Accepted Manuscript

The neural bases of the pseudohomophone effect: phonological constraints onlexico-semantic access in reading

Mario Braun, Florian Hutzler, Thomas F. Münte, Michael Rotte, MichaelDambacher, Fabio Richlan, Arthur M. Jacobs

PII: S0306-4522(15)00262-6DOI: http://dx.doi.org/10.1016/j.neuroscience.2015.03.035Reference: NSC 16145

To appear in: Neuroscience

Accepted Date: 17 March 2015

Please cite this article as: M. Braun, F. Hutzler, T.F. Münte, M. Rotte, M. Dambacher, F. Richlan, A.M. Jacobs,The neural bases of the pseudohomophone effect: phonological constraints on lexico-semantic access in reading,Neuroscience (2015), doi: http://dx.doi.org/10.1016/j.neuroscience.2015.03.035

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customerswe are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, andreview of the resulting proof before it is published in its final form. Please note that during the production processerrors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

1

The neural bases of the pseudohomophone effect:

phonological constraints on lexico-semantic access

in reading.

Mario Braun1, Florian Hutzler

1, Thomas F. Münte

3, Michael Rotte

4, Michael

Dambacher2,5

, Fabio Richlan1 and Arthur M. Jacobs

2,6

1Centre for Cognitive Neuroscience, University of Salzburg, Austria

2Institute of Psychology, Freie Universität Berlin, Germany

3Dept. of Neurology and Institute of Psychology II, University of Lübeck, Germany

4Novartis Pharma, Basle, Switzerland

5Dept. of Psychology, University of Konstanz, Germany

6Dahlem Institute for Neuroimaging of Emotion, Berlin, Germany

Keywords: Models of visual word recognition, fMRI, lexical decision, phonological mediation,

lexico-semantic access, spelling-check.

Address correspondence to:

Mario Braun

Centre for Cognitive Neuroscience

Universität Salzburg

Hellbrunner Straße 34

5020 Salzburg Austria

phone: ++43.662.8044.5176

fax: ++43.662.8044.5126

email: mario.braun@sbg.ac.at

2

Abstract

We investigated phonological processing in normal readers to answer the question to what

extent phonological recoding is active during silent reading and if or how it guides lexico-semantic

access. We addressed this issue by looking at pseudohomophone and baseword frequency

effects in lexical decisions with event-related functional magnetic resonance imaging (fMRI). The

results revealed greater activation in response to pseudohomophonesthan for well controlled

pseudowords in the left inferior/superior frontal and middle temporal cortex, left insula, and left

superior parietal lobule. Furthermore, we observed a baseword frequency effect for

pseudohomophones (e.g., FEAL) but not for pseudowords (e.g., FEEP). This baseword frequency

effect was qualified by activation differences in bilateral angular and left supramarginal, and

bilateral middle temporal gyri for pseudohomophones with low- compared to high-frequency

basewords. We propose that lexical decisions to pseudohomophones involves phonology-driven

lexico-semantic activation of their basewords and that this is converging neuroimaging evidence

for automatically activated phonological representations during silent reading in experienced

readers.

3

Introduction

Almost everyone who learns to read is already able to speak and to understand spoken language.

The first language code with which we are confronted is of phonological nature. In the process of

learning to read, written letters are mapped to their phonological representations by a process of

phonological recoding. In reading aloud phonological recoding is a necessary condition, since the

orthographic code has to be translated into its phonological counterpart to allow for speech

output. Most researchers agree that phonological recoding takes place in both, spoken and

written language processing, but it is still a matter of debate to what extent it is also active in

experienced readers (e.g., Ziegler et al, 2013; Hawelka et al., 2013; Frisson et al., 2014) and

whether phonological recoding is automatically activated and guides access to lexico-semantic

representations. In principle, phonological recoding could be abandoned after stable

representations of visual word forms have developed and a skilled reader could determine the

meaning of words directly from knowledge of its spelling. However, there is behavioral,

electrophysiological, and neuroimaging evidence that the first language code with which we are

confronted remains active in silent reading (e.g., Van Orden, 1987; Ziegler et al., 2000; Ziegler et

al., 2001a; Braun et al., 2009; Briesemeister et al., 2009; Wheat et al., 2010; Pammer et al., 2004;

Perrone-Bertolotti et al., 2012; Jäncke and Shah, 2004; Alexander and Nygaard, 2008; Yao, et al.,

2011).

Direct Access vs. Phonological Mediation

Two main views exist on how we gain access to the meaning of words in silent reading. The direct

access hypothesis states that there is a direct pathway from orthography to meaning (e.g.,

Seidenberg, 1985), and that if phonological recoding is performed in silent reading it must be

4

post-lexical (i.e., after meaning is computed). In contrast, the phonological mediation hypothesis

(e.g., Van Orden, 1987) claims that phonological activation is a necessary condition to gain access

to word meaning. Accordingly, phonology is assumed to be activated automatically prior to

lexical access of word meaning and thus taking place early during visual word recognition (for

reviews see Berent and Perfetti, 1995; Frost, 1998).

Pseudohomophone Effect

Evidence for phonological mediation is provided by the pseudohomophone effect (Rubenstein et

al., 1971; McCann et al., 1988; Seidenberg et al., 1996) which describes the fact that a nonword

(e.g., FEAL) which shares phonology but not orthography with a word (e.g., FEEL) leads to slower

RTs compared to another nonword (e.g., FEEP). "FEEP" differs in phonology but its orthography is

equally similar to the real word "FEEL". The pseudohomophone effect is used as a marker of

phonological activation in reading development (e.g., Goswami et al., 2001) and provides major

constraints for computational models of visual word recognition (Jacobs and Grainger, 1994;

Jacobs et al., 1998; Seidenberg et al., 1996; Ziegler and Jacobs, 1995; Ziegler et al., 2001b). The

standard explanation for the pseudohomophone effect is that a given pseudohomophone

contacts the lexical representation of its phonologically identical baseword in a mental lexicon,

which in turn activates semantic information associated with that representation. Thus, the

'phonological lexicon' and co-activated semantics of the baseword signal the presence of a word,

whereas the 'orthographic lexicon' signals its absence. In lexical decision, it is assumed that

resolving this conflict takes time and therefore participants show longer response times for

pseudohomophones compared to pseudowords (Jacobs et al., 1998; Ziegler et al., 2001b).

Furthermore, baseword frequency effects provide evidence for a lexical locus of the

5

pseudohomophone effect (see below).

Baseword Frequency

The baseword frequency effect states that pseudohomophones with low frequency basewords

lead to slower response times than those with high baseword frequencies in lexical decision (e.g.,

Ziegler et al., 2001b; Reynolds and Besner, 2005), but also in semantic categorization and

proofreading (Ziegler et al., 1997). A baseword frequency effect for pseudohomophones could be

interpreted as reflecting the activation of frequency sensitive representations of the basewords

through the pseudohomophone’s identical phonology suggesting phonology driven lexical access

(e.g., Forster and Chambers, 1973).

Models of Visual Word Recognition

The first computational models simulating the pseudohomophone effect, were the multiple read-

out model including phonology (MROM-P; Jacobs et al., 1998) and the dual-route cascaded model

(DRC; Coltheart et al., 2001). Both explain the pseudohomophone effect in lexical decision in

terms of higher summed global lexical activity in a phonological lexicon which prolongs a temporal

deadline generating longer response times for pseudohomophones compared to pseudowords.

Both make the same predictions concerning the baseword frequency effect. Pseudohomophones

with low frequency basewords should lead to faster responses because pseudohomophones with

high frequency basewords should elicit higher activation leading to stronger word present signals

resulting in more errors and slower responses in lexical decisions. However, the empirical findings

show reversed effects: responses to pseudohomophones with high frequency basewords are

faster than to those with low frequency basewords (e.g., Ziegler et al., 2001). In the light of this

finding a spelling-check was proposed to account for the baseword frequency effect (e.g., Paap et

6

al., 1982). The idea is that pseudohomophones activate high frequency basewords faster, because

these have higher resting levels and are therefore accessed more easily in an orthographic lexicon

allowing for a faster spelling-check. Unfortuneately, the models mostly do not directly speak to

brain regions assumed to process certain kinds of information or to the processing strength of

specific stimuli in specific tasks. Nevertheless, recent neuroimaging research has made progress to

more directly link the output of computational models to brain activation (Braun et al., 2006; Levy

et al., 2009; Taylor et al., 2013; Graves et al., 2014; Hofmann and Jacobs, 2014).

