See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/49756667

Do patients with pure alexia suffer from a

specific word form processing deficit? Evidence
from 'wrods with trasnpsoed letetrs'

Article in Neuropsychologia April 2011

DOI: 10.1016/j.neuropsychologia.2011.01.012 Source: PubMed

CITATIONS READS

6 62

6 authors, including:

Tobias Pflugshaupt
Luzerner Kantonsspital
36 PUBLICATIONS 620 CITATIONS

SEE PROFILE

All content following this page was uploaded by Tobias Pflugshaupt on 29 September 2014.

The user has requested enhancement of the downloaded file. All in-text references underlined in blue are added to the original document
and are linked to publications on ResearchGate, letting you access and read them immediately.
 Neuropsychologia 49 (2011) 12941301

Contents lists available at ScienceDirect

Neuropsychologia
journal homepage: www.elsevier.com/locate/neuropsychologia

Do patients with pure alexia suffer from a specic word form processing decit?
Evidence from wrods with trasnpsoed letetrs
Tobias Pugshaupt a,b, , Julia Suchan a , Marc-Andr Mandler a , Alexander N. Sokolov c,d ,
Susanne Trauzettel-Klosinski c , Hans-Otto Karnath a
a
Section of Neuropsychology, Centre for Neurology, Hertie-Institute for Clinical Brain Research, Eberhard Karls University, Tbingen, Germany
b
Unit of Neuropsychology, Clinic of Neurology, University Hospital, Zrich, Switzerland
c
Centre for Ophthalmology, Low Vision Clinic and Research Laboratory, Eberhard Karls University, Tbingen, Germany
d
Department of Psychosomatic Medicine and Psychotherapy, University of Tbingen, Medical School, Tbingen, Germany

a r t i c l e i n f o a b s t r a c t

Article history: It is widely accepted that letter-by-letter reading and a pronounced increase in reading time as a function
Received 20 October 2010 of word length are the hallmark features of pure alexia. Why patients show these two phenomena with
Received in revised form respect to underlying cognitive mechanisms is, however, much less clear. Two main hypotheses have been
13 December 2010
proposed, i.e. impaired discrimination of letters and decient processing of word forms. While the former
Accepted 6 January 2011
decit can easily be investigated in isolation, previous ndings favouring the latter seem confounded.
Available online 12 January 2011
Applying a word reading paradigm with systematically manipulated letter orders in two patients with
pure alexia, we demonstrate a word form processing decit that is not attributable to sublexical letter
Keywords:
Reading
discrimination difculties. Moreover, pure alexia-like xation patterns could be induced in healthy adults
Pure alexia by having them read sentences including words with transposed letters, so-called jumbled words. This
Visual word form further corroborates a key role of decient word form processing in pure alexia. With regard to basic
Letter transposition reading research, the present study extends recent evidence for relative, rather than precise, encoding of
Letter position encoding letter position in the brain.
Eye movements 2011 Elsevier Ltd. All rights reserved.

1. Introduction
Pure alexia is a rare neurological disorder characterised by
severe reading difculties in the absence of other language-related
impairments. A pronounced increase in reading time as a function
of word length (Behrmann, Plaut, & Nelson, 1998) and a com-
pensatory letter-by-letter (LBL) reading strategy (Coslett, 2000)
are the clinical hallmark features of this disorder. LBL reading is
also shown by patients suffering from alexia with agraphia (e.g.
Rapcsak & Beeson, 2004; Sakurai et al., 2000). However, these lat-
ter patients suffer from both reading and writing problems. Most
pure alexics reported in the literature show damage to left occipito-
temporal brain areas (Montant & Behrmann, 2000). Specically,
lesions affecting the paraventricular white matter of the left occip-
ital lobe (Damasio & Damasio, 1983), the left inferior temporal and
fusiform gyri (Binder & Mohr, 1992; Leff, Spitsyna, Plant, & Wise,
2006; Pugshaupt et al., 2009), or the left occipito-temporal sulcus
Corresponding author at: Unit of Neuropsychology, Clinic of Neurology, Univer-
sity Hospital, Frauenklinikstrasse 26, CH-8091 Zrich, Switzerland.
Tel.: +41 44 255 55 77; fax: +41 44 255 44 29.
E-mail address: tobias.pugshaupt@usz.ch (T. Pugshaupt).
(Cohen et al., 2003) have been proposed as the anatomical corre-
late of pure alexia. The three most recent of these studies linked
the common lesion site found in pure alexics with the location of
the visual word form area (VWFA) in healthy adults (Cohen et al.,
2003; Leff et al., 2006; Pugshaupt et al., 2009). This suggests that
the cognitive pathomechanism underlying pure alexia might have
something to do with the computation or processing of the visual
word form, dened as an abstract representation of ordered let-
ter strings (Warrington & Shallice, 1980) that is invariant across
changes in spatial location, case, or font (Cohen et al., 2000).
In fact, several hypotheses attributing pure alexia to impaired
word form processing have been postulated. For instance, patients
may have lost the ability to encode letters in parallel (Rayner &
Johnson, 2005) or map the percept of all the letters in a string onto
the visual word form (Leff et al., 2001). Damage to visual word
forms might also cause the disorder (Warrington & Langdon, 1994).
Experimental evidence in support of these hypotheses comes from
tasks intended to increase the necessity for whole word reading,
as opposed to sequential LBL reading. For example, patients with
pure alexia showed particularly pronounced reading difculties
when words were written in script relative to print or when
their exposure duration was very brief (Warrington & Shallice,
1980). Detecting single letters added to words, or parsing unspaced

