Cognitive mechanisms underlying reading and spelling development in five European orthographies

lable at ScienceDirect

Learning and Instruction 29 (2014) 65e77

Contents lists avai

Learning and Instruction

journal homepage wwwelsevier comlocate learninstruc

Cognitive mechanisms underlying reading and spelling developmentin five European orthographies

Kristina Moll a b c e Franck Ramus d Juumlrgen Bartling e Jennifer Bruder e Sarah Kunze eNina Neuhoff e Silke Streiftau e Heikki Lyytinen f g Paavo HT Leppaumlnen f gKaisa Lohvansuu f g Deacutenes Toacuteth h Ferenc Honbolygoacute h Valeacuteria Cseacutepe hCaroline Bogliotti d i Steacutephanie Iannuzzi j k l Jean-Franccedilois Deacutemonet k mEmilie Longeras d n o Sylviane Valdois n o Florence George p Isabelle Soares-Boucaud q rMarie-France Le Heuzey s Catherine Billard l Michael OrsquoDonovan t Gary Hill tJulie Williams t Daniel Brandeis u v w x Urs Maurer u x y z Enrico Schulz u x ASanne van der Mark x B Bertram Muumlller-Myhsok C D Gerd Schulte-Koumlrne eKarin Landerl a b

a Department of Psychology University of Salzburg Austriab Department of Psychology University of Graz Austriac Department of Psychology University of York York UKd Laboratoire de Sciences Cognitives et Psycholinguistique Ecole Normale Supeacuterieure CNRS EHESS Paris Francee Department of Child and Adolescent Psychiatry and Psychotherapy University Hospital Munich Munich Germanyf Department of Psychology University of Jyvaumlskylauml Finlandg Agora Center University of Jyvaumlskylauml Finlandh Institute for Psychology Hungarian Academy of Sciences Hungaryi Universiteacute Paris Ouest Nanterre la Deacutefense MODYCO CNRS Francej Uniteacute de Neurologie Peacutediatrique Hocircpital des Enfants Toulouse Francek Inserm U825 Hocircpital Purpan Toulouse Francel Centre de Reacutefeacuterence sur les Troubles des Apprentissages Hocircpital Bicecirctre APHP Paris Francem Leenards Memory Center Centre Hospitalier Universitaire Vaudois Universiteacute de Lausanne Switzerlandn Laboratoire de Psychologie et NeuroCognition UMR 5105 CNRS Universiteacute Pierre Mendegraves France Grenoble Franceo Centre Reacutefeacuterent pour le Diagnostic des Troubles du Langage et des Apprentissages Deacutepartement de Peacutediatrie CHU Nord Grenoble Francep Centre de Reacutefeacuterence des Troubles drsquoapprentissages CHU Timone Marseille Franceq Centre de Reacutefeacuterence pour les Troubles des Apprentissages Hospices Civils de Lyon Hocircpital E Herriot Lyon Francer Centre Hospitalier Le Vinatier Bron Frances Service de Psychopathologie de lrsquoenfant et de lrsquoadolescent Hocircpital Robert Debreacute APHP Paris Francet MRC Centre for Neuropsychiatric Genetics and Genomics Department of Psychological Medicine and Neurology Cardiff University Cardiff UKu Department of Child and Adolescent Psychiatry University of Zuumlrich Zuumlrich Switzerlandv Center for Integrative Human Physiology University of Zuumlrich Zuumlrich Switzerlandw Department of Child and Adolescent Psychiatry and Psychotherapy Central Institute of Mental Health Medical Faculty MannheimHeidelberg UniversityMannheim Germanyx Neuroscience Center Zurich University of Zurich ETH Zurich Switzerlandy Sackler Institute for Developmental Psychobiology Weill Medical College of Cornell University New York USAz Institute of Psychology University of Zuumlrich Zuumlrich SwitzerlandA Department of Neurology Technische Universitaumlt Muumlnchen Muumlnchen GermanyB MR-Center University Childrenrsquos Hospital University of Zuumlrich Zuumlrich SwitzerlandC Max Planck Institute of Psychiatry Munich GermanyD Munich Cluster for Systems Neurology (SyNergy) Munich Germany

a r t i c l e i n f o

Article historyReceived 15 November 2012Received in revised form

Corresponding author Department of Child and2380337 Munich Germany Tel thorn49 (0)89 4522 903

E-mail address kristinamollgmailcom (K Moll)

0959-4752$ e see front matter 2013 Elsevier Ltdhttpdxdoiorg101016jlearninstruc201309003

a b s t r a c t

This paper addresses the question whether the cognitive underpinnings of reading and spelling areuniversal or languageorthography-specific We analyzed concurrent predictions of phonological pro-cessing (awareness and memory) and rapid automatized naming (RAN) for literacy development in a

Adolescent Psychiatry Psychosomatics and Psychotherapy Ludwig-Maximilians-University Munich Waltherstraszlige1

All rights reserved

K Moll et al Learning and Instruction 29 (2014) 65e7766

26 August 2013Accepted 3 September 2013

KeywordsReading developmentCross-linguisticOrthographic consistencyPhonological awarenessRapid automatized naming

large European sample of 1062 typically developing elementary school children beyond Grade 2acquiring five different alphabetic orthographies with varying degrees of graphemeephoneme consis-tency (English French German Hungarian Finnish) Findings indicate that (1) phonological processingand RAN both account for significant amounts of unique variance in literacy attainment in all five or-thographies Associations of predictors with reading speed reading accuracy and spelling are differen-tial in general RAN is the best predictor of reading speed while phonological processing accounts forhigher amounts of unique variance in reading accuracy and spelling (2) the predictive patterns arelargely comparable across orthographies but they tend to be stronger in English than in all otherorthographies

2013 Elsevier Ltd All rights reserved

1 Introduction

Recently considerable research interest has been generated bythe question whether the cognitive underpinnings of readingacquisition vary between orthographies or whether they are largelysimilar All known orthographic systems represent languagehowever there is a large degree of variance in the consistency of themapping between spoken and written language and consequentlyin the transparency of these mappings for the young learner Themain principle of all alphabetic orthographies that are used in theWesternworld is that graphic symbols (letters) represent the soundstructure of the spoken word However few orthographies closelyadhere to this alphabetic principle of simple 11 relationships be-tween letters and phonemes (like Finnish) while most alphabetsprovide the reader with a certain degree of inconsistency or ir-regularity The English orthographic system with its many com-plexities is probably on the most extreme end of this continuum oforthographic consistency Both theoretical conceptions (Katz ampFrost 1992 Ziegler amp Goswami 2005) and empirical evidence(see Landerl 2005 for a review) indicate that the development ofdecoding skills (ie the systematic translation of graphemes intophonemes) takes considerably longer in English than in moreconsistent orthographies Thus the complicated and opaque map-ping system of English orthography seems to cause particularproblems to the young learner It is probably no coincidence thatthe investigation of reading acquisition in English strongly domi-nates the research field However the question then ariseswhether the outlier status of English orthographic complexity isreflected in the cognitive mechanisms underpinning the readingprocess which would seriously limit the relevance of such anldquoAnglocentric viewrdquo (Share 2008) for other orthographies Thisissue is not only of high theoretical interest but has importantimplications for reading instruction as the relevant cognitive pre-dictors are used to identify children who are at risk for readingfailure

11 Cognitive predictors of literacy skills

Two cognitive skills that are closely associated with the complexprocess of reading and spelling acquisition are phonological pro-cessing and rapid automatized naming (RAN) Phonological pro-cessing refers to the ability to perceive store and manipulatespeech sounds and includes phonological awareness and phono-logical working memory In a typical phonological awareness taska child might be asked to delete a certain sound from a word ornonword pronunciation (eg ldquoSaygulstwithout thelrdquo) The childthen has to maintain the sound sequence in working memoryidentify thel-sound in the phoneme string delete it from thepronunciation and blend the remaining sound parts Thus it isobvious that although such tasks are taken to measure phonolog-ical awareness they usually also require working memory capacityPhonological awareness enables the child to understand and sys-tematically exploit the mappings between graphic symbols and the

sound structure of spoken language It is crucial whenever thegraphemes of words or nonwords are decoded during reading andalso when words are segmented into their constituent phonemesduring spelling Thus phonological awareness plays an importantrole during early literacy development across alphabetic orthog-raphies (eg Byrne 1998 Wagner amp Torgesen 1987) however inconsistent orthographies competent graphemeephoneme andphonemeegrapheme translation is typically achieved earlier andgrowth of literacy skills is faster than in inconsistent orthographieslike English (eg Caravolas Lervaringg Defior Seidlovaacute Maacutelkovaacute ampHulme 2013 Seymore Aro amp Erskine 2003) Beyond these earlyphases of literacy development phonological awareness is sup-posed to exert its influence on building-up word-specific repre-sentations (Ehri 1992 Perfetti 1992) According to this theoreticalview an efficient storage of orthographic patterns depends onmultiple associations between phonological segments of a spokenword and the corresponding graphemes of its written form Word-specific orthographic representations enable direct word recogni-tion during reading and correct orthographic spelling Once againthe degree of consistency of grapheme- as well as phonemeegrapheme correspondences can be assumed to play an importantrole Coping with the many irregularities and inconsistenciesinherent in the English orthographic system may particularlychallenge the phonological system of the learner This would implythat the relevance of phonological processing skills should be lowerin consistent than in less consistent orthographies

