ArticlePDF Available

Unpredictability and complexity of print-to-speech correspondences increase reliance on lexical processes: more evidence for the orthographic depth hypothesis

Authors:

Abstract

The Orthographic Depth Hypothesis [Katz, L., & Frost, R. (1992). The reading process is different for different orthographies: The orthographic depth hypothesis. In R. Frost & L. Katz (Eds.), Orthography, phonology, morphology, and meaning (pp. 67–84). Amsterdam: Elsevier Science] proposes cross-linguistic differences in the involvement of lexical processing during reading. In orthographies with complex, inconsistent, and/or incomplete sublexical correspondences, decoding is more difficult and therefore slower. This gives more time to the lexical route to retrieve information, and leads to a greater ratio of lexical processing. We test whether this mechanism applies both for words with inconsistent (in English) and for words with complex (in French) correspondences. As complex correspondences are sufficient to derive a correct pronunciation, an increase in lexical processing may not occur. In a reading-aloud task, we used the frequency effect to measure lexical processing. The data showed stronger involvement of lexical processing for inconsistent compared to consistent words, and for complex compared to simple words. The results confirm that Katz and Frost’s proposed mechanism applies to different sources of orthographic depth.
Full Terms & Conditions of access and use can be found at
http://www.tandfonline.com/action/journalInformation?journalCode=pecp21
Download by: [Aix-Marseille Université], [Dr Eddy Cavalli] Date: 09 May 2016, At: 02:57
Journal of Cognitive Psychology
ISSN: 2044-5911 (Print) 2044-592X (Online) Journal homepage: http://www.tandfonline.com/loi/pecp21
Unpredictability and complexity of print-to-speech
correspondences increase reliance on lexical
processes: more evidence for the orthographic
depth hypothesis
Xenia Schmalz, Elisabeth Beyersmann, Eddy Cavalli & Eva Marinus
To cite this article: Xenia Schmalz, Elisabeth Beyersmann, Eddy Cavalli & Eva Marinus (2016):
Unpredictability and complexity of print-to-speech correspondences increase reliance on
lexical processes: more evidence for the orthographic depth hypothesis, Journal of Cognitive
Psychology, DOI: 10.1080/20445911.2016.1182172
To link to this article: http://dx.doi.org/10.1080/20445911.2016.1182172
Published online: 06 May 2016.
Submit your article to this journal
Article views: 18
View related articles
View Crossmark data
Unpredictability and complexity of print-to-speech correspondences
increase reliance on lexical processes: more evidence for the orthographic
depth hypothesis
Xenia Schmalz
a,b
, Elisabeth Beyersmann
c,d
, Eddy Cavalli
c,e
and Eva Marinus
b,f
a
Dipartimento di Psicologia dello Sviluppo e della Socializzazione (DPSS), Università degli Studi di Padova, Padova, Italy;
b
ARC Centre
of Excellence in Cognition and its Disorders, Macquarie University, Sydney, Australia;
c
Laboratoire de Psychologie Cognitive, Aix-
Marseille Université, Marseille, France;
d
Brain and Language Research Institute, Marseille;
e
Centre National de la Recherche
Scientique, France;
f
Department of Cognitive Science, Macquarie University, Sydney, Australia
ABSTRACT
The Orthographic Depth Hypothesis [Katz, L., & Frost, R. (1992). The reading process is
different for different orthographies: The orthographic depth hypothesis. In R. Frost &
L. Katz (Eds.), Orthography, phonology, morphology, and meaning (pp. 6784).
Amsterdam: Elsevier Science] proposes cross-linguistic differences in the involvement
of lexical processing during reading. In orthographies with complex, inconsistent, and/
or incomplete sublexical correspondences, decoding is more difcult and therefore
slower. This gives more time to the lexical route to retrieve information, and leads to a
greater ratio of lexical processing. We test whether this mechanism applies both for
words with inconsistent (in English) and for words with complex (in French)
correspondences. As complex correspondences are sufcient to derive a correct
pronunciation, an increase in lexical processing may not occur. In a reading-aloud task,
we used the frequency effect to measure lexical processing. The data showed stronger
involvement of lexical processing for inconsistent compared to consistent words, and
for complex compared to simple words. The results conrm that Katz and Frosts
proposed mechanism applies to different sources of orthographic depth.
ARTICLE HISTORY
Received 1 February 2016
Accepted 19 April 2016
KEYWORDS
Dual-route model; cross-
linguistic; French; English
Previous research has shown that the cognitive pro-
cesses underlying reading differ across orthogra-
phies. This is true for both adult reading (Frost, Katz,
& Bentin, 1987; Rau, Moll, Snowling, & Landerl,
2015; Schmalz et al., 2014) and reading development
(Aro & Wimmer, 2003; Landerl, Wimmer, & Frith, 1997;
Mann & Wimmer, 2002; Marinus, Nation, & de Jong,
2015; Seymour, Aro, & Erskine, 2003). From a theoreti-
cal perspective, it is important to isolate how and why
cognitive mechanisms of reading differ across ortho-
graphies. This will provide insight into how the uni-
versal perceptual systems interact with specic
properties of each language and orthography, and
lay out benchmarks for models of reading (Frost,
2012; Share, 2008).
From the onset of research on cross-linguistic
differences in reading, the concept that has received
the most attention is orthographic depth (Bridge-
man, 1987;Frostetal.,1987;Katz&Feldman,
1983; Turvey, Feldman, & Lukatela, 1984). Broadly
speaking, orthographic depth refers to the cross-lin-
guistic variability in the closeness of the relationship
between orthographic word forms and their pro-
nunciations. Within most models of reading, print-
to-speech correspondences are important for a
process called sublexical decoding, whereby the pro-
nunciation of a word or nonword is assembled using
the knowledge of the regularities that underlie print-
to-speech conversion in a given orthography
(Coltheart, Rastle, Perry, Langdon, & Ziegler, 2001;
Perry, Ziegler, & Zorzi, 2007; Plaut, McClelland, Seiden-
berg, & Patterson, 1996). In shallow orthographies,
such as Italian, this is a relatively simple process. The
Italian grapheme a, for example, almost always
maps onto the phoneme /a/. Conversely, in English,
a prototypical deep orthography, the grapheme a
can be pronounced as in cat,”“nation,”“wasp,or
false.Part of this ambiguity can be resolved by devel-
oping sensitivity to more complex regularities that
exist in the orthography. For example, the
© 2016 Informa UK Limited, trading as Taylor & Francis Group
CONTACT Xenia Schmalz xenia.schmalz@gmail.com
JOURNAL OF COGNITIVE PSYCHOLOGY, 2016
http://dx.doi.org/10.1080/20445911.2016.1182172
Downloaded by [Aix-Marseille Université], [Dr Eddy Cavalli] at 02:57 09 May 2016
pronunciation of the word wasp is predictable,
because an apreceded by a wtends to be pronounced
as /ɔ/ (Schmalz et al., 2014; Treiman, Kessler, & Bick,
2003). In English orthography, however, there are
also instances where the pronunciation is not predict-
able based on any sublexical information, such as in
the words yacht or colonel. Therefore, there are two
reasons why English orthography might be con-
sidered deep: First, the relatively high degree of com-
plexity of the print-speech correspondences
compared to orthographies such as German, Italian,
and Dutch, and second, a high degree of unpredict-
ability, even when those complex rules are applied
(van den Bosch, Content, Daelemans, & de Gelder,
1994; Schmalz, Marinus, Coltheart, & Castles, 2015;
Ziegler, Perry, & Coltheart, 2000).
Corpus analyses have shown that, across ortho-
graphies, unpredictability and complexity are disso-
ciable on a linguistic level (van den Bosch et al.,
1994; Schmalz et al., 2015): Although orthographies
with simple correspondences tend to also have a
high degree of predictability, these two concepts
are not perfectly correlated. A particularly interest-
ing example is French orthography, as there is a dis-
crepancy between the degree of complexity and
unpredictability. Specically, French is high in com-
plexity, because it contains multi-letter rules (au
/o/) and context-sensitive rules (c[a,o,u]/k/, c[e,i]
/s/), but low in unpredictability (van den Bosch
et al., 1994; Schmalz et al., 2015).
To date, it is unclear whether complexity and
unpredictability of the sublexical correspondences
act as separate sources of orthographic depth, or if
they affect reading processes in the same way on
a behavioural level. An existing hypothesis on ortho-
graphic depth is the orthographic depth hypothesis
(hereafter: ODH; Katz & Frost, 1992). Here, the
authors offer both a well-specied denition of
orthographic depth, and propose a specic cogni-
tive mechanism that drives cross-linguistic differ-
ences as a function of orthographic depth. In deep
orthographies, they describe print-to-speech con-
version code as characterised by complex,inconsist-
ent, and/or incomplete sublexical information. This
makes the sublexical conversion process more dif-
cult in deep compared to shallow orthographies. As
a result, the sublexical conversion process is
impaired in one way or another, which gives more
time for a lexical look-up mechanism to derive the
correct pronunciation. This leads to a higher
overall ratio of lexical-to-sublexical processing, as a
function of degree of orthographic depth.
It is particularly noteworthy that Katz and Frost
(1992)list three different properties that underlie
the sublexical regularities of deep versus shallow
orthographies: complexity, consistency, and incom-
pleteness. The concepts of complexity and consist-
ency map onto the distinction between complexity
and unpredictability proposed by Schmalz et al.
(2015; see also van den Bosch et al., 1994). Yet
despite the theoretical and linguistic work that has
shown a distinction between these multiple con-
structs underlying orthographic depth, whether
these may differentially affect reading processes
on the behavioural level has not been previously
empirically tested.
The rst construct proposed by both Katz and
Frost (1992) and Schmalz et al. (2015) is complexity.
An orthography with complex correspondences is
characterised by multi-letter rules, where several
letters are required to denote a single phoneme (e.
g., augh /o:/ in English, aient /ε/ in French)
and/or context-sensitive regularities, where sur-
rounding letters affect a graphemes pronunciation
(e.g., in English, ais pronounced as /ɔ/ when pre-
ceded by a w,asinswan; in French, a gis pro-
nounced as /ʒ/ when followed by an ior e,asin
gélatine). When words contain complex correspon-
dences, the sublexical information is sufcient to
access full information about the words phonology
and semantics, once these complex rules are
applied. However, evidence exists that applying
multi-letter rules slows down the sublexical pro-
cedure, as they cause a conict between the pro-
nunciation of the single letters (e.g., in English, t
and h) and the graphemes pronunciation, th /θ/
(Marinus & de Jong, 2010; Rastle & Coltheart, 1998;
Rey, Jacobs, Schmidt-Weigand, & Ziegler, 1998).
The second construct that has been described by
both Katz and Frost (1992) and Schmalz et al. (2015)
is inconsistency, or unpredictability. Inconsistency is
the presence of two or more pronunciations for the
same orthographic unit. Conventionally, this is
dened at the level of a wordsbody (e.g., the
body ear is inconsistent because it can be pro-
nounced as in hearor bear), but the same
measure can also be applied to graphemes (e.g.,
the grapheme th is inconsistent, because it can be
pronounced as in thistle,”“this,or thyme). For
this source of depth, the sublexical information is
not sufcient to derive the correct pronunciation.
For example, the English words tough,”“though
and throughhave nearly identical sublexical
information, but each of them has a different
2X. SCHMALZ ET AL.
Downloaded by [Aix-Marseille Université], [Dr Eddy Cavalli] at 02:57 09 May 2016
pronunciation of the grapheme ough, which cannot
be derived without knowledge of the whole word
(see Schmalz et al., 2015, for an in-depth discussion).
According to rule-based computational models,
such as the Dual Route Cascaded (DRC) model
(Coltheart et al., 2001), such words need to be read
aloud via the lexical route for a correct response,
because the sublexical route will give a regularised,
or rule-based response for words which do not
comply to the rules (e.g., /θaim/ for the word
thyme). In a connectionist framework (Perry et al.,
2007; Plaut et al., 1996), such words require the
reliance on larger units, which in the case of unpre-
dictable words coincide with a whole word (e.g.,
ough is pronounced as in toughwhen preceded
by a tand as in thoughwhen preceded by a th).