0028-3932/$ see front matter 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.neuropsychologia.2011.01.012
 T. Pugshaupt et al. / Neuropsychologia 49 (2011) 12941301
words also seem specically impaired (Warrington & Langdon,
1994).
However, the idea that pure alexia results from impaired word
form processing is not undisputed. Others have argued that the
main decit of patients concerns the encoding of letters on a sub-
lexical level (Arguin & Bub, 1993). Indeed, many patients with pure
alexia display letter confusion, which mainly affects visually simi-
lar letters (e.g. Patterson & Kay, 1982; Perri, Bartolomeo, & Silveri,
1996) and is enhanced when several letters have to be identied
simultaneously (Arguin & Bub, 2005). Strong experimental support
for a fundamental role of letter confusion in pure alexia comes from
the observation that the characteristic increase in reading time as
a function of word length the so-called word-length effect (WLE)
can be eliminated by matching the summed confusability of con-
stituent letters across different word lengths (Fiset, Arguin, Bub,
Humphreys, & Riddoch, 2005).
In light of the latter nding, one might even ask whether dif-
culties in letter discrimination explain all reading problems of
pure alexia patients, so that the disorder should be relabelled
letter confusability dyslexia, as suggested by Fiset et al. (2005).
Consequently, the mere existence of a specic word form pro-
cessing decit one that is not attributable to more fundamental
letter discrimination problems is put into question. That word
form processing by denition includes letter encoding consider-
ably complicates matters. For example, it seems unclear whether
pure alexics more pronounced difculties in reading script as
opposed to print (Warrington & Shallice, 1980) should be attributed
to impaired word form processing, enhanced letter confusability
when words are written in a different font, or both. Investigating
a specic word form processing decit in pure alexia thus requires
experimental paradigms that allow manipulating word forms while
keeping letter information constant.
This requirement can be met by modifying the order of let-
ters within words, thereby creating so-called jumbled words.
Provoked by a fake email message about research conducted
at Cambridge University, jumbled words have recently gained
attention (Grainger & Whitney, 2004; Rayner, White, Johnson, &
Liversedge, 2006). The message stated that we can read them with
relative ease, as long as rst and last letters are in the right place.
It was proposed that this jumbled word effect sheds light on how
the brain encodes the position of letters within words (Grainger &
Whitney, 2004).
While classical models of visual word recognition assume
precise-position encoding (Coltheart, Rastle, Perry, Langdon, &
Ziegler, 2001; McClelland & Rumelhart, 1981) which is at odds
with the jumbled word effect and other experimental ndings (e.g.
Gomez, Ratcliff, & Perea, 2008) Grainger and Whitney (2004) pro-
pose relative-position coding based on ordered letter pairs labelled
open bigrams. These units are open insofar as they code the posi-
tion of a given letter somewhere left or right of another letter, with
a maximum of two intervening letters (Grainger & Van Heuven,
2003). For instance, reading the word FOREST is assumed to acti-
vate the units representing FO, FR, FE, OR, OE, OS, RE, RS, RT, ES, ET,
and ST. Analysing its jumbled variant FOERST nicely exemplies
why the open-bigram approach is considered more appropriate in
explaining the jumbled word effect than precise-position encod-
ing: the variant still activates 92% of the open bigram units (11 of
12) but only shows 67% overlap in precise letter positions (4 of 6).
Jumbled word paradigms might thus be helpful not only in examin-
ing the cognitive basis of pure alexia but also in testing assumptions
about how the brain encodes words.
The aim of the present study was to investigate the supposed
word form processing decit of pure alexics in three steps. First, we
intended to replicate two main ndings from the seminal work by
Warrington and Shallice (1980), i.e. patients enhanced difculties
when words are written in script or only briey presented. Sec-
1295
ond, the limitation that these two difculties seem ambiguous with
regard to the underlying letter and/or word form processing decit
should be overcome by applying a jumbled word paradigm. Finally,
and similar to previous studies proposing that a certain mechanism
is the main cognitive decit in pure alexia (Fiset, Gosselin, Blais, &
Arguin, 2006; Rayner & Johnson, 2005), we attempted to induce
LBL reading in healthy adults by having them read sentences with
jumbled words while their eye movements were recorded. The
task design was based on the nding that jumbling initial letters
of words provokes more pronounced visuo-motor cost in healthy
readers than jumbling internal letters of words, relative to nor-
mal text (Rayner et al., 2006; White, Johnson, Liversedge, & Rayner,
2008).
2. Methods
2.1. Participants
Two patients (PS, RR) participated in the study. Both fullled the main diagnos-
tic criteria for pure alexia: presence of a signicant WLE when reading single words
(Fig. 1; Table 1) in the absence of other language-related impairments. The latter
was examined with an aphasia screening test, which evaluates spontaneous speech,
oral repetition, reading of single words and sentences, colour and picture naming,
writing of single words and sentences, auditory comprehension, and reading com-
prehension (Weniger, 2006). Moreover, both patients were able to identify local
and global aspects of a complex visual scene as part of the same aphasia screening
test (Weniger, 2006), correctly named both global and local elements of compound
hierarchical letters (Navon, 1977), and were unimpaired in the identication of over-
lapping gures (Poppelreuter, 1917). In contrast, their reading was extremely slow
even with short function words and characterised by overt LBL reading. More-
over, they displayed minor letter naming confusion which exclusively concerned
visually similar letters and a trend towards prolonged reaction times during sin-
gle letter naming (Table 1B). With regard to aetiology, both patients had suffered
a stroke in the supply territory of the left posterior cerebral artery (PCA), causing
brain damage predominantly in the ventral parts of the left occipital and temporal
lobe (Table 1A). Fig. 2A depicts that the lesion of one patient (RR) included the coor-
dinate assigned to the centre of the VWFA in healthy controls (Jobard, Crivello, &
Tzourio-Mazoyer, 2003), while the other patient showed brain damage close to this
coordinate.
Fig. 1. Reading times from a single word reading task during which participants
had to read single words of different lengths (6, 8, or 10 letters) as accurately and
quickly as possible. Words were written in Courier new (font size: 40 pixels, cor-
responding to 3.72 , 4.96 , and 6.20 of visual angle) and shown one-by-one on a
notebook screen placed at 60 cm from participants, to the left of a preceding central
xation point. Overall, 20 frequency-matched German nouns were displayed for
every word length and presented in a pseudo-randomised sequence. Participants
were instructed to press a key as soon as they identify the word, and to read the
word aloud thereafter. Bars show means, error bars display standard deviations of
the control group. Reading accuracy was faultless in both patients and all control
participants.
1296 T. Pugshaupt et al. / Neuropsychologia 49 (2011) 12941301

Table 1A
Demographic and clinical data of the two patients.