Rapid automatized naming (RAN) refers to the speed withwhich an individual can pronounce the names of a sequentially andrepeatedly presented limited set of stimuli like letters Arabic digitscolor patches or pictures of familiar objects Performing RAN taskscertainly requires phonological skills (accessing the phonologicaloutput programs of the required word pronunciations as quickly aspossible) and is therefore sometimes seen as a third subcomponentof phonological processing (Torgesen Wagner amp Rashotte 1994Torgesen Wagner Rashotte Burgess amp Hecht 1997 VaessenGerretsen amp Blomert 2009) However there is now ample evi-dence that ldquonaming speed is phonological but not only phono-logicalrdquo (Kirby Georgiou Martinussen amp Parrila 2010 p 356) andconstitutes a second cognitive mechanism underpinning readingdevelopment that is largely independent from phonologicalawareness and memory First the correlation between phonolog-ical awareness and RAN is typically only low to moderate (38 in ameta-analysis of 35 studies that were almost exclusively carried outin English Swanson Trainin Necoechea amp Hammill 2003) Sec-ond although phonological awareness and RAN contribute someamount of shared variance both components have consistentlybeen shown to make unique contributions to the variance of liter-acy skills above and beyond the other one Third and most impor-tantly these unique contributions seem to be differentialphonological awareness and RAN have been demonstrated to showspecific relationships with particular subcomponents of literacyprocessing While phonological skills seem to be most stronglyrelated to literacy skills that involve decoding (most importantly

K Moll et al Learning and Instruction 29 (2014) 65e77 67

nonword reading accuracy) RAN has been found to be moststrongly related to the fluency with which different types ofreading material (words nonwords texts) can be read (Kirby et al2010 Manolitsis Georgiou amp Parrila 2011 Swanson et al 2003)

Although the relationship between RAN and reading is aconsistent finding the mechanisms underlying this association areunder debate (see Kirby et al 2010 for a current review) Onetheoretical explanation for the RAN-reading relationship that iscurrently discussed is that RAN is a reflection of orthographicprocessing (Bowers Golden Kennedy amp Young 1994 Bowers ampNewby-Clark 2002 Conrad amp Levy 2007 Manis Seidenberg ampDoi 1999) According to this view the build-up of an efficientorthographic lexicon depends on the precise integration of visualinformation about letter sequences in words When letter identi-fication is slowed down as indexed by poor RAN representations oforthographic patterns (ie whole words) cannot be reliably storedFollowing this argument RAN should be particularly stronglyrelated to orthographic spelling and should be a better predictor ofword than of nonword reading as orthographic processes are ofrelatively low relevance in nonword reading However previousstudies did not consistently confirm this pattern (eg MollFussenegger Willburger amp Landerl 2009) Two other explana-tions for the RAN-literacy relationship that have been put forwardare that RAN indicates the efficiency of visual-verbal integrationprocesses (eg Moll et al 2009) or that RAN captures variance inphonological lexical retrieval (Decker Roberts amp Englund 2013)Integration of visual and verbal information is relevant in order tofluently read any reading material including nonwords which ismore in line with the available evidence (Kirby et al 2010) Pre-dictions of the latter two explanations (fastefficient mapping andlexical retrieval) for orthographic spelling are less clear but bothaccounts predict some contribution of RAN to spelling First visual-verbal integration is likely to support the storage of orthographicpatterns based on multiple associations between sounds and let-ters Secondly correlations between orthographic and phonologicallexical retrieval may explain the RAN-spelling relationship How-ever both accounts predict a more important role of RAN duringfluent reading than during spelling processes

It has been suggested that RAN may be a better predictor ofreading development in consistent than in inconsistent orthogra-phies (eg de Jong amp van der Leij 1999 Di Filippo et al 2005Georgiou Parrila amp Papadopoulos 2008 Landerl amp Wimmer2008 Mann amp Wimmer 2002 Mayringer Wimmer amp Landerl1998) However this may mostly be due to the fact that readingattainment is usually measured in terms of reading speed inconsistent orthographies as accuracy levels are generally high anddo not sufficiently differentiate between good and poor readersOnce the same literacy components are considered in orthogra-phies with different degrees of consistency it is conceivable thatthe many complexities of the English orthography place higherdemands on the cognitive components that are assumed to bemeasured by RAN than more consistent orthographies In Englishthe learner needs to hold a number of letter-sound or sound-lettercorrespondences active during reading and spelling while in moreconsistent orthographies the number of orthographic patterns orvisual-verbal associations is clearly lower

12 Cross-linguistic studies

Due to the inclusion of differentmeasures it is often problematicto compare findings across studies carried out in different orthog-raphies A number of studies have attempted to tackle this problemby investigating the cognitive underpinnings of reading develop-ment in two or more orthographies within the very same researchdesign As differences in the predictive patterns are presumably

most prominent in the early phases of literacy development themajority of these cross-linguistic studies examined the cognitivepredictors of reading in the first or second school year

In these early phases of reading development phonologicalawareness was consistently found to be a reliable concurrent(Caravolas Volin amp Hulme 2005 Georgiou et al 2008 Vaessenet al 2010 Ziegler et al 2010) and longitudinal (Caravolas et al2012 Furnes amp Samuelsson 2011) predictor of reading skills (ac-curacy and speed) across different orthographies However find-ings are mixed with respect to the relative importance of phonemeawareness as a function of orthographic consistency While somestudies showed that the impact of phonological awareness onreading is stronger in less than in more consistent orthographies(Mann amp Wimmer 2002 Vaessen et al 2010 Ziegler et al 2010)others reported an equally strong prediction of phonologicalawareness in English and in more transparent orthographies(Caravolas et al 2005 2012) Differences between orthographieshave also been suggested with respect to the impact of phonolog-ical awareness on reading over time It has been argued that inmore consistent orthographies the predictive strength of phono-logical awareness decreases after about one year of reading in-struction (Furnes amp Samuelsson 2011 Georgiou et al 2008Vaessen et al 2010) because decoding skills are then alreadysufficiently acquired In inconsistent orthographies phonologicalawareness remains a strong predictor beyond Grade 1 reflectingthe fact that the development of decoding skills takes longer ininconsistent compared to consistent orthographies

Cross-linguistic findings on the predictive pattern of RAN in theearly phases of reading development are mixed Some studies re-ported that RAN predicts reading in consistent as well as incon-sistent orthographies (Caravolas et al 2012 Furnes amp Samuelsson2011 Georgiou et al 2008 Vaessen et al 2010) In contrast othersfound associations between RAN and reading at this age only inconsistent orthographies (Mann amp Wimmer 2002) or reportedgenerally weak associations between RAN and reading across or-thographies (Ziegler et al 2010) A plausible explanation for thesemixed findings is that RAN has been shown to be specifically linkedto fluent word and text reading (see Kirby et al 2010 for review)Reading fluency is usually assessed by list or text reading para-digms Especially during the early phases of reading developmentsuch paradigms are of limited validity if young readersrsquo readingfluency is constrained by problems to read the presented stimuliaccurately Indeed Vaessen et al (2010) reported an increase of theimpact of RAN on reading fluency between Grades 1 and 4 Due tothe relatively low reading accuracy assessing reading fluency inyoung readers is especially problematic in inconsistent orthogra-phies This might explain why during the early phases of literacydevelopment RAN was found to be a better predictor in consistentthan in inconsistent orthographies

Only a few studies included spelling as a criterion measure andfindings indicate that both phonological awareness and RAN pre-dict spelling skills in consistent as well as inconsistent orthogra-phies (Caravolas et al 2005 2012 Furnes amp Samuelsson 2011)However findings are again mixed with respect to the relativeimportance of RAN and phonological awareness for reading incomparison to spelling skills While Furnes and Samuelsson (2011)could confirm a differential prediction with RAN being a strongerpredictor for reading and phonological awareness being a strongerpredictor for spelling the predictive patterns of these cognitivemeasures were similar for reading and spelling in the Caravolaset al study (2012) Up to date evidence on cross-linguistic differ-ences in the cognitive underpinnings of spelling development islimited to the first two years of formal instruction and studiescomparing the predictors of orthographic spelling between or-thographies beyond Grade 2 are lacking On the one hand larger

sylvianevaldois
Texte surligneacute
sylvianevaldois
Texte surligneacute


differences than for reading development could be expected as thecorrect reproduction of word spellings probably requires a thor-ough understanding of the function of orthographic markers thatare specific to a particular writing system On the other hand mostalphabetic orthographies are characterized by a good deal ofinconsistency in phonemeegrapheme correspondences In order tospell a word correctly word-specific knowledge is indispensable(eg Desimoni Scalisi amp Orsolini 2012) Therefore the spellingprocess may be more similar across orthographies than the readingprocess and as a consequence cognitive underpinnings should becomparable as well

In summary during the early phases of literacy developmentphonological awareness and RAN have been found to predictreading and spelling in a variety of orthographies in cross-sectionalas well as longitudinal designs that assessed phonological aware-ness and RAN before the onset of formal reading instructionImportantly Caravolas et al (2012) showed that the predictivepattern even holds when controlling for the autoregressor (readingor spelling) at the beginning of Grade 1 indicating that phonolog-ical awareness and RAN predict growth in reading and spellingHowever findings are mixed with respect to the relative impor-tance of predictors as a function of orthographic consistency andwith respect to the relative importance of predictors for readingcompared to spelling skills Phonological memory was assessed inmost cross-linguistic studies but was generally reported to play arather minor role

While individual differences in reading accuracy and phono-logical spelling are probably most prominent in these early yearsdifferences in reading fluency and orthographic spelling dominatelater developmental phases when the first hurdles of cracking thealphabetic code are alreadymastered To this date only three cross-linguistic studies have investigated the cognitive underpinnings ofreading beyond Grade 2 Patel Snowling and de Jong (2004)compared reading skills in 67 English and 40 Dutch speakingchildren aged 6e11 and found a similar pattern for the two or-thographies phonological awareness was a significant predictor inboth languages while RAN did not enter the regression model as asignificant predictor The uncommon finding that RAN did not evenpredict reading speed may be due to the relatively large age rangein associationwith a relatively small sample size and to the fact thatthe timed phonological awareness measure included speed vari-ance otherwise picked up by RAN Vaessen et al (2010) investigatedconcurrent predictions of reading fluency cross-sectionally forGrades 1e4 and found that the impact of RAN increases with gradelevel while the impact of phonological awareness was limited tothe lower grades and was weaker in consistent orthographies (espHungarian) than in the more inconsistent ones (French and Por-tuguese) None of the two studies examined the cognitive un-derpinnings of orthographic spelling Finally in a large Europeanstudy of developmental dyslexia overlapping with the present one(NEURODYS) the prediction of dyslexia status by phonologicalawareness and RAN was found to increase with orthographiccomplexity (Landerl et al 2013) Phonological memory played acomparatively minor role in the prediction of dyslexia status