Arguably, the fact that the orthographic unit
coincides with a whole word makes its processing
qualitatively different from processing sublexical
orthographic units, as whole words have direct con-
nections to their semantic information, whereas sub-
lexical units do not (for a discussion, see Schmalz
et al., 2015).
The third construct proposed by Katz and Frost
(1992) is incompleteness. This construct is of high
relevance to Semitic orthographies, where vowels
are not always represented. In pointed Hebrew,
the sublexical information is complete, because all
phonemes are represented; vowels are represented
as diacritics. Generally, however, texts are written
in unpointed Hebrew, without vowel markings.
Here, vowel information is incomplete, and the pro-
nunciation needs to be derived via semantic context
by uent readers. Incompleteness is not of high rel-
evance for European orthographies, however. In the
European alphabetic scripts, the orthographic (sub-
lexical and whole-word) information is mostly suf-
cient to assemble a full phonological representation
and to use this to access a words semantics. There
are some examples of words with incomplete lexical
and sublexical information, namely heterophonic
homographs. For a word like present,semantic
context is needed to derive both a pronunciation,
and to access different semantic information depend-
ing on whether this word occurs as a verb or a noun.
By denition, lexical-semantic processing is required
when the sublexical correspondences are incomplete.
According to the ODH, complexity, inconsistency,
and incompleteness result in a higher ratio of lexical
and/or semantic to sublexical processing. The notion
of an independent lexical and sublexical route is the
basis of the dual-route framework (Coltheart et al.,
2001; Perry et al., 2007). Here, the lexical and sublex-
ical routes operate in parallel to obtain a pronuncia-
tion from an orthographic input. The longer the
sublexical route takes, the more the nal pronuncia-
tion will be inuenced by excitatory connections
from the orthographic lexicon to the phonological
lexicon and to the phonological output buffer. If
the sublexical information can be processed
quickly, the phonological output will be driven to a
greater extent by phoneme activation from the sub-
lexical units.
Previous research has provided support for a
stronger lexical inuence for deep compared to
shallow scripts, as predicted by the ODH. Frost
et al. (1987) showed, in a between-language com-
parison, that lexical and semantic marker effects
increase as a function of depth in Serbo-Croatian
(a shallow orthography), English (medium), and
unpointed Hebrew (deep). In a further study, Frost
(1994), took advantage of the presence of both the
shallow pointed and the deep unpointed script in
Hebrew. This allows for a within-item design,
where the same words can be presented with and
without diacritics. Again, Frost (1994) showed stron-
ger lexical (word frequency) and semantic (semantic
priming) effects for the deep compared to the
shallow script.
Both studies support the view that incomplete-
ness increases the reliance on lexical processing, as
both report a comparison of unpointed Hebrew
with a complete orthography (pointed Hebrew,
and English and Serbo-Croatian). The comparison
between English and Serbo-Croatian, however, can
be interpreted in different ways, because these
two orthographies differ from each other both in
terms of complexity and unpredictability. The rst
possibility, which is in line with the ODH, is that
complex correspondences slow down the process
of sublexical decoding. Thus, while the sublexical
output is in principle sufcient for a correct response
to occur, the slowdown will allow more time for the
lexical route to contribute to the nal phonological
output. This would mean that any source of ortho-
graphic depth (i.e., complexity, unpredictability, or
incompleteness) should increase the relative contri-
bution of the lexical route.
Alternatively, it is possible that there is a quali-
tatively different impact of unpredictability and
incompleteness as compared to complexity: As
unpredictability and incompleteness make it
impossible for the reader to compute a pronuncia-
tion from the sublexical information, the nal
JOURNAL OF COGNITIVE PSYCHOLOGY 3
Downloaded by [Aix-Marseille Université], [Dr Eddy Cavalli] at 02:57 09 May 2016
response of the sublexical route will be either incor-
rect or partial. In this case, a correct reading-aloud
response cannot occur until the lexical route has
provided enough activation to the phonological
output buffer. This is different for words with
complex correspondences: Here, the sublexical
information is, in principle, sufcient for a correct
pronunciation. Any slowdown associated with the
presence of complex correspondences might not
be sufcient to result in a substantial effect on the
relative amount of lexical processing.
The existing studies do not allow us to differen-
tiate between the two possibilities. To our knowl-
edge, all comparisons of lexical-semantic marker
effects used orthography pairs which differ both in
terms of complexity and unpredictability, such as
English and Serbo-Croatian (Frost et al., 1987)or
English and German (Frith, Wimmer, & Landerl,
1998; Rau et al., 2015). The main aim of the current
study was to distinguish between these two possibi-
lities. We use two orthographies, where the corre-
spondences reect two different sources of depth,
namely unpredictability in English, and complexity
in French (van den Bosch et al., 1994; Schmalz
et al., 2015). We chose reading aloud rather than
silent reading as the experimental task, because
the ODH is specically concerned with the process
of deriving speech from print. Lexical decision is
considered to be less sensitive to this sublexical
process, as high accuracy on this task can be
achieved purely by relying on lexical access
(Coltheart et al., 2001).
The unpredictability measure
Dening unpredictability is not straightforward,
because existing models of reading make different
assumptions about the way in which the sublexical
route assembles a pronunciation (Coltheart et al.,
2001; Plaut et al., 1996; for a discussion, see
Schmalz et al., 2015). Given that there is no consen-
sus about the type of information that is used to
assemble a pronunciation, it is also unclear what
kinds of words would be considered to have an
unpredictable pronunciation. To ensure that the
results are meaningful beyond the assumptions of
a specic model, we use a denition which is compa-
tible with both connectionist and rule-based
models: We classify a word as unpredictable, if (1)
it is both irregular (by the set of grapheme
phoneme correspondence rules implemented
within the rule-based computational model, DRC;
Coltheart et al., 2001) and inconsistent (i.e., if the
body has more than one possible pronunciation),
such as the word ghost,or (2) if it is irregular,
and does not have any body neighbours, such as
the word debt.Thus, neither grapheme
phoneme correspondences nor body-rime corre-
spondences can be reliably used to read aloud
these words correctly.
The concepts of irregularity and inconsistency are
strongly correlated, but reect theoretically different
constructs and can be manipulated to vary orthog-
onally (Andrews, 1982; Cortese & Simpson, 2000;
Jared, 1997,2002; Jared, McRae, & Seidenberg,
1990). Here, we classied words as predictable if
the pronunciation was predictable both from gra-
phemephoneme correspondence rules (i.e.,
regular) and from body-rime correspondences (i.e.,
consistent), and as unpredictable when neither
source could be used to read aloud the words cor-
rectly. We excluded words that are regular but
inconsistent (e.g., mint,which is regular but has
the enemy pint) or irregular but consistent (e.g.,
walk,which should be pronounced as /wælk/
according to the DRC). Recent behavioural data
suggests that participants rely on information from
various types of sources to predict a novel words
pronunciation (Schmalz et al., 2014); as it is not yet
clear how the cognitive system merges conicting
information from different sources, we excluded
these types of words for the current purposes.
It is unconventional to use predictability as a vari-
able in psycholinguistic research. To date, the litera-
ture has focused predominantly on contrasting the
effects of regularity with those of consistency
(Andrews, 1982; Cortese & Simpson, 2000; Jared,
1997,2002;Jared et al., 1990). Rule-based models,
such as DRC, predict effects of regularity, because
a lack of compliance to graphemephoneme corre-
spondence rules should impair the reading-aloud
process via the sublexical route. Connectionist
models, such as the triangle or connectionist dual
processing (CDP) models (Perry et al., 2007; Plaut
et al., 1996), use a learning algorithm to extract the
relationships between print and speech, which
becomes more difcult when a given orthographic
pattern can map onto multiple pronunciations (e.g.,
the body ost, which can be pronounced as in
ghostor as in lost). Thus, connectionist
models predict an effect of consistency, but not
regularity. While previous studies have shown
that inconsistent words and nonwords are read
aloud more slowly than matched consistent items
4X. SCHMALZ ET AL.
Downloaded by [Aix-Marseille Université], [Dr Eddy Cavalli] at 02:57 09 May 2016
(Andrews, 1982;Cortese&Simpson,2000;Glushko,
1979;Jared,1997,2002; Jared et al., 1990), other
data suggests that participants also rely on print-
to-speech rules, especially for unusual orthographic
patterns (Andrews & Scarratt, 1998; Pritchard,
Coltheart, Palethorpe, & Castles, 2012;Robidoux&
Pritchard, 2014).
The current study
In Experiment 1, we compare the frequency effect
for English words with predictable versus unpre-
dictable correspondences. The English orthography
is used, because the relatively high degree of both
complexity and unpredictability allows us to
manipulate frequency and predictability. If there
is stronger involvement of lexical processing
when the pronunciation of a word is unpredictable,
we expect a frequency-by-predictability inter-
action, where the frequency effect is larger for
unpredictable compared to predictable words.
This study serves as a conceptual replication of
the nding of Frost et al. (1987) that there is a stron-
ger relative involvement of the lexical route when
the correspondences are unpredictable compared
to when they are predictable.
In Experiment 2, we use the French orthogra-
phy, which has a high degree of complexity,
while being highly predictable. This allows us to
manipulate frequency and complexity in a
within-subject design and without unpredictabil-
ity as a confounding variable. We aim to establish
whether there is a stronger frequency effect for
words containing complex correspondences, com-
pared to words which contain only simple corre-
spondences (i.e., there is a one-to-one
correspondence between letters and sounds). If
the complexity of the correspondences slows
down the assembly process, we expect to nd a
frequency-by-complexity interaction, as lexical
processing should be stronger for words with
complex than simple correspondences according
to the ODH. If we obtain this pattern, it would indi-
cate that both complexity and unpredictability
affect reading processes in adults in the same
way as incompleteness in the previous studies
(Frost, 1994;Frostetal.,1987). If we do not nd
a frequency-by-complexity interaction, this
means that cross-linguistic differences in the rela-
tive reliance on lexical processing are driven by
unpredictability and incompleteness, but not
complexity.
Experiment 1: Unpredictability in English
Method
Participants
Twenty undergraduate students at an Australian
university participated in the experiment. All were
native speakers of English and received course
credit for their participation.
Items
To justify the use of the predictability metric rather
than the more conventional consistency metric, we
rst veried that predictability reects a psychologi-
cally valid construct. We created two models from
the full data set that is analysed in Experiment 1
(see below). The models were nearly identical to
those described in the Results section. The indepen-
dent variables were predictability (coded as a binary
contrast) or consistency (centred ratio of friends to
enemies), centred log frequency, and the two-way
interaction. Note that we centred all continuous
independent variables (by subtracting each value
from the mean) and contrast-coded dichotomous
conditions (as 0.5 and 0.5) because Linear Mixed
Effect (LME) models provide parameter estimates
as deviations from the point closest to zero rather
than deviations from the mean. The dependent vari-
able was trimmed inverse RT (for more details about
the trimming procedure, see the Results section).
Items and participants were included as random
effects, and the slope of the frequency effect was
allowed to vary across participants (Barr, Levy,
Scheepers, & Tily, 2013). The model with predictabil-
ity as the independent variable yielded a numeri-
cally better t than the model with the consistency
ratio as the independent variable (AIC = 4,620 for
the former, AIC = 4,658 for the latter). A Bayesian
analysis, where the two models were contrasted,
provided support for the model which used predict-
ability as the independent variable over the model
using consistency, with a Bayes Factor value
>1,000,000 (for a description of how we interpret
Bayes Factors, see below). This justies the use of
predictability as an independent variable, and
suggests that predictability has stronger psychologi-
cal validity than consistency.