Patient Gender Handedness Age in Education Aetiology Time post onset Lesion size Affected brain regionsb
years in years in months in ccma (exclusively in the left hemisphere)

PS Male Right 53 13 Ischaemic stroke 13 63.90 Fusiform gyrus; lingual gyrus;

inferior temporal gyrus;
hippocampus; parahippocampal
gyrus; inferior, middle, and
superior occipital gyri; calcarine
sulcus; cuneus.
RR Male Right 70 13 Haemorrhagic stroke 112 99.00 Fusiform gyrus; lingual gyrus;
inferior and middle temporal gyri;
parahippocampal gyrus; inferior,
middle, and superior occipital gyri;
calcarine sulcus; cuneus.
a
Measured with MRIcroN (www.mricro.com/mricron) on the normalised brain.
b
Based on the Automated Anatomical Labelling (AAL) map (Tzourio-Mazoyer et al., 2002).

Tuebingen automated perimetry (TAP 2000; Oculus, Wetzlar, Germany)
revealed a right upper quadrantanopia in both patients (Fig. 2B). Applying a
previously described procedure (Trauzettel-Klosinski & Reinhard, 1998), fundus-
controlled perimetry with a scanning laser ophthalmoscope (SLO 101; Rodenstock,
Munich, Germany) showed that macular sparing along the horizontal meridian was
0.5 in one patient (RR) and absent in the other. Apart from this visual eld defect
and severe reading problems, standard neurological status examination revealed
no further impairment in either patient. Their experimental results were separately
compared with those of ten healthy adults (6 males; all right-handed; age: mean
66 years, range 5781 years), applying modied t tests specically designed for
single case studies (Crawford & Howell, 1998). Differences between experimental
conditions within the control group were analysed with paired-samples t tests or

Table 1B
Single letter naming and word-length effect (WLE).

Patient Single letter naminga Word-length effect (WLE)c

Mean reaction t (df = 9) P (one-tailed)b Letter naming errors WLE in ms/letterd t (df = 9) P (one-tailed)e
time in ms (stimulus/response)

PS 780 1.584 0.074 q/p, j/y, b/d 283 21.532 <0.001

RR 783 1.605 0.071 q/p 245 18.513 <0.001
a
Based on a task during which patients had to name all 26 lowercase letters of the German alphabet which appeared one-by-one on a notebook screen, to the left of a
preceding central xation point as accurately and quickly as possible. Letters were written in Arial, font size was 125 pixels, corresponding to 1.4 of visual angle on average.
Participants were instructed to press a key as soon as they identify the letter, and to name it aloud thereafter.
b
One-tailed error probability of a modied t test specically designed to compare the value of a single patient with those of a small control group (Crawford & Howell,
1998). Values are compared with those of the healthy control group (N = 10; mean = 554 ms, SD = 136 ms).
c
This task is described in the caption to Fig. 1.
d
Word-length effect (WLE) in ms per additional letter.
e
One-tailed error probability of a modied t test specically designed to compare the value of a single patient with those of a small control group (Crawford & Howell,
1998). Values below 0.05 conrm the hypothesis that the WLE of a given patient is signicantly larger than corresponding values from the healthy control group (N = 10;
mean = 12 ms, SD = 12 ms).

