elicited more correct
responses than words of low imageability. There were no
eVects of stem imageability [W (1) D 1.51, n.s.], word frequency [W (1) D .79, n.s.], stem frequency [W (1) D 1.13, n.s.],
or session [W (1) D .88, n.s.] on the production of correct
responses.
Although there was no inXuence of condition on the
production of correct responses, there was a numerical
diVerence across these conditions. Therefore, stem errors
were analyzed as a subset of the incorrect responses. Crucially, these analyses did reveal a signiWcant eVect of condition on the production of stem errors [W (2) D 16.26,
p < .001]: genuinely suYxed words elicited more stem errors
than both pseudosuYxed words [W (1) D 16.26, p < .001]
and embedded words [W (1) D 6.06, p D .014]. The production of stem errors was also inXuenced by stem imageability
[W (1) D 6.34, p D .012] and marginally by stem frequency
[W (1) D 3.52, p D .061] indicating that when targets were
too low in imageability to be named correctly, they were
likely to yield stem errors if their stems were high in imageability and/or frequency. There were no eVects of session
[W (1) D .23, n.s.], word frequency [W (1) D 1.32, n.s.], or
word imageability [W (1) D 2.15, n.s.] on the production of
stem errors.
The preceding analysis establishes that DE’s production
of stem errors was inXuenced by the morphological status
of words that he read aloud: DE made more stem errors
when words were genuinely suYxed than when they were
pseudosuYxed or contained embedded words. We therefore sought to learn more about the nature of the eVect of
condition on stem error production. Table 3 shows the percentage of cases across the three conditions in which DE
produced ‘deletion’ and ‘substitution’ errors. ‘Deletion’
errors occurred when the stem was produced alone (e.g.,
swimmer ! swim); ‘substitution’ errors occurred when the
stem was produced with an incorrect aYx (e.g.,
madly ! madness). Occasionally, DE produced both a dele-

193

Table 3
Percentages of deletion errors and substitution errors for each condition
as a subset of the incorrect responses

SuYxed
PseudosuYxed
Embedded

Deletion (e.g.,
swimmer ! swim) (%)

Substitution
(e.g., tally >taller) (%)

19.40
18.57
30.51

62.68
20.00
18.64

tion error and a substitution error for a single item (e.g.,
tallish ! tall, taller); in these rare instances, the stem error
was counted as both a deletion and a substitution. It is
important to note that the Wgures in Table 3 are expressed
as a percentage of the incorrect responses made in each
condition. These Wgures do not sum to 100% because they
do not include DE’s other errors (e.g., semantic errors).
Logistic regression analyses on the incorrect responses,
which controlled for the four predictor variables plus session, revealed clearly the source of the eVect of condition on
stem error production. While deletion errors were not inXuenced by condition [W(2) D 2.04, n.s.], the analysis of substitution errors revealed a highly signiWcant eVect of condition
[W(2) D 28.42, p < .001]: genuinely suYxed words yielded
more substitution errors than either pseudosuYxed words
[W(1) D 20.75, p < .001] or embedded words [W(1) D 18.84,
p < .001].1
6. General discussion
Morphological errors in reading aloud (e.g.,
soloist ! solo, swimmer ! swimming) have long been considered a central feature of the symptom-complex known as
deep dyslexia (see e.g., Coltheart et al., 1980/1987). Some
investigators, however (e.g., Funnell, 1987, 2000; Plaut &
Shallice, 1993), have questioned whether these errors genuinely reveal morphologically structured representations in
the reading system, and have argued instead that they may
arise from the same sources as the visual and semantic
errors with which they co-occur. In particular, it has been
claimed that morphological errors are a type of visual error
that arises when a target word is too low in imageability
and/or frequency to be read aloud correctly. In these circumstances, the target word may be altered by adding, subtracting, or substituting letters such that a word higher in
imageability and/or frequency is formed. While the result
may appear to be a morphological error (e.g., soloist
! solo), it is argued that careful scrutiny of “stem errors”
in matched morphologically simple target words (e.g.,
1
It is interesting to note that while this eVect of morphological status
was observed only on DE’s substitution errors (of which there were a substantial number), JG (Funnell, 1987) made mostly errors of deletion and
showed no eVect of morphological status. It is diYcult to compare these
two patients directly because they were tested on diVerent stimuli. However, we Wnd it unlikely that an appropriate investigation of JG’s few substitution errors would have yielded an eVect of morphological status. Further
research is needed to determine whether morphological errors are restricted to a subset of deep dyslexic patients.