Reading Network

There is neuroimaging evidence for an indirect (sublexical) and a direct (lexico-semantic) route for

lexical access in line with the dual-route account of the DRC. The former is thought to correspond

to a dorsal occipito-parietal to inferior-frontal circuit (Jobard et al., 2003; Price, 2012). The dorsal

circuit is believed to be involved in the conversion of orthography to phonology and to comprise

superior temporal and inferior parietal regions with the supramarginal gyrus (SMG) and the

opercular part of the IFG. The direct route comprises a ventral occipito-temporal circuit involving

the ventral occipito-temporal cortex (vOT), the inferior temporal (ITG; Kronbichler et al., 2004) and

posterior middle temporal gyrus (pMTG; Noonan et al., 2013), as well as the triangular part of the

IFG and is assumed to be involved in lexico-semantic processing. Furthermore, parts of the angular

gyrus (AG) and the SMG seem to be involved in the processing of semantic information (e.g.,

Binder et al., 2003; Binder et al., 2005; Binder and Desai, 2011; Binder et al., 2009). But these

results do not directly speak to the question about the fate of phonology after stable sound-to-

letter correspondences have been established as in experienced readers. Below we report brain

regions involved in phonological and lexico-semantic processing. We expect activity in these

7

regions in case pseudohomophones’ phonology activate their basewords and in turn allow for

lexico-semantic access.

Phonological Processing

The specific regions believed to signal phonological processing comprise the left IFG (Fiebach et

al., 2002; Ischebeck et al., 2004; Heim et al., 2009; Heim et al., 2013), bilateral insula (Mechelli et

al., 2007; Borowsky et al., 2006), the STG (Booth et al., 2002; Mesulam, 1990; Rumsey et al.,

1997), the SMG and AG (Joubert et al., 2004; Binder et al., 2003; Paulesu et al., 1993; Kronbichler

et al., 2007; Ischebeck et al., 2004; Church et al., 2008), and the supplementary motor area

(SMA; Carreiras et al., 2006). Activation of the left IFG and the insula has been proposed to serve

grapheme-phoneme conversion (pars opercularis) and/or processing for lexico-semantic access

(i.e., pars triangularis and orbitalis; Fiebach et al., 2002; Poldrack et al., 1999). Such functional

roles are also assigned to parts of the inferior parietal lobule (IPL; e.g., Booth et al., 2002). In

their meta-analysis Jobard et al. (2003) suggested that phonological processing involves left

lateralized brain structures such as superior temporal areas, the SMG and the opercular part of

the left IFG. Jobard et al. (2003) identified three clusters of which two were located in the

anterior and posterior part of the STG and the third in the MTG. Furthermore, Bitan et al. (2005)

reported activation in left MTG for rhyming, but not for spelling judgments (see also Indefrey

and Levelt, 2004; Graves et al., 2010; Graves et al., 2014; Simos et al., 2009; Newman and

Joanisse, 2011 for further evidence of phonological processing in MTG).

Semantics

The meta-analysis of 120 functional neuroimaging studies of Binder et al. (2009) reported

seven main regions for semantic processing including the posterior IPL (AG and SMG), the

8

lateral temporal cortex (MTG and ITG), and the left IFG. Furthermore, activation in pars

orbitalis of the left ventral IFG was linked to semantic processing. Devlin et al., (2003) applied

transcranial magnetic stimulation (TMS) to the anterior IFG during semantic and perceptual

decision tasks. TMS slowed participants’ reaction time on the semantic but not on the control

task, supporting a causal role for the anterior IFG in semantic processing (see also Poldrack et

al., 1999; Wagner et al., 2001). A large body of evidence links MTG activation to semantic

processing (e.g., Gold et al., 2005; Noonan et al., 2013; Newman and Joanisse, 2011; Indefrey

and Levelt, 2004; Booth et al., 2007; Price et al., 1997; Vigneau et al., 2006; Richlan et al., 2009;

Whitney et al., 2011). Thus, research mainly suggests a lexico-semantic reading pathway that

integrates the left vOT with the left ventral IFG, and a probably non-semantic, phonological

decoding pathway linking the superior temporal and inferior parietal cortices to the dorsal

precentral gyrus. Until now it remains unclear how these pathways overlap or dissociate in the

reading system (Price, 2012). This is especially true for the MTG where evidence suggests that

activation either signals phonological processing involving the dorsal path (Graves et al., 2010;

Indefrey and Levelt, 2004; Graves et al., 2014; Simos et al., 2009; Newman and Joanisse, 2011)

or semantic processing involving the ventral path (Gold et al., 2005; Noonan et al., 2013;

Newman and Joanisse, 2011; Indefrey and Levelt, 2004; Booth et al., 2007; Price et al., 1997;

Vigneau et al., 2006; Richlan et al., 2009; Whitney et al., 2011).

Present Study

The aim of the present study was to investigate the pseudohomophone and the baseword

frequency effect in lexical decision by means of fMRI. The demonstration of pseudohomophone

and baseword frequency effects in lexical decision, a task which does not necessarily require

9

phonological processing (Grainger and Jacobs, 1996; Seidenberg, 1985) would provide evidence

for an automatic activation of phonological representations which constrains access to meaning

(Van Orden, 1987; Edwards et al., 2005). We assume that lexical decisions to pseudohomophones

involve the activation of the baseword’s phonological code which signals the presence of a word

while the orthographic code signals its absence. This should result in greater activation in left

hemispheric regions involved in the computation of these codes such as the insula, pars

opercularis of the IFG, the SMA, the STG or SMG in case of phonological processing, and vOT and

parts of the IFG in case of orthographic (Cohen et al., 2000), and pars orbitalis and triangularis of

the IFG and inferior parietal and middle temporal areas for semantic processing (Gold et al., 2005;

Noonan et al., 2013).

Paralleling the logic used in behavioral studies, brain activation should be modulated by the

pseudohomophone – pseudoword contrast. In addition, if phonological processing constrains

lexico-semantic access at the whole-word level, neural activation in response to

pseudohomophones should be influenced by their baseword frequency. In particular, a baseword

frequency effect would support models of visual word recognition implementing whole-word

phonological as well as frequency sensitive representations. We thus expect that this kind of

processing activates areas believed to be involved in lexico-semantic processing like the AG and

MTG, or the pars triangularis and orbitalis of the left IFG (Bitan et al., 2005; Gold et al., 2005; Lau

et al., 2008; Binder et al., 2009).

While the basis of the baseword frequency effect in lexical decision is still under debate, the most

parsimonious explanation seems the assumption of a spelling-check which is probably more

effortful for low frequency basewords (Ziegler et al., 2001b) and might be due to lower resting

levels of the orthographic codes (Grainger and Jacobs, 1996). A harder spelling-check for low

10

frequency items could lead to greater activation in regions concerned with orthographic

processing like the IFG or the vOT (Purcell et al., 2011). We carefully controlled our

pseudohomophones and pseudowords for a number of lexical and sublexical measures known to

influence visual word processing (see Table 1). Pseudohomophones and pseudowords did not

differ in orthographic similarity from each other nor to their identical basewords. Thus, if

pseudohomophones and pseudowords differed in response times and BOLD responses the effect

could not be due to an orthographic similarity confound. Furthermore, if participants in lexical

decision made only use of the direct ortographic-semantic route both, pseudohomophones and

pseudowords should activate the meaning of their basewords to the same extent, leading to

similar brain activation in the above mentioned 'lexico-semantic' networks.

11

Experimental Procedures

All procedures were cleared by the local ethical review board.

Participants

Fourteen right-handed students (three men, mean age 23 years, range 20 to 30) participated in

the study. All participants were native German speakers, had normal or corrected to normal

vision, were free of any current or past neuropsychiatric disorders and did not take psychoactive

medication. Data from all participants were used in the analyses. Participants were compensated

for their time with 24 Euro.