Fig. 2. (A) Brain lesions of the two patients with pure alexia. Lesions were marked on T2-weighted MR-scans (FLAIR sequence) using the freely available MRIcroN software
(www.mricro.com/mricron; Rorden, Karnath, & Bonilha, 2007), then normalised with SPM5 (www.l.ion.ucl.ac.uk/spm) and plotted on the CH2 template brain in MRIcroN.
The selection of axial slices corresponds to that of previous studies on pure alexia (Leff et al., 2006; Pugshaupt et al., 2009). White crosshairs show the coordinate assigned
to the centre of the VWFA in healthy adults (Jobard et al., 2003). (B) Central visual elds (radius = 30 ) of the two patients, based on automated static perimetry. Field defects
are highlighted by a superimposed semi-transparent grey area.
 T. Pugshaupt et al. / Neuropsychologia 49 (2011) 12941301
repeated-measurement ANOVA. Finally, corrected visual acuity of the better eye
at reading distance was 0.63 in one patient (RR) and 1.0 in the other (PS). These
acuity values did not signicantly differ from those of the ten control participants
(mean = 0.83, SD = 0.23, range 0.401.25).
2.2. Tasks
2.2.1. Print versus script reading
Fifty German seven-letter nouns written in Arial or Vereinfachte Ausgangsschrift
which is one of the three ofcial script types in Germany (downloaded from
www.pelikan.com) were presented in a pseudo-randomised sequence on a note-
book placed at 60 cm from participants. Font size was 40 pixels corresponding to
about 4 of visual angle on average (mean = 3.95 , SD = 0.52 ) and both font types
were equally often used. Moreover, print and script words were matched with regard
to frequency (Baayen, Piepenbrock, & van Rijn, 1993), concreteness, and imageability
(Hager & Hasselhorn, 1994). During each trial, a central xation point appeared on
the screen for one second, followed by a word stimulus shown on the left to minimise
the inuence of patients right-sided visual defect. Participants were instructed to
press a key which terminated the stimulus presentation as soon as they identi-
ed the word, and to read the word aloud thereafter. If no keypress occurred within
20 s, the stimulus automatically disappeared, and reaction time was not logged for
this trial. The experimenter made notes of oral responses, while trial presentation
and reaction time logging were performed by the E-PrimeTM software (Psychology
Software Tools Inc., Pittsburgh, PA, USA). Four practise trials were conducted in each
participant.
2.2.2. Reading briey presented words
Based on the same experimental setup, a lexical decision task was used to exam-
ine word form processing under brief exposure conditions. Thirty German ve-letter
nouns and 30 ve-letter non-words were presented, one-by-one, in a fully ran-
domised sequence. Non-words were constructed by replacing two internal letters
in 30 frequency-matched German nouns (Baayen et al., 1993), similar to a previously
described procedure (Behrmann, Nelson, & Sekuler, 1998). All words and non-words
were written in Courier new, applying a font size of 40 pixels, with each stimulus sub-
tending 3.10 of visual angle. Every trial started with a central xation point shown
for one second, followed by a word or non-word stimulus displayed to the left of
xation. More precisely, there was one character space between the nal letter of
the word/non-word and the centre of the screen. With reference to previous stud-
ies (Coslett & Saffran, 1989; Coslett, Saffran, Greenbaum, & Schwartz, 1993), every
stimulus was shown for 250 ms and then masked by random visual noise. Subse-
quently, participants had to indicate whether a word or a non-word was displayed
by pressing one of two predened keys. In addition, and prior to ve practise trials,
they were instructed that solely the accuracy of their responses, but not reaction
times, were analysed.
2.2.3. Reading jumbled words
Sixty German six-letter nouns written in Courier new were evenly assigned
to the following three experimental conditions: unjumbled, slightly jumbled (let-
ter orders: 124536 (e.g. Ruribk instead of Rubrik), 125346 (e.g. Gelhat instead of
Gehalt), 134256 (e.g. Slakve instead of Sklave), or 142356 (e.g. Frabik instead of Fab-
rik); 75% open bigram overlap; 50% letter position overlap), and heavily jumbled
(letter orders: 143526 (e.g. Aetinl instead of Anteil), 152436 (e.g. Soybml instead
of Symbol), or 135426 (e.g. Pnezar instead of Panzer); 50% open bigram and letter
position overlap). Font size was 40 pixels, corresponding to 3.72 of visual angle.
In addition to examining a word form processing decit in pure alexia, selecting
these particular letter orders allowed testing whether the open bigram approach
(Grainger & Whitney, 2004) can explain the jumbled word effect better than classi-
cal word recognition models (Coltheart et al., 2001; McClelland & Rumelhart, 1981).
If healthy participants performed better with slightly as opposed to heavily jumbled
words, this nding would be attributable to a difference in open bigram overlap,
but obviously not to a difference in letter position overlap. The three groups of
words were matched with regard to frequency (Baayen et al., 1993), concreteness,
imageability (Hager & Hasselhorn, 1994), and both summed and mean confusabil-
ity of constituent letters (Bouma, 1971; Geyer, 1977; Gilmore, Hersh, Caramazza,
& Grifn, 1979; Heijden, Malhas, & van den Roovaart, 1984). All further aspects of
the task design were identical to the script reading task with one crucial exception:
participants were told that the order of letters was changed in some of the word
stimuli but that this manipulation did not affect initial and nal letters and that
they should try to identify the base word in these cases.
2.2.4. Reading sentences with jumbled words
Twenty-one of the 24 standardised German sentences developed by Radner et al.
(2002) were the stimuli of the sentence reading task. These sentences are highly sim-
ilar in terms of grammatical complexity and read at comparable speeds by healthy
participants (Radner et al., 2002). Each sentence starts with a main clause followed
by a relative clause, together comprising 14 words. The present task included three
experimental conditions: normal letter order, jumbled internal letters, and jumbled
initial letters. Concerning the two latter conditions, we consistently jumbled three
target words per sentence, i.e. the main verb and the 10-letter compound noun
present in all main clauses as well as the main verb of the relative clause. Sentences
1297
were evenly assigned to conditions, thereby matching word frequency (Baayen et al.,
1993) and letter confusability (Bouma, 1971; Geyer, 1977; Gilmore et al., 1979; van
der Heijden et al., 1984) of target words.
All sentences were written in Courier new (font size 30 pixels) and displayed as
two text lines on a 17-inch CRT computer screen placed at 60 cm from participants.
Approximately two character spaces corresponded to 1 of visual angle. Initially,
a starting point (i.e. a black dot) was shown to indicate the position of the rst
letter of the subsequently appearing sentence. Participants were instructed to look
at that point and to initiate text presentation with a button press. Then they had
to silently read the sentence before stopping its presentation with another button
press. The experimenter thereafter tested comprehension by asking participants
three questions per sentence, referring to each target word separately.
Eye movements during sentence reading were recorded with an infrared-
based video-oculographic tracking system (EyeLink IITM , SR Research, Osgoode/ON,
Canada). Concerning the identication of saccades and xations from the raw data
recorded by the eye tracking system, the EyeLink saccade detection algorithm sug-
gested for reading studies by the manufacturer was used, including the following
thresholds: a saccade velocity threshold of 30 s1 , a saccade acceleration threshold
of 8000 s2 , and a saccade motion threshold of 0.15 . Moreover, a minimal xa-
tion duration of 100 ms was applied. Prior to blocks of three sentences, the system
was calibrated with a 9-point target grid, and drift correction was performed before
the presentation of every single sentence. A chin rest was used to ensure constant
viewing distance and minimise head movements. Moreover, head movement com-
pensation was performed by the tracking system. For the quantitative analysis of
eye movement behaviour, we selected two xational parameters, i.e. xation fre-
quency expressed as xation-to-character ratio and mean xation duration. These
two parameters are not only suitable to quantify reading difculties in pure alexia
but also less inuenced by the usual right-sided visual eld defect of patients (Leff
et al., 2001) than other visuo-motor variables such as the amplitude of saccades
(Pugshaupt et al., 2009).
3. Results
3.1. Single word reading tasks
Table 2 shows that both patients needed more time to read
words written in script as opposed to words written in print
and that this difference was signicantly larger than corresponding
values of the control group. In one (RR) patient, the script read-
ing decit concerned accuracy as well. When the lexical status (i.e.
word vs. non-word) of briey presented words had to be assessed,
accuracy of both patients was (a) signicantly lower than that of
controls (PS: 61.67%, t = 15.110, P < 0.001; RR: 46.70%, t = 21.437,
P < 0.001; controls: mean = 97.33%, SD = 2.25%), and (b) not signif-
icantly different from chance according to binominal tests (PS:
P = 0.092, RR: P = 0.699).
Degrading the visual word form by jumbling internal letters of
single words elicited a marked loss of reading accuracy, but not
reading duration, in patients. Averaged across experimental con-
ditions, both patients showed a mean decrease in accuracy that
was signicantly larger than corresponding values of the control
group (Table 3). Within the control group, signicant main effects
of experimental condition on accuracy (F(2,18) = 36.516, P < 0.001)
and on reading speed (F(2,18) = 90.847, P < 0.001) were revealed.
Further analyses showed that control participants had more prob-
lems when they read heavily as opposed to slightly jumbled words,
and this concerned both accuracy (t = 6.332, df = 9, P < 0.001) and
reading time (t = 7.866, df = 9, P < 0.001).
3.2. Sentence reading task
Results of the sentence reading task are summarised in Table 4.
Both patients showed unimpaired comprehension for normal
sentences, in contrast to signicant comprehension decits for
sentences containing jumbled words. Within the control group,
a signicant main effect of experimental condition on compre-
hension (F(2,18) = 4.519, P = 0.026) was revealed. Further analyses
showed that sentences containing target words with internally
jumbled letters (t = 2.753, df = 9, P = 0.022) and those contain-
ing target words with initially jumbled letters (t = 2.377, df = 9,
P = 0.041) were both more difcult to understand than normal text,
but did not signicantly differ from one another in this regard.
1298 T. Pugshaupt et al. / Neuropsychologia 49 (2011) 12941301

Table 2
Results of the print versus script reading task.