194

K. Rastle et al. / Brain and Language 97 (2006) 189–199

irony ! iron) suggests otherwise (Funnell, 1987). Our data
are inconsistent with this claim. DE’s reading aloud performance demonstrates that the morphological status of a
word does inXuence the production of stem errors. Genuinely suYxed words yield more stem errors than either
pseudosuYxed words or words containing other embeddings—an eVect of morphological status that cannot be
attributed to the relative levels of target and stem imageability and/or frequency.
Although our data indicate that morphological errors in
deep dyslexic reading cannot be regarded as visual errors,
some may argue that there are other means of accounting
for these data without reference to morphologically structured representations in the reading system. One possibility
is that morphological errors in deep dyslexia reXect a type
of mixed visual and semantic error (e.g., Plaut & Shallice,
1993), in which a word close in orthography and meaning
to the target is produced when the target cannot be read
aloud. These errors would occur more often for morphologically complex words than for morphologically simple
words because—by deWnition—for all morphologically
complex words there is at least one other word similar in
orthography and meaning (e.g., cleaner ! clean;
darkness ! dark). While we cannot evaluate this possibility
fully, we do not believe that our data are entirely consistent
with it. To be speciWc, 28% of DE’s substitution errors were
novel morphological constructions (e.g., arsonist ! arsoner,
illness ! illy, brainless ! brainly, goddess ! goddery) in
which the appropriate aYx was replaced by an inappropriate one. If DE were simply selecting an orthographically
and semantically similar word in cases in which targets
could not be read aloud, then it is unclear how a nonword
response would be produced.
Rather, these types of errors implicate a level of
representation at which semantically transparent complex words are analyzed in terms of their morphemic constituents. This claim is nicely consistent with a growing
body of evidence for a level of ‘morpho-semantic’ decomposition in unimpaired readers, which is revealed when
tasks that tap central-semantic levels of the reading system (e.g., visual priming with fully visible primes) are
used. Under these conditions, for example, Rastle et al.
(2000) reported robust priming of stem targets (e.g., dark)
by morphologically related words (e.g., darkness) relative
to unrelated controls. Conversely, priming of stem targets
(e.g., corn, broth) by pseudosuYxed words (e.g., corner) or
words with other non-morphological embeddings (e.g.,
brothel) was non-signiWcant. ‘Morpho-semantic’ decomposition has been modeled in both classical localist and
distributed-connectionist terms as a level of representation that resides between orthographic and semantic
representations—a level of representation at which the
local or distributed representations of semantically
transparent complex words overlap the local or distributed representations of their stems (e.g., Giraudo &
Grainger, 2000; Plaut & Gonnerman, 2000; Raveh &
Rueckl, 2000).