13 Aims of the present study

The first analysis of the NEURODYS sample (Landerl et al 2013)was limited to rather coarse-grained comparisons of dyslexicversus typically developing readers in alphabetic European or-thographies grouped into three levels of orthographic complexityIn the current paper we aimed to extend the analysis of the largeEuropean NEURODYS sample by providing a more fine-grainedanalysis of the concurrent predictive mechanisms underlyingdifferent literacy components (reading speed reading accuracy

and spelling) in typically developing readers acquiring five or-thographies varying in consistency (English French GermanHungarian and Finnish)

The following research questions will be investigated

RQ 1 To what extent do phonological processing (phonologicalawareness and memory) and RAN differentially influencedifferent measures of literacy (reading speed reading accuracyand orthographic spelling) beyond Grade 2H11Basedonpreviousfindingswehypothesize thatphonologicalawareness and RAN constitute two separate factors that inde-pendently predict different literacy skills across orthographiesH12 We assume that the predictive patterns for the three lit-eracy measures are differential with RAN being the strongestpredictor of reading speed and phonological processing (iephonological awareness) being the best predictor of readingaccuracy For orthographic spelling predictions are less clearPhonological awareness should be a strong predictor of ortho-graphic spelling given its role in building-up orthographicrepresentations The association between RAN and orthographicspelling should be especially strong if RAN reflects orthographicprocessing In contrast if RAN captures visual-verbal integrationor lexical retrieval it should be less important for spelling thanfor fluent readingRQ2 To what extent does the absolute influence of phonologicalprocessing and RAN on each literacy measure vary with ortho-graphic complexity beyond Grade 2H21 We predict that because of the generally higher demandsthat inconsistent orthographies place on the cognitive processesof the learner the total amount of variance explained byphonological processing and RAN is higher in inconsistent thanin consistent orthographiesRQ3 To what extent is the relative influence (irrespective of thetotal amount of variance explained) of phonological processingand RAN on each literacy measure determined by orthographiccomplexity As the majority of findings in this field are based onEnglish speaking samples it is of special importance to identifyany differences between English and the more consistentorthographiesH31 We assume that the relative predictive pattern is similaracross orthographies once the same literacy measures arecompared

2 Method

21 Rationale

Large-scale cross-linguistic comparisons have to deal withparticular methodological problems First quantifying the differ-ences between orthographies is extremely difficult and all availableattempts have serious methodological limitations (see Protopapasamp Vlahou 2009 for a critical discussion) Although there isnotable agreement onwhere to place particular writing systems ona continuum of orthographic complexity (eg Borgwaldt Hellwigamp De Groot 2005 Caravolas 2005 Seymour Aro amp Erskine2003) the adequate levels of description and their quantificationare still under discussion For instance Borgwaldtrsquos entropy mea-sure is based on word onsets only and is therefore missing most ofthe irregularities in many languages In the current project wedecided to use amore conservative classification by ranking the fiveorthographies according to their consistency of mappings fromgraphemes to phonemes (reading direction) and from phonemes tographemes (spelling direction)

The five orthographies covered the full continuum of ortho-graphic consistency The highly complex orthography of English is


on the one end of the continuum as it is characterized by high levelsof inconsistency in both the reading and spelling direction1 Similarto English French has complex relationships from phonemes tographemes (spelling direction) but is in general more rule-basedin the reading direction French vowels are more consistent thanEnglish vowels For example Ziegler Jacobs and Stone (1996) re-ported that 79 of monosyllabic French words are inconsistent inthe spelling direction (ie there exists more than one way to spell apronunciation) while only 12 are inconsistent in the reading di-rection (ie there exists more than one way to pronounce aspelling) German has highly consistent graphemeephoneme cor-respondences but less consistent phonemeegrapheme corre-spondences and represented a medium level of orthographiccomplexity Hungarian and Finnish comprised the lowest level oforthographic complexity as both languages are characterized byhighly consistent relationships between letters and sounds in bothreading and spelling direction Especially Finnish represents theextreme other end of orthographic complexity as in addition tosimple 11 relationships between phonemes and graphemes multi-letter graphemes do not exist and consonant clusters are highlyexceptional Note that our ranking order (English French GermanHungarian and Finnish) is fully consistent with the complexitysequences provided by Seymour et al (2003) based on readingaccuracy at the end of Grade 1 and Borgwaldt et al (2005) based onword-initial letter-to-phoneme mappings

Another methodological issue concerns the selection ofadequate tasks to measure the relevant cognitive and literacyconstructs Compatibility across languages was relatively easy toachieve for verbal and nonverbal IQ and for phonological memoryas standardized versions of the relevant WISC subtests wereavailable in each language Naming speed was measured by lan-guage-specific RAN paradigms requiring children to name asquickly as possible lists of single digits and highly familiar picturedobjects that correspond to short high frequency nouns (eg dogcar fish) Thus although different stimuli were used across lan-guages task format and selection criteria for the words that had tobe named were matched

With respect to phonological awareness we followed theexample of earlier cross-linguistic studies (Caravolas et al 2005Vaessen et al 2010 Ziegler et al 2010) and administeredphoneme deletion thus ensuring reasonable comparability offindings across studies Phoneme deletion is a standard paradigmwhich is sufficiently difficult in order to pick up individual differ-ences in higher grades and in samples acquiring consistent or-thographies As the five languages involved differ largely in theirlinguistic structure devising one task with exactly the same itemsfor all participants was not viable Specifying the linguistic struc-ture of presented items across languages might have inducedhigher typicality in some languages than others (eg consonantclusters are atypical in Finnish but very frequent in Germanwhereas polysyllabic words are frequent in Finnish but less typicalin English) Thus it was decided to leave the language-specificcharacteristics to individual partners who were advised to selectitems with typical linguistic structure and to ask children to deletea specified phoneme (eg ldquoSaygulstwithoutlrdquo)

All partners measured word and nonword reading accuracy aswell as speed with language-specific standardized reading testsReading speed could be reliably assessed as children had at least

1 Note that most alphabetic orthographies show an asymmetry in orthographicconsistency with higher consistency in the reading direction (graphemeephonemecorrespondences) compared to the spelling direction (phonemeegrapheme corre-spondences) The only exception is Finnish which is highly consistent in bothdirections

two years of reading instruction in the more consistent orthogra-phies (Finnish Hungarian and German) and three years in the lessconsistent orthographies (French and English) Reading accuracywas assessed under speeded conditions in order to avoid error ratesat ceiling for consistent orthographies

In summary the main advantage of the current joint Europeanresearch effort is that data collection was parallelized as much aspossible across orthographies and that the same constructs wereassessed by all partners so that a major problem of earlier researchcarried out on the cognitive underpinnings of literacy developmentcould be overcome namely the low compatibility of findings fromdifferent studies

22 Participants2

Participants were native speakers of five different languages andcame from seven European countries (English UK French FranceGerman Germany Austria Switzerland Hungarian HungaryFinnish Finland) Data came from the EU-NEURODYS-study whichcomprises large samples of dyslexic and typically developingelementary school children across the European Union The currentanalysis is mostly regression based and treating two differentreading level groups as a homogenous sample seemed methodo-logically problematic It would also have artificially increased thevariances within each national sample as the lower end of thedistribution of reading skills is clearly overrepresented in the fullNEURODYS sample Thus it was decided to base the current anal-ysis on the national samples of typically developing readers whichwere selected by each partner lab based on a standardizedlanguage-specific test of word recognition (Table A1) with thelimitation that performance should not be more than one standarddeviation below the age or grade level normWe are aware that thisprocedure somewhat reduces the variance of reading skills in oursample and that findings are mostly informative with respect totypical reading which we consider an interesting perspective withrespect to the question of cognitive mechanisms underlyingdifferent literacy components across orthographies

Written informed consent was obtained from parents beforetesting Children in less consistent orthographies were slightlyolder which accounts for the fact that literacy development takeslonger in less consistent compared to consistent orthographies(eg Seymour et al 2003) and to ensure that fluent reading andorthographic spelling can be reliably assessed Data for all relevantmeasures were available for 1062 children ranging from Grade 2 toGrade 5 for the three consistent orthographies (Finnish Hungarianand German) and from Grade 3 to Grade 7 for the less consistentorthographies (French and English) The number of children bygrade for each country is listed in the Appendix (Table A2)

23 Tasks

231 Word and nonword readingIn each country reading accuracy and speed for words and

nonwords were assessed by presenting language-specific materialunder a speeded instruction (ldquoRead as quickly as possible withoutmaking mistakesrdquo) The relevant measures were the total numberof items read per minute (reading speed) and the percentage ofitems read correctly based on the total number of items read

2 The present dataset overlaps with the data reported by Landerl et al (2013)(954 participants in common) which focused on predictors of dyslexia acrossgroups of dyslexic and control children but did not analyze literacy skills There is afurther overlap of 44 Hungarian participants with Ziegler et al (2010) and of 178Hungarian children with Vaessen et al (2010)


(reading accuracy) Grade-specific z-scores for word and nonwordreading speed and accuracy were calculated based on nationalnorms

232 SpellingLanguage-specific standardized spelling tests were given by

each partner All tests required to spell single words dictated insentence frames Grade-specific z-scores for the percentage ofwords spelled correctly were calculated based on language-specificnorms

233 IQVerbal and nonverbal IQ were estimated based on the subtests

lsquoSimilaritiesrsquo and lsquoBlock Designrsquo from the Wechsler IntelligenceScale for Children (WISC III-R or IV depending on availability ineach country Wechsler 1992 2003)