Variable Condition PS RR Controls mean (SD) t (df = 9) P (one-tailed)a

Accuracy (%) Print 100.00 96.00 100.00 (0.00)

Script 100.00 70.83 99.60 (1.26)
Accuracy difference (%) (Print script) 0.00 0.40 (1.26) 0.303 0.385
25.17 0.40 (1.26) 18.744 <0.001

Mean reaction time (s) Print 2.756 2.365 0.522 (0.085)

Script 3.189 3.736 0.550 (0.093)
Mean reaction time difference (s) (Script print) 0.434 0.028 (0.029) 13.348 <0.001
1.370 0.028 (0.029) 44.122 <0.001

Notes: Reaction time data loss due to responses longer than 20 s was 0% in the control group as well as in PS and 4% in RR. Moreover, reaction times from both correct and
incorrect responses were used to calculate mean reaction times per condition.
a
One-tailed error probability of a modied t test specically designed to compare the value of a single patient with those of a small control group (Crawford & Howell,
1998).

Table 3
Results of the jumbled word reading task.

Variable Condition PS RR Controls mean (SD) t (df = 9) P (one-tailed)c

Accuracy (%) Normal 100.00 100.00 100.00 (0.00)

Slightly jumbled 83.33 65.00 93.50 (7.84)
Heavily jumbled 21.10 25.00 72.50 (12.96)
Accuracy decrease (%)a 39.47 13.75 (6.48) 3.784 0.002
37.50 13.75 (6.48) 2.391 0.020

Mean reaction time (s) Normal 2.958 4.293 0.928 (0.179)

Slightly jumbled 8.808 8.768 2.506 (0.744)
Heavily jumbled 8.738 11.860 7.370 (1.880)
Mean reaction time increase (s)b 2.890 3.221 (0.899) 0.351 0.367
3.784 3.221 (0.899) 0.596 0.283

Notes: Reaction time data loss due to responses longer than 20 s was 0% in the control group as well as in RR and 5% in PS. Moreover, reaction times from both correct and
incorrect responses were used to calculate mean reaction times per condition.
a
Mean decrease in accuracy per 25% reduction in open bigram overlap, which is 100% for normal words, 75% for slightly jumbled words, and 50% for heavily jumbled
words.
b
Mean increase in reaction time per 25% reduction in open bigram overlap.
c
One-tailed error probability of a modied t test (Crawford & Howell, 1998).

Taking into account all xations made during the task, fur-
ther signicant main effects were found in the control group
when the frequency (F(2,18) = 14.864, P < 0.001) and mean dura-
tion (F(2,18) = 59.838, P < 0.001) of xations were compared
between experimental conditions. Pairwise comparisons revealed
the expected pattern: sentences containing target words with ini-
tially jumbled letters i.e. the most difcult condition according
to previous studies (Rayner et al., 2006; White et al., 2008)
evoked higher xation frequency (t = 3.563, df = 9, P = 0.006) and
longer mean xation duration (t = 5.000, df = 9, P = 0.001) than sen-
tences containing target words with internally jumbled letters.
In turn, the latter sentences were less uently read than normal
text by controls, as indicated by respective increases in xation
frequency (t = 3.080, df = 9, P = 0.013) and mean xation duration

Table 4
Results of the sentence reading task.

Variable Conditiona PS RR Controls mean (SD) t (df = 9) P (one-tailed)b

Comprehension accuracy (%) Normal 95.24 98.57 (2.30) 1.380 0.100

95.24 98.57 (2.30) 1.380 0.100
Internal 76.19 94.76 (6.13) 2.888 0.009
55.56 94.76 (6.13) 6.097 <0.001
Initial 61.90 90.00 (12.18) 2.200 0.028
38.10 90.00 (12.18) 4.063 0.001

Fixation frequency (xation-to-character ratio) Normal 1.64 0.30 (0.09) 14.019 <0.001
0.96 0.30 (0.09) 6.957 <0.001
Internal 2.81 0.43 (0.18) 12.989 <0.001
2.70 0.43 (0.18) 12.362 <0.001
Initial 3.19 0.62 (0.29) 8.338 <0.001
2.97 0.62 (0.29) 7.634 <0.001

Mean xation duration (ms) Normal 303 226 (28) 2.653 0.013
288 226 (28) 2.158 0.030
Internal 315 247 (27) 2.375 0.021
313 247 (27) 2.290 0.024
Initial 315 262 (35) 1.451 0.090
318 262 (35) 1.539 0.079
a
Experimental conditions: normal = all words with normal letter order, internal = the three target words had jumbled internal letters, initial = the three target words had
jumbled initial letters.
b
One-tailed error probability of a modied t test (Crawford & Howell, 1998).
 T. Pugshaupt et al. / Neuropsychologia 49 (2011) 12941301 1299

Fig. 3. Analysing xations made during the sentence reading task. (A) Fixation frequency displayed as xation-to-character ratio. (B) Mean xation duration in ms. Normal = all
words with normal letter order, internal = the three target words had jumbled internal letters, initial = the three target words had jumbled initial letters. Bars show means,
error bars display standard deviations of the control group. The shaded grey areas in the background of the two left-sided panels depict patient ranges for the normal condition
to highlight the similarity in magnitude with control data of the initial condition with regard to xations made on the three target word in each sentence.