Although DE’s performance appears to suggest a level
of ‘morpho-semantic’ representation for which there is
also evidence from unimpaired readers, we see some diYculties in explaining his behaviour in terms of damage to
the normal reading system. Some investigators (e.g., Morton and Patterson, 1980/1987; Plaut & Shallice, 1993)
have argued that deep dyslexia may be explained by multiple lesions to the left-hemisphere reading system, which
leave the patient reading solely through a damaged
semantic system with no other lexical or non-lexical
means of converting orthography to phonology. There
are reservations about this view of deep dyslexia, however, the most serious of which is the observation that the
multiple behavioural features of deep dyslexia (presumed
to arise from multiple functional lesions; e.g., Morton
and Patterson, 1980/1987) do not appear to dissociate.
These reservations have prompted other investigators
(e.g., Coltheart, 1980b/1987b; Coltheart, 2000; SaVran,
Bogyo, Schwartz, & Marin, 1980/1987; Weekes, Coltheart, & Gordon, 1997) to argue that deep dyslexia
reXects access to a right-hemisphere reading system not
normally used by unimpaired readers. On this theory, the
deep dyslexic reader activates orthographic lexical
entries in the right hemisphere, which in turn activate
right-hemisphere semantic representations presumed to
be relatively impoverished for abstract words (Coltheart
et al., 1980/1987; SaVran et al., 1980/1987). Right-hemisphere theories are not committed to a view on the locus
of phonological representations for speech output. The
right-hemisphere hypothesis is argued to explain the core
features of deep dyslexia (see Coltheart et al., 1980/1987;
SaVran et al., 1980/1987), and has garnered support from
a variety of sources including lateralized presentation to
split-brain patients (see Coltheart et al., 1980/1987;
Michel, HenaV, & Intrilligator, 1996; Saffran etal.,1980/
1987), neuroimaging of deep dyslexic readers (Coltheart,
2000; Weekes et al., 1997), and the study of left-hemispherectomy patients (e.g., Patterson, Vargha-Khadem,
& Polkey, 1987).
How might DE’s performance in reading morphologically complex words aloud be explained on the right-hemisphere theory of deep dyslexia? Our data suggest, Wrst of all,
that DE’s deletion errors do not depend on the morphological status of target words. Many of these errors could
therefore be visual errors (Funnell, 1987), and ascribed to
the same mechanism that underlies visual errors in the right
hemisphere (Coltheart et al., 1980/1987; see also Shallice &
Warrington, 1975). Our data also suggest, however, that
DE makes substitution errors that do depend on the morphological status of targets. If deep dyslexia does reXect
right-hemisphere reading, then these data suggest that the
right hemisphere is characterized by a form of lexical representation that captures the morphological properties of
semantically transparent complex words. One speciWc possibility is that the right hemisphere is characterized by a
level of morpho-semantic representation that resides
between right-hemisphere orthographic and semantic

K. Rastle et al. / Brain and Language 97 (2006) 189–199

lexical representations, at which semantically transparent
complex words are analyzed in terms of their morphemic
constituents. Substitution errors might then arise because
the aYx component (e.g., -ist) of a decomposed stimulus
may not activate right-hemisphere semantic representations
(presumed to be impoverished for abstract words;
Coltheart et al., 1980/1987) suYciently to drive speech production. In such cases, an alternative aYx may be activated
at the morpho-semantic level and produced together with
the stem (cf., Shallice & Warrington’s, 1975 account of
visual errors). Further research is clearly needed, however,
to establish fully this account of morphological errors in
deep dyslexia. In particular, it would be desirable to seek
further independent evidence that right-hemisphere lexical
representations—like left-hemisphere lexical representations—are structured morphologically.

195

In summary, we have oVered data that argue against the
claims made by Funnell (1987) concerning the nature of
morphological errors in deep dyslexia. These errors are not
always a type of visual error that occurs when a target word
that cannot be read aloud is modiWed (by the addition, subtraction, or substitution of letters) to yield a word higher in
imageability and/or frequency that can be read aloud.
Rather, these errors can be inXuenced signiWcantly by the
morphological status of target words. Irrespective of the
particular account of deep dyslexia oVered (e.g., Coltheart
et al., 1980/1987; Morton and Patterson, 1980/1987), these
data implicate a level of lexical representation that is organized on the basis of morphological relationships. Further
research is needed to determine whether morphological
errors are a general feature of deep dyslexia, or whether they
are observed in only a subset of patients (see Funnell, 1987).

Appendix A
Stimuli and DE’s reading aloud responses
Target

Condition

Response session 1

Response session 2

Word
frequency

Word
imageability

Stem
frequency

Stem
imageability

option
partly
blandly
worthless
goodness
crabby
wordy
teaser
brainless
sickish
thickly
grower
sexist
meanness
madly
smoothly
swiftly
willowy
lovely
buVer
childish
stockist
arsonist
chilly
killer
tallish
bulky
talker
illness
crusty
birdie
washer
sweetie
scabby
goddess
snowy
junkie
buzzer
woolly
curly