For the experiment, we used only monosyllabic
words, because traditionally, measures of regularity
and consistency, which form the basis of the predict-
ability construct, have been dened for monosylla-
bic words only (but see Chateau & Jared, 2003;
Kearns et al., 2014; Yap & Balota, 2009, for an
JOURNAL OF COGNITIVE PSYCHOLOGY 5
Downloaded by [Aix-Marseille Université], [Dr Eddy Cavalli] at 02:57 09 May 2016
extension of the consistency measure to multisylla-
bic words). We extracted all monosyllabic words
from the British Lexicon Project (BLP; Keuleers,
Lacey, Rastle, & Brysbaert, 2012). We retained all
words with log frequencies between 0 and 2,
because analyses of large-scale lexical databases
have shown that the frequency effect is most
robust for this log frequency range (Balota et al.,
2007; Brysbaert et al., 2011; Ferrand et al., 2010; Keul-
eers, Diependaele, & Brysbaert, 2010). All words had
a lexical decision accuracy >80%, suggesting that
the words should be familiar to the majority of
undergraduate students. We classied the words
as predictable or unpredictable based on the
above criteria.
We selected a total of 376 words. Half of these
were predictable (e.g., forge)andhalfwere
unpredictable (e.g., ghost), and they were
chosen to vary in frequency, as half had a relatively
low frequency (log frequency of 01) and the other
half a relatively high frequency (log frequency of 1
2).Notethatwetreatfrequencyasacontinuum
rather than a dichotomy throughout the paper to
increase experimental power. Frequency, as well
as orthographic Ncounts, are based on the subtitle
counts provided by the BLP (van Heuven, Mandera,
Keuleers, & Brysbaert, 2014; Keuleers et al., 2012).
Linear models were performed to assess whether
any of the item characteristics co-varied with fre-
quency or unpredictability. In separate analyses,
each centred potential covariate was used as the
dependent variable; centred frequency, contrast-
coded predictability, and their interaction were
used as predictor variables. The outcomes of this
set of analyses are shown in Table 1. The individual
items and their full descriptives, as well as the raw
dataandtheRscriptusedforthecurrentstudy,can
be found here: osf.io/hm8fw. Note that ortho-
graphic neighbourhood and Phonological Levensh-
tein Distance (the number of phoneme
substitutions, deletions, or additions which are
required to reach the nearest 20 neighbours; see
Yarkoni, Balota, & Yap, 2008) co-vary with fre-
quency, and the ratio of letters to phonemes
differs across predictable and unpredictable
words. However, the critical comparison in the
current experiment is the interaction between pre-
dictability and frequency, and none of the covari-
ates show a stronger manipulation for the
predictable than unpredictable condition, all p>.3
for the interaction. We therefore do not include
any of them as covariates in the main analysis. To
conrm that these potential confounds do not
inuence the results, however, we present a covari-
ate analysis in a post-hoc test.
Procedure
Item presentation was controlled with DMDX
(Forster & Forster, 2003). The words were shown,
one at a time, in random order, for 2.5 s or until
the voice-key was triggered. The participants were
instructed to read aloud each item as quickly and
accurately as possible.
Results and discussion
The reading-aloud responses were scored ofine
with the software CheckVocal (Protopapas, 2007),
as correct, incorrect, or no response. Response
latencies were readjusted using CheckVocal, based
on the onset of the sound waves, in the case of pre-
mature of late voice-key triggers. This removes
potential biases associated with rst phonemes.
The data were further analysed using the software
R, both withLME models (Baayen, Davidson, & Bates,
2008) and with Bayes Factors (Morey & Rouder,
2014; Rouder, Speckman, Sun, Morey, & Iverson,
2009). LMEs allow us to obtain an estimate of the
slope (which serve as descriptives given the use of a
continuous measure of frequency). We provide the
results of t-tests, and p-values, when appropriate, to
provide a point of reference for those unfamiliar
with Bayesian analyses.
We report Bayes Factors for all theoretically inter-
esting comparisons (i.e., for the critical interactions),
and base our conclusions on them. Unlike frequen-
tist statistics, Bayes Factors allow us to quantify the
evidence for (or against) an effect or interaction of
interest, given a prior belief. Therefore, they argu-
ably provide a closer link to the conclusions that
can be drawn from the data. Here, we use the
default prior of the BayesFactor package, which
assumes a Cauchy distribution with the width par-
ameter r= 0.5 (Morey & Rouder, 2014). We interpret
the results according to a set of guidelines described
in Rouder et al. (2009): Bayes Factor values smaller
than 1/3 provide evidence against an effect or inter-
action, values between 1/3 and 1 and between 1 and
3 are considered to provide anecdotal or equivocal
evidence against or for it, respectively, values
larger than 3 provide some evidence for the effect
or interaction, and values larger than 10 provide
strong evidence. Thus, throughout the paper,
smaller values provide evidence for a null
6X. SCHMALZ ET AL.
Downloaded by [Aix-Marseille Université], [Dr Eddy Cavalli] at 02:57 09 May 2016
hypothesis, and larger values provide evidence for
the alternative hypothesis.
For the LME model, we used inverse RTs as the
dependent variables. For the independent variables,
we used centred log frequency as a continuous pre-
dictor, and predictability, contrast-coded as 0.5 (pre-
dictable) and 0.5 (unpredictable), as a binary
predictor. The model also included previous RT
(Baayen, 2008). Participants and items were included
as random factors, and the frequency slope was
allowed to vary across participants (Barr et al., 2013).
There were seven non-responses (0.01% of all
data), and overall accuracy was 97.2%. The accuracy
rates ranged from 93.4% for low-frequency unpre-
dictable words to 99.3% for low-frequency predict-
able words. An LME model with accuracy as the
dependent variable showed a main effect of predict-
ability, β= 1.5, z=3.9, p< .0001, reecting higher
accuracy for the predictable (99.1%) than unpredict-
able (95.3%) conditions.
1
The interaction between fre-
quency and predictability was signicant, β=1.7, z
=2.5, p= .013, indicating a facilitatory frequency
slope for unpredictable (β= 1.5) but not predictable
(β=0.3) words. The main effect of frequency was
not signicant, p= .1. The results are broadly in line
with the RT results discussed below. As the error
rate was relatively high, and there was not a lot of
variability between the conditions, we draw con-
clusions based on the RT results only.
Before conducting the RT analyses, we excluded
all incorrect responses, and trials with latencies
<300 ms (0.2% of the data) and >1,200 ms (0.1% of
the data). This yielded an approximately normal dis-
tribution of inverse RTs. When we articially dichot-
omise frequency into high (log frequency >1) and
low (log frequency <1), the averages RTs of the
trimmed data set are 494.2 (SD = 97.8) and 494.1
(SD = 97.2), respectively, for the predictable words,
and 503.6 (SD = 97.5) and 528.3 (SD = 115.1), respect-
ively, for the unpredictable words.
The LME showed a signicant main effect of pre-
dictability, β=0.08, t=5.3, p< .0001, reecting
shorter RTs for predictable (forge) than unpredict-
able (ghost) words, and a main effect of frequency,
β=0.05, t=3.5, p= .0006, indicating shorter RTs
for words of higher frequencies. The interaction
was also signicant, β= 0.07, t= 2.3, p=.02, with a
steeper frequency slope for unpredictable compared
to predictable words. In a comparison of the full
model against an additive one that included the
main effects of predictability and frequency, the Bayes
Factor provided anecdotal evidence for the presence
of the interaction, BF = 1.7 (±1.2%).
Follow-up analyses
To potentially strengthen the case for the inter-
action, we retrieved all trial-level data for our items
from the English Lexicon Project (ELP) reading-
aloud database (Balota et al., 2007). Note that we
included data from the ELP and not the BLP
because the ELP has both lexical decision and
reading-aloud data, whereas the BLP only has
lexical decision. As both the ELP and our experiment
employed a standardised reading-aloud procedure,
we can increase the amount of evidence by collap-
sing the two data sets.
The ELP contains trial-level reading-aloud data for
375 of the original 376 words. These include 10,342
valid and correct trials, with an average of 27.6 par-
ticipants per word. We combined the data from our
experiment with data of the ELP. The trimming pro-
cedure of this bigger item set was identical to that of
the original data, as was the model, except that pre-
vious RT was not included, as it was unavailable in
Table 1. Potential moderators, and how they co-vary with the critical manipulations of Experiment 1.
Potential covariate
Main effect of
frequency
Main effect of
predictability
Interaction of predictability and
frequency
Overall average and standard
deviation
Number of letters t=1.51, p= .13 t=0.94, p= .35 t=0.91, p= .36 4.99 (0.93)
Number of phonemes t=1.58, p= .12 t= 1.52, p= .13 t=0.26, p= .79 3.78 (0.82)
Orthographic Nt= 2.02, p= .04* t= 1.04, p= .30 t=0.23, p= .82 5.82 (4.35)
Bigram frequency t= 0.49, p= .62 t= 0.12, p= .91 t=0.90, p= .37 68.91 (73.76)
Ratio of letters to phonemes t=0.55, p= .58 t=3.10, p< .01* t=0.80, p= .42 1.36 (0.32)
Phonological Levenshtein
Distance
t=3.80, p< .01* t=0.06, p=.95 t=0.74, p= .46 1.42 (0.31)
Note: Asterisks (*) show a signicant relationship at the 0.05-level. Frequency counts are based on the English subtitle corpus (New, Brysbaert, Veronis,
& Pallier, 2007); orthographic Ncounts are retrieved from the British Lexicon Project (Keuleers et al., 2012); bigram frequency is from the MCWord
database (Medler & Binder, 2005); Phonological Levenshtein Distance is retrieved from the English Lexicon Project (Balota et al., 2007).
1
In the accuracy analyses, we did not allow slope to vary across participants, because this model failed to converge. This is a common issue with
maximal models (see Bates, Kliegl, Vasishth, & Baayen, 2015). However, as the accuracy data were not interpreted, this is not an issue for the
current study.
JOURNAL OF COGNITIVE PSYCHOLOGY 7
Downloaded by [Aix-Marseille Université], [Dr Eddy Cavalli] at 02:57 09 May 2016
the ELP database. Dichotomising frequency, the
average RTs for the four types of items were 561.9
ms (SD = 125.4) for high-frequency predictable
words, 570.2 (SD = 125.7) for high-frequency unpre-
dictable; 575.3 (SD = 132.4) for low-frequency pre-
dictable, and 597.6 (SD = 141.5) for low-frequency
unpredictable. Note that p-values are not reported
for any of the follow-up analyses in this paper, as
due to the multiple comparisons, the Type-I error
rate increases and is no longer 5% with a cut-off of
α= 0.05 (Cramer et al., 2015; Simmons, Nelson, &
Simonsohn, 2011).
The LME results showed the same pattern as the
original data, with shorter RTs for more frequent
compared to less frequent words, with a slope of
β=0.07, t=10.0, with shorter latencies for pre-
dictable than unpredictable words, β=0.05, t=
6.6, and a steeper frequency slope for unpredict-
able than predictable words by β= 0.05, t= 3.3.
Importantly, the Bayes Factor now provided evi-
dence for the presence of the interaction between
frequency and predictability, BF = 9.6 (±0.6%).
In an additional post-hoc analysis, we ensured
that the obtained results remain stable after taking
into account potential confounds. As shown in
Table 1, some of the psycholinguistic variables co-
varied with our manipulations. The model was iden-
tical to the one above, but we also included main
effects of orthographic Nand PLD20 (which differ
as a function of frequency), and the ratio of letters
to phonemes (which differs as a function of predict-
ability). The adjusted means when frequency is
dichotomised are 606.5 ms for high-frequency pre-
dictable words, 612.6 ms for high-frequency unpre-
dictable words, 624.7 ms for low-frequency
predictable words, and 646.5 ms for low-frequency
unpredictable words. The results of the full model
can be downloaded from the OSF folder (osf.io/
hm8fw). The patterns of results did not change:
The LME showed a main effect of frequency, β=
0.07, t=10.0, predictability, β=0.04, t=5.8,
and the interaction, β=0.05, t= 3.8. The Bayes
Factor provided evidence for the presence of the
critical interaction, BF = 50.7 (±0.88%).
In sum, we found strong evidence for the pre-
dicted interaction between frequency and predict-
ability, where the frequency effect is stronger for
unpredictable than predictable words. This provides
a conceptual replication of previous experiment by
Frost and colleagues (Frost, 1994; Frost et al., 1987),
and evidence for the ODH. Specically, the results
suggest that unpredictability of print-to-speech
correspondences impairs sublexical processing,
which results in stronger lexical involvement com-
pared to words with predictable correspondences.