234 Phonological short-term and working memoryWISC digit span (forward and backwards) were given by each

partner (Wechsler 1992 2003) Scaled scores (Mfrac14 10 SDfrac14 3) werecalculated based on national norms

235 Phonological awareness (PA)In each country a phoneme deletion task was administered

requiring the child to pronounce a sound sequence after deleting aspecified sound (eg say ldquogulstwithoutlrdquo) Language-specifictasks were constructed with comparable difficulty levels inconsistent and inconsistent orthographies (see Table 1)

236 Rapid automatized naming (RAN)Two RAN tasks (digit and picture naming) were administered

Childrenwere asked to name as quickly and accurately as possible amatrix of digits and pictures of simple objects respectively Therelevant measure was the time to name the lists Correlations be-tween the two RAN tasks were moderate to high and the compo-nent analysis (Section 25) revealed that both RAN tasks loadedhighly on the same factor Therefore a composite RAN score wasused for all further analyses

24 Calculation of z-scores

For all variables but the word reading inclusion variable and theWISC subtests (which were already on a standardized scale) rawscores were converted into z-scores within each country and each

Table 1Gender ratios and means (SD) for age IQ cognitive predictors and literacy skills forthe five languages

English French German Hungarian Finnish

N 60 86 473 195 248Gender

[ boys)717 442 512 518 508

Age [months] 1297 (171) 1201 (123) 1144 (108) 1132 (100) 1114 (64)PIQa 105 (24) 112 (24) 110 (26) 119 (26) 104 (26)VIQa 120 (29) 131 (28) 128 (28) 124 (28) 115 (34)Digit spana 98 (37) 96 (27) 102 (25) 109 (25) 92 (26)PA [ correct] 882 (108) 916 (117) 783 (164) 811 (152) 890 (141)RAN composite

[itemsmin]990 (211) 969 (179) 943 (179) 963 (143) 675 (127)

Reading speed[z-score]

02 (86) 02 (91) 03 (96) 00 (95) 16 (92)

Readingaccuracy[ correct]

773 (138) 865 (98) 965 (39) 960 (37) 907 (93)

Spellingaccuracy[ correct]

709 (153) 736 (202) 783 (164) 674 (174) 654 (218)

a Standardized scaled score (mean frac14 10 SD frac14 3)

grade level As expected some variables in some countries hadhighly skewed distributions (ie reading accuracy in consistentorthographies) thus we further applied the following procedureEach variable in each country was converted into ranks thenrescaled on a 0e100 interval then applied the normal distributionfunction to convert them into grade-specific z-scores This proce-dure reduced the skew of distributions and made them morecomparable between measures and between countries (Landerlet al 2013)

25 Component analysis

Two principal component analyses (Varimax rotation withKaiser Normalization) were carried out to reduce the number offive outcome measures and four predictor variables to a theoreti-cally meaningful number of factors that could be included in theregression model The first analysis included the five literacyattainment measures speed and accuracy for word and nonwordreading as well as spelling performance The results indicated a twocomponent solution with clear loadings of word and nonwordreading speed on the first component (speed factoreigenvalue frac14 238 both factor loadings frac14 90) and word andnonword reading accuracy on the second component (accuracyfactor eigenvalue frac14 110 factor loadings frac14 76 and 80 respec-tively) Correlations revealed a notable association between wordand nonword reading for all languages (30e61 for accuracy and46e79 for speed) These correlations were higher than the corre-lations between accuracy and speed within one item categoryconfirming the two component solution with a speed and an ac-curacy factor

The spelling measure loaded higher on the accuracy componentthan on the speed component (64 versus 40) but the loadingswere not as clear-cut as for the reading measures Therefore wedecided to analyze spelling separately resulting in the followingthree outcome measures reading speed (composite mean z-scorefor word and nonword reading speed) reading accuracy (com-posite mean z-score for word and nonword reading accuracy) andspelling

The second analysis included the four predictor variablesphonological awareness (phoneme deletion) phonological mem-ory (digit span) RAN digits and RAN pictures The results showedagain a two component solution with phoneme deletion and digitspan loading on the first component (phonology factoreigenvalue frac14 113 factor loadings frac14 81 and 77 respectively) andthe two RAN measures loading on the second component (RANfactor eigenvalue frac14 164 factor loadings frac14 87 and 83 respec-tively) Correlations between the two RAN measures were moder-ate to highwith 47 for thewhole sample ranging from 41 to 72 forthe different languages The correlations between the two phono-logical tasks were lower than for the two RANmeasures (28 for thewhole sample ranging from 09 to 44) As a consequence a com-posite score was calculated for the two RANmeasures whereas thetwo phonology measures were investigated separately in thefollowing analyses

3 Results

31 Descriptives and correlation analyses by language

The descriptive statistics for age IQ cognitive measures andliteracy skills are presented in Table 1 The results show that per-formance in all measures is similar across languages

Table 2 presents the simple correlations (based on grade-specific scores) between all predictor and literacy measures sepa-rately for each language Reading speed showed higher associations

Table 2Correlations (based on grade-specific z-scores) between predictor variables and outcome measures for the five languages

READ speed READ acc SPELL Digit span PA RAN digit RAN picture RAN comp

EnglishREAD speed 298 185 196 126 370 378 426READ acc 330 576 221 483 383 330 401SPELL 356 612 297 403 517 255 432Phon mem 096 317 295 308 088 093 103PA 153 343 214 094 092 084 100RAN digit 452 080 048 263 041 543 878RAN picture 366 074 071 080 049 717 879RAN comp 442 083 012 185 049 927 926

FrenchGerman

READ speed 272 516 108 306 465 309 458READ acc 166 404 170 376 044 081 074SPELL 254 366 190 375 120 205 192Phon mem 097 270 446 268 014 124 081PA 152 360 496 440 073 209 166RAN digit 440 001 143 081 140 430 847RAN picture 442 120 296 133 284 489 844RAN comp 511 070 254 124 246 863 862

HungarianFinnish

READ speed 371 329 338 286 178 291 279READ acc 467 216 378 013 085 043SPELL 228 293 101 132 138Phon mem 247 039 155 116PA 133 192 194RAN digit 412 839RAN picture 842RAN comp


with RAN than with the two phonological measures in all orthog-raphies apart from Finnish where the associations with the threepredictor components were roughly equal For reading accuracyhowever the correlations were highest for phonological awarenessfollowed by phonological memory and were not significant forRAN The only exception was the English sample where RANcorrelated moderately with reading accuracy A similar patternwasobserved for spelling with higher correlations for phonologicalawareness and memory than for RAN Again the English sampleshowed different associations between the predictor variables andspelling with comparable correlations for the three predictors

32 Prediction of literacy skills separately for each language

The concurrent predictions of phonological processing (memoryand awareness) and RAN with the three dependent literacy mea-sures (reading speed reading accuracy and spelling) were exam-ined separately for each language in a series of stepwise regressionanalyses In all analyses step 1 controlled for differences in age andIQ (verbal and nonverbal) In Step 2 the three theoretically inter-esting proximal factors (phonological memory phonologicalawareness and RAN) were entered simultaneously For each factorwe calculated the percentage of variance in the dependent variableexplained by a specific predictor variable above and beyond theother predictors These regression analyses allowed investigatingwhether the predictive patterns for the three literacy measures aredifferential (question 1) and to assess the absolute influence ofpredictors on each literacy measure separately for each language(question 2)

All regression models reported in this section were calculatedfor a composite RAN score as well as separately for RAN digits andRAN pictures In general the predictive pattern for both RANmeasures was similar to the pattern reported for the compositeRAN score therefore only the models for the composite score willbe reported

321 Reading speedTable 3 shows an impressively consistent pattern of prediction

of reading speed across languages RAN explained clearly higheramounts of unique variance in reading speed than the twophonological predictor measures in English French German andHungarian (167e228) Only in Finnish the contribution of allproximal factors was largely equal and low compared to the otherorthographies (43 and 31 for phonological measures and 35 forRAN) Phonological processing made unique but comparably smallcontributions to reading speed in English (phonological awarenessand memory) and German (phonological awareness only) butexplained hardly any unique variance in French and Hungarian

322 Reading accuracyTable 4 shows that overall the proximal predictors phonological

processing and RAN could account for higher amounts of variancein reading accuracy in English (39) than in all other orthographies(113e147) The variance explained by phonological memory wascomparably small (04e69) and not always significantWhile RANwas found the best unique predictor of reading speed phonologicalawareness accounted for more variance in reading accuracy (52e176) The only orthography in which RAN could account for asubstantial amount of variance above and beyond phonologicalprocessing was English (142)

323 SpellingAs it was found for reading accuracy Table 5 shows that the

proximal predictors (phonological awareness memory and RAN)accounted for clearly higher amounts of variance in English (347)than in all other orthographies (89e162) Phonological memoryseems to be somewhat more important for spelling than for thereading measures as it could account for significant amounts ofvariance in Finnish French and Hungarian Phonological aware-ness accounted for significant amounts of variance in all orthog-raphies (41e89) apart from French (22) Interestingly English

Table 3Regression analyses for the five languages with reading speed as dependentvariable

Reading speed

R2-change p B SE B

English N [ 60Step 1 AgeIQ 92 143Step2 Unique phon memory 58 036 238 111

Unique PA 44 066 212 113Unique RAN 172 000 440 118Variance step 2 249 001Total variance 340

French N [ 86Step 1 AgeIQ 54 204Step2 Unique phon memory 02 674 041 097


German N [ 473Step 1 AgeIQ 10 189Step2 Unique phon memory 00 761 012 040


Hungarian N [ 195Step 1 AgeIQ 34 085Step2 Unique phon memory 00 745 022 068


Finnish N [ 247Step 1 AgeIQ 90 000Step 2 Unique phon memory 43 000 208 056


Table 4Regression analyses for the five languages with reading accuracy as dependentvariable