SuYxed
SuYxed
SuYxed
SuYxed
SuYxed
SuYxed
SuYxed
SuYxed
SuYxed
SuYxed
SuYxed
SuYxed
SuYxed
SuYxed
SuYxed
SuYxed
SuYxed
SuYxed
SuYxed
SuYxed
SuYxed
SuYxed
SuYxed
SuYxed
SuYxed
SuYxed
SuYxed
SuYxed
SuYxed
SuYxed
SuYxed
SuYxed
SuYxed
SuYxed
SuYxed
SuYxed
SuYxed
SuYxed
SuYxed
SuYxed

optimum, no idea
parted
L and B, land, dk
worthily
good goodly
crabbing
words
tears
brain, dk, something
sickering
thicken, thickening
grown
sexy
meaning, mean
madness
smootly, same again
swifty
willows
c
bluV
child, dk
stocking
arsoner, dk
c
kill, killer
tall, taller, wrong
c
talking
illy
c
c
washing
sweeties
scabs
godistest
c
c
bell, buzz
c
c

opting
c
blankly
worthery, lot of money
goodly
c
words
tea, dk
brainly
sickly
thickens
grown
sexy
dk
madness
smoother
c
trees, can’t say it
c
c
childly, dk
dk
arson, dk
chills
kill
c
c
talking
illery
c
c
washing
sweeties
scrabble, wrong
godery, dk
c
c
buzzing
c
c

16
72
3
5
15
1
0
0
1
0
5
0
3
2
5
12
16
1
54
2
14
0
0
6
12
0
6
2
36
1
1
2
1
0
8
2
1
2
3
4

1.74
1.79
2
2.11
2.47
2.5
2.58
2.63
2.74
2.74
2.79
2.82
2.84
2.95
2.95
3
3
3.16
3.32
3.37
3.47
3.53
3.56
3.58
3.84
3.89
4.21
4.32
4.37
4.53
4.61
4.63
4.68
4.84
5
5.11
5.11
5.16
5.47
5.68

0
1.55
487
3.25
6
2
99
2.5
911
2.73
5
6.75
212
5
5
6.36
70
6.17
71
4.08
69
3.92
96
3.73
129
5.33
296
2.58
49
3.75
38
4.5
12
3.73
4
6.67
367
3.25
1
3.09
440
6.58
64
4
2
4.17
10
3.5
86
4
68
5.58
22
3.25
248
4.17
57
3.08
7
5.58
44
6.92
43
4.67
48
5.25
1
5.82
22
3.33
59
6.33
8
4.83
3
3.91
23
6.55
6
5.25
(continued on next page)

196

K. Rastle et al. / Brain and Language 97 (2006) 189–199

Appendix A (continued)
Target

Condition

Response session 1

Response session 2

Word
frequency

Word
imageability

Stem
frequency

Stem
imageability

tourist
hilly
artist
auntie
cabbie
actor
tanker
zipper
swimmer
election
exactly
computer
cower
adder
tuber
solder
burnish
rasher
bunion
tarnish
teeter
muster
pasty
testy
vanish
cater
caper
potion
ponder
booty
beaker
stingy
putty
ruby
tally
mutter
lotion
tailor
wander
husky
punish
trillion
tenor
bully
gingerly
rotor
legion
butcher
analogy
portion
irony
mayor
fury
temper
shower
brandy
belly
mission
passion
billion
master
forty
career

SuYxed
SuYxed
SuYxed
SuYxed
SuYxed
SuYxed
SuYxed
SuYxed
SuYxed
SuYxed
SuYxed
SuYxed
PseudosuYxed
PseudosuYxed
PseudosuYxed
PseudosuYxed
PseudosuYxed
PseudosuYxed
PseudosuYxed
PseudosuYxed
PseudosuYxed
PseudosuYxed
PseudosuYxed
PseudosuYxed
PseudosuYxed
PseudosuYxed
PseudosuYxed
PseudosuYxed
PseudosuYxed
PseudosuYxed
PseudosuYxed
PseudosuYxed
PseudosuYxed
PseudosuYxed
PseudosuYxed
PseudosuYxed
PseudosuYxed
PseudosuYxed
PseudosuYxed
PseudosuYxed
PseudosuYxed
PseudosuYxed
PseudosuYxed
PseudosuYxed
PseudosuYxed
PseudosuYxed
PseudosuYxed
PseudosuYxed
PseudosuYxed
PseudosuYxed
PseudosuYxed
PseudosuYxed
PseudosuYxed
PseudosuYxed
PseudosuYxed
PseudosuYxed
PseudosuYxed
PseudosuYxed
PseudosuYxed
PseudosuYxed
PseudosuYxed
PseudosuYxed
PseudosuYxed