Experiment 2: Complexity in French
For French, the aim was to assess whether the fre-
quency effect for words containing complex print-
to-speech correspondences is stronger than for
words where the pronunciation can be deciphered
based on simple single-letter correspondences.
This would provide further support for the ODH,
and insights about the orthographic characteristics
that may lead to a script being classied as deep
or shallow. A lack of an interaction between fre-
quency and complexity would suggest that
complex correspondences are processed qualitat-
ively differently to unpredictable correspondences.
Method
Participants and procedure
The participants were 24 students from a university
in France. All were native speakers of French and
received course credit in exchange for their partici-
pation. The procedure was identical to Experiment 1.
Items
We retrieved words and their corresponding infor-
mation from the Lexique 2 database (New, Pallier,
Brysbaert, & Ferrand, 2004) and the French Lexicon
Project (FLP; Ferrand et al., 2010). For frequency,
we relied on subtitle counts (Brysbaert et al., 2011;
Brysbaert & New, 2009; New et al., 2007). We again
removed words with log frequencies of <0 or >2.
To classify words as complex or simple, we used
the ratio of letters to phonemes in each word: The
presence of multi-letter correspondences means
that multiple letters correspond to a single
phoneme, thus a complex word has a letter-to-
phoneme ratio >1. Simple words were those with
a letter-to-phoneme ratio of one (e.g., garnir),
and words with a ratio of greater than one were con-
sidered complex (e.g., gâteau). In the database, this
procedure classied 280 words (8.9%) as simple,
and 2,852 (91.1%) as complex.
We selected 384 words, half with complex corre-
spondences (gâteau) and half with simple corre-
spondences (garnir). The words were chosen to
vary in frequency, where half the items had frequency
counts lower than 1, and the other half higher than
1. In addition, the items were chosen such that they
8X. SCHMALZ ET AL.
Downloaded by [Aix-Marseille Université], [Dr Eddy Cavalli] at 02:57 09 May 2016
did not differ, across conditions, on average gra-
pheme consistency (Lété, Sprenger-Charolles, &
Colé, 2004), suggesting that there were no differences
in the degree of unpredictability. Overall, the French
orthography has a high degree of predictability
once complex rules are taken into account (Schmalz
et al., 2015; Ziegler, Perry, & Coltheart, 2003).
However, there are some words with ambiguous pro-
nunciations (e.g., femme,where the second letter is
pronounced as /a/ rather than the default /ε/). While,
to our knowledge, there is no quantication method
of regularity that can be applied to polysyllabic words
in French, the Manulex database contains average
grapheme consistency ratings (Lété et al., 2004). We
use these as a measure of unpredictability, as words
with unpredictable pronunciations necessarily have
graphemes that can be pronounced in multiple ways.
All items had a lexical decision accuracy, accord-
ing to the FLP, of >80%. The descriptive statistics
are listed in Table 2. For the full item set with indi-
vidual word characteristics, as well as the raw data
and R scripts, see here: osf.io/hm8fw. Again, in the
results section, we will follow up with covariate
analyses to ensure that the results cannot be
explained by the variables that differ as a function
of the manipulation.
Results and discussion
The data were scored with CheckVocal as correct,
incorrect, or no response, and the RTs were adjusted
when the voice-key had been triggered prematurely
or late (again, adjusting for potential biases associ-
ated with rst phonemes). As for the English analyses,
we used inverse RTs as the dependent variables, con-
tinuous centralised frequency and binary contrast-
coded complexity (0.5 = simple) as independent
variables, and previous RT. Participants and items
were included as random factors, and the effect of
frequency was allowed to vary across participants.
As the items were matched on the number of
letters (as is common in studies on multi-letter
rules; see Rastle & Coltheart, 1998; Rey et al., 1998),
thesimple(garnir) condition had, by denition,
more phonemes than the complex (gâteau) con-
dition. This also resulted in a lower number of sylla-
bles for complex (average = 2.0) compared to
simple (average = 2.8) words. It was therefore
decided, a priori, that the number of syllables
should be included in the model, to act as a covariate.
Overall, there were no non-responses, and the
accuracy rate was 97.5%. Accuracy was very high
and evenly distributed across conditions (ranging
from 95.7% for low-frequency simple words to
98.7% for high-frequency complex words). An LME
on the accuracy rates showed a main effect of fre-
quency, β= 0.5, z= 4.0, p< .0001, reecting higher
accuracy for high- than low-frequency words.
Neither the effect of complexity nor the complex-
ity-by-frequency interaction reached signicance,
p> .1. This is likely to be reect the overall high accu-
racy rates and lack of variability across conditions.
For this reason, as for in Experiment 1, we draw con-
clusions from the RT data.
For the RT analyses, we removed one data point
with RT <300 ms, which yielded an approximately
normal distribution of inverse RTs. When articially
dichotomising frequency (high: log frequency >1;
low: log frequency <1), the average RTs are 608.5
ms (SD = 146.9) and 620.4 ms (SD = 156.4), respect-
ively, for simple words, and 565.9 ms (SD = 114.5)
and 603.5 ms (SD = 140.5), respectively, for
complex words. Adjusting these means for the
number of syllables yields, for simple words,
600.3and 608.0 ms, for high- and low-frequency
words, respectively, and for complex words, 578.4
ms and 616.0, respectively.
The latency analyses showed a main effect of fre-
quency, β=0.05, t=5.3, p< .0001. The main
effect of the number of syllables, which was included
as a covariate, was also signicant, β=0.07,t= 9.8, p
< .0001. The main effect of complexity was not signi-
cant, β=0.01, t=0.9, p= .4, but the critical inter-
action between frequency and complexity was,
indicating a steeper frequency slope for complex
than simple words, β=0.06, t=3.5, p= .0005.
The Bayes Factor provided strong evidence for
the presence of this interaction, BF = 37.9 (±1.1%).
Follow-up analyses
An unexpected nding in Experiment 2 is the absence
of a signicant main effect of complexity. As the
explanation of the complexity-by-frequency inter-
action, in the Orthographic Depth Hypothesis frame-
work, is based on the assumption that complex
words are more difcult to process by the sublexical
route than simple words, this nding might compro-
mise our conclusion. A possible explanation is the
inclusion of relatively high-frequency words in our
item set. Previous research has shown that the com-
plexity effect is diminished for high- compared to
low-frequency words (Rey et al., 1998). LME provides
the slope estimates at the point where the indepen-
dent variables equal to zero. As we used centred
JOURNAL OF COGNITIVE PSYCHOLOGY 9
Downloaded by [Aix-Marseille Université], [Dr Eddy Cavalli] at 02:57 09 May 2016
log frequency as an independent variable, it is poss-
ible that the slope estimate of the complexity effect
is based on a point where the frequency is too high
to show a complexity effect. To test this possibility,
we conducted follow-up tests of the effect of com-
plexity separately for low-frequency (log frequency
<1) and high-frequency (log frequency >1) words.
Indeed, the data showed slower RTs for complex
than simple items for low-frequency words, β=0.03,
t= 1.6, and faster RTs for complex than simple items
for high-frequency words, β=0.05, t=2.8. The
Bayes Factors provided equivocal evidence for the
expected inhibitory complexity effect for low-fre-
quency words, BF = 0.4 (±1.1%), and weak evidence
for the unexpected facilitatory complexity effect for
high-frequency words, BF = 4.9 (±0.9%).
As the facilitatory effect (faster RTs for complex
than simple words) for high-frequency words goes
both against the existing literature and existing
models of reading, we considered possible confounds
that could be driving this counter-intuitive pattern. In
matching items with complex correspondences
against items with simple correspondences, it is cus-
tomary to match for the number of letters, not pho-
nemes (Rastle & Coltheart, 1998; Rey et al., 1998).
This is a conservative approach: As complex words
contain more letters than phonemes, the complex
condition necessarily has fewer phonemes than the
simple condition. This could be counteracting the
complexity effect in our analysis. Indeed, when
adding the number of phonemes as an additional pre-
dictor, we still get the predicted inhibitory effect
(numerically) for low-frequency words, β= 0.03, t=
1.6 (adjusted means: 579.9 and 598.5 ms for complex
and simple words, respectively), and the unexpected
numerically facilitatory effect for high-frequency
words, β=0.04, t=2.3 (adjusted means: 622.2
and 603.0 ms, for complex and simple words, respect-
ively), but now the Bayes Factor provides weak
evidence for the expected inhibitory effect for low-fre-
quency words, BF = 3.9 (±0.8%), and equivocal evi-
dence for the unexpected facilitatory effect for high-
frequency words, BF = 1.6 (±0.8%). This means that
the current data does not give us any conclusive evi-
dence about whether or not there is a complexity
effect for high-frequency words after taking into
account the number of syllables and phonemes as a
covariate,butsuggeststhattheremightbethe
expected inhibitory effect for low-frequency words.
Note that in a post-hoc analysis of the full French
data set which includes the number of phonemes as
well as the number of syllables as covariates, we con-
tinue to get evidence for a frequency-by-complexity
interaction, β=0.07, t=3.7, BF = 7.3 (±0.8%),
suggesting that the key result is robust.
As with the English analyses, we performed one
nal post-hoc test to ensure that none of the poten-
tial covariates from Table 2 compromise our results.
We repeated the analyses while including the main
effect of OLD20 and PLD20 (which co-varied with
complexity), and the main effect of bigram fre-
quency and its interactions with frequency and com-
plexity. For bigram frequency, missing values were
replaced with the global mean. As in the previous
model, we also included the number of phonemes,
number of syllables, and the critical two main
effects of frequency, complexity, and their inter-
action. Again, the pattern of results remained
stable, with a main effect of frequency, β=0.05,
t=5.1, an unexpected facilitatory effect of com-
plexity, β=0.04, t=2.4, and the critical inter-
action, β=0.07, t=3.5. The adjusted mean RTs
are 574.2 and 604.3 ms for high-frequency
complex and simple words, respectively, and 612.0
and 612.6 ms for low-frequency complex and
simple words, respectively. The evidence for the
critical interaction between complexity and fre-
quency was BF = 13.6 (±0.86%).
Table 2. Potential moderators, and how they co-vary with the critical manipulations of Experiment 2.
Potential covariate
Main effect of
frequency
Main effect of
complexity
Interaction of complexity and
frequency
Overall average and standard
deviation
Number of letters t=0.12, p= .91 t= 1.59, p= .11 t= 1.59, p= .11 6.62 (1.60)
Number of phonemes t=0.93, p= .35 t=13.22, p< .01* t= 0.99, p= .32 5.54 (1.70)
Orthographic Levenshtein
Distance
t=1.58, p= .12 t=2.76, p< .01* t=0.71, p= .48 2.03 (0.47)
Bigram frequency t= 1.13, p= .26 t= 5.57, p< .01* t=3.42, p< .01* 1,128.61 (603.59)
Grapheme consistency t=1.24, p= .22 t=0.32, p= .75 t= 0.63, p= .53 84.14 (9.44)
Phonological Levenshtein
Distance
t=0.59, p= .56 t=7.79, p<.01* t=0.76, p= .45 1.82 (0.60)
Note: Asterisks (*) show a signicant relationship at the 0.05-level. Log subtitle frequency, Orthographic Levenshtein Distance, and Phonologicial
Levenshtein Distance are retrieved from Lexique (New et al., 2004); grapheme consistency and bigram frequency from Manulex (Lété et al.,
2004). Note that Manulex has the bigram frequency for only 314 out of the 384 words; missing cells were excluded for the analysis which included
bigram frequency as the dependent variable.
10 X. SCHMALZ ET AL.
Downloaded by [Aix-Marseille Université], [Dr Eddy Cavalli] at 02:57 09 May 2016
In sum, we found evidence for the critical inter-
action, showing that the frequency effect is stronger
for words with complex compared to words with
simple correspondences. This suggests that, like
unpredictability, complexity acts as a source of ortho-
graphic depth by impairing the sublexical route. This
leads to a relative increase in the degree to which
the lexical route contributes to the nal output.