Reading accuracy

R2-change p B SE B

English N [ 59Step 1 AgeIQ 02 989Step 2 Unique phon memory 07 440 087 112


French N [ 86Step 1 AgeIQ 115 018Step 2 Unique phon memory 69 008 236 087


German N [ 473Step 1 AgeIQ 15 063Step 2 Unique phon memory 04 129 055 036


Hungarian N [ 195Step 1 AgeIQ 86 001Step 2 Unique phon memory 23 022 139 061





was once again the only orthography where RAN could make asignificant e and indeed the highest e contribution to variance inspelling

33 Prediction of literacy skills across orthographies

In order to identify any differences in the cognitive mechanismsassociated with literacy skills between orthographies (question 3)we resorted to multilevel analyses using the R-package lme4(lme4_0999999-0 httpCRANR-projectorgpackagefrac14lme4) Wenested children within the variable language (model M1) andallowed for fixed (model M2) and both fixed and random effects(effectively allowing the predictor to vary by level of the nestingvariable) for each of the three predictors (models M3) In additionIQ and age were included as covariates in each model Thus threemodels were specified for each predictor based on the wholesample of 1062 children These models were run for the threedependent measures (reading speed reading accuracy andspelling)

To test the fixed effect we compared the likelihood ratio be-tween models M1 and M2 The distributional property of the teststatistic was verified using a sample of 10000 permutations of thedependent measure for each of the nine predictoreliteracy com-binations In each case the fit of the distribution was found to

adhere very well to theoretical expectations hence p-valuesderived on asymptotic theory are provided

The test for heterogeneity between languages in the estimates(again bymeans of a likelihood ratio test this time betweenmodelsM2 and M3) showed severe deviation from the expected distribu-tion of test statistics again tested by means of a permutation of thenesting variable language The test was very severely conservativetherefore we tested the heterogeneity of random effect estimatesby means of permutations (n frac14 10000 for each combination) Thetest statistic used was the sum of the Euclidean distances betweenthe random effect estimates for each of the languages In thosecases where the test for heterogeneity of the random effect esti-mates for a predictor reached significance pair-wise comparisonsbetween the languages were tested to identify the source of theheterogeneity again using the permutations performed Estimatesand confidence intervals from these analyses are provided inTables 6e8 with estimates being obtained from model M3

In sum the results of the multilevel analyses support our thirdprediction that the predictive pattern is to a great extent similaracross orthographies once the same literacy skills are comparedwith the following exceptions

(1) For reading speed phonological awareness showed some evi-dence for heterogeneity between languages (pfrac14 021) howeverdifferences between consistent and inconsistent languages

Table 5Regression analyses for the five languages with spelling as dependent variable

Spelling

R2-change p B SE B











Table 6Multilevel analyses for the three predictors (aec) for reading speed

Reading speed

Estimates 95 CI p-Value

(a) Phon memoryFixed effect estimates for phon memory

127 [070 to 188] lt001Random effect estimates phon memory by languageEnglish 005 [072 to 085]French 024 [142 to 048]German 032 [111 to 031]Hungarian 039 [148 to 035]Finnish 089 [005 to 228]Global heterogeneity of random effects 127

(b) PAFixed effect estimates for PA

152 [068 to 220] lt001Random effect estimates PA by languageEnglish 102 [320 to 002]French 020 [146 to 115]German 093 [003 to 206]Hungarian 041 [131 to 047]Finnish 071 [013 to 194]Global heterogeneity of random effects 021

(c) RANFixed effect estimates for RAN

439 [372 to 512] lt001Random effect estimates RAN by languageEnglish 004 [105 to 108]French 001 [109 to 126]German 053 [027 to 146]Hungarian 059 [018 to 172]Finnish 107 [243 to 015]Global heterogeneity of random effects 086


were not clear-cut which might to some extent reflect that theimpact of phonological awareness for reading speed wasgenerally low (Section 321)

(2) For reading accuracy RAN showed clear evidence for hetero-geneity between languages (p frac14 008) reflecting that RAN ismore important in English than in the other languages Pair-wise comparisons revealed that this difference was signifi-cant for all four languages

(3) For spelling there was a tendency for a higher contribution ofRAN in the English sample compared to more consistent or-thographies (see regression analysis separately for each lan-guage (Section 323)) However this difference was notsignificant in the direct language comparison based on themultilevel analysis

4 Discussion

The current paper compares the associations of phonologicalprocessing (awareness and memory) and RAN with reading (speedand accuracy) and spelling in five alphabetic orthographiescovering the full range of orthographic consistency Obviously thepresented concurrent analyses do not allow strong conclusions onthe directions of causality nevertheless the findings add to thecross-linguistic literature as this is the first analysis of cognitivemechanisms underlying reading as well as orthographic spellingbeyond the initial stages of literacy development

First we asked to what extent phonological processing and RANdifferentially influence different measures of literacy (RQ1) In linewith our predictions (H11) our analyses indicate that phonologicalprocessing and RAN constitute two separate factors which bothaccount for significant amounts of unique variance in literacyattainment in all five orthographies In contrast to other studies(Kirby et al 2010) we did not find a clear difference between al-phanumeric and non-alphanumeric RAN (digits vs pictures) bothconditions showed similar predictive patterns As we had predicted(H12) the associations of the predictor measures with readingspeed reading accuracy and spelling were differential In generalRAN was the best predictor of reading speed while phonologicalprocessing (phonological awareness and memory) accounted forhigher amounts of unique variance in reading accuracy andspelling

Next we asked to what extent the absolute (RQ2) and relative(RQ3) influence of phonological processing and RAN on each lit-eracy measure varies with orthographic complexity A centralquestion was whether the outlier orthography of English (Share2008) behaves in any crucial aspects differently from more regu-lar and consistent orthographies Such an orthographic differencewould seriously limit the generalizability of the rich Englishresearch literature to other orthographies In summary our find-ings confirm the assumption (H31) that the predictive pattern issimilar across orthographies once the same literacy componentsare compared English behaves like more consistent alphabeticorthographies to a large extent but with two notable differences(1) As assumed (H21) the overall predictive power of the cognitiveskills of interest on literacy measures was higher in English (25e39) than in more consistent orthographies (9e26) (2) The as-sociation between RAN and reading accuracy as well as between

sylvianevaldois
Texte surligneacute

Table 7Multilevel analyses for the three predictors (aec) for reading accuracy

Reading accuracy







084 [009 to 156] 081Random effect estimates RAN by languageEnglish 214 [000 to 464]French 075 [246 to 024]German 029 [110 to 056]Hungarian 041 [142 to 055]Finnish 069 [177 to 032]Global heterogeneity of random effects 008

Table 8Multilevel analyses for the three predictors (aec) for spelling

Spelling





325 [262 to 386] lt001Random effect estimates PA by LanguageEnglish 007 [047 to 121]French 027 [207 to 013]German 009 [050 to 088]Hungarian 036 [002 to 179]Finnish 025 [124 to 024]Global heterogeneity of random effects 268


185 [109 to 271] lt001Random effect estimates RAN by LanguageEnglish 013 [018 to 300]French 019 [302 to 028]German 009 [070 to 111]Hungarian 011 [051 to 155]Finnish 015 [179 to 054]Global heterogeneity of random effects 435


RAN and spelling was negligible in more consistent orthographies(0e2) whereas in English RAN turned out to be a significantpredictor for reading accuracy and spelling (14 and 16respectively)

41 Predictive pattern for reading speed

The predictive pattern for reading speed was highly similar inconsistent and less consistent alphabetic orthographies with RANbeing a strong and consistent predictor in all five orthographies afinding that is very much in line with Vaessen et alrsquos (2010) recentcross-linguistic analysis of the cognitive mechanisms underpinningreadingfluency inPortugueseDutch andHungarian Interestingly itwas not English but Finnish which is located on the extreme otherend of the continuum of orthographic complexity that turned out tobehave like an outlier orthography in the current study Although therelative influence of RAN on reading speed did not differ betweenlanguages the absolute contribution of RAN in explaining individualdifferences in reading fluency in the Finnish sample was relativelysmall and not larger than that of the two phonological predictorsZiegler et al (2010) recently also found a relatively minor impact ofRANon Finnish childrenrsquos reading attainment suggesting that in thishighly transparent orthographic system reading skills may be morestrongly dependent on other factors like reading experience than onRAN This effect may be limited to typically developing readers as anumber of earlier studies on Finnish including poor or dyslexicreaders consistently reported RAN to be the strongest predictor ofreading speed (Holopainen Ahonen amp Lyytinen 2001 LepolaPoskiparta Laakkonen amp Niemi 2005 Puolakanaho et al 2007Torppa Lyytinen Erskine Eklund amp Lyytinen 2010) One importantconclusion that canbedrawnbasedon the similarityof thepredictive

pattern for reading speed is that the consistent inclusion of readingspeedmeasures in English studies is highly desirable andwould helpto increase the comparability of findings across orthographies

42 Predictive pattern for reading accuracy

Currently the central measure of reading attainment in Englishstudies is usually reading accuracy in especially number of items(words or nonwords) read correctly This measure is not alwaysuseful in more consistent orthographies due to reduced variance inaccuracy scores In the current study reading accuracy wasmeasured under speeded conditions (ldquoRead as fast as possiblerdquo)which helped to induce reasonably distributed numbers of incor-rect readings in all languages As language-specific tests were usedand scores were z-standardized separately for each subsample adirect comparison of reading accuracy across orthographies is notfeasible in the current design Instead of analyzing absolute per-formance scores we compared the relative importance of pre-dictors between languages The overall variance in accuracyaccounted for by phonological processing and RAN tended to belarger in English than in the other orthographies whichmay at leastpartly be due to the greater variance of the reading accuracymeasure in the English sample The most obvious differencehowever was that English was the only orthography where RANcould account for a significant amount of variance in reading ac-curacy above and beyond phonological processing