tourists
c
c
c
australian, cab, american
c
c
zip
swimming
vote
dk
computed
no idea, cow
c
tubbing
soldier
burnly
bacon
bunions
varnish
dk, teeth, wrong
muscles, no idea
pastry
tester
c
caterer, cat, kitten, holiday
cap, ing, caping
portion
c
c
bleak
sting
c
c
taller
c
cream
c
wagner, wrong, dk
hussy
punishment
c
tender
c
ginger
c
legends
butchers
can’t say it
c
iron something
c
furry
c
c
c
c
miss, sign, gun, war
dk, can’t say
billions
c
c
c

c
c
c
c
c
c
c
zip
swimming
elect, dk
ex, sing something
c
dk
c
tube
soldier
burnly
bacon
c
c
c
muster, wrong, like musty
cake, beforehand, pastry, dk
tester
c
c
cap
motion
pondly
boots
squeak, wrong
dk
c
c
taller
but, dk
cream, ointment
c
dk
hussy
punishment
dk
ten something, or, ten, or
c
ginger
c
legions
c
dk
c
iron
c
c
c
c
c
bellows
dk
c
million billion
c
c
dk

19
2
40
4
0
46
1
1
2
71
135
59
0
0
0
0
0
0
0
0
0
1
1
1
2
1
1
1
1
1
1
1
1
1
1
2
2
6
2
2
2
2
3
3
5
4
6
5
6
12
13
16
16
16
17
17
18
31
32
43
40
50
58

5.74
5.84
5.95
6.05
6.06
6.11
6.21
6.26
6.89
3.3
1.9
4.8
3.11
5.32
4.18
4.06
2.94
4.95
4.72
4.32
3
1.72
4.58
1.89
3.53
2.84
4.11
4.68
2.67
3.68
6.72
2.68
4.94
5.84
2.5
3.61
5.32
6.26
2.95
3.16
6.21
2.84
4.47
4.42
3.11
3.94
3.26
6.68
1.39
3.95
1.58
5.95
4.79
3.11
6.47
6.47
6.53
3.63
3.53
3.79
4.11
3.79
2.11

40
72
166
31
10
189
22
2
25
5
31
0
23
84
15
60
27
11
4
1
0
925
276
79
57
44
29
24
15
10
5
4
3
13
68
0
259
32
2
2
1
0
222
21
5
8
67
1
0
26
71
824
20
0
234
11
28
41
100
56
3
24
182

3.36
6.64
5.17
5.75
6.58
3.17
6.5
6.75
5.27
2.2
1.2
3
6.92
3
6.33
2.67
4.83
5.58
6.58
4.78
5.67
2
2.75
3.42
6.5
6.83
6.83
6.36
6.73
6.5
6.58
4.75
4.42
3.67
2.55
5.33
2.83
6.5
6.5
5.42
1.67
3.17
4
6.82
5.33
4.91
6.5
4.73
1.73
6
6.42
1.92
6.58
3.82
3.75
3.92
6.67
3.58
3.45
6.25
6.45
6.25
3.08

K. Rastle et al. / Brain and Language 97 (2006) 189–199

197

Appendix A (continued)
Target

Condition

Response session 1

Response session 2

Word
frequency

Word
imageability

Stem
frequency

Stem
imageability

proper
shoulder
brother
corner
study
army
door
party
million
pastor
luster
manger
lasso
cornea
dingo
legume
ramble
billow
haggle
blurb
warble
caret
bungle
armada
addict
pastel
mayhem
bellow
prowl
barb
tallow
rote
millet
ripple
twitch
riddle
capsule
brothel
doe
lapse
batch
punch
ribbon
furnace
rubble
puberty
menu
potent
booth
cellar
barley
prophet
push
china
pencil
passenger
catch
tackle
portrait
tennis
studio
crown
temple