General discussion
Although orthographic depth has been studied exten-
sively throughout the past decades, it is unclear
whether the complexity and the unpredictability of
the sublexical correspondences affect skilled reading
processes in the same way, or whether these two con-
structs have a differential effect on the cognitive pro-
cesses (Schmalz et al., 2015). The Orthographic Depth
Hypothesis proposes that in deep orthographies, the
lexical route becomes relatively more important,
because the sublexical information is less efcient in
retrieving a correct pronunciation (Katz & Frost, 1992).
We hypothesised that this may not be the case for
orthographies with complex but predictable sublexical
correspondences, such as French, because here the
sublexical information is, in principle, sufcient to
derive a correct pronunciation. However, increased
lexical processing may be observed if complex corre-
spondences slow down the sublexical route, thus
allowing more time for the lexical route to retrieve
the relevant phonological information. We found
support for the latter possibility inasmuch as the fre-
quency effect (a marker of lexical processing) was
greater for words with complex sublexical correspon-
dences than for those with simple correspondences.
Predictability within models of reading
Experiment 1 indicates that, in a within-experiment
manipulation, the frequency effect is stronger for
English unpredictable (ghost) than predictable
(forge) words. In line with the ODH (Katz & Frost,
1992) and with previous research (Frost et al.,
1987), this suggest that unpredictability increases
the relative reliance on lexical processing, as the
sublexical processing cannot be resolved without
lexical knowledge.
Note that, within a rule-based model of reading,
the theoretical explanation of a predictability-by-fre-
quency interaction is slightly different from what is
likely to happen in the case of complexity, even
though they result in an identical behavioural
pattern(Coltheartetal.,2001). If the sublexical route
uses a set of print-to-speechconversionrules,thesub-
lexical output for words with irregular correspon-
dences, which do not comply to the rules, will be an
incorrect response (e.g., in English /dept/ instead of
/dεt/ for the written word debt). A conict would
then take place in the phonological buffer, when com-
bining the output of the lexical and the sublexical
routes. Such a conict may be resolved by postponing
the initiation of the verbal response, until sufcient
activation from the lexical route has accumulated to
trump the incorrect phonemic activation from the sub-
lexical route. This would explain the main effect of
unpredictability, because the pronunciation of unpre-
dictable or irregular words is delayed, due to the con-
ict between the two routes. Furthermore, this conict
does not occur for words with predictable or regular
correspondences, therefore the pronunciation does
not need to be delayed until the lexical route trumps
the activation of the sublexical route. As a result, rela-
tively stronger lexical involvement is needed to resolve
the pronunciation of unpredictable words. For predict-
able words, the sublexical route does not need to be
suppressed for a correct pronunciation.
Within a connectionist framework (Perry et al.,
2007; Plaut et al., 1996), the sublexical route would
be predicted to operate more slowly for unpredict-
able compared to predictable words. Unpredictable
words, by denition, contain inconsistent correspon-
dences (e.g., in the word ghost,the grapheme ois
inconsistent, as it can also be pronounced as in
lost). It is possible that phonemic activation associ-
ated with inconsistent graphemes is slower than the
activation of consistent graphemes (e.g., sh /ʃ/
would be activated faster than th /θ/). In this
case, unpredictability (or, more specically, inconsis-
tency) would lead to an overall slowdown of the
sublexical route, thus giving more time for the
lexical or semantic information to contribute to the
verbal output. Thus, in contrast to rule-based
models, connectionist models would suggest that
the mechanism responsible for the predictability-
by-frequency interaction is very similar to the mech-
anism underlying the complexity-by-frequency
interaction.
Complexity within models of reading
Experiment 2 examined whether the frequency
effect would be stronger, for French, in words con-
taining complex (multi-letter) correspondences
(gâteau), compared to words with simple
JOURNAL OF COGNITIVE PSYCHOLOGY 11
Downloaded by [Aix-Marseille Université], [Dr Eddy Cavalli] at 02:57 09 May 2016
correspondences only (garnir). Again, there was
evidence for an interaction, suggesting that com-
plexity, like unpredictability, increases the relative
importance of lexical processing.
As previous studies on the ODH have used cross-lin-
guistic comparisons of pairs of orthographic systems
that differed in both complexity and unpredictability
(e.g., Serbo-Croatian/English, German/English), our
study is the rst to suggest that complexity affects
the ratio of lexical-to-sublexical processing. Presum-
ably, this is due to a slowdown of the sublexical
decoding process, which is caused by the application
of complex multi-letter rules. More specically,
complex rules could lead to a conict between the
activation of the phoneme corresponding to a multi-
letter grapheme and the phonemes corresponding
to its underlying individual letters, as proposed by
the dual-route cascaded (DRC) model of reading
aloud (Rastle & Coltheart, 1998). In a word like
garnir,each letter maps onto its default phoneme.
A simple word would lead to faster activation of the
phonemes in the output buffer from the sublexical
route, thus reducing the relative contribution of the
lexical route in achieving the nal pronunciation. For
a word with complex correspondences, like gâteau,
the activation of the phonemes of the individual
letters, e(/ε/),a(/a/),and u(/y/), would
cause a conict within the sublexical route, as the
three letters need to be combined into a single gra-
pheme and mapped onto the correct phoneme /o/.
This would slow down the output of the sublexical
route, such that the lexical route has a larger contri-
bution to the nal output.
We did not nd a main effect of complexity, thus
failing to replicate the results of Rastle and Coltheart
(1998)andReyetal.(1998). Including articulatory vari-
ables, namely the number of syllables and the number
of phonemes, as covariates, provided a more coherent
picture. Here, there was some evidence for an effect of
complexity in the low-frequency condition, though this
emerged only in the covariate analysis that included
both the number of syllables and the number of pho-
nemes. Thus, it seems that articulatory processes coun-
teract the effect of complexity. Articulatory processes
affect reading-aloud latencies at a post-lexical stage
(Cholin & Levelt, 2009; Cholin, Schiller, & Levelt, 2004),
which results in in facilitation of the verbal response,
driven by a smaller number of phonemes and syllables
for all types of words, regardless of frequency.
The effect of complexity counteracts this facilitatory
articulation-level effect especially for low-frequency
words, as complexity operates on the sublexical level.
Notwithstanding the lack of a main effect of com-
plexity, Experiment 2 provided strong evidence for
an interaction between complexity and frequency
in French. This suggests that sublexical information
plays a role in determining the net ratio of lexical-
to-sublexical processing, even if the output is
driven to a great extent by the lexical route. While
the process of reading aloud appears to happen at
the same rate for complex as for simple words,
there is relatively more contribution from the
lexical than the sublexical route.
The frequency effect: Lexical, semantic, or
sublexical marker?
Finally, it is worth expanding on our central assump-
tion that the frequency effect is a marker of lexical
processing. While it is generally assumed that fre-
quency reects some kind of threshold of the acti-
vation of entries in a mental orthographic lexicon
(e.g., Coltheart et al., 2001; Taft, 1991), there are
alternative views of how the frequency effect
works. First, it is possible that frequency effects
reect other constructs that are strongly correlated,
such as imageability (Strain, Patterson, & Seiden-
berg, 1995), age-of-acquisition (Zevin & Seidenberg,
2002), or contextual diversity (Adelman, Brown, &
Quesada, 2006). We did not match for these vari-
ables, as it would have substantially limited the
choice of items. The norms for these variables are
not available for the majority of our items, therefore
we can also not include them as post-hoc covariates
in follow-up analyses. This does not present a
problem for our conclusions, however, as these vari-
ables reect lexical-semantic activation and thus
measure processes which occur broadly within the
lexical route. The DRC (Coltheart et al., 2001) and
CDP+ (Perry et al., 2007; Perry, Ziegler, & Zorzi,
2010) models make a distinction between an ortho-
graphic lexicon and a purely semantic route. The
semantic route can be reached either by activation
from the orthographic lexicon or the phonological
lexicon, and in turn sends activation to the non-
semantic lexical components. These models would
therefore predict a close link between non-semantic
and semantic lexical processes. Triangle models do
not make a distinction between a semantic and a
non-semantic route, as there is no purely ortho-
graphic representation of whole words, thus there
is an even closer link between semantic marker
effects and the lexical route (Plaut et al., 1996; Sei-
denberg & McClelland, 1989).
12 X. SCHMALZ ET AL.
Downloaded by [Aix-Marseille Université], [Dr Eddy Cavalli] at 02:57 09 May 2016
As a second alternative explanation of the fre-
quency effect, it is possible that it reects the fre-
quency not of the whole word, but of the letters
and letter clusters which are contained in the word.
Thus, in a connectionist model, it is possible to
show word frequency effects in the absence of an
orthographic lexicon, because frequent letter clusters
and their pronunciations are easier to learn (Plaut
et al., 1996). This would imply that the frequency
effect is a measure of sublexical processing. If a sub-
lexical mechanism, reecting the frequency of letter
clusters, drives the interactions with complexity or
predictability, one would expect the interaction to dis-
appear once bigram frequency is taken into account.
However, in Experiment 1 we found the frequency-
by-predictability interaction while the manipulations
did not co-vary with bigram frequency, and in Exper-
iment 2, the frequency-by-consistency interaction
remained robust after taking into account bigram fre-
quency as a covariate. Thus, we can exclude the possi-
bility that the frequency effect in our study reects a
sublexical process.
Conclusion
The current study is the rst, to our knowledge, to
empirically address the hypothesis that ortho-
graphic depth consists of various components
that differentially affect skilled reading processes.
The experiments reported here suggest that both
complexity and unpredictability independently
increase relative reliance on the lexical route. This
provides support for the ODH, and the cognitive
mechanism that Katz and Frost (1992)proposed
as driving the cross-linguistic differences associ-
ated with orthographic depth: complexity and
unpredictability both act to impair the efciency
of the sublexical route, which allows for a relatively
greater inuence of the lexical route in retrieving
the words pronunciation.
Acknowledgements
We thank Melvin Yap and David Balota for making the
trial-level ELP data available to us, and Sachiko Kinoshita
for valuable discussions.
Disclosure statement
No potential conict of interest was reported by the
authors.
References
Adelman, J. S., Brown, G. D., & Quesada, J. F. (2006).
Contextual diversity, not word frequency, determines
word-naming and lexical decision times. Psychological
Science,17(9), 814823. doi:10.1111/j.1467-9280.2006.
01787.x
Andrews, S. (1982). Phonological recoding: Is the regularity
effect consistent? Memory & Cognition,10(6), 565575.
doi:10.3758/BF03202439
Andrews, S., & Scarratt, D. R. (1998). Rule and analogy
mechanisms in reading nonwords: Hough dou peapel
rede gnew wirds? Journal of Experimental Psychology:
Human Perception and Performance,24(4), 10521086.
doi:10.1037//0096-1523.24.4.1052
Aro, M., & Wimmer, H. (2003). Learning to read: English in
comparison to six more regular orthographies. Applied
Psycholinguistics,24, 621635 doi:10.1017/
S0142716403000316.
Baayen, R. H. (2008). Analyzing linguistic data: A practical
introduction to statistics using R. Cambridge:
Cambridge University Press.
Baayen, R. H., Davidson, D. J., & Bates, D. M. (2008). Mixed-
effects modeling with crossed random effects for sub-
jects and items. Journal of Memory and Language,59
(4), 390412. doi:10.1016/j.jml.2007.12.005
Balota, D. A., Yap, M. J., Cortese, M. J., Hutchison, K. A.,
Kessler, B., Loftis, B., Treiman, R. (2007). The English
Lexicon Project. Behavior Research Methods,39(3),
445459. doi:10.3758/Bf03193014
Barr, D. J., Levy, R., Scheepers, C., & Tily, H. J. (2013). Random
effects structure for conrmatory hypothesis testing:
Keep it maximal. Journal of Memory and Language,68
(3), 255278. doi:10.1016/J.Jml.2012.11.001
Bates, D., Kliegl, R., Vasishth, S., & Baayen, H. (2015).