43 Predictive pattern for spelling

This is the first study that investigated the cognitive un-derpinnings of spelling development in different orthographiesbeyond Grade 2 As predicted (H31) no significant interactions

sylvianevaldois
Texte surligneacute


between predictors and orthographic structure were observedHowever as it was found for reading accuracy the predictivepattern was only partly consistent across orthographies Onceagain a higher amount of variance could be explained in Englishthan in the other orthographies (H21) Phonological processingwas the better proximal predictor of spelling in all orthographiesexcept English where RAN accounted for more variance thanphonological awareness Within the two phonology measuresmemory was somewhat more important for spelling than forreading speed and accuracy This finding probably reflects that inmost orthographies letter-sound correspondences are less consis-tent in the spelling than in the reading direction As a consequencestoring word-specific knowledge in memory is crucial in order toproduce orthographically correct spellings However in general thissubcomponent played a rather minor role a finding that is in linewith most earlier studies (Caravolas et al 2012 Vaessen et al2010 Ziegler et al 2010)

In summary the current large-scale analysis of the associations ofphonological processing and RAN with reading and spelling indifferent alphabetic orthographies allows the conclusion that thecommonalities of cognitive underpinnings of literacy developmentbetween these orthographies are obviously prevailing Previousstudies (eg Caravolas et al 2012 2013) reported similar predictivepatterns across orthographies in the very early phases of literacydevelopment The current study complements these findings byshowing that similarities between languages can also be observedlateron inprimary school Still there are also anumberoffine-graineddifferences that warrant further investigation in more detailedresearch designs

44 Educational implications

Our findings have a number of practical implications concerningthe assessment of literacy skills

(1) Literacy skills do not represent a single construct as the un-derlying cognitive mechanisms vary depending on the literacycomponent (reading speed reading accuracy or spelling) thatis assessed Comprehensive assessment batteries need todifferentiate between these literacy components

(2) Most assessment batteries that include cognitive measuresassociated with literacy skills focus on phonological processingwhereas performance in RAN is not always assessed Whilephonological processing is a reliable predictor of individualdifferences in spelling it is a less useful predictor of readingskills especially in more consistent orthographies wherereading speed (not accuracy) is the relevant measure todifferentiate between good and poor readers Assessment toolsshould therefore include both phonological processing andRAN given that both cognitive skills are significant and uniquepredictors of literacy performance across orthographies

(3) In line with a number of other studies (Landerl et al 2013Vaessen et al 2010 Ziegler et al 2010) our findings indicatethat phonological processing andRANare generally less powerfulin explaining performance in reading and spelling in consistent(ie in Finnish) compared to inconsistent orthographies Thisimplies that children with low performance in phonologicalprocessing or RAN have a better chance to develop adequate lit-eracy skills in consistent than in inconsistent orthographies

45 Limitations and implications for future research

The reader should be aware that although the integrated Euro-pean research initiative NEURODYS enabled the systematic direct

comparison of predictive patterns for an unprecedented number ofalphabetic orthographies and literacy measures there are certainmethodological limitations that result from this approach Firstsome measures required normalization in order to allow theintended cross-linguistic comparisons The predictive patterns re-ported here are based on rank ordered data and may therefore notbe directly comparable with earlier studies based on raw or stan-dard scores Second the samples sizes for the five languagesdiffered considerably The German sample was especially large dueto the fact that three German speaking countries were involved inthis European network In comparisonwith the German sample theEnglish and French samples were rather small but still of reason-able size In order to reduce the effect of sample size on the resultspredictive patterns rather than absolute performances werecompared between language groups

It should also be noted that variance was reduced in the cur-rent analysis as the sample did not include children whose readinglevel was more than one standard deviation below the age normDifferences in predictive patterns between orthographies mightoverall be larger across the whole range of literacy skills Theadvantage of this approach however is that the current analysis isinformative with respect to cognitive underpinnings of readingand spelling development in standard classrooms (see Section 44)and therefore goes beyond group comparisons between dyslexicand control readers Note that Landerl et alrsquos (2013) finding thatbased on phonological processing and RAN more participantswere correctly classified as dyslexic or typical reader in complexthan in less complex orthographies indicates that the predictivepattern reported here does extend to the whole range of readingskills

Finally the aim of the NEURODYS research initiative was toinvestigate concurrent predictions and longitudinal patterns ofprediction may differ This may be less of a problem for the RAN-reading relationship which seems to be mostly unidirectionalRAN predicts growth in reading while reading development doesnot seem to have a relevant impact on RAN performance (Lervaringg ampHulme 2009) For phonological memory a recent analysis (Nationamp Hulme 2011) even suggests that it is reading development thatdrives improvement in nonword repetition while nonword repe-tition cannot predict growth in reading skills The picture is mostcomplex for phonological awareness which obviously develops inclose interaction with the acquisition of an alphabetic orthographyThe apparently largely consistent phonological awarenesseliteracyrelationship that we found across orthographies may largely reflectthis close interaction Longitudinal studies in consistent orthogra-phies repeatedly found that the prediction of preschool phono-logical awareness is mostly limited to the early stages of readingdevelopment (de Jong amp van der Leij 1999 Landerl amp Wimmer2008) The longitudinal studies comparing the early prediction ofphonological awareness and RAN in a consistent orthography withEnglish corroborate this evidence (Furnes amp Samuelsson 2011Georgiou et al 2008 BUT see also Caravolas et al 2012 for nodifferential pattern across orthographies in the first 10 months ofliteracy instruction) Further research applying more fine-grainedresearch designs and following children beyond Grade 2 will benecessary to finally settle this question

Acknowledgments

This project was funded by the European Commission FP6grant to the Neurodys consortium as well as by Agence Nationalede la Recherche Ville de Paris Austrian Science Fund (project no18351-B02) SNSF grant 32-108130 and Stiftung fuumlr wissen-schaftliche Forschung an der Universitaumlt Zuumlrich The Finnish au-thors are members of the Finnish Center of Excellence Program


213486 for 2006e2011 supported by the Academy of Finland theNiilo Maumlki Foundation and the University of Jyvaumlskylauml We thankall the children and their families for their participation Wefurther thank the following people for their help Charles HulmeEdith Willburger Yves Chaix Camille Chabernaud Laure BricoutJohannes Ziegler Isabelle Comte-Gervais Sophie Jery GeneviegraveveLina Christine Nossent Corinne Bruchet Judith Kerleroux Van-nessa Bongiovanni Lionel Collet Ying Wang Marie-ChristineMouren

Appendix

Table A1Standardized reading tests applied in the five languages for sample selection

Language Reading test

English Elliot C Smith P amp McCulloch K (1997) British AbilityScales II Windsor NFERNelson

French Jacquier-Roux M Valdois S amp Zorman M (2005) OdeacutedysOutil de deacutepistage des dyslexiques (version 2) GrenobleLaboratoire Cognisciences

German Moll K amp Landerl K (2010) SLRT-II e Verfahren zurDifferentialdiagnose von Stoumlrungen der Teilkomponentendes Lesens und Schreibens Bern Huber

Hungarian Toacuteth D Cseacutepe V Vaessen A Blomert L (in press)3DM-H A diszlexia differenciaacuteldiagnoacutezisa Az olvasaacutes eacuteshelyesiacuteraacutes kognitiacutev elemzeacutese Nyiacuteregyhaacuteza Kogentum

Finnish Haumlyrinen T Serenius-Sirve S amp Korkman M (1999)Lukilasse Helsinki Psykologien Kustannus Oy

Table A2Number of participants by grade and country

Language Country Grade N

English United Kingdom 3 74 115 126 157 15Total 60

French France 3 234 265 226 15Total 86

German Germany 3 1244 96Total 220

Switzerland 2 83 104 25 25Total 45

Austria 2 593 884 61Total 208

Hungarian Hungary 2 753 634 57Total 195

Finnish Finland 2 873 1604 1

Total 248

References

Borgwaldt S R Hellwig F M amp De Groot A M B (2005) Onset entropy mattersletter-to-phoneme mappings in seven languages Reading and Writing 18 211e229 httpdxdoiorg101007s11145-005-3001-9

Bowers P Golden J Kennedy A amp Young A (1994) Limits upon orthographicknowledge due to processes indexed by naming speed In V W Berninger (Ed)The varieties of orthographic knowledge I Theoretical and developmental issues(pp 173e218) Dordrecht The Netherlands Kluwer Academic Press httpdxdoiorg101007978-94-017-3492-9_6

Bowers P G amp Newby-Clark E (2002) The role of naming speed within a model ofreading acquisition Reading and Writing An Interdisciplinary Journal 15 109e126 httpdxdoiorg101023A1013820421199

Byrne B (1998) The foundation of literacy The childrsquos acquisition of the alphabeticprinciple Hove UK Psychology Press

Caravolas M (2005) The nature and causes of dyslexia in different languages InM J Snowling amp C Hulme (Eds) The science of reading (pp 336e355) OxfordBlackwell

Caravolas M Lervaringg A Defior S Seidlovaacute Maacutelkovaacute G amp Hulme C (2013)Different patterns but equivalent predictors of growth in reading in consistentand inconsistent orthographies Psychological Science 24 1398e1407 httpdxdoiorg1011770956797612473122

Caravolas M Lervaringg A Mousikou P Efrim C Litavskyacute M Onochi-Quintanilla Eet al (2012) Common patterns of prediction of literacy development indifferent alphabetic orthographies Psychological Science 23 678e686 httpdxdoiorg1011770956797611434536

Caravolas M Volin J amp Hulme C (2005) Phoneme awareness is a key componentof alphabetic literacy skills in consistent and inconsistent orthographies evi-dence from Czech and English children Journal of Experimental Child Psychology92 107e139 httpdxdoiorg101016jjecp200504003