(t = 6.745, df = 9, P < 0.001). The two patients displayed LBL reading
in all experimental conditions, as indicated by xation-to-character
ratios of about 1 or higher. Relative to controls, they showed signi-
cantly enhanced xation frequencies and partly prolonged xation
durations (Table 4).
Most importantly, region of interest area analyses including only
those xations that fell on the three target words per sentence
revealed that the xation behaviour of healthy controls during the
initial condition was very similar to that of patients during the
normal condition (Fig. 3). This concerned both frequency (Fig. 3A)
and mean duration (Fig. 3B) of xations. Moreover, controls mean
xation-to-character ratio in the initial condition was 1.01, which
is almost precisely the magnitude expected for LBL reading, given
strict LBL reading implies looking at every single character once.
Restated in different terms, initially jumbled words evoked pure
alexia-like xation behaviour in healthy controls. This nding is
also illustrated by the scanning pattern examples depicted in Fig. 4.
4. Discussion
It has been suggested that the main cognitive decit of patients
with pure alexia concerns impaired word form processing. For
example, they may have lost the ability to map the percept of let-
ters onto the mental representation of words (Leff et al., 2001) or
process multiple letters in parallel (Rayner & Johnson, 2005). Con-
sistent with this suggestion, most pure alexics show brain damage
at or near the coordinate assigned to the centre of the visual word
form area (Leff et al., 2006; Pugshaupt et al., 2009), a region in
the left fusiform gyrus that is specialised in identifying visual letter
strings (Dehaene, Cohen, Sigman, & Vinckier, 2005). The experi-
mental evidence supporting a word form processing decit in pure
alexia is mainly based on the nding that reading difculties of
patients worsen during tasks intended to increase the necessity for
whole word reading, for example reading script or briey presented
words (Warrington & Langdon, 1994; Warrington & Shallice, 1980).
1300 T. Pugshaupt et al. / Neuropsychologia 49 (2011) 12941301
Fig. 4. Typical scanpaths of a control participant and a patient (RR) from the sentence reading task. Arrows depict saccades, circles represent xations with circle diameter
being proportional to xation duration and the shaded grey rectangles indicate the interest areas (IA) dened for each of the three target words per sentence (i.e. main
verbs of the main and relative clauses, 10-letter noun in the main clause). Normal = all words with normal letter order, internal = the three target words had jumbled internal
letters, initial = the three target words had jumbled initial letters.
Interestingly, both of our patients also displayed these two clas-
sical signs of a word form processing decit. However, enhanced
difculties when reading script or briey presented words can-
not unambiguously be attributed to a word form processing decit,
but may also occur due to sublexical impairment in letter discrim-
ination, as proposed by others (Arguin & Bub, 1993; Fiset et al.,
2005).
The aim of the present study was to investigate this problem
by systematically modifying the order of letters within words,
thereby creating so-called jumbled words (Grainger & Whitney,
2004). Restated in different terms, we manipulated word forms
while keeping letter information constant. A rst main nding was
that during single word reading, the accuracy of two pure alexia
patients decreased with more severely jumbled words, and their
decrease was signicantly larger than that of healthy controls. This
may be the rst empirical demonstration of a word form processing
decit in pure alexia that is not suspected of being confounded by
more fundamental letter discrimination difculties. In addition, our
nding clearly contradicts the proposal that all reading problems
of pure alexics might be attributable to impaired letter discrimina-
tion, so that the disorder ought to be relabelled letter confusability
dyslexia (Fiset et al., 2005).
It is, however, important to note that both of our patients
showed minor letter confusion and a trend towards prolonged
reaction times during single letter naming. Similar to previously
reported cases (Patterson & Kay, 1982; Perri et al., 1996), the confu-
sion concerned visually similar letters such as p and q. Interestingly,
RR made fewer letter confusion errors than PS, which might be
related to the relatively larger sparing of posterior occipital cortex
in RR, as intact function of this cortical area has been associated
with the ability to correctly identify letters in other lesion studies
(e.g. Sakurai, 2004; Sakurai et al., 2008). These ndings also t well
in a recently proposed model assumption about our brain encodes
written words (Dehaene et al., 2005): it has been suggested that
part of the occipito-temporal what pathway may be tuned to word
recognition. This part might form a hierarchy of local combination
detectors sensitive to increasingly larger fragments of words i.e.
oriented bars, contours, shapes, letters, bigrams, words which
are represented in more and more anterior/temporal brain regions.
Returning to the present study, we conclude that the reading prob-
lems of PS and RR are not solely a consequence of decient word
form processing, but rather reect combined letter discrimination
and word form processing difculties.
Another important question is whether the reading impairment
of our two patients may result from an underlying visuo-perceptive
decit affecting linguistic and non-linguistic stimuli alike. For
example, one might think of a general inability to process multi-
ple forms or objects in parallel, a decit that is considered the main
cognitive problem of patients with simultanagnosia (e.g. Huberle
& Karnath, 2006; Kinsbourne & Warrington, 1962; Nyffeler et al.,
2005; Wolpert, 1924). Since we did not experimentally manipulate
non-linguistic stimuli, it is not possible to conclusively resolve this
issue with the present study. However, as both patients showed
unimpaired performances in standard simultanagnosia screening
tests for example, they were able to correctly identify and describe
overlapping gures and both the local and global aspects of a com-
plex visual scene we assume that their impairment affects, at least