Parsimonious mixed models. arXiv preprint
arXiv:1506.04967.
van den Bosch, A., Content, A., Daelemans, W., & de Gelder,
B. (1994). Measuring the complexity of writing systems.
Journal of Quantitative Linguistics,1(3), 178188. doi:10.
1080/09296179408590015
Bridgeman, B. (1987). Is the dual-route theory possible in pho-
netically regular languages. Behavioral and Brain Sciences,
10(2), 331332. doi:10.1017/S0140525X00048184
Brysbaert, M., Buchmeier, M., Conrad, M., Jacobs, A. M.,
Bölte, J., & Böhl, A. (2011). The word frequency effect:
A review of recent developments and implications for
the choice of frequency estimates in German.
Experimental Psychology,58(5), 412. doi:10.1027/1618-
3169/a000123
Brysbaert, M., & New, B. (2009). Moving beyond Kučera and
Francis: A critical evaluation of current word frequency
norms and the introduction of a new and improved
word frequency measure for American English.
Behavior Research Methods,41(4), 977990. doi:10.
3758/BRM.41.4.977
Chateau, D., & Jared, D. (2003). Spellingsound consistency
effects in disyllabic word naming. Journal of Memory
and Language,48(2), 255280. doi:10.1016/S0749-
596X(02)00521-1
JOURNAL OF COGNITIVE PSYCHOLOGY 13
Downloaded by [Aix-Marseille Université], [Dr Eddy Cavalli] at 02:57 09 May 2016
Cholin, J., & Levelt, W. J. (2009). Effects of syllable prep-
aration and syllable frequency in speech production:
Further evidence for syllabic units at a post-lexical
level. Language and Cognitive Processes,24(5), 662
684. doi:10.1080/01690960802348852
Cholin, J., Schiller, N. O., & Levelt, W. J. (2004). The prep-
aration of syllables in speech production. Journal of
Memory and Language,50(1), 4761. doi:10.1016/j.jml.
2003.08.003
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. doi:10.1037//0033-295x.108.1.204
Cortese, M. J., & Simpson, G. B. (2000). Regularity effects in
word naming: What are they? Memory & Cognition,28
(8), 12691276. doi:10.3758/Bf03211827
Cramer, A. O., van Ravenzwaaij, D., Matzke, D.,
Steingroever, H., Wetzels, R., Grasman, R. P.,
Wagenmakers, E.-J. (2015). Hidden multiplicity in
exploratory multiway ANOVA: Prevalence and reme-
dies. Psychonomic Bulletin & Review,18. doi:10.3758/
s13423-015-0913-5
Ferrand, L., New, B., Brysbaert, M., Keuleers, E., Bonin, P.,
Méot, A., Pallier, C. (2010). The French Lexicon
Project: Lexical decision data for 38,840 French words
and 38,840 pseudowords. Behavior Research Methods,
42(2), 488496. doi:10.3758/BRM.42.2.488
Forster, K. I., & Forster, J. (2003). DMDX: A Windows display
program with millisecond accuracy. Behavior Research
Methods, Instuments & Computers,35, 116124. doi:10.
3758/BF03195503
Frith, U., Wimmer, H., & Landerl, K. (1998). Differences in
phonological recoding in German- and English-speak-
ing children. Scientic Studies of Reading,2(1), 3154.
doi:10.1207/s1532799xssr0201_2
Frost, R. (1994). Prelexical and postlexical strategies in
reading: Evidence from a deep and a shallow orthogra-
phy.. Journal of Experimental Psychology: Learning,
Memory & Cognition,20(1), 116129. doi:10.1037/0278-
7393.20.1.116
Frost, R. (2012). Towards a universal model of reading.
Behavioral and Brain Sciences,35(5), 263279. doi:10.
1017/S0140525X11001841
Frost, R., Katz, L., & Bentin, S. (1987). Strategies for visual
word recognition and orthographical depth: A multilin-
gual comparison.. Journal of Experimental Psychology:
Human Perception & Performance,13(1), 104115.
doi:10.1037/0096-1523.13.1.104
Glushko, R. (1979). The organization and activation of ortho-
graphic knowledge in reading aloud. Journal of
Experimental Psychology: Human Perception and
Performance,5(4), 674691. doi:10.1037/0096-1523.5.4.674
van Heuven, W., Mandera, P., Keuleers, E., & Brysbaert, M.
(2014). SUBTLEX-UK: A new and improved word fre-
quency database for British English. The Quarterly
Journal of Experimental Psychology,67(6), 11761190.
doi:10.1080/17470218.2013.850521
Jared, D. (1997). Spelling-sound consistency affects the
naming of high-frequency words. Journal of Memory
and Language,36(4), 505529. doi:10.1006/jmla.1997.
2496
Jared, D. (2002). Spelling-sound consistency and
regularity effects in word naming. Journal of Memory
and Language,46(4), 723750. doi:10.1006/jmla.2001.
2827
Jared, D., McRae, K., & Seidenberg, M. (1990). The basis of
consistency effects in word naming. Journal of Memory
and Language,29(6), 687715. doi:10.1016/0749-596X
(90)90044-Z
Katz, L., & Feldman, L. (1983). Relation between pronuncia-
tion and recognition of printed words in deep and
shallow orthographies. Journal of Experimental
Psychology: Learning Memory and Cognition,9(1), 157
166. doi:10.1037/0278-7393.9.1.157
Katz, L., & Frost, R. (1992). The reading process is different
for different orthographies: The orthographic depth
hypothesis. In R. Frost & L. Katz (Eds.), Orthography, pho-
nology, morphology, and meaning (pp. 6784).
Amsterdam: Elsevier Science.
Kearns, D. M., Steacy, L. M., Compton, D. L., Gilbert, J. K.,
Goodwin, A. P., Cho, E., Collins, A. A. (2014).
Modeling polymorphemic word recognition exploring
differences among children with early-emerging and
late-emerging word reading difculty. Journal of
Learning Disabilities.doi:10.1177/0022219414554229
Keuleers, E., Diependaele, K., & Brysbaert, M. (2010).
Practice effects in large-scale visual word recognition
studies: A lexical decision study on 14,000 Dutch
mono-and disyllabic words and nonwords. Language
Sciences,1, 174. doi:10.3389/fpsyg.2010.00174
Keuleers, E., Lacey, P., Rastle, K., & Brysbaert, M. (2012). The
British Lexicon Project: Lexical decision data for 28,730
monosyllabic and disyllabic English words. Behavior
Research Methods,44(1), 287304. doi:10.3758/S13428-
011-0118-4
Landerl, K., Wimmer, H., & Frith, U. (1997). The impact of
orthographic consistency on dyslexia: A German-
English comparison. Cognition,63, 315334. doi:10.
1016/S0010-0277(97)00005-X
Lété, B., Sprenger-Charolles, L., & Colé, P. (2004).
MANULEX: A grade-level lexical database from
French elementary school readers. Behavior Research
Methods, Instruments, & Computers,36(1), 156166.
doi:10.3758/BF03195560
Mann, V., & Wimmer, H. (2002). Phoneme awareness and
pathways into literacy: A comparison of German and
American children. Reading and Writing: An
Interdisciplinary Journal,15, 653682. doi:10.1023/
A:1020984704781
Marinus, E., & de Jong, P. F. (2010). Variability in the word-
reading performance of dyslexic readers: Effects of
letter length, phoneme length and digraph presence.
Cortex,46(10), 12591271. doi:10.1016/j.cortex.2010.06.
005
Marinus, E., Nation, K., & de Jong, P. (2015). Density and
length in the neighbourhood: Explaining cross-linguis-
tic differences in learning to read in English and
Dutch. Journal of Experimental Child Psychology,139,
127147. doi:10.1016/j.jecp.2015.05.006
Medler, D. A., & Binder, R. J. (2005). MCWord: An on-line
orthographic database of the English language.
Retrieved from http://www.neuro.mcw.edu/mcword/
14 X. SCHMALZ ET AL.
Downloaded by [Aix-Marseille Université], [Dr Eddy Cavalli] at 02:57 09 May 2016
Morey, R. D., & Rouder, J. N. (2014). Package BayesFactor.
Retrieved from http://cran.r-project.org/web/packages/
BayesFactor/BayesFactor.pdf
New, B., Brysbaert, M., Veronis, J., & Pallier, C. (2007). The
use of lm subtitles to estimate word frequencies.
Applied Psycholinguistics,28(04), 661677. doi:10.1017/
S014271640707035X
New, B., Pallier, C., Brysbaert, M., & Ferrand, L. (2004).
Lexique 2: A new French lexical database. Behavior
Research Methods, Instruments, & Computers,36(3),
516524. doi:10.3758/BF03195598
Perry, C., Ziegler, J., & Zorzi, M. (2007). Nested incremental
modeling in the development of computational theories:
The CDP+ model of reading aloud. Psychological Review,
114(2), 273315. doi:10.1037/0033-295X.114.2.273
Perry, C., Ziegler, J., & Zorzi, M. (2010). Beyond single sylla-
bles: Large-scale modeling of reading aloud with the
connectionist dual process (CDP++) model. Cognitive
Psychology,61(2), 106151. doi:10.1016/j.cogpsych.
2010.04.001
Plaut, D. C., McClelland, J. L., Seidenberg, M., & Patterson, K.
(1996). Understanding normal and impaired word
reading: Computational principles in quasi-regular
domains. Psychological Review,103(1), 56115. doi:10.
1037/0033-295x.103.1.56
Pritchard, S. C., Coltheart, M., Palethorpe, S., & Castles, A.
(2012). Nonword reading: Comparing dual-route cas-
caded and connectionist dual-process models with
human data. Journal of Experimental Psychology:
Human Perception and Performance,38(5), 12681288.
doi:10.1037/A0026703
Protopapas, A. (2007). Checkvocal: A program to facilitate
checking the accuracy and response time of vocal
responses from DMDX. Behavior Research Methods,39
(4), 859862. doi:10.3758/bf03192979
Rastle, K., & Coltheart, M. (1998). Whammies and double
whammies: The effect of length on nonword reading.
Psychonomic Bulletin & Review,5(2), 277282. doi:10.
3758/BF03212951
Rau, A. K., Moll, K., Snowling, M. J., & Landerl, K. (2015).
Effects of orthographic consistency on eye movement
behavior: German and English children and adults
process the same words differently. Journal of
Experimental Child Psychology,130,92105. doi:10.
1016/j.jecp.2014.09.012
Rey, A., Jacobs, A. M., Schmidt-Weigand, F., & Ziegler, J. C.
(1998). A phoneme effect in visual word recognition.
Cognition,68(3), B71B80. doi:10.1016/S0010-0277(98)
00051-1
Robidoux, S., & Pritchard, S. C. (2014). Hierarchical cluster-
ing analysis of reading aloud data: A new technique for
evaluating the performance of computational models.
Frontiers in Psychology,5.doi:10.3389/Fpsyg.2014.
00267
Rouder, J. N., Speckman, P. L., Sun, D. C., Morey, R. D., &
Iverson, G. (2009). Bayesian ttests for accepting and
rejecting the null hypothesis. Psychonomic Bulletin &
Review,16(2), 225237. doi:10.3758/Pbr.16.2.225
Schmalz, X., Marinus, E., Coltheart, M., & Castles, A. (2015).
Getting to the bottom of orthographic depth.
Psychonomic Bulletin and Review,22(6), 16141629.
doi:10.3758/s13423-015-0835-2
Schmalz, X., Marinus, E., Robidoux, S., Palethorpe, S.,
Castles, A., & Coltheart, M. (2014). Quantifying the
reliance on different sublexical correspondences in
German and English. Journal of Cognitive Psychology,
26(8), 831852. doi:10.1080/20445911.2014.968161
Seidenberg, M., & McClelland, J. L. (1989). A distributed,
developmental model of word recognition and
naming. Psychological Review,96(4), 523568. doi:10.