Conrad N J amp Levy B A (2007) Letter processing and the formation of memoryrepresentations in children with naming speed deficits Reading and Writing AnInterdisciplinary Journal 20 201e223 httpdxdoiorg101007s11145-006-9028-8

Decker S L Roberts A M amp Englund J A (2013) Cognitive predictors of rapidpicture naming Learning and Individual Differences 25 141e149 httpdxdoiorg101016jlindif201303009

Desimoni M Scalisi T G amp Orsolini M (2012) Predictive and concurrent relationsbetween literacy skills in grades 1 and 3 a longitudinal study of Italian childrenLearning and Instruction 22 340e353 httpdxdoiorg101016jlearninstruc201202002

Di Filippo G Brizzolara D Chilosi A De Luca M Judica A Pecini C et al (2005)Rapid naming not cancellation speed or articulation rate predicts reading in anorthographically regular language (Italian) Child Neuropsychology 11 349e361httpdxdoiorg10108009297040490916947

Ehri L C (1992) Reconceptualizing the development of sight word reading and itsrelationship to recoding In P B Gough L C Ehri amp R Treiman (Eds) Readingacquisition (pp 107e143) Hillsdale NJ Erlbaum

Furnes B amp Samuelsson S (2011) Phonological awareness and rapid automatizednaming predicting early development in reading and spelling results from across-linguistic longitudinal study Learning and Individual Differences 21 85e95 httpdxdoiorg101016jlindif201010005

Georgiou G K Parrila R amp Papadopoulos T C (2008) Predictors of worddecoding and reading fluency across languages varying in orthographic con-sistency Journal of Educational Psychology 100 566e580 httpdxdoiorg1010370022-06631003566

Holopainen L Ahonen T amp Lyytinen H (2001) Predicting delay in readingachievement in a highly transparent language Journal of Learning Disabilities34 401e413

de Jong P F amp van der Leij A (1999) Specific contributions of phonological abil-ities to early reading acquisition results from a Dutch latent variable longitu-dinal study Journal of Educational Psychology 91 450e476 httpdxdoiorg1010370022-0663913450

Katz L amp Frost R (1992) Reading in different orthographies the orthographicdepth hypothesis In R Frost amp L Katz (Eds) Orthography phonologymorphology and meaning (pp 67e84) Amsterdam Elsevier

Kirby J R Georgiou G K Martinussen R amp Parrila R (2010) Review of researchnaming speed and reading from prediction to instruction Reading ResearchQuarterly 45 341e362 httpdxdoiorg101598RRQ4534

Landerl K (2005) Reading acquisition in different orthographies evidence fromdirect comparisons In M Joshi amp P G Aaron (Eds) Handbook of orthographyand literacy (pp 513e530) Hove Erlbaum

Landerl K Ramus F Moll K Lyytinen H Leppaumlnen P H T Lohvansuu K et al(2013) Predictors of developmental dyslexia in European orthographies withvarying complexity Journal of Child Psychology and Psychiatry 54 686e694httpdxdoiorg101111jcpp12029

Landerl K amp Wimmer H (2008) Development of word reading fluency andorthographic spelling in a consistent orthography an 8-year follow-up Journalof Educational Psychology 100 150e161 httpdxdoiorg1010370022-06631001150

Lepola J Poskiparta E Laakkonen E amp Niemi P (2005) Development of andrelationship between phonological motivational processes and naming speedin predicting word recognition in grade 1 Scientific Studies of Reading 9 367e399 httpdxdoiorg101207s1532799xssr0904_3


Lervaringg A amp Hulme C (2009) Rapid automatized naming (RAN) taps a mechanismthat places constraints on the development of early readingfluency PsychologicalScience 20 1040e1048 httpdxdoiorg101111j1467-9280200902405x

Manis F R Seidenberg M S amp Doi L M (1999) See Dick RAN rapid naming and thelongitudinal prediction of reading subskills in first and second graders ScientificStudies of Reading 3 129e157 httpdxdoiorg101207s1532799xssr0302_3

Mann V amp Wimmer H (2002) Phoneme awareness and pathways into literacy acomparison of German and American children Reading and Writing An Inter-disciplinary Journal 15 653e682 httpdxdoiorg101023A1020984704781

Manolitsis G Georgiou G K amp Parrila R (2011) Revisiting the home literacy modelof reading development in an orthographically consistent language Learning andInstruction 21 496e505 httpdxdoiorg101016jlearninstruc201006005

Mayringer H Wimmer W amp Landerl K (1998) Phonological skills and literacyacquisition in German In P Reitsma amp L Verhoeven (Eds) Problems and in-terventions in literacy development (pp 147e161) Dordrecht The NetherlandsKluwer Academic

Moll K Fussenegger B Willburger E amp Landerl K (2009) RAN is not a measureof orthographic processing Evidence from the asymmetric German orthog-raphy Scientific Studies of Reading 13 1e25 httpdxdoiorg10108010888430802631684

Nation K amp Hulme C (2011) Learning to read changes childrenrsquos phonologicalskills evidence from a latent variable longitudinal study of reading andnonword repetition Developmental Science 14 649e659 httpdxdoiorg101111j1467-7687201001008x

Patel T K Snowling M J amp de Jong P F (2004) A cross-linguistic comparison ofchildren learning to read in English and Dutch Journal of Educational Psychol-ogy 96 785e797 httpdxdoiorg1010370022-0663964785

Perfetti C A (1992) The representation problem in reading acquisition InP B Gough L C Ehri amp R Treiman (Eds) Reading acquisition (pp 145e174)Hillsdale NJ Erlbaum

Protopapas A amp Vlahou E L (2009) A comparative quantitative analysis of Greekorthographic transparency Behavior Research Methods 41 991e1008 httpdxdoiorg103758BRM414991

Puolakanaho A Ahonen T Aro M Eklund K Leppanen P H Poikkeus A Met al (2007) Very early phonological and language skills estimating individualrisk of reading disability Journal of Child Psychology and Psychiatry 48 923e931httpdxdoiorg101111j1469-7610200701763x

Seymour P H Aro M amp Erskine J M (2003) Foundation literacy acquisition inEuropean orthographies British Journal of Psychology 94 143e174 httpdxdoiorg101348000712603321661859

Share D L (2008) On the Anglocentricities of current reading research and prac-tice the perils of overreliance on an ldquooutlierrdquo orthography PsychologicalBulletin 134 584e615 httpdxdoiorg1010370033-29091344584

Swanson H L Trainin G Necoechea D M amp Hammill D D (2003) Rapid namingphonological awareness and reading a meta-analysis of the correlation evi-dence Review of Educational Research 73 407e440 httpdxdoiorg10310200346543073004407

Torgesen J K Wagner R K amp Rashotte C A (1994) Longitudinal studies ofphonological processing and reading Journal of Learning Disabilities 27 276e286 httpdxdoiorg101177002221949402700503

Torgesen J K Wagner R K Rashotte C A Burgess S R amp Hecht S A (1997)Contributions of phonological awareness and rapid automatic naming abilityto the growth of word-reading skills in second- to fifth-grade children Sci-entific Studies of Reading 1 161e185 httpdxdoiorg101207s1532799xssr0102_4

Torppa M Lyytinen P Erskine J Eklund K amp Lyytinen H (2010) Languagedevelopment literacy skills and predictive connections to reading in Finnishchildrenwith and without family risk of dyslexia Journal of Learning Disabilities43 308e321 httpdxdoiorg1011770022219410369096

Vaessen A Bertrand D Toacuteth D Cseacutepe V Faiacutesca L Reis A et al (2010) Cognitivedevelopment of fluent word reading does not qualitatively differ betweentransparent and opaque orthographies Journal of Educational Psychology 102827e842 httpdxdoiorg101037a0019465

Vaessen A Gerretsen P amp Blomert L (2009) Naming problems do not reflect asecond independent core deficit in dyslexia double deficits explored Journal ofExperimental Child Psychology 103 202e221 httpdxdoiorg101016jjecp200812004

Wagner R K amp Torgesen J K (1987) The nature of phonological processing and itscausal role in the acquisition of reading skills Psychological Bulletin 101 192e212 httpdxdoiorg1010370033-29091012192

Wechsler D (1992) The Wechsler intelligence scale for children (3rd ed) LondonThe Psychological Corporation

Wechsler D (2003) Wechsler intelligence scale for children (4th ed) San AntonioTX Psychological Corporation

Ziegler J C Bertrand D Toacuteth D Cseacutepe V Reis A Faiacutesca L et al (2010)Orthographic depth and its impact on universal predictors of reading a cross-language investigation Psychological Science 21 551e559 httpdxdoiorg1011770956797610363406

Ziegler J C amp Goswami U (2005) Reading acquisition developmentaldyslexia and skilled reading across languages a psycholinguistic grain sizetheory Psychological Bulletin 131 3e29 httpdxdoiorg1010370033-290913113

Ziegler J C Jacobs A M amp Stone G O (1996) Statistical analysis of the bidirec-tional inconsistency of spelling and sound in French Behavior Research MethodsInstruments and Computers 28 504e515 httpdxdoiorg103758BF03200539


26 August 2013Accepted 3 September 2013

KeywordsReading developmentCross-linguisticOrthographic consistencyPhonological awarenessRapid automatized naming

large European sample of 1062 typically developing elementary school children beyond Grade 2acquiring five different alphabetic orthographies with varying degrees of graphemeephoneme consis-tency (English French German Hungarian Finnish) Findings indicate that (1) phonological processingand RAN both account for significant amounts of unique variance in literacy attainment in all five or-thographies Associations of predictors with reading speed reading accuracy and spelling are differen-tial in general RAN is the best predictor of reading speed while phonological processing accounts forhigher amounts of unique variance in reading accuracy and spelling (2) the predictive patterns arelargely comparable across orthographies but they tend to be stronger in English than in all otherorthographies