Stimuli

The stimulus set contained 480 stimuli (240 words and 240 nonwords). Of the 240 word stimuli

120 served as fillers. Of the 240 non-words half were pseudohomophones and half were

pseudowords. Pseudohomophones and pseudowords were derived from the same base words by

replacement of one letter. Where possible, the letter was changed at the same position and a

vowel was replaced by another vowel and a consonant by another consonant.

Pseudohomophones had the same phonology, but differed in spelling from their basewords.

Pseudowords differed in spelling and also in phonology from their base words. All pseudowords

were pronounceable, for example, ’SAHL’ is a pseudohomophone derived from the baseword

’SAAL’ (hall) and ’SARL’ is the corresponding pseudoword. Of the pseudohomophones and the

pseudowords one third had three, one third had four and one third had five letters. Half of the

pseudohomophones and pseudowords of each length were derived from high frequency

basewords (more than 20 occurrences per million, mean 854.3). The other half of the

pseudohomophones and pseudowords had low frequency basewords (less than 20 occurrences

12

per million, mean 6.8). Frequency estimates were taken from the CELEX database (Baayen et al.,

1993). To rule out orthographic similarity as a potential source of the pseudohomophone effect,

pseudohomophones and pseudowords were matched according to the strong criteria put forward

by Martin (1982). Pseudohomophones and pseudowords were controlled for number of letters

(let, F(3,236) = 0, p = 1, bigram frequency (BF, F(3,236) = 1.705, p = .167), number of bigram

neighbors (BN, F(2,236) = 0.198, p = .898), number of orthographic neighbors (N, F(3,236) = .086, p

= .968), number of higher frequency numbers (HFN, F(3,236) = 2.121, p = .098) and summed

frequency of orthographic neighbors (FN, F(3,236) = 0.666, p = .574). Thus, none of these variables

differed for items of low and high frequency within each nonword condition and also not between

conditions which should rule out sublexical and global lexical activity as a potential source of

differences between pseudohomophones and pseudowords. Table 1 shows the controlled

variables for pseudohomophones and pseudowords and the statistics for words. Of the 120 word

stimuli, one third had three, one third had four, and one third had five letters. Half of the word

stimuli of each word length were of high frequency (more than 11 occurrences per million, mean

421.2) and the other half were of low frequency (less than 11 occurrences per million, mean 3.3).

Words of high and low frequency were matched on bigram count (BC, F(1,118) = 0.985, p = .323),

number of letters (let, F(1,118) = 0, p = 1), summed frequency of the neighbors (FN, F(1,118) =

0.714, p = .400) and summed frequency of higher frequency neighbors (FHFN, F(1,118) = 0.649, p =

.422).

<< insert Table 1 and 2 about here >>

Procedure

Before participants were placed in the scanner they were given written instructions and were

13

further orally informed about the experimental procedure. They were informed that they would

see letter strings, some of which were German words and some were nonwords. Participants

were instructed to indicate via button press as rapidly as possible, but not to the expense of

accuracy, whether the stimulus was a German word or not. Two magnet-compatible response

boxes (one in each hand) were used. The response hands were counterbalanced across

participants.

Participants performed 15 practice trials to familiarize themselves with the task. The experiment

was divided into three runs. The experimental trials were presented in randomized order for

each participant. Each stimulus was presented for 1000 ms with an inter-stimulus interval of

2000 ms during which a fixation cross was presented – resulting in a total trial duration of 3000

ms. Responses with reaction times below 200 ms and above 2000 ms were excluded (0.39%).

The stimuli were displayed in yellow on a grey background using upper case letters set in Courier

48pt font. Visual images were back-projected onto a screen by an LED-projector and participants

viewed the images through a mirror on the head coil. The whole experiment took about 40

minutes. The experiment was implemented using Presentation experimental software

(Neurobehavioral Systems Inc, Albany, CA).

Image Acquisition

Functional and structural imaging was performed with a General Electric 1.5 Tesla Signa scanner

(General Electric, Fairfield, CT, USA) using a standard head-coil. Conventional high-resolution

structural images (rf-spoiled GRASS sequence, 60 slice sagittal, 2.8 mm thickness) were followed

by three runs with 308 volumes each of functional images sensitive to BOLD contrast acquired

with a T2* weighted gradient echo EPI sequence (TR = 2000ms, TE = 29 ms, flip angle = 80°,

14

number of slices = 34, slice thickness = 3.5 mm, 64 × 64 matrix, FOV 224 mm). Low frequency

noise was removed with a high-pass filter (128s).

fMRI Analysis

Data were analyzed using the general linear model in SPM8 (Statistical Parametric Mapping;

Wellcome Department of Cognitive Neurology, London, UK: http://www.fil.ion.ucl.ac.uk/spm).

The first three scans of each run were discarded to avoid magnetic saturation effects. Firstly, data

were pre-processed for each subject: after separate realignment procedures for the three runs, a

co-registration step was performed to adjust the images of each subjects to the first invidiual run.

Parameters for spatial normalization were determined by normalizing the mean realigned image

to a standard EPI template. Subsequently, functional images were resampled to isotropic 3 x 3 x 3

mm voxels and spatially smoothed with an 8 mm full-width at half-maximum isotropic Gaussian

kernel. All experimental conditions were modeled by a canonical hemodynamic response

function and its temporal time derivatives. Errors to pseudohomophones and pseudowords were

low (3%; corresponding to an average of 7 items per subject) and were therefore not modeled

separately. Statistical analysis was performed in a two-stage mixed effects model. On the subject-

specific 1st level model conditions of interest were contrasted against the fixation baseline. These

subject-specific contrast images were used for the 2nd level group analysis. Direct contrasts

between pseudohomophones and pseudowords with high and low baseword frequency were

calculated with 2 x 2 repeated measures ANOVA and paired t-tests. Stereotaxic coordinates for

voxels with maximal z-values within activation clusters are reported in the MNI coordinate

system.

Results

15

Behavioral Results

Response times to words were about 150 ms faster than to nonwords. Percentage of errors to

words and nonwords were low and nearly the same (2.89% vs. 3.05%) and therefore not further

analyzed. Items with response times above 2000ms were excluded from the analysis. Words were

responded to fastest followed by responses to pseudowords and pseudohomophones. Response

times were slower to low than to high frequency items in each category. Table 3 shows the mean

response times and error rates for the different stimuli categories. For response times, separate

2x2 ANOVAs with homophony (pseudohomophones vs. pseudowords) and frequency (high vs.

low) as factors were performed for subjects (F1) and items (F2). In the repeated measures

ANOVA by subjects homophony and frequency were treated as within-subject factors. The

analyses revealed effects of homophony (F1(1,13) = 18.85, p < .001; F2(1,236) = 11.55, p < .001)

and frequency (F1(1,13) = 7.89, p = .015; F2(1,236) = 4.69, p = .031) and also a significant

interaction (F1(1,13) = 8.40, p = .012; F2(1,236) = 3.89, p = .049). Response times were slower in

response to pseudohomophones derived from low compared to high frequency basewords (t = -

3.610, df = 13, p = .003). In contrast, response times to pseudowords did not differ with regard to

frequency (t = -0.559, df = 13, p = .586). Additional contrasts for pseudohomophones and

pseudowords separately calculated for items with high and low frequency basewords showed a

significant difference for low (t = 5.187, df = 13, p < .001) and a marginally significant difference

for high frequency items (t = 1.873, df =13, p = .083). Thus, the behavioral analysis revealed a

pseudohomophone effect and a baseword frequency effect for pseudohomophones but not for

pseudowords. Both effects seem to rely stronger on the processing of pseudohomophones with

low frequency basewords.

16

<< insert Table 3 about here >>

Imaging Results

First, images for the contrasts of interests were calculated separately for each subject. In the next

step pseudohomophones and pseudowords were separately contrasted with the fixation

baseline. Pseudohomophones elicited greater activation compared to fixation in posterior

occipital, occipito-temporal, parietal, and temporo-parietal regions. There were also extended

activations in left inferior frontal regions including the insula, pars opercularis and triangularis.