1037/0033-295X.96.4.523
Seymour, P., Aro, M., & Erskine, J. (2003). Foundation lit-
eracy acquisition in European orthographies. British
Journal of Psychology,94,143174. doi:10.1348/
000712603321661859
Share, D. (2008). On the anglocentricities of current
reading research and practice: The perils of overreliance
on an outlierorthography. Psychological Bulletin,134
(4), 584615. doi:10.1037/0033-2909.134.4.584
Simmons, J. P., Nelson, L. D., & Simonsohn, U. (2011). False-
positive psychology undisclosed exibility in data col-
lection and analysis allowspresenting anything as sig-
nicant. Psychological Science,22(11), 13591366.
doi:10.1177/0956797611417632
Strain, E., Patterson, K., & Seidenberg, M. S. (1995).
Semantic effects in single-word naming. Journal of
Experimental Psychology: Learning, Memory, and
Cognition,21(5), 1140. doi:10.1037/0278-7393.21.5.1140
Taft, M. (1991). Reading and the mental lexicon. Hove, UK:
Lawrence Erlbaum.
Treiman, R., Kessler, B., & Bick, S. (2003). Inuence of conso-
nantal context on the pronunciation of vowels: A com-
parison of human readers and computational models.
Cognition,88(1), 4978. doi:10.1016/s0010-0277(03)
00003-9
Turvey, M., Feldman, L., & Lukatela, G. (1984). The Serbo-
Croatian orthography constrains the reader to a phono-
logically analytic strategy. In L. Henderson (Ed.),
Orthographies and reading: Perspectives from cognitive
psychology, neuropsychology, and linguistics (pp. 81
89). Hillsdale, NJ: Lawrence Erlbaum.
Yap, M. J., & Balota, D. A. (2009). Visual word recognition of
multisyllabic words. Journal of Memory and Language,
60(4), 502529. doi:10.1016/j.jml.2009.02.001
Yarkoni, T., Balota, D., & Yap, M. (2008). Moving beyond
Colthearts N: A new measure of orthographic similarity.
Psychonomic Bulletin & Review,15(5), 971979. doi:10.
3738/Pbr.15.5.971
Zevin, J. D., & Seidenberg, M. S. (2002). Age of acquisition
effects in word reading and other tasks. Journal of
Memory and Language,47(1), 129. doi:10.1006/jmla.
2001.2834
Ziegler, J., Perry, C., & Coltheart, M. (2000). The DRC model of
visual word recognition and reading aloud: An extension
to German. European Journal of Cognitive Psychology,12
(3), 413430. doi:10.1080/09541440050114570
Ziegler, J., Perry, C., & Coltheart, M. (2003). Speed of lexical
and nonlexical processing in French: The case of the
regularity effect. Psychological Bulletin & Review,10(4),
947953. doi:10.3758/BF03196556
JOURNAL OF COGNITIVE PSYCHOLOGY 15
Downloaded by [Aix-Marseille Université], [Dr Eddy Cavalli] at 02:57 09 May 2016
... However, to the extent that they are present within any orthography, all the preceding factors reduce spelling-sound and sound-spelling consistency. Evidence from a variety of European languages indicates that while orthographic complexity and inconsistency at the letter-sound mapping level are separable influences on word-level literacy performance, it is inconsistency that has the more deleterious effects on the acquisition, growth rate, and fluent reading of words (e.g., Marinelli, Romani, Burani, McGowan, & Zoccolotti, 2016;Schmalz et al., 2016). ...
... This raises the possibility that the observed variations in word reading usually attributed to differences in preferred grain sizes due to orthographic consistency might be attributable to other factors. At the grapheme-phoneme level, effects of grapheme complexity (Schmalz et al., 2016) and to a greater extent, of consistency have been reported for reading in the inconsistent orthographies of English and French (e.g., Chateau & Jared, 2003;Kessler & Treiman, 2001;Schmalz et al., 2016). Interestingly, studies of skilled readers of Italian and German (e.g., Burani, Barca, & Ellis, 2006;De Simone et al., 2021) show that graphotactic inconsistency -which can be reliably disambiguated by contextual and/or letter sequencing regularities -impedes reading performance even in relatively consistent orthographies. ...
... This raises the possibility that the observed variations in word reading usually attributed to differences in preferred grain sizes due to orthographic consistency might be attributable to other factors. At the grapheme-phoneme level, effects of grapheme complexity (Schmalz et al., 2016) and to a greater extent, of consistency have been reported for reading in the inconsistent orthographies of English and French (e.g., Chateau & Jared, 2003;Kessler & Treiman, 2001;Schmalz et al., 2016). Interestingly, studies of skilled readers of Italian and German (e.g., Burani, Barca, & Ellis, 2006;De Simone et al., 2021) show that graphotactic inconsistency -which can be reliably disambiguated by contextual and/or letter sequencing regularities -impedes reading performance even in relatively consistent orthographies. ...
Chapter
Full-text available
This chapter begins with an overview of the relevant characteristics of alphabetic orthographies, noting those that are known to influence both skilled reading and its development. It presents theories of reading that include a cross‐linguistic perspective. The chapter summarizes current understanding of the processes involved in skilled word reading in the light of variations in grapho‐phonemic consistency and considers the available evidence regarding word‐reading profiles of adults with developmental reading disorders. Skilled word reading involves recognizing words fluently in the service of extracting meaning from written language. Most of the preceding effects are moderated by orthographic consistency. A central issue in cross‐linguistic research on reading development concerns the nature and timing of influences of orthographic consistency on reading behavior and its cognitive underpinnings. The growing consensus indicates that the foundations of literacy are universal across alphabetic orthographies.
... t pronunciation of novel and unfamiliar words. Conversely, readers of opaque orthographies have been found to read primarily via the lexical route by using groups of letters, such as bodies or morphemes, and other types of lexical information specific to each word (Miller, 2018). It has been found that orthographic depth affects both reading aloud X. Schmalz et al., 2016) and silent reading (A. K. Rau et al., 2015;Schroeder et al., 2022), as well as reading in skilled and beginning readers (Caravolas et al., 2019;C. V. Marinelli et al., 2016;A. K. Rau et al., 2015A. K. Rau et al., , 2016, and also determines specific symptoms associated with developmental dyslexia (C. V. Marinelli et al., 2023;Provazza e ...
... graphic transparency within visual word recognition, which is reflected by the authors' choice of comparing German and French. In fact, although German orthography is considered to be more transparent than French (Seymour et al., 2003), the orthographic complexity of French does not cause it to be unpredictable in the grapheme-to-phoneme direction X. Schmalz et al., 2016). This means that once readers of French have learnt the numerous sets of complex rules that govern the translation from print to sound, they can apply them virtually unmistaken. Therefore, German and French do not differ much in terms of orthographic depth in the reading direction, as readers of both can rely on predictable rules (as co ...
... In the DRC, lexical activation is necessary to obtain a correct pronunciation for irregular words; for complex words, sublexical processing is sufficient. Increased top-down lexical activation may be required for complex words, as compared to simple words, as well, in order to disambiguate between different parsing options (Schmalz et al., 2016); however, assuming that a reader knows the correct parsing, they can generate a correct pronunciation even while relying on sublexical knowledge only. ...
... This provides a justification for empirical studies which would aim to identify cognitive and behavioral consequences of each of the dimensions separately. Previous studies have either used artificial orthographies to dissociate the behavioral consequences of complexity versus unpredictability (Schmalz et al., 2022), or relied on French as the only orthography where complexity and unpredictability can be dissociated (De Simone et al., 2021;Schmalz et al., 2016). ...
Preprint
Full-text available
Cross-linguistic reading research often focuses on the effect of orthographic depth - the closeness of the relationship between print and speech. To understand its effect on reading, we need to be able to objectively quantify the level of orthographic depth of a given orthography. Previous work has suggested that different dimensions underlie orthographic depth, and it is not always clear if and how existing quantifications map onto these underlying dimensions. Here, we first examine how different measures relate conceptually to underlying theoretical dimensions. Then, we quantify the relative depth of 8 European orthographies. We use both 11 existing methods and two new approaches which have not been previously used to quantify orthographic depth: Distance-based measures relying on the closeness of the phonology of orthographically similar words, and mutual information, as a theory-neutral approach. The relationship between the different measures suggests that they map on two separate dimensions: the size of the orthographic units that map onto phonology and the systematicity of the mapping, in line with previous theoretical work which drew a distinction between complexity and unpredictability. The measures derived based on different theoretical assumptions largely show agreement. From a theoretical perspective, this prevents us from making differential predictions based on different approaches. From a practical perspective, this suggests that different measures may yield comparable results, as long as they tap into the same underlying dimension of orthographic depth.
... Besides, the association of statistical learning and reading ability in different modalities has been mostly confirmed in studies of speakers of languages with opaque orthographies (such as Chinese, Hebrew and English), and less is known about this association in languages with transparent orthographies (such as Hungarian). As Hungarian has shallow orthography, readers might rely on the phonological route more than the lexical route in reading compared to languages with more opaque orthographies [66][67][68], making acoustic and visual pattern learning abilities similarly important in reading. These factors make testing segmentation abilities both in the acoustic and visual modalities well motivated. ...
... The importance of acoustic statistical learning might be especially pronounced in the case of shallow orthographies like Hungarian. As in such systems, grapheme-phoneme correspondences are consistent, and there are few irregularities, readers in shallow orthography systems may rely primarily on the phonological route in reading compared to more opaque orthography systems, and knowledge about phoneme patterns extracted through statistical segmentation might contribute more strongly to reading abilities [66][67][68]. However, testing this relationship at earlier ages when reading skills are still developing may yield better insights into the relationship of statistical learning and reading skills (both in the visual and the acoustic modality). ...
Article
Full-text available
The vulnerability of statistical learning has been demonstrated in reading difficulties in both the visual and acoustic modalities. We examined segmentation abilities of Hungarian speaking adolescents with different levels of reading fluency in the acoustic verbal and visual nonverbal domains. We applied online target detection tasks, where the extent of learning is reflected in differences between reaction times to predictable versus unpredictable targets. Explicit judgments of well-formedness were also elicited in an offline two-alternative forced choice (2AFC) task. Learning was evident in both the acoustic verbal and visual nonverbal tasks, both in online and offline measures, but learning effects were larger both in online and offline tasks in the verbal acoustic condition. We haven’t found evidence for a significant relationship between statistical learning and reading fluency in adolescents in either modality. Together with earlier findings, these results suggest that the relationship between reading and statistical learning is dependent on the domain, modality and nature of the statistical learning task, on the reading task, on the age of participants, and on the specific language. The online target detection task is a promising tool which can be adapted to a wider set of tasks to further explore the contribution of statistical learning to reading acquisition in participants from different populations.
... Also, from the very first publications on the data has been used to decide which was the best word frequency measure to be used in French word processing research (see also Gimenes et al., 2016). Other authors used the data to evaluate the quality of a word processing measure combining speed and accuracy (Bruyer & Brysbaert, 2011), to build a measure of reliability for experiments with repeated measures (Courrieu et al., 2012), to test a computational model of word recognition (Norris & Kinoshita, 2012), to examine the effect of the syllabic structure of words (Chetail, 2014), to study the role of embodied cognition in word recognition (Bonin et al., 2015;Lalancette et al., 2024), to investigate the effect of the complexity of the print-to-speech conversion in word naming (Schmalz et al., 2016), to examine the importance of conceptual familiarity in word recognition (Chedid et al., 2019), to model the detailed effect of word length in lexical decision (Ginestet et al., 2019), to examine the effect of repeated letters in visual word recognition (Trifonova & Adelman, 2019), to select stimuli for word association experiments (Lakhzoum et al., 2023), to evaluate procedures for analyzing reaction times (Miller, 2023;Morís Fernández & Vadillo, 2020;Rousselet & Wilcox, 2018), and to select stimuli for a clinical memory battery (Monetta et al., 2024). ...
Preprint
Full-text available
This article reviews three large-scale French megastudies investigating word processing. These studies employed various paradigms, including word naming, lexical decision, and perceptual identification, to measure the accuracy and speed of word processing. The primary goal was to examine the effects of variables such word frequency and word length, on word processing. The findings from these studies have contributed significantly to our understanding of word recognition, providing valuable insights into methodological issues, theoretical questions, and practical applications.