PseudosuYxed
PseudosuYxed
PseudosuYxed
PseudosuYxed
PseudosuYxed
PseudosuYxed
PseudosuYxed
PseudosuYxed
PseudosuYxed
PseudosuYxed
PseudosuYxed
Embedded
Embedded
Embedded
Embedded
Embedded
Embedded
Embedded
Embedded
Embedded
Embedded
Embedded
Embedded
Embedded
Embedded
Embedded
Embedded
Embedded
Embedded
Embedded
Embedded
Embedded
Embedded
Embedded
Embedded
Embedded
Embedded
Embedded
Embedded
Embedded
Embedded
Embedded
Embedded
Embedded
Embedded
Embedded
Embedded
Embedded
Embedded
Embedded
Embedded
Embedded
Embedded
Embedded
Embedded
Embedded
Embedded
Embedded
Embedded
Embedded
Embedded
Embedded
Embedded

pellow
c
c
corners
student
c
c
c
c
pasta, spaghetti, wrong dk
hustle
manager
c
corn, dk
australian, dingo, dog, hyena
no idea, leg
c
bill, billow, no
haggis, wrong
c
war something
jewel, stone, caret
dk, bun something
arm, dk
adding something
pasta
c
bellows
c
barble, Wsh hook
tall, low
rot
mill, let, no idea
ripples
itching
c
capsules
c
c
c
c
c
c
fur, furren, no idea
c
pub something, dk
c
pot, tent, no idea
c
c
c
dk
c
c
c
people, customers
c
c
c
c
c
c
c

pellows, can’t say
c
c
c
student
c
c
part
millions
passion, wrong
dk
manager
lassoon
corn something
your place, can’t say it
legs something
c
bills something
haggis, wrong
blob
war something
car something
bunery
arm something, dk
adding
pale
may, dk
c
c
c
tall something
rot
pets
ripples
twickle
c
c
brother
c
c
c
c
c
Wr, but bigger, ferno, dk
rubbish
public
c
pots, dk
c
c
c
propet something
c
c
c
c
c
c
c
c
c
c
c

61
71
89
105
106
113
347
377
199
4
0
0
0
0
0
0
0
0
0
0
0
0
0
1
2
2
1
1
1
1
1
1
2
2
3
3
3
3
3
3
4
7
6
6
6
7
8
7
9
11
10
11
20
12
16
16
23
22
20
22
22
23
25

1.58
6.32
5
5.42
4.79
5.58
7
6.16
4.05
—
2.69
5.22
5.06
4.78
4.47
4.25
3.58
3.44
3.16
3.06
2.89
2.6
2.39
5.28
4.89
4.21
4.17
4.05
4
3.94
2.64
1.5
5.22
5
4.84
2.47
6.11
5.32
5.06
1.89
3.22
5.16
7
6.28
4.72
3.42
6.21
2.11
5.44
6.11
5.26
4.68
4.16
6.11
7
5.42
4.47
3.84
6.11
6.37
6.26
6.84
6.32

3
4.75
772
1.75
2
5.67
25
6.67
2
5.33
114
6.67
1928
1.75
487
2.36
10
5.58
276
2.67
10
4.42
1012
6.5
1
5.83
25
6.67
0
3.17
67
6.5
5
6.75
56
6
1
5.33
3
4.24
343
5
182
6.82
4
6.75
114
6.73
84
2.58
276
2.67
824
2
28
6.58
3
4.63
68
6.67
68
5.18
8
4.42
10
5.92
5
4.67
1
5.08
0
1.83
29
6.42
2
5.67
1928
1.75
19
5.25
9
6.58
1
1.92
2
5.33
20
3.4
13
4.25
21
6.83
686
6.33
24
6.67
10
6.42
36
5.58
68
6.67
6
5.17
2
5.58
27
6.75
19
7
100
3.17
44
6.83
2
4.58
26
5.75
222
4
2
5.67
2
6.83
0
3.82
(continued on next page)

198

K. Rastle et al. / Brain and Language 97 (2006) 189–199

Appendix A (continued)
Target

Condition

Response session 1

Response session 2

Word
frequency

Word
imageability

Stem
frequency

Stem
imageability

drama
soldier
missile
partner
fortune
kitchen
diet
control
article

Embedded
Embedded
Embedded
Embedded
Embedded
Embedded
Embedded
Embedded
Embedded

actors
c
missiles
parners
fortunes
cooking, kitchen
c
trolls
artist, art, clothes, furniture

dk
c
c
c
c
c
c
trolls
clothes, argyle

22
27
28
27
30
111
56
197
41

4.16
6.68
6.58
4.21
4.37
6.68
2.4
3.2
4.1

1
60
41
487
24
9
80
1
166

4.67
2.91
3.08
2.75
6.75
3.36
4.2
2.2
5.1

Note. c, correct; dk, don’t know.