Furthermore, pre- and postcentral regions and posterior cingulate and right SPL/AG were more

activated by pseudohomophones compared to fixation baseline. Greater activation for

pseudowords compared to fixation were found in nearly the same regions as for

pseudohomophones. In contrast to pseudohomophones, pseudowords showed less extended

inferior frontal and inferior and superior parietal activation (see Figure 1). Since the aim of the

study was to investigate the neural basis of pseudohomophone and baseword frequency effects

the following analyses directly contrasted the activity in response to pseudohomophones vs.

pseudowords and items with high vs. low baseword frequencies.

<< insert Figure 1 about here>>

Pseudohomophone Effect

The whole-brain analysis revealed that no region showed greater activation for pseudowords than

for pseudohomophones. In contrast, activation to pseudohomophones was greater than for

pseudowords in the left hemisphere comprising the insula and the inferior orbito-frontal cortex

including pars orbitalis and pars triangularis, pre-motor, middle/superior frontal gyrus,

inferior/superior parietal regions and in inferior/middle temporal gyri on the right (see Table 4).

17

Activation in the insula und pars orbitalis showed the strongest activation in response to low

frequency pseudohomophones followed by high frequency pseudohomophones. Pseudowords

produced only little activation or deactivation. In the superior frontal gyrus all stimulus conditions

showed greater activation for low frequency items with low frequency pseudohomophones

eliciting the strongest activation. The separately performed 2x2 ANOVAs with the beta estimates

of the peak values with homophony (pseudohomophones, pseudowords) and baseword frequency

(high, low) as within-subject factors for regions showing greater activation for pseudohomophones

than for pseudowords revealed main effects of homophony (insula: F(1,13) = 17.85, p < .001; pars

orbitalis: F(1,13) = 11.70, p = .005; superior frontal gyrus (SFG): F(1,13) = 17.40, p = .001; MTG:

F(1,13) = 16.27, p = .001; SPL: F(1,13) = 6.96, p = .021), no main effect of frequency, and no

interactions. Thus, regions showing stronger activation in response to pseudohomophones did not

show significant effects of baseword frequency (see Figure 2 and Table 4).

<< insert Figure 2 and Table 4 about here >>

Baseword Frequency Effect

To see if there were brain regions showing differential activation for items of high and low

baseword frequency, we performed a whole brain analysis for the contrasts of high vs. low

frequency pseudohomophones and high vs. low frequency pseudowords. There were no regions

showing an effect of baseword frequency for pseudowords. In contrast, the results showed an

effect of baseword frequency for pseudohomophones. Pseudohomophones with high frequency

basewords showed the highest and pseudohomophones with low frequency basewords showed

always the lowest activation. The four clusters showing a difference in activation for

pseudohomophones with high compared to pseudohomophones with low frequency

18

basewords comprised the left inferior and superior lateral occipital cortex including the AG, parts

of the left MTG and STG and the posterior SMG. Furthermore, the right AG and SMG showed

greater activation for high frequency pseudohomophones compared to pseudohomophones with

low frequency basewords. Also the left and right MTG and STG showed greater activation for

pseudohomophones with high compared to those with low frequency basewords (see Figure 3).

To further show that the baseword frequency effect was only obtained for pseudohomophones and not

for pseudowords we additionally calculated the whole brain analyses for the contrasts of high vs.

low frequency pseudohomophones by exclusively masking (p < .05, uncorrected) it with the

contrast of high vs. low frequency pseudowords. The results revealed a reliable baseword frequency

effect for pseudohomophones but not for pseudowords (see Table 5).

<<insert Figure 3 and Table 5 about here >>

The separately performed 2x2 ANOVAs with the beta estimates of the peak values with

homophony (pseudohomophones, pseudowords) and baseword frequency (high, low) as within-

subject factors for regions showing greater activation for pseudohomophones with high

compared to pseudohomophones with low frequency basewords revealed main effects of

frequency (left AG: F(1,13) = 19.63, p < .001; left MTG: F(1,13) = 11.01, p < .001; right AG: F(1,13)

= 6.45, p = .025; right MTG: F(1,13) = 14.67, p < .001) and interactions of frequency and

homophony (left AG: F(1,13) = 9.19, p = .001; left MTG: F(1,13) = 7.54, p = .017; right AG: F(1,13)

= 8.41, p = .012; right MTG: F(1,13) = 11.30, p < .001) and no main effect of homophony. To

further investigate the interaction paired t-tests for the comparisons of high and low frequency

pseudohomomphones and high and low frequency pseudowords were calculated. The results

showed that the effect of baseword frequency was significant for pseudohomophones (left AG: t

19

= 6.921, df = 13, p < .001; rAG: t = 4.402, df = 13, p < .001; left MTG: t = 3.613, df = 13, p = .003;

right MTG: t = 4.799, df = 13, p < .001) but not for pseudowords (left AG: t = 1.003, df = 13, p =

.334; right AG: t = -0.402, df = 13, p = .694; right MTG: t = 1.586, df = 13, p = .137; left MTG: t =

1.437, df = 13, p = .174).

Discussion

In the introduction, we outlined the empirical criteria for evidence of phonology triggered

lexico-semantic access in silent reading i.e. the pseudohomophone effect (differences in

response times and brain activation for pseudohomophones and pseudowords derived from

the same basewords), and the baseword frequency effect (differences in response times and

brain activation for pseudohomophones derived from basewords of different lexical

frequency). Our behavioral and neuroimaging results unequivocally demonstrate both effects

and thus provide evidence for phonological mediation in visual word recognition (e.g.,

Coltheart et al., 2001; Frost, 1998; Van Orden, 1987). Responses to pseudohomophones were

42 ms slower than to pseudowords. Furthermore, pseudohomophones showed a baseword

frequency effect with slower responses (50 ms) to pseudohomophones with low than with high

frequency basewords. In contrast, there was no baseword frequency effect for pseudowords.

Our study thus is the first to demonstrate both, a pseudohomophone and a baseword

frequency effect in the scanner.

Imaging: Pseudohomophone Effects

In the hemodynamic responses the pseudohomophone effect was evident in greater activation

for pseudohomophones compared to pseudowords in left hemispheric regions as the insula, pars

triangularis and orbitalis, SMA and superior frontal gyrus (SFG). Furthermore, a right

20

inferior/middle temporal cluster showed greater activation for pseudohomophones. Thus,

pseudohomophones showed activation in inferior and superior frontal regions involved in

phonological processing (e.g., Bokde et al., 2001; Jobard et al., 2003; Mechelli et al., 2007;

Cattinelli et al., 2013; Binder et al., 2009; Newman and Joanisse, 2011; Pillay et al., 2014) but also

in semantic processing (e.g., Fiebach et al., 2002; Ischebeck et al., 2004).

Imaging: Baseword Frequency Effects

Whereas pseudowords did not show baseword frequency effects for the imaging data, the

comparison of pseudohomophones with high and low frequency basewords revealed activation

differences in bilateral IPL including the AG, posterior SMG and in bilateral STG and MTG with

greater activation for pseudohomophones with low frequency basewords. Previous research

suggests that activation in MTG is involved in semantic processing (e.g., Gold et al., 2005; Noonan

et al., 2013; Newman and Joanisse, 2011; Indefrey and Levelt, 2004; Booth et al., 2007; Price et al.,

1997; Vigneau et al., 2006; Richlan et al., 2009; Whitney et al., 2011) but there is also evidence

suggesting a role in phonological processing (e.g., Indefrey and Levelt, 2004; Bitan et al., 2005;

Simos et al., 2009; Newman and Joanisse, 2011; Graves et al., 2014), or orthographic-phonological

mapping (e.g., Graves et al., 2010). Many studies report that left MTG is consistently more active

during the processing of words than during the processing of pseudowords (e.g., Binder et al.,

2003; Fiebach et al., 2002).

Interpretation of Effects

In the light of these findings we suggest that the greater activation for pseudohomophones in

comparison to pseudowords reflects stronger phonological and semantic activation of the

basewords. Since pseudohomophones and pseudowords are visually equally similar to their

21

basewords orthography cannot explain the obtained activation differences. Thus, information

signaling the presence of a word is low in case of a pseudoword (some orthographic similarity) and

high in case of a pseudohomophone (same phonology and semantics and some orthographic

similarity) and must rely on phonological features. Greater activation for identifying

pseudohomophones reflects greater access to and mapping with stored representations.