predominantly, the encoding of linguistic stimuli.
As a second main nding of our study, it was possible to provoke
pure alexia-like xation patterns in healthy adults by jumbling a
subset of words in a sentence reading task. Restricting the anal-
ysis to xations made on words that were jumbled in two of the
three experimental conditions i.e. those containing either initially
or internally jumbled words we showed that the frequency and
mean duration of healthy controls xations in the initial condi-
tion was remarkably similar to corresponding values of pure alexia
patients who read normal text. Moreover, when controls read ini-
tially jumbled words, their xation frequency precisely equalled
the magnitude expected for LBL reading, which is a hallmark feature
of pure alexia (Coslett, 2000). Restated in general terms, induc-
ing word form processing difculties in healthy adults led to pure
alexia-like xation behaviour, so that it seems legitimate to assume
that pure alexia is associated with impaired word form processing.
The present results also have implications for basic reading
research. Grainger and Whitney (2004) have suggested that the
jumbled word effect i.e. the relative ease with which we can
read words with transposed letter order sheds light on how the
brain encodes the position of letters within words. In contrast to
precise-position coding implemented in classical word recognition
models (e.g. Coltheart et al., 2001; McClelland & Rumelhart, 1981),
they propose relative-position coding based on ordered letter pairs.
These open bigrams code the position of a given letter up to three
slots left or right of another letter. Grainger and Whitney (2004)
assume that in many cases, the open bigrams of a jumbled word still
shows considerable overlap with the open bigrams of its unjum-
bled variant thereby facilitating correct recognition while the
respective overlap in precise letter position is comparatively low.
The ndings from one of our single word reading tasks corrob-
orate their proposal. Healthy adults displayed signicantly higher
reading accuracy and faster reading speed with slightly (75% open
bigram overlap with the unjumbled variant) as opposed to heavily
(50% open bigram overlap) jumbled words, although all jumbled
word stimuli showed exactly 50% precise letter position over-
lap with their unjumbled variants. Hence, our results empirically
validate the suggested superiority of the open bigram approach
(Grainger & Whitney, 2004) over classical word recognition mod-
 T. Pugshaupt et al. / Neuropsychologia 49 (2011) 12941301
els (e.g. Coltheart et al., 2001; McClelland & Rumelhart, 1981) in
explaining the jumbled word effect. Congruent with recent ndings
from cognitive psychology (e.g. Gomez et al., 2008), they thereby
provide further evidence for relative, rather than precise, encoding
of letter position in the brain.
Finally, we suggest that the relative ease with which we can
read jumbled words must be put into perspective. It has already
been shown that jumbling a subset of words in a sentence pro-
vokes signicant visuo-motor cost in healthy adults, as indicated
by more frequent and prolonged xations (Rayner et al., 2006;
White et al., 2008). Not only did our ndings conrm this extra eye
movement effort, they extend it by adding a cost in comprehension.
Evident in both single word and sentence reading, the compre-
hension of healthy adults decreased with more difcult jumbling
conditions.
Acknowledgments
HOK is supported by the Deutsche Forschungsgemeinschaft and
the Bundesministerium fr Bildung und Forschung, TP received
a grant from the Novartis Foundation, and ANS is supported by
the DFG, the Werner Reichardt Centre for Integrative Neuro-
science (CIN), University of Tbingen, and the Willy Robert Pitzer
Foundation. We also thank Prof. W. Radner (Medical University
of Vienna, Austria) for providing us the stimuli for the sentence
reading task and Hanna Eisenstein (Low Vision Clinic and Research
Laboratory Tbingen, Germany) for measuring the visual elds in
both patients.
References
Arguin, M., & Bub, D. N. (1993). Single-character processing in a case of pure alexia.
Neuropsychologia, 31(5), 435458.
Arguin, M., & Bub, D. (2005). Parallel processing blocked by letter similarity in letter
by letter dyslexia: A replication. Cognitive Neuropsychology, 22(5), 589602.
Baayen, R., Piepenbrock, R., & van Rijn, H. (1993). The CELEX lexical data base [CD-
ROM]. Philadelphia: Linguistic Data Consortium, University of Pennsylvania.
Behrmann, M., Nelson, J., & Sekuler, E. B. (1998). Visual complexity in letter-by-letter
reading: pure alexia is not pure. Neuropsychologia, 36(11), 11151132.
Behrmann, M., Plaut, D. C., & Nelson, J. (1998). A literature review and new data
supporting an interactive account of letter-by-letter reading. Cognitive Neu-
ropsychology, 15(12), 751.
Binder, J. R., & Mohr, J. P. (1992). The topography of callosal reading pathways. A
case-control analysis. Brain, 115, 18071826.
Bouma, H. (1971). Visual recognition of isolated lower-case letters. Vision Research,
11(5), 459474.
Cohen, L., Dehaene, S., Naccache, L., Lehricy, S., Dehaene-Lambertz, G., Hnaff, M. A.,
& Michel, F. (2000). The visual word form area: Spatial and temporal character-
ization of an initial stage of reading in normal subjects and posterior split-brain
patients. Brain, 123(2), 291307.
Cohen, L., Martinaud, O., Lemer, C., Lehricy, S., Samson, Y., Obadia, M., Slachevsky,
A., & Dehaene, S. (2003). Visual word recognition in the left and right hemi-
spheres: Anatomical and functional correlates of peripheral alexias. Cerebral
Cortex, 13(12), 13131333.
Coltheart, M., Rastle, K., Perry, C., Langdon, R., & Ziegler, J. (2001). DRC: A dual route
cascaded model of visual word recognition and reading aloud. Psychological
Review, 108(1), 204256.
Coslett, H. B. (2000). Acquired dyslexia. Seminars in Neurology, 20(4), 419426.
Coslett, H. B., & Saffran, E. M. (1989). Evidence for preserved reading in pure alexia.
Brain, 112, 327359.
Coslett, H. B., Saffran, E. M., Greenbaum, S., & Schwartz, H. (1993). Reading in pure
alexia. The effect of strategy. Brain, 116, 2137.
Crawford, J. R., & Howell, D. C. (1998). Comparing an individuals test score
against norms derived from small samples. The Clinical Neuropsychologist, 12(4),
482486.
Damasio, A. R., & Damasio, H. (1983). The anatomic basis of pure alexia. Neurology,