References
Baayen, R. H., Piepenbrock, R., & van Rijn, H. (1993). The CELEX lexical
database (CD-ROM). Philadelphia, PA: Linguistic Data Consortium,
University of Pennsylvania.
Badecker, W., & Caramazza, A. (1987). The analysis of morphological
errors in a case of acquired dyslexia. Brain and Language, 32, 278–305.
Castles, A., Coltheart, M., Savage, G., Bates, A., & Reid, L. (1996). Morphological processing and visual word recognition: Evidence from
acquired dyslexia. Cognitive Neuropsychology, 13, 1041–1057.
Coltheart, M. (1980a/1987a). Deep dyslexia: A review of the syndrome. In
M. Coltheart, K. Patterson, & J. C. Marshall (Eds.), Deep dyslexia.
London: Routledge and Kegan Paul.
Coltheart, M. (1980b/1987b). Deep dyslexia: A right-hemisphere hypothesis. In M. Coltheart, K. Patterson, & J. C. Marshall (Eds.), Deep dyslexia. London: Routledge & Kegan Paul.
Coltheart, M. (1981). MRC psycholinguistic database. Quarterly Journal
of Experimental Psychology, 33A, 497–505.
Coltheart, M. (2000). Deep dyslexia is right hemisphere reading. Brain and
Language, 71, 299–309.
Coltheart, M., Patterson, K., & Marshall, J. C. (1980/1987). Deep dyslexia.
London: Routledge & Kegan Paul.
Frost, R., & Grainger, J. (2000). Cross linguistic perspectives on morphological processing. New York: Psychology Press.
Frost, R., Grainger, J., & Rastle, K. (2005). Current issues in morphological
processing. New York: Psychology Press.
Frost, R., Forster, K. I., & Deutsch, A. (1997). What can we learn from the
morphology of Hebrew? A masked-priming investigation of morphological representation. Journal of Experimental Psychology: Learning,
Memory and Cognition, 23, 829–856.
Funnell, E. (1987). Morphological errors in acquired dyslexia: A case of
mistaken identity. Quarterly Journal of Experimental Psychology, 39A,
497–539.
Funnell, E. (2000). Deep dyslexia. In E. Funnell (Ed.), Case studies in the
neuropsychology of reading. New York: Psychology Press.
Giraudo, H., & Grainger, J. (2000). EVects of prime word frequency and
cumulative root frequency in masked morphological priming. Language and Cognitive Processes, 15, 421–444.
Job, R., & Sartori, G. (1984). Morphological decomposition—evidence
from crossed phonological dyslexia. Quarterly Journal of Experimental
Psychology, 36A, 435–458.
Kay, J., Lesser, R., & Coltheart, M. (1996). Psycholinguistic assessments of
language processing in aphasia (PALPA): An introduction. Aphasiology, 10, 159–180.
Levelt, W. J. M., Roelofs, A., & Meyer, A. S. (1999). A theory of lexical
access in speech production. Behavioral and Brain Sciences, 22, 1–38.
Marshall, J. C., & Newcombe, F. (1966). Syntactic and semantic errors in
paralexia. Neuropsychologia, 4, 169–176.
Marshall, J. C., & Newcombe (1973). Patterns of paralexia. Journal of Psycholinguistic Research, 2, 175–199.
Marslen-Wilson, W. D., & Tyler, L. K. (1997). Dissociating types of mental
computation. Nature, 387, 592–594.