... In their study, higher pronunciation variability was predicted by lower spelling-sound consistency in sub-lexical orthographic components and a smaller orthographic neighbourhood [31]. This therefore aligns with the construct of unpredictability, identified by Schmalz et al. as being one of the components of orthographic depth [32,33]. ...
Article
Full-text available
High-quality lexical representations depend on robust representations of written form (orthography), spoken form (phonology) and meaning (semantics), and strong bonds between them. Quality of lexical representations may be affected by amount of print exposure and the form of individual words. Words that are harder to decode (print-to-sound) may lead to fuzzy representations of the orthographic and phonological forms, potentially creating less stable foundations for semantic knowledge. These factors are difficult to disentangle in natural language research; in this registered report, we experimentally manipulated decoding ease and exposure at the item level. Adults read paragraphs describing invented meanings of pseudowords. Pseudowords appeared two or six times in a paragraph, and had easy (e.g. bamper ) or hard (e.g. uzide ) to decode spelling–sound mappings. Post-tests assessed word-form knowledge, orthography–semantic mappings and semantic–phonology mappings. Results showed that greater decoding ease improved learning of word forms and consequently also impacted on word meanings. Higher exposure frequency improved learning of word forms but not meanings. Exposure frequency also modulated the effect of decoding ease on word-form learning, with a stronger effect of decoding ease for fewer exposures. Disentangling effects of decoding ease from print exposure has important implications for understanding potential barriers to vocabulary learning.
... Various attempts to develop a measure of orthographic depth have been developed over the latter half of the twentieth century. Schmalz et al. (2015Schmalz et al. ( , 2016 reviewed several of these and showed that it was important to dissociate the concepts of complexity and unpredictability. Using an approach taken from the Dual Route Cascaded (DRC) theory (Gibson et al., 2001), the authors collated results for Dutch, English, French, German and Italian from several different sources (Schmalz et al., 2015(Schmalz et al., , p. 1620, although they did not report an irregularity score for Italian (see Table 4). ...
Thesis
Full-text available
Reading is one of the most important skills for children to master during their time in school. It is strongly connected to life outcomes, and as such, education ministries place it at the centres of their education policies. English is one of the most challenging alphabetic languages to learn to read, and governments of anglophone countries have spent many years working to improve the effectiveness of their literacy education. However, when examining International Large-Scale Assessments, it is notable that although students in anglophone countries are able to achieve among the highest reading levels, their poorest readers lag much further behind than the poorest readers in similarly successful non-anglophone countries. This study made use of data from PIRLS (2016) and PISA (2018) to investigate possible relationships between range of reading ability and language complexity variables related to orthography and morphology, as well as between range of reading ability and home-language disparity in anglophone countries. Pearson correlational analyses showed that orthographic complexity and morphological complexity were moderately correlated with range of reading ability in both datasets. Orthographic transparency was found to be strongly correlated with range of reading ability in the PISA dataset and very strongly correlated in the PIRLS dataset. Morphological unpredictability was not found to be correlated with either dataset. Home-language disparity was not shown to be connected with range of reading ability in the PISA dataset, but in the PIRLS dataset, students who never spoke English at home were shown to have a wider range of reading ability than other students.
... This seems to be, indeed the case (e.g., Martensen, Maris, & Dijkstra, 2003;Schmalz, Porshnev, & Marinus, 2017). However, evidence suggests that complexity continues to exert an effect in reading aloud latencies, even in skilled readers (e.g., Rastle & Coltheart, 1998; but see Schmalz, Beyersmann, Cavalli, & Marinus, 2016). Such an effect on latencies does not contradict the possibility that skilled readers approach ceiling knowledge of context-sensitive rules: the explanation for the effect on reading aloud latencies is that there is a conflict between the pronunciation of the single letters and of the multi-letter cluster. ...
Article
Full-text available
Orthographies vary in complexity (the number of multi-letter grapheme-phoneme rules describing print-to-speech regularities) and unpredictability (the number of words which cannot be read correctly, even with at-ceiling knowledge of the rules). To assess how these constructs affect reading acquisition, we used an artificial orthography learning paradigm, where participants learn to read pseudowords written in unfamiliar symbols, and subsequently read aloud novel words written in the same symbols (generalisation). In three experiments (third experiment pre-registered), we manipulated the consistency of symbol-to-sound mappings: in the first inconsistent condition, vowel pronunciation depended on the subsequent letter (condition complexity), and in the second inconsistent condition, vowel pronunciation was unpredictable from the context (condition unpredictability). Across experiments, we found that pseudowords with inconsistent mappings are more difficult to learn than pseudowords with consistent mappings only, regardless of whether the inconsistency is due to complexity or unpredictability. Numerically, participants learning orthographies containing unpredictable correspondences seem to be less likely to form rules, either for simple or for complex correspondences. We propose that rule extraction and distributional learning happens simultaneously during reading acquisition: in a mathematical model, we show that distributional learning may lead to more complete knowledge than rule extraction for orthographies that are high in unpredictability.
... This seems to be, indeed the case (e.g., Martensen, Maris, & Dijkstra, 2003;Schmalz, Porshnev, & Marinus, 2017). However, evidence suggests that complexity continues to exert an effect in reading aloud latencies, even in skilled readers (e.g., Rastle & Coltheart, 1998; but see Schmalz, Beyersmann, Cavalli, & Marinus, 2016). Such an effect on latencies does not contradict the possibility that skilled readers approach ceiling knowledge of context-sensitive rules: the explanation for the effect on reading aloud latencies is that there is a conflict between the pronunciation of the single letters and of the multi-letter cluster. ...
Preprint
Full-text available
Orthographies vary in complexity (the number of multi-letter grapheme-phoneme rules describing print-to-speech regularities) and unpredictability (the number of words which cannot be read correctly, even with at-ceiling knowledge of the rules). To assess how these constructs affect reading acquisition, we used an artificial orthography learning paradigm, where participants learn to read pseudowords written in unfamiliar symbols, and subsequently read aloud novel words written in the same symbols (generalisation). In three experiments (third experiment pre-registered), we manipulated the consistency of symbol-to-sound mappings: in the first inconsistent condition, vowel pronunciation depended on the subsequent letter (condition complexity), and in the second inconsistent condition, vowel pronunciation was unpredictable from the context (condition unpredictability). Across experiments, we found that pseudowords with inconsistent mappings are more difficult to learn than pseudowords with consistent mappings only, regardless of whether the inconsistency is due to complexity or unpredictability. Numerically, participants learning orthographies containing unpredictable correspondences seem to be less likely to form rules, either for simple or for complex correspondences. We propose that rule extraction and distributional learning happens simultaneously during reading acquisition: in a mathematical model, we show that distributional learning may lead to more complete knowledge than rule extraction for orthographies that are high in unpredictability.
... Therefore, Hebrew offers an excellent natural context to study letter-position processing in learning to read (Friedmann et al., 2007). Other examples are the influence of vowel letters on the pronunciation of consonant letters in Russian (Schmalz et al., 2017) and complexity vs. irregularity effects in French and English (Schmalz et al., 2016). By strategically conducting experimental studies in different languages, capitalizing on their specific features and challenges, we can systematically isolate and better understand the different processes which the reading and spelling system in the brain is capable of or how it has adapted itself to the linguistic structures of different languages. ...
Article
Full-text available
Because of its regularity, it is relatively easy to learn to read and spell in Finnish. However, a specific hurdle in spelling acquisition seems to be the doubling of consonant letters. In this study on consonant letter doubling spelling in Finnish children (91 Grade 1 and 191 Grade 2 children), we asked two questions. First, are items with double consonant letters (e.g., “kissa” [ˈkisːɑ] ‘cat’) indeed harder to spell than single consonant items (e.g., “kisa” [ˈkisɑ] ‘contest’)? Second, is consonant doubling harder for stop consonants (e.g., “takki” [ˈtɑkːi] ‘coat’) than for continuant consonants (e.g., “kissa” [ˈkisːɑ] ‘cat’)? We found that Finnish children made more errors on items with double consonant letters than on items with single consonant letters and that this effect was larger for stop than for continuant consonant letters. Exploratory analyses showed that these effects were stronger for younger and poorer spellers. Post hoc analyses of the errors made on double consonant items showed that the children predominantly made nonlexical errors (> 90%). When they did make a lexical error, these errors typically did not map on the type of errors that would be expected from a corpus analysis of the higher-frequency orthographic neighbors. Overall, lexical influences on spelling of Finnish children seem to be minimal and unpredictable. We discuss two potential reasons why it is more difficult to spell items with double consonant letters than with single consonant letters and suggest how these could be investigated in future research.
Article
Full-text available
The data includes measures collected for the two experiments reported in “False-Positive Psychology” [1] where listening to a randomly assigned song made people feel younger (Study 1) or actually be younger (Study 2). These data are useful because they illustrate inflations of false positive rates due to flexibility in data collection, analysis, and reporting of results. Data are useful for educational purposes.
Article
Full-text available
Orthographic depth has been studied intensively as one of the sources of cross-linguistic differences in reading, and yet there has been little detailed analysis of what is meant by orthographic depth. Here we propose that orthographic depth is a conglomerate of two separate constructs: the complexity of print-to-speech correspondences and the unpredictability of the derivation of the pronunciations of words on the basis of their orthography. We show that on a linguistic level, these two concepts can be dissociated. Furthermore, we make different predictions about how the two concepts would affect skilled reading and reading acquisition. We argue that refining the definition of orthographic depth opens up new research questions. Addressing these can provide insights into the specific mechanisms by which language-level orthographic properties affect cognitive processes underlying reading. Getting to the bottom of orthographic depth.. Available from: https://www.researchgate.net/publication/275217141_Getting_to_the_bottom_of_orthographic_depth [accessed Apr 29, 2015].
Article
Full-text available
Many empirical researchers do not realize that the common multiway analysis of variance (ANOVA) harbors a multiple comparison problem. In the case of two factors, three separate null hypotheses are subject to test (i.e., two main effects and one interaction). Consequently, the probability of at least one Type I error (if all null hypotheses are true) is 14% rather than 5% if the three tests are independent. We explain the multiple comparison problem and demonstrate that researchers almost never correct for it. We describe one of several correction procedures (i.e., sequential Bonferroni), and show that its application alters at least one of the substantive conclusions in 45 out of 60 articles considered. An additional method to mitigate the multiplicity in multiway ANOVA is preregistration of hypotheses.
Article
Full-text available
The type of sublexical correspondences employed during non-word reading has been a matter of considerable debate in the past decades of reading research. Non-words may be read either via small units (graphemes) or large units (orthographic bodies). In addition, grapheme-to-phoneme correspondences may involve context-sensitive correspondences, such as pronouncing an “a” as /ɔ/ when preceded by a “w”. Here, we use an optimisation procedure to explore the reliance on these three types of correspondences in non-word reading. In Experiment 1, we use vowel length in German to show that all three sublexical correspondences are necessary and sufficient to predict the participants' responses. We then quantify the degree to which each correspondence is used. In Experiment 2, we present a similar analysis in English, which is a more complex orthographic system.
Article
Full-text available
The current study investigated the time course of cross-linguistic differences in word recognition. We recorded eye movements of German and English children and adults while reading closely matched sentences, each including a target word manipulated for length and frequency. Results showed differential word recognition processes for both developing and skilled readers. Children of the two orthographies did not differ in terms of total word processing time, but this equal outcome was achieved quite differently. Whereas German children relied on small-unit processing early in word recognition, English children applied small-unit decoding only upon rereading possibly when experiencing difficulties in integrating an unfamiliar word into the sentence context. Rather unexpectedly, cross-linguistic differences were also found in adults in that English adults showed longer processing times than German adults for nonwords. Thus, although orthographic consistency does play a major role in reading development, cross-linguistic differences are detectable even in skilled adult readers.
Article
Linear mixed-effects models (LMEMs) have become increasingly prominent in psycholin-guistics and related areas. However, many researchers do not seem to appreciate how random effects structures affect the generalizability of an analysis. Here, we argue that researchers using LME