Activation in pars orbitalis and triangularis probably signals processing for meaning, and activation

in SMA (Purcell et al., 2011) and the insula (Borowsky et al., 2006) could signal processes of

orthographic-phonological mapping. Activation differences for pseudohomophones with high

compared to those with low frequency in the IPL, STG and MTG could signal faster and easier

access to stored lexico-semantic representations for pseudohomophones with high frequency

basewords, probably due to higher resting levels. In contrast, pseudowords did not activate their

high and low frequency basewords differentially, probably because they fail to activate whole-

word phonology and semantic representations.

Spelling-Check

The aim of our study was to investigate the neural bases of pseudohomophone and baseword

frequency effects in lexical decision. Current models of visual word recognition either do not

predict both effects or predict the wrong direction (longer RTs in response to pseudohomophones

with high compared to low frequency basewords). To account for this finding it was suggested that

making lexical decisions to pseudohomophones involves orthographic processing in the form of a

spelling-check (Paap et al., 1982; Ziegler et al., 2001). Such a spelling-check would, when

successfully performed, allow for a correct nonword decision for pseudohomophones and

pseudowords. Based on the current results we suggest that the pseudohomophone and the

baseword frequency effect originate from different mechanisms and to rely on different brain

22

regions. The pseudohomophone effect is probably driven by a spelling-check involving inferior and

superior frontal and superior parietal brain regions. Activation in the SPL is shown to be associated

with mental imagery (Ganis et al., 2004), visual attention (Wojciulik and Kanwisher, 1999; Simon et

al., 2004; Gottlieb, 2007), and the spelling of words (Bitan et al., 2005, 2007; Purcell et al., 2011).

Such a spelling-check probably operates in a serial fashion (e.g., Cohen et al., 2008; Gaillard et al.,

2006). In contrast, the baseword frequency effect is signaled by activation differences in inferior

parietal and middle temporal regions, probably driven by the need for greater allocation of

attention and the suppression of ongoing processing (Binder et al., 2003) necessary to process the

phonological (Indefrey and Levelt, 2004; Bitan et al., 2005; Simos et al., 2009; Newman and

Joanisse, 2011; Graves et al., 2014) and semantic (Gold et al., 2005; Noonan et al., 2013; Newman

and Joanisse, 2011; Indefrey and Levelt, 2004; Booth et al., 2007; Price et al., 1997; Vigneau et al.,

2006; Richlan et al., 2009; Whitney et al., 2011; Mechelli et al., 2007) properties of

pseudohomophones with high and low frequency basewords. This interpretation is in line with

results showing that greater attention to task demands and higher short-term memory load is

associated with reduced BOLD signals in right AG (McKiernan et al., 2003) and right temporal

parietal junction (Todd et al., 2005).

Model - Pathways

As outlined in the introduction, current models of visual word recognition cannot fully account

for the pseudohomophone and baseword frequency effects. Models with implemented

interactive activation processes (e.g., DRC and MROM-P), assume that higher activation in a

phonological lexicon leads to longer processing times for pseudohomophones with high

compared to those with low frequency basewords. This is because high frequency

pseudohomophones are believed to elicit stronger 'word present' signals and thus should show

23

greater activation in brain regions known to process phonological and/or lexcico-semantic

information. However, the behavioral and imaging results of the present study are against this

interpretation. Pseudohomophones with high compared to those with low frequency basewords

were rejected faster and showed less deactivation in bilateral MTG, AG and SMG.

Since pseudohomophones and pseudowords did not differ in orthographic similarity to each

other and also not to their same basewords we would have expected a baseword frequency

effect also for pseudowords if orthographic processing was the basis of the baseword frequency

effect (Newman and Joanisse, 2011). The fact that there was a baseword frequency effect only

for pseudohomophones argues against the activation of meaning via a direct orthographic route

as proposed by the triangle model.

Furthermore, the fact that pseudohomophones with high and low baseword frequencies activate

the dorsal visual pathway with AG, SMG, and STG (Sandak et al., 2004; McCandliss et al., 2003) but

also the MTG of the ventral visual pathway (Schurz et al., 2010; Jobard et al., 2003; Sandak et al.,

2004) is evidence against the proposal that pseudohomophones, like other nonwords are

exclusively processed by an indirect route and by converting sublexical phonology to sublexical

orthography as proposed by dual-route models of visual word recognition.

The current results thus speak for a close interaction of dorsal and ventral pathways in visual word

recognition in the case of pseudohomophones and therefore for models assuming parallel and

serial processing pathways as the DRC and the CDP+. For example, the CDP+ implements a parallel

lexical pathway and a serial grapheme–phoneme conversion pathway. When reading nonwords

the sublexical pathway converts a string of letters into graphemes by serial left-to-right processing,

which is probably associated with activation in dorsal-parietal regions and also used by a potential

24

spelling-check. But this serial mechanism does probably not account for the obtained baseword

frequency effect for pseudohomophones.

Recent research has shown that there are probably multiple routes used in skilled reading

depending on the task and the stimuli (Jobard et al., 2011; Richardson et al., 2011; Seghier et al.,

2012; Cummine et al., 2013; Graves et al., 2014). Richardson et al. (2011) identified three potential

processing streams from occipital to temporal cortical areas: a ventral lexico-semantic route, a

dorsal phonological route, and an intermediate route encompassing partially both ventral and

dorsal routes. The results of the DCM analysis were interpreted to show that connections between

the posterior inferior occipital gyrus (pIO) and the posterior superior temporal sulcus (pSTS) were

involved in linking orthography to phonology. Connections between anterior and posterior STS

were suggested to link phonology to semantics, and the ventral path from pIO over vOT to

anterior STS is suggested to link orthography to semantics. Activation in pSTS was interpreted to

signal phonological processing which could be influenced by early inferior occipital and late visual

processing (vOT). This was hypothesized to be consistent with the idea that orthography and

phonology are linked over different levels of processing, i.e. different grain sizes, as proposed by

Ziegler et al. (2001a) and Ziegler and Goswami (2005, 2006). Furthermore, information between

these areas are probably transmitted by the inferior longitudinal fasciculus which extends to the

ventral and dorsal path and hypothesized that the posterior MTG is probably involved as an

intermediate region which may relay information transferred from pIO to pSTS.

Extending the hypothesis of Richardson et al. (2011) we propose that the MTG may not only relay

information from early visual to phonological areas but also to be involved in the integration of

orthographic, phonological and semantic information. We suggest that the MTG could be a region

25

where multiple sources of information converge and are integrated into a coherent percept of the

input in support of access to stored lexico-semantic representations.

Conclusion

In sum, the present study used pseudohomophones and pseudowords in lexical decision to

investigate phonological processing in visual word recognition. Both, behavioral and neuroimaging

results revealed a pseudohomophone effect as well as a baseword frequency effect for

pseudohomophones but not for pseudowords. We interpret these results as evidence that reading

pseudohomophones in contrast to pseudowords activate whole-word representations of their

phonologically identical basewords triggering access at the lexico-semantic level. This conclusion is

supported by the obtained baseword frequency effect showing activation differences for

pseudohomophones with high and low frequency basewords. Both effects therefore support

models of visual word recognition, which implement frequency sensitive representations and

assume phonologically mediated access to semantic representations in visual word recognition.

We suggest that reading pseudohomophones comprises the same stages of processing as with

words: a prelexical analysis of the visual word form and early lexical processing in ventral occipito-

temporal areas; access to orthographic, phonological, and semantic features in MTG, STG, AG and

SMG; and a manipulation of these features in the IFG involving processes of orthographic-

phonological mapping, lexico-semantic search, retrieval and selection. Processing specific to

pseudohomophones and pseudowords presumably involves a spelling-check which is harder for

pseudohomophones and likely to be associated with activation in the SFG and the SPL.

26

Acknowledgments

Thomas F. Münte was supported by grants of the DFG and BMBF.

27

References

Alexander JD, Nygaard LC (2008), Reading voices and hearing text: talker-specific auditory imagery in

reading. J Exp Psychol Human 34:446–446.

Baayen RH, Piepenbrock R, van Rijn H (1993), The CELEX lexical database.

Berent I, Perfetti CA (1995), A rose is a REEZ: The two-cycles model of phonology assembly in reading

English. Psychol Rev 102:146.