33(12), 15731583.
Dehaene, S., Cohen, L., Sigman, M., & Vinckier, F. (2005). The neural code for written
words: A proposal. Trends in Cognitive Science, 9(7), 335341.
Fiset, D., Arguin, M., Bub, D., Humphreys, G. W., & Riddoch, M. J. (2005). How to make
the word-length effect disappear in letter-by-letter dyslexia: Implications for an
account of the disorder. Psychological Science, 16(7), 535541.
Fiset, D., Gosselin, F., Blais, C., & Arguin, M. (2006). Inducing letter-by-letter dyslexia
in normal readers. Journal of Cognitive Neuroscience, 18(9), 14661476.
Geyer, L. H. (1977). Recognition and confusion of the lowercase alphabet. Perception
and Psychophysics, 22(5), 487490.
View publication stats
1301
Gilmore, G. C., Hersh, H., Caramazza, A., & Grifn, J. (1979). Multidimensional letter
similarity derived from recognition errors. Perception and Psychophysics, 25(5),
425431.
Gomez, P., Ratcliff, R., & Perea, M. (2008). The overlap model: A model of letter
position coding. Psychological Review, 115(3), 577600.
Grainger, J., & Van Heuven, W. (2003). Modeling letter position coding in printed
word perception. In P. Bonin (Ed.), The mental lexicon (pp. 124). New York:
Nova Science Publishers.
Grainger, J., & Whitney, C. (2004). Does the huamn mnid raed wrods as a wlohe?
Trends in Cognitive Science, 8(2), 5859.
Hager, W., & Hasselhorn, M. (1994). Handbuch deutschsprachiger Wortnormen. Gt-
tingen: Hogrefe Verlag.
van der Heijden, A. H., Malhas, M. S., & van den Roovaart, B. P. (1984). An empir-
ical interletter confusion matrix for continuous-line capitals. Perception and
Psychophysics, 35(1), 8588.
Huberle, E., & Karnath, H.-O. (2006). Global shape recognition is modulated by the
spatial distance of local elements-evidence from simultanagnosia. Neuropsy-
chologia, 44(6), 905911.
Jobard, G., Crivello, F., & Tzourio-Mazoyer, N. (2003). Evaluation of the dual route
theory of reading: A metanalysis of 35 neuroimaging studies. Neuroimage, 20(2),
693712.
Kinsbourne, M., & Warrington, E. K. (1962). A disorder of simultaneous form per-
ception. Brain, 85, 461486.
Leff, A. P., Crewes, H., Plant, G. T., Scott, S. K., Kennard, C., & Wise, R. J. (2001). The
functional anatomy of single-word reading in patients with hemianopic and
pure alexia. Brain, 124, 510521.
Leff, A. P., Spitsyna, G., Plant, G. T., & Wise, R. J. (2006). Structural anatomy of pure
and hemianopic alexia. Journal of Neurology, Neurosurgery & Psychiatry, 77(9),
10041007.
McClelland, J. L., & Rumelhart, D. E. (1981). An interactive activation model of context
effects in letter perception: Part I. An account of basic ndings. Psychological
Review, 88(5), 375407.
Montant, M., & Behrmann, M. (2000). Pure alexia. Neurocase, 6(4), 265294.
Navon, D. (1977). Forest before trees: The precedence of global features in visual
perception. Cognitive Psychology, 9, 353383.
Nyffeler, T., Pugshaupt, T., Hofer, H., Baas, U., Gutbrod, K., von Wartburg, R., Hess,
C. W., & Mri, R. M. (2005). Oculomotor behaviour in simultanagnosia: A longi-
tudinal case study. Neuropsychologia, 43(11), 15911597.
Patterson, K., & Kay, J. (1982). Letter-by-letter reading: Psychological descriptions
of a neurological syndrome. Quarterly Journal of Experimental Psychology, 34A,
411441.
Perri, R., Bartolomeo, P., & Silveri, M. C. (1996). Letter dyslexia in a letter-by-letter
reader. Brain and Language, 53(3), 390407.
Pugshaupt, T., Gutbrod, K., Wurtz, P., von Wartburg, R., Nyffeler, T., de Haan, B.,
Karnath, H.-O., & Mueri, R. M. (2009). About the role of visual eld defects in
pure alexia. Brain, 132, 19071917.
Poppelreuter, W. (1917). Die psychischen Schdigungen durch Kopfschuss im Kriege.
Leipzig: Von Leopold Voss.
Radner, W., Obermayer, W., Richter-Mueksch, S., Willinger, U., Velikay-Parel, M., &
Eisenwort, B. (2002). The validity and reliability of short German sentences for
measuring reading speed. Graefes Archive for Clinical and Experimental Ophthal-
mology, 240(6), 461467.
Rapcsak, S. Z., & Beeson, P. M. (2004). The role of left posterior inferior temporal
cortex in spelling. Neurology, 62, 22212229.
Rayner, K., & Johnson, R. L. (2005). Letter-by-letter acquired dyslexia is due to the
serial encoding of letters. Psychological Science, 16(7), 530534.
Rayner, K., White, S. J., Johnson, R. L., & Liversedge, S. P. (2006). Raeding wrods with
jubmled lettres: There is a cost. Psychological Science, 17(3), 192193.
Rorden, C., Karnath, H.-O., & Bonilha, L. (2007). Improving lesion-symptom mapping.
Journal of Cognitive Neuroscience, 19(7), 10811088.
Sakurai, Y. (2004). Varieties of alexia from fusiform, posterior inferior temporal and
posterior occipital gyrus lesions. Behavioral Neurology, 15(12), 3550.
Sakurai, Y., Takeuchi, S., Takada, T., Horiuchi, E., Nakase, H., & Sakuta, M. (2000).
Alexia caused by a fusiform or posterior inferior temporal lesion. Journal of the
Neurological Sciences, 178(1), 4251.
Sakurai, Y., Terao, Y., Ichikawa, Y., Ohtsu, H., Momose, T., Tsuji, S., & Mannen, T.
(2008). Pure alexia for kana. Characterization of alexia with lesions of the inferior
occipital cortex. Journal of the Neurological Sciences, 268(12), 4859.
Trauzettel-Klosinski, S., & Reinhard, J. (1998). The vertical eld border in human
hemianopia and its signicance for xation behavior and reading. Investigative
Ophthalmology & Visual Science, 39(11), 21772186.
Tzourio-Mazoyer, N., Landeau, B., Papathanassiou, D., Crivello, F., Etard, O., Delcroix,
N., Mazoyer, B., & Joliot, M. (2002). Automated anatomical labeling of activations
in SPM using a macro-scopic anatomical parcellation of the MNI MRI single-
subject brain. Neuroimage, 15(1), 273289.
Warrington, E. K., & Langdon, D. (1994). Spelling dyslexia: A decit of the visual
word-form. Journal of Neurology, Neurosurgery & Psychiatry, 57(2), 211216.
Warrington, E. K., & Shallice, T. (1980). Word-form dyslexia. Brain, 103, 99112.
Weniger, D. (2006). Aphasie. In H.-O. Karnath, W. Hartje, & W. Ziegler (Eds.), Kognitive
Neurologie (pp. 4864). Stuttgart: Thieme.
White, S. J., Johnson, R. L., Liversedge, S. P., & Rayner, K. (2008). Eye movements
when reading transposed text: The importance of word-beginning letters.
Journal of Experimental Psychology: Human Perception and Performance, 34(5),
12611276.
Wolpert, I. (1924). Die Simultanagnosie. Strung der Gesamtauffassung. Zeitschrift
fr die gesamte Neurologie und Psychiatrie, 93, 397415.