Marslen-Wilson, W. D., Tyler, L. K., Waksler, R., & Older, L. (1994). Morphology and meaning in the English mental lexicon. Psychological
Review, 101, 3–33.
Melinger, A. (2003). Morphological structure in the lexical representation
of preWxed words: Evidence from speech errors. Language and Cognitive Processes, 18, 335–362.
Michel, F., HenaV, M. A., & Intrilligator, J. (1996). Two diVerent readers
in the same brain after a posterior callosal lesion. NeuroReport, 7,
786–788.
Morton, J., & Patterson, K. (1980). A new attempt at an interpretation, or,
an attempt at a new interpretation. In M. Coltheart, K. Patterson, & J.
C. Marshall (Eds.), Deep dyslexia. London: Routledge & Kegan Paul.
Patterson, K. (1978). Phonemic dyslexia: Errors of meaning and the meaning of errors. Quarterly Journal of Experimental Psychology, 30, 587–
601.
Patterson, K. (1979). What is right with ‘deep’ dyslexics? Brain and Language, 8, 111–129.
Patterson, K. (1980). Derivational errors. In M. Coltheart, K. Patterson, &
J. C. Marshall (Eds.), Deep dyslexia. London: Routledge & Kegan Paul.
Patterson, K. E., & Marcel, A. J. (1977). Aphasia, dyslexia, and phonological coding of written words. Quarterly Journal of Experimental Psychology, 29, 307–318.
Patterson, K., Vargha-Khadem, F., & Polkey, C. (1987). Reading with one
hemisphere. Brain, 112, 39–63.
Plaut, D. C., & Gonnerman, L. M. (2000). Are non-semantic morphological eVects incompatible with a distributed connectionist approach to
language processing? Language and Cognitive Processes, 15, 445–485.
Plaut, D. C., & Shallice, T. (1993). Deep dyslexia—A case-study of connectionist neuropsychology. Cognitive Neuropsychology, 10, 377–500.
Rastle, K., Davis, M. H., Marslen-Wilson, W., & Tyler, L. K. (2000). Morphological and semantic eVects in visual word recognition: A time
course study. Language and Cognitive Processes, 15, 507–538.
Rastle, K., Davis, M. H., & New, B. (2004). The broth in my brother’s
brothel: Morpho orthographic segmentation in visual word recognition. Psychonomic Bulletin & Review, 11, 1090–1098.
Raveh, M., & Rueckl, J. G. (2000). Equivalent eVects of inXected and derived
primes: Long-term morphological priming in fragment completion and
lexical decision. Journal of Language and Memory, 42, 103–119.
Roelofs, A., & Baayen, H. (2002). Morphology by itself in planning
the production of spoken words. Psychonomic Bulletin & Review, 9,
132–138.
SaVran, E. M., Bogyo, LC., Schwartz, M. F., & Marin, O. S. M. (1980/
1987). Does deep dyslexia reXect right-hemisphere reading. In M. Coltheart, K. Patterson, & J. C. Marshall (Eds.), Deep dyslexia. London:
Routledge & Kegan Paul.
Shallice, T., & Warrington, E. K. (1975). Word recognition in a phonemic
dyslexic patient. Quarterly Journal of Experimental Psychology, 27,
187–199.
Taft, M., & Forster, K. I. (1975). Lexical storage and retrieval for preWxed
words. Journal of Verbal Learning and Verbal Behavior, 14, 638–647.
Tyler, L. K. (1992). Spoken language comprehension: An experimental
approach to disordered and normal processing. Cambridge: MIT Press.

K. Rastle et al. / Brain and Language 97 (2006) 189–199
Tyler, L. K., de Mornay-Davies, P., Anokinah, R., Longworth, C., Randall,
B., & Marslen-Wilson, W. D. (2002). Dissociations in processing past
tense morphology: Neuropathology and behavioral studies. Journal of
Cognitive Neuroscience, 14, 79–94.
Tyler, L. K., Moss, H. E., & Jennings, F. (1995). Abstract word deWcits in
aphasia—Evidence from semantic priming. Neuropsychology, 9, 354–363.

199

Tyler, L. K., Randall, B., & Marslen-Wilson, W. D. (2002). Phonology and
neuropsychology of the English past tense. Neuropsychologia, 40,
1154–1166.
Weekes, B., Coltheart, M., & Gordon, E. (1997). Deep dyslexia and righthemisphere reading—A regional cerebral blood Xow study. Aphasiology, 11, 1139–1158.