Binder JR, Desai RH, Graves WW, Conant LL (2009), Where is the semantic system? A critical review

and meta-analysis of 120 functional neuroimaging studies. Cereb Cortex 19:2767–96.

Binder JR, McKiernan KA, Parsons ME, Westbury CF, Possing ET, Kaufman JN, Buchanan L (2003),

Neural Correlates of Lexical Access during Visual Word Recognition. J Cognitive Neurosci

15:372–393.

Binder JR, Westbury CF, McKiernan KA, Possing ET, Medler DA (2005), Distinct brain systems for

processing concrete and abstract concepts. J Cogn Neurosci 17:905–917.

Bitan T, Booth JR, Choy J, Burman DD, Gitelman DR, Mesulam M-M (2005), Shifts of effective

connectivity within a language network during rhyming and spelling. J Neurosci 25:5397–5403.

Bitan T, Cheon J, Lu D, Burman DD, Gitelman DR, Mesulam M-M, Booth JR (2007), Developmental

changes in activation and effective connectivity in phonological processing. Neuroimage

38:564–575.

Bokde AL, Tagamets MA, Friedman RB, Horwitz B (2001), Functional interactions of the inferior

frontal cortex during the processing of words and word-like stimuli. Neuron 30:00:00 609–17.

Booth JR, Bebko G, Burman DD, Bitan T (2007), Children with reading disorder show modality

independent brain abnormalities during semantic tasks. Neuropsychologia 45:775–83.

Booth JR, Burman DD, Meyer JR, Gitelman DR, Parrish TB, Mesulam MM (2002), Functional Anatomy

28

of Intra- and Cross-Modal Lexical Tasks. Neuroimage 16:7–22.

Borowsky R, Cummine J, Owen WJ, Friesen CK, Shih F, Sarty GE (2006), FMRI of ventral and dorsal

processing streams in basic reading processes: insular sensitivity to phonology. Brain Topogr

18:233–239.

Braun M, Hutzler F, Ziegler JC, Dambacher M, Jacobs AM (2009), Pseudohomophone effects provide

evidence of early lexico-phonological processing in visual word recognition. Hum Brain Mapp

30:1977–89.

Braun M, Jacobs AM, Hahne A, Ricker B, Hofmann M, Hutzler F (2006), Model-generated lexical

activity predicts graded ERP amplitudes in lexical decision. Brain Res 1073-1074:431–439.

Briesemeister BB, Hofmann, MJ, Tamm, S, Kuchinke, L, Braun, M, Jacobs, AM (2009), The

pseudohomophone effect: evidence for an orthography-phonology-conflict. Neurosci Lett

455:124–8.

Carreiras M, Mechelli A, Price CJ (2006), Effect of word and syllable frequency on activation during

lexical decision and reading aloud. Hum Brain Mapp 27:963–972.

Cattinelli I, Borghese, NA, Gallucci, M, Paulesu, E (2013), Reading the reading brain: a new meta-

analysis of functional imaging data on reading. J Neurolinguist 26:214–238.

Church JA, Coalson RS, Lugar HM, Petersen SE, Schlaggar BL (2008), A Developmental fMRI Study of

Reading and Repetition Reveals Changes in Phonological and Visual Mechanisms Over Age.

Cereb Cortex 18:2054–2065.

Cohen L, Dehaene S, Vinckier F, Jobert A, Montavont A (2008), Reading normal and degraded words:

Contribution of the dorsal and ventral visual pathways. Neuroimage 40:353–366.

Cohen L, Martinaud O, Lemer C, Lehericy S, Samson Y, Obadia M (2003), Visual word recognition in

the left and right hemispheres: Anatomical and functional correlates of peripheral alexias.

Cereb Cortex 13:1313–1333.

29

Coltheart M, Rastle K, Perry C, Langdon R, Ziegler JC (2001), DRC: A dual route cascaded model of

visual word recognition and reading aloud. Psychol Rev 108:204–256.

Cummine J, Gould L, Zhou C, Hrybouski S, Siddiqi Z, Chouinard B, Borowsky R (2013), Manipulating

instructions strategically affects reliance on the ventral-lexical reading stream: converging

evidence from neuroimaging and reaction time. Brain Lang 125:203–214.

Devlin JT, Matthews PM, Rushworth MF (2003), Semantic processing in the left inferior prefrontal

cortex: a combined functional magnetic resonance imaging and transcranial magnetic

stimulation study. J Cognitive Neurosci 15:71–84.

Edwards JD, Pexman PM, Goodyear BG, Chambers CG (2005), An fMRI investigation of strategies for

word recognition. Cognitive Brain Res 24:648–662.

Fiebach C, Friederici A, Müller K, von Cramon D (2002), fMRI evidence for dual routes to the mental

lexicon in visual word recognition. J Cognitive Neurosci 14:11–23.

Forster KI, Chambers SM (1973), Lexical access and naming time. Journal of Verbal Learning and

Verbal Behavior, 12:627–635.

Frisson S, Koole H, Hughes L, Olson A, Wheeldon L (2014), Competition between orthographically

and phonologically similar words during sentence reading: Evidence from eye movements. J

Mem Lang 73:148–173.

Frost R (1998), Towards a strong phonological theory of visual word recognition: true issues and

false trails. Psychol Bull 123:71–99.

Gaillard R, Naccache L, Pinel P, Clémenceau S, Volle E, Hasboun D (2006), Direct intracranial, FMRI,

and lesion evidence for the causal role of left inferotemporal cortex in reading. Neuron

50:191–204.

Ganis G, Thompson, WL, Kosslyn, SM (2004), Brain areas underlying visual mental imagery and visual

perception: an fMRI study. Cognitive Brain Res 20:226–241.

30

Gold BT, Balota DA, Kirchhoff BA, Buckner, RL (2005), Common and dissociable activation patterns

associated with controlled semantic and phonological processing: evidence from FMRI

adaptation. Cereb Cortex 15 1438–50

Goswami U, Ziegler JC, Dalton L, Schneider W (2001), Pseudohomophone effects and phonological

recoding procedures in reading development in English and German. J Mem Lang 45:648-664.

Gottlieb J (2007), From thought to action: the parietal cortex as a bridge between perception, action,

and cognition. Neuron 53:9–16.

Grainger J, Jacobs AM (1996), Orthographic processing in visual word recognition: A multiple read-

out model. Psychol Rev 103:518–565.

Graves WW, Binder JR, Desai RH, Humphries C, Stengel BC, Seidenberg MS (2014), Anatomy is

strategy: Skilled reading differences associated with structural connectivity differences in the

reading network. Brain Lang 133C: 1–13.

Graves WW, Desai R, Humphries C, Seidenberg MS, Binder JR (2010), Neural Systems for Reading

Aloud: A Multiparametric Approach. Cereb Cortex 20:1799–1815.

Hawelka S, Schuster S, Gagl B, Hutzler F (2013), Beyond single syllables: The effect of first syllable

frequency and orthographic similarity on eye movements during silent reading. Lang Cognitive

Proc 28:1134–1153.

Heim S, Eickhoff SB, Ischebeck AK, Friederici AD, Stephan KE, Amunts K (2009), Effective connectivity

of the left BA 44, BA 45, and inferior temporal gyrus during lexical and phonological decisions

identified with DCM. Hum Brain Mapp 30:392-402.

Heim S, Wehnelt A, Grande M, Huber W, Amunts K (2013), Effects of lexicality and word frequency

on brain activation in dyslexic readers. Brain Lang 125:194–202.

Hofmann MJ, Jacobs AM (2014), Interactive activation and competition models and semantic

context: From behavioral to brain data. Neurosci Biobehav R 46: 85-104.

31

Indefrey P, Levelt WJ (2004), The spatial and temporal signatures of word production components

Cognition 92:101-144.

Ischebeck A, Indefrey P, Usui N, Nose I, Hellwig F, Taira M (2004), Reading in a regular orthography:

an FMRI study investigating the role of visual familiarityJ Cogn Neurosci 16 727-41.

Jacobs AM, Grainger J (1994), Models of visual word Recognition - Sampling the state of the Art. J

Exp Psychol Human 20:1311–1334.

Jacobs AM, Rey A, Ziegler JC, Grainger J (1998), MROM-P: An interactive activation, multiple read-

out model of orthographicand phonological processes in visual word recognition. In: Local

Connectionists Approaches to Human Cognition (Grainger J, Jacobs AM, eds), pp 147–187.

Mahwah, New Jersey: Lawrence Erlbaum Associates.

Jäncke L, Shah NJ (2004), “Hearing” syllables by “seeing” visual stimuli. Eur J Neurosci 19:2603-2608.

Jobard G, Crivello F, Tzourio-Mazoyer N (2003), Evaluation of the dual route theory of reading: a

metanalysis of 35 neuroimaging studies. Neuroimage 20:693–712.

Jobard G, Vigneau M, Simon G, Tzourio-Mazoyer N (2011), The weight of skill: interindividual

variability of reading related brain activation patterns in fluent readers. J Neurolinguist

24:113–132.

Joubert SA, Beauregard M, Walter N, Bourgouin P, Beaudoin G, Leroux J-M, Lecours AR (2004),

Neural correlates of lexical and sublexical processes in reading. Brain Lang 89:9–20.

Kronbichler M, Bergmann JRV, Hutzler F, Staffen W, Mair A, Ladurner G (2007), Taxi vsTaksi: On

Orthographic Word Recognition in the Left Ventral Occipitotemporal Cortex. J Cognitive

Neurosci 19:1584-1594.

Kronbichler M, Hutzler F, Wimmer H, Mair A, Staffen W, Ladurner G (2004), The visual word form

area and the frequency with which words are encountered: evidence from a parametric fMRI

study. NeuroImage 21:946–53.

32

Lau EF, Phillips C, Poeppel D (2008), A cortical network for semantics: (de)constructing the N400. Nat

Rev Neurosci 9:920-933.

Levy J, Pernet C, Treserras S, Boulanouar K, Aubry F, Démonet J-F, Celsis P (2009), Testing for the

dual-route cascade reading model in the brain: an fMRI effective connectivity account of an

efficient reading style. PLoS One 4:e6675-e6675.

McCandliss B, Cohen L, Dehaene S (2003), The visual word form area: expertise for reading in the

fusiform gyrus. Trends Cogn Sci 7:293–299.

McCann RS, Besner D, Davelaar E (1988), Word recognition and identification: Do word-frequency

effects reflect lexical access? J Exp Psychol Human 14:693-706.

McKiernan KA, Kaufman JN, Kucera-Thompson J, Binder JR (2003), A parametric manipulation of

factors affecting task-induced deactivation in functional neuroimaging. J Cogn Neurosci

15:394–408.

Mechelli A, Josephs O, Ralph L, Matthew A, McClelland JL, Price CJ (2007), Dissociating stimulus-

driven semantic and phonological effect during reading and naming. Hum Brain Mapp 28:205–

217.

Mesulam MM (1990), Large-scale neurocognitive networks and distributed processing for attention,

language, and memory. Ann Neurol 28:597–613.

Newman RL, Joanisse MF (2011), Modulation of brain regions involved in word recognition by

homophonous stimuli: an fMRI study. Brain Res 1367:250–64.

Noonan KA, Jefferies E, Visser M, Lambon Ralph MA (2013), Going beyond inferior prefrontal

involvement in semantic control: evidence for the additional contribution of dorsal angular

gyrus and posterior middle temporal cortex. J Cognitive Neurosci 25:1824–1850.

Paap K, Newsome SL, McDonald JE, Schvaneveldt RW (1982), An activation-verification model for

letter and word recognition: The wordsuperiority effect. Psychol Rev 89:573–589.

33

Pammer K, Hansen PC, Kringelbach ML, Holliday I, Barnes G, Hillebrand A, Cornelissen PL (2004),

Visual word recognition: the first half second. Neuroimage 22:1819–1825.

Perrone-Bertolotti M, Kujala J, Vidal JR, Hamame CM, Ossandon T, Bertrand O, Lachaux J-P (2012),

How silent is silent reading? Intracerebral evidence for top-down activation of temporal voice

areas during reading. J Neurosci 32:17554–62

Pillay SB, Stengel BC, Humphries C, Book DS, Binder JR (2014), Cerebral localization of impaired

phonological retrieval during rhyme judgment. Ann Neurol 76:738-746.

Poldrack RA, Wagner AD, Prull MW, Desmond JE, Glover GH, Gabrieli JD (1999), Functional

specialization for semantic and phonological processing in the left inferior prefrontal cortex.

Neuroimage 10:15–35.

Price CJ (2012), A review and synthesis of the first 20 years of PET and fMRI studies of heard speech,

spoken language and reading. Neuroimage 62:816–47

Price CJ, Moore C, Humphreys G, Wise R (1997), Segregating semantic from phonological processes

during reading. J Cognitiv Neurosci 9:727–733.

Purcell JJ, Turkeltaub PE, Eden GF, Rapp B (2011), Examining the central and peripheral processes of

written word production through meta-analysis. Front Psychol 2:1-16.

Reynolds M, Besner D (2005), Basic processes in reading: A critical review of pseudohomophone

effects in reading aloud and a new computational account. Psychon B Rev 12:622–646.

Richardson FM, Seghier ML, Leff AP, Thomas MSC, Price CJ (2011), Multiple routes from occipital to

temporal cortices during reading. J Neurosci 31:8239–47

Richlan F, Kronbichler M, Wimmer H (2009), Functional abnormalities in the dyslexic brain: a

quantitative meta-analysis of neuroimaging studies. Hum Brain Mapp30:3299–308.

Rubenstein H, Lewis SS, Rubenstein M (1971), Homographic entries in the internal lexicon: Effects of

systemacity and relative frequency of meaningsJournal of Verbal Learning and Verbal Behavior

34

10 57–62.

Rumsey J, Horwitz B, Donohue B, Nace K, Maisog J, Andreason P (1997), Phonological and

orthographic components of word recognitionA PET-rCBF studyBrain 120 739–759.

Sandak R, Mencl WE, Frost SJ, Pugh KR (2004), The neurobiological basis of skilled and impaired

reading: Recent findings and new directionsScientific Studies of Reading 8 273–292.

Schurz M, Sturm D, Richlan F, Kronbichler M, Ladurner G, Wimmer H (2010), A dual-route

perspective on brain activation in response to visual words: evidence for a length by lexicality

interaction in the visual word form area (VWFA). NeuroImage 49 2649–61.

Seghier ML, Neufeld NH, Zeidman P, Leff AP, Mechelli A, Nagendran A, Price, CJ (2012), Reading

without the left ventral occipito-temporal cortex. Neuropsychologia 50:3621–3635.

Seidenberg MS (1985), The time cours of phonological code activation in two writing systems.

Cognition 19:1–30.

Seidenberg MS, Petersen A, McDonald MC, Plaut DC (1996), Pseudohomophone effects and models

of word recognition. J Exp Psychol Learn 22:48–62.

Simon O, Kherif F, Flandin G, Poline J-B, Rivière D, Mangin J-F, Dehaene S (2004), Automatized

clustering and functional geometry of human parietofrontal networks for language, space, and

number. Neuroimage 23:1192–1202.

Simos PG, Pugh K, Mencl E, Frost S, Fletcher JM, Sarkari S, Papanicolaou AC (2009), Temporal course

of word recognition in skilled readers: a magnetoencephalography study. Behav Brain Res 197

45–54.

Taylor JSH, Rastle K, Davis MH (2013), Can cognitive models explain brain activation during word and

pseudoword reading? A meta-analysis of 36 neuroimaging studies. Psychol Bull 139:766–91.

Todd JJ, Fougnie D, Marois R (2005), Visual short-term memory load suppresses temporo-parietal

junction activity and induces inattentional blindness. Psychol Sci 16:965–972.

35

Van Orden GC (1987), A ROWS is a ROSE: Spelling, sound, and reading. Mem Cognition 15: 181–198.

Vigneau M, Beaucousin V, Hervé PY, Duffau H, Crivello F, Houdé O, Tzourio-Mazoyer N (2006), Meta-

analyzing left hemisphere language areas: phonology, semantics, and sentence processing.

NeuroImage 30:1414–32.

Wagner AD, Paré-Blagoev EJ, Clark J, Poldrack RA (2001), Recovering meaning: left prefrontal cortex

guides controlled semantic retrieval. Neuron 31:329–338.

Wheat KL, Cornelissen PL, Frost SJ, Hansen PC (2010), During visual word recognition, phonology is

accessed within 100 ms and may be mediated by a speech production code: evidence from

magnetoencephalography. J Neurosci 30:5229–5233.

Whitney C, Kirk M, O’Sullivan J, Lambon Ralph MA, Jefferies E (2011), The neural organization of

semantic control: TMS evidence for a distributed network in left inferior frontal and posterior

middle temporal gyrus. Cereb Cortex 21:1066-1075.

Wojciulik E, Kanwisher N (1999), The generality of parietal involvement in visual attention. Neuron

23:747–764.

Yao B, Belin P, Scheepers C (2011), Silent reading of direct versus indirect speech activates voice-

selective areas in the auditory cortex. J Cogn Neurosci 23:3146–52.

Ziegler JC, Bertrand D, Lété B, Grainger J (2013), Orthographic and Phonological Contributions to

Reading Development: Tracking Developmental Trajectories Using Masked Priming. Dev

Psychol 50:1026–1026.

Ziegler JC, Besson M, Jacobs AM, Nazir T, Carr TH (1997), Word pseudoword, and nonword

processing: A multitask comparison using eventrelated potentials. J Cognitive Neurosci 9:758–

775.

Ziegler JC, Goswami U (2006), Becoming literate in different languages: similar problems, different

solutions. Developmental Sci 9:429–436.

36

Ziegler JC, Jacobs AM (1995), Phonological information provides early sources of constraint in the

processing of letter stringsJournal of Mem Lang 34:567–593.

Ziegler JC, Jacobs AM, Klueppel D (2001b), Pseudohomophone Effects in Lexical Decision: Still a

Challenge for Current Word Recognition Models. J Exp Psychol Human 27:547–599.

Ziegler JC, Perry C, Jacobs AM, Braun M (2001a), Identical words are not processed the same in

different languages: The grainsize hypothesis. Psychol Sci 12:379–384.

Ziegler JC, Tan LH, Perry C, Montant M (2000), Phonology matters: The phonological frequency effect

in written Chinese. Psychol Sci 11:234–238.

37

Tables

Table 1. Descriptive statistics for pseudohomophones, pseudowords and words

Frequency

low high mean

pseudohomophones

bwF 6.8(6.6) 854.3(3390) 430.5

BF 2623(5038) 3713(4419) 3157.5

BN 17 (11) 16 (9) 16.5

FN 3242(16083) 1231(3609) 2236.5

HFN 3.6(2.3) 3.4(2.5) 3.5

Letters 4(0.84) 4(0.82) 4

N 3.6(2.3) 3.4(2.5) 3.5

pseudowords

bwF 6.8(6.6) 854.3(3390) 430.5

BF 5150(15883) 8051(21930) 6600.5

BN 16(11) 15(10) 15.5

FN 3416(15727) 5324(22117) 4370

HFN 2.8(2.2) 2.8(1.9)

Letters 4(0.84) 4(0.82) 4

N 3.6(2.3) 3.4(2.5) 3.5

words

F 3.3(2.7) 421.2(295) 212.3

BC 42(49) 52(55) 47.5

BF 4025(5492) 13672(23806) 8848.5

BN 15(10) 20(10) 17.5

N 2.9(2.5) 4.3(2.8) 3.6

FHFN 2778(15869) 5653(22539) 4215.5

FN 2722(14595) 5496(20745) 4109

HFN 1.8(1.9) 0.5(0.7) 1.18

Letters 4(0.82) 4(0.82) 4

Syl 1.5(0.5) 1.3(0.5) 1.4 Note. BC = summed bigram count, BF = summed bigram frequency; BN = number of bigram neighbors, F = word

frequency/million, FHFN= summed frequency of higher frequency orthographic neighbors, FN = summed frequency of

orthographic neighbors; HFN = number of higher frequency orthographic neighbors; Letter= number of letters; N =

number of neighbors; Syl = number of syllables. Numbers in brackets represent standard deviations..

38

Table 2. Sample stimuli

pseudohomophones pseudowords basewords

Length LF HF LF HF LF HF

3-letters ZEE ALD ZER ALZ ZEH ALT

4-letters EKKE FOLK EFKE BOLK ECKE VOLK

5-letters ZISD RAICH ZWISP REUCH ZWIST REICH

Note. LF = low baseword frequency, HF = high baseword frequency

Table 3. Mean response times (with standard error) and error rates for pseudohomophones,

pseudowords and words

Response Time (ms) Error rates (%)

LF HF LF HF

PSH 959(15) 909(12) 1.68 0.67

PSW 893(10) 891(12) 0.40 0.30

WOR 792(16) 748(11) 1.23 1.66

Note. PSH = pseudohomophones, PSW = pseudowords and WOR = words. LF = low (base)word frequency, HF = high

(base)word frequency.

39

Table 4. Brain regions showing greater activation for pseudohomophones compared to

pseudowords (voxel-level uncorrected p < .001, FDR cluster-level corrected p < 0.05).

Brain Region Brodmann Area hem x y z mm³ Zmax

pseudohomophones > pseudowords

Insular Cortex, Frontal

Orbital Cortex

47,13 L -27 20 -5 44 4.54

Frontal Pole, pars orbitalis 47 L -42 44 -5 31 4.39

Supplementary Motor Area,

Paracingulate Gyrus,

Superior Frontal Gyrus

8 L -6 23 46 73 4.01

Middle Temporal Gyrus,

temporooccipital part,

Inferior Temporal Gyrus,

temporooccipital part

20 R 60 -43 -11 28 3.95

Lateral Occipital Cortex,

superior division, SPL

7 L -30 -70 58 43 3.87

Note. x, y, z = coordinates according to MNI stereotactic space.

40

Table 5. Brain regions showing greater activation for pseudohomophones with high compared to

pseudohomophones with low baseword frequency (voxel-level uncorrected p < .005, FDR

cluster-level corrected p < .05) exclusively masked by the contrast of pseudowords with high >

pseudowords with low baseword frequency (p > .05, uncorrected).

Brain Region Brodmann Area hem x y z mm³ Zmax

pseudohomophones (high) > pseudohomophones (low)

Lateral Occipital Cortex,

inferior/superior division,

Angular Gyrus

39 L -54 -61 28 363 4.56

Angular Gyrus 39/40 R 60 -52 37 212 4.26

Middle Temporal Gyrus,

posterior division, Superior

Temporal Gyrus, posterior

division, Supramarginal

Gyrus, posterior division,

Middle Temporal Gyrus,

temporo-occipital part

21 66 -34 1 199 4.08

Middle Temporal Gyrus,

posterior division, Middle

Temporal Gyrus, temporo-

occipital part, posterior

Superior Temporal Gyrus

21 L -63 -40 -2 89 3.80

Note. x, y, z = coordinates according to MNI stereotactic space

41

Figure Captions

Figure 1: Activation for pseudohomophones and pseudowords compared to fixation baseline

(voxel level uncorrected p < .001; extent threshold k = 5 voxels). Note. Pseudohomophones in red,

pseudowords in blue, overlapping regions in pink. SFG = superior frontal gyrus, MTG = middle

temporal gyrus, SPL = superior parietal lobule.

Figure 2: (A) Axial slices showing regions stronger activated by pseudohomophones (PSH)

compared to pseudowords (PSW). (B) Signal change for high and low frequency

pseudohomophones and pseudowords for regions higher activated for pseudohomophones. Error

bars show standard error of the mean. Note: SFG = superior frontal gyrus, MTG = middle temporal

gyrus, SPL = superior parietal lobule.

Figure 3: (A) Axial slices showing regions differentially activated by pseudohomophones (PSH) with

high and low baseword frequencies. (B) Signal change for high and low frequency

pseudohomophones and pseudowords (PSW) for regions higher activated for

pseudohomophones. Error bars show standard error of the mean. Note: AG = angular gyrus, MTG

= middle temporal gyrus, STG = superior temporal gyrus.

42

44

45

46

• Pseudohomophone effect associated with a spelling-check in superior parietal lobule

• Baseword frequency effect in middle temporal gyrus suggests lexico-semantic access

• Phonological recoding is active during silent reading in experienced readers