2013 Elsevier Ltd All rights reserved

1 Introduction

Recently considerable research interest has been generated bythe question whether the cognitive underpinnings of readingacquisition vary between orthographies or whether they are largelysimilar All known orthographic systems represent languagehowever there is a large degree of variance in the consistency of themapping between spoken and written language and consequentlyin the transparency of these mappings for the young learner Themain principle of all alphabetic orthographies that are used in theWesternworld is that graphic symbols (letters) represent the soundstructure of the spoken word However few orthographies closelyadhere to this alphabetic principle of simple 11 relationships be-tween letters and phonemes (like Finnish) while most alphabetsprovide the reader with a certain degree of inconsistency or ir-regularity The English orthographic system with its many com-plexities is probably on the most extreme end of this continuum oforthographic consistency Both theoretical conceptions (Katz ampFrost 1992 Ziegler amp Goswami 2005) and empirical evidence(see Landerl 2005 for a review) indicate that the development ofdecoding skills (ie the systematic translation of graphemes intophonemes) takes considerably longer in English than in moreconsistent orthographies Thus the complicated and opaque map-ping system of English orthography seems to cause particularproblems to the young learner It is probably no coincidence thatthe investigation of reading acquisition in English strongly domi-nates the research field However the question then ariseswhether the outlier status of English orthographic complexity isreflected in the cognitive mechanisms underpinning the readingprocess which would seriously limit the relevance of such anldquoAnglocentric viewrdquo (Share 2008) for other orthographies Thisissue is not only of high theoretical interest but has importantimplications for reading instruction as the relevant cognitive pre-dictors are used to identify children who are at risk for readingfailure

11 Cognitive predictors of literacy skills

Two cognitive skills that are closely associated with the complexprocess of reading and spelling acquisition are phonological pro-cessing and rapid automatized naming (RAN) Phonological pro-cessing refers to the ability to perceive store and manipulatespeech sounds and includes phonological awareness and phono-logical working memory In a typical phonological awareness taska child might be asked to delete a certain sound from a word ornonword pronunciation (eg ldquoSaygulstwithout thelrdquo) The childthen has to maintain the sound sequence in working memoryidentify thel-sound in the phoneme string delete it from thepronunciation and blend the remaining sound parts Thus it isobvious that although such tasks are taken to measure phonolog-ical awareness they usually also require working memory capacityPhonological awareness enables the child to understand and sys-tematically exploit the mappings between graphic symbols and the

sound structure of spoken language It is crucial whenever thegraphemes of words or nonwords are decoded during reading andalso when words are segmented into their constituent phonemesduring spelling Thus phonological awareness plays an importantrole during early literacy development across alphabetic orthog-raphies (eg Byrne 1998 Wagner amp Torgesen 1987) however inconsistent orthographies competent graphemeephoneme andphonemeegrapheme translation is typically achieved earlier andgrowth of literacy skills is faster than in inconsistent orthographieslike English (eg Caravolas Lervaringg Defior Seidlovaacute Maacutelkovaacute ampHulme 2013 Seymore Aro amp Erskine 2003) Beyond these earlyphases of literacy development phonological awareness is sup-posed to exert its influence on building-up word-specific repre-sentations (Ehri 1992 Perfetti 1992) According to this theoreticalview an efficient storage of orthographic patterns depends onmultiple associations between phonological segments of a spokenword and the corresponding graphemes of its written form Word-specific orthographic representations enable direct word recogni-tion during reading and correct orthographic spelling Once againthe degree of consistency of grapheme- as well as phonemeegrapheme correspondences can be assumed to play an importantrole Coping with the many irregularities and inconsistenciesinherent in the English orthographic system may particularlychallenge the phonological system of the learner This would implythat the relevance of phonological processing skills should be lowerin consistent than in less consistent orthographies

Rapid automatized naming (RAN) refers to the speed withwhich an individual can pronounce the names of a sequentially andrepeatedly presented limited set of stimuli like letters Arabic digitscolor patches or pictures of familiar objects Performing RAN taskscertainly requires phonological skills (accessing the phonologicaloutput programs of the required word pronunciations as quickly aspossible) and is therefore sometimes seen as a third subcomponentof phonological processing (Torgesen Wagner amp Rashotte 1994Torgesen Wagner Rashotte Burgess amp Hecht 1997 VaessenGerretsen amp Blomert 2009) However there is now ample evi-dence that ldquonaming speed is phonological but not only phono-logicalrdquo (Kirby Georgiou Martinussen amp Parrila 2010 p 356) andconstitutes a second cognitive mechanism underpinning readingdevelopment that is largely independent from phonologicalawareness and memory First the correlation between phonolog-ical awareness and RAN is typically only low to moderate (38 in ameta-analysis of 35 studies that were almost exclusively carried outin English Swanson Trainin Necoechea amp Hammill 2003) Sec-ond although phonological awareness and RAN contribute someamount of shared variance both components have consistentlybeen shown to make unique contributions to the variance of liter-acy skills above and beyond the other one Third and most impor-tantly these unique contributions seem to be differentialphonological awareness and RAN have been demonstrated to showspecific relationships with particular subcomponents of literacyprocessing While phonological skills seem to be most stronglyrelated to literacy skills that involve decoding (most importantly











sylvianevaldois
Texte surligneacute
sylvianevaldois
Texte surligneacute










2 Method

21 Rationale











22 Participants2



23 Tasks




















N 60 86 473 195 248Gender

[ boys)717 442 512 518 508


[itemsmin]990 (211) 969 (179) 943 (179) 963 (143) 675 (127)


02 (86) 02 (91) 03 (96) 00 (95) 16 (92)


773 (138) 865 (98) 965 (39) 960 (37) 907 (93)


709 (153) 736 (202) 783 (164) 674 (174) 654 (218)







3 Results







FrenchGerman


HungarianFinnish














Reading speed

R2-change p B SE B












Reading accuracy

R2-change p B SE B





















Spelling

R2-change p B SE B












Reading speed












4 Discussion




sylvianevaldois
Texte surligneacute


Reading accuracy









Spelling

















sylvianevaldois
Texte surligneacute














Acknowledgments




Appendix

















Total 248

References




























































sylvianevaldois
Texte surligneacute
sylvianevaldois
Texte surligneacute










2 Method

21 Rationale











22 Participants2



23 Tasks




















N 60 86 473 195 248Gender

[ boys)717 442 512 518 508


[itemsmin]990 (211) 969 (179) 943 (179) 963 (143) 675 (127)


02 (86) 02 (91) 03 (96) 00 (95) 16 (92)


773 (138) 865 (98) 965 (39) 960 (37) 907 (93)


709 (153) 736 (202) 783 (164) 674 (174) 654 (218)







3 Results







FrenchGerman


HungarianFinnish














Reading speed

R2-change p B SE B












Reading accuracy

R2-change p B SE B





















Spelling

R2-change p B SE B












Reading speed












4 Discussion




sylvianevaldois
Texte surligneacute


Reading accuracy









Spelling

















sylvianevaldois
Texte surligneacute














Acknowledgments




Appendix

















Total 248

References



























































2 Method

21 Rationale











22 Participants2



23 Tasks




















N 60 86 473 195 248Gender

[ boys)717 442 512 518 508


[itemsmin]990 (211) 969 (179) 943 (179) 963 (143) 675 (127)


02 (86) 02 (91) 03 (96) 00 (95) 16 (92)


773 (138) 865 (98) 965 (39) 960 (37) 907 (93)


709 (153) 736 (202) 783 (164) 674 (174) 654 (218)







3 Results







FrenchGerman


HungarianFinnish














Reading speed

R2-change p B SE B












Reading accuracy

R2-change p B SE B





















Spelling

R2-change p B SE B












Reading speed












4 Discussion




sylvianevaldois
Texte surligneacute


Reading accuracy









Spelling

















sylvianevaldois
Texte surligneacute














Acknowledgments




Appendix

















Total 248

References


























































22 Participants2



23 Tasks




















N 60 86 473 195 248Gender

[ boys)717 442 512 518 508


[itemsmin]990 (211) 969 (179) 943 (179) 963 (143) 675 (127)


02 (86) 02 (91) 03 (96) 00 (95) 16 (92)


773 (138) 865 (98) 965 (39) 960 (37) 907 (93)


709 (153) 736 (202) 783 (164) 674 (174) 654 (218)







3 Results







FrenchGerman


HungarianFinnish














Reading speed

R2-change p B SE B












Reading accuracy

R2-change p B SE B





















Spelling

R2-change p B SE B












Reading speed












4 Discussion




sylvianevaldois
Texte surligneacute


Reading accuracy









Spelling

















sylvianevaldois
Texte surligneacute














Acknowledgments




Appendix

















Total 248

References


































































N 60 86 473 195 248Gender

[ boys)717 442 512 518 508


[itemsmin]990 (211) 969 (179) 943 (179) 963 (143) 675 (127)


02 (86) 02 (91) 03 (96) 00 (95) 16 (92)


773 (138) 865 (98) 965 (39) 960 (37) 907 (93)


709 (153) 736 (202) 783 (164) 674 (174) 654 (218)







3 Results







FrenchGerman


HungarianFinnish














Reading speed

R2-change p B SE B












Reading accuracy

R2-change p B SE B





















Spelling

R2-change p B SE B












Reading speed












4 Discussion




sylvianevaldois
Texte surligneacute


Reading accuracy









Spelling

















sylvianevaldois
Texte surligneacute














Acknowledgments




Appendix

















Total 248

References





















































FrenchGerman


HungarianFinnish














Reading speed

R2-change p B SE B












Reading accuracy

R2-change p B SE B





















Spelling

R2-change p B SE B












Reading speed












4 Discussion




sylvianevaldois
Texte surligneacute


Reading accuracy









Spelling

















sylvianevaldois
Texte surligneacute














Acknowledgments




Appendix

















Total 248

References



















































Reading speed

R2-change p B SE B












Reading accuracy

R2-change p B SE B





















Spelling

R2-change p B SE B












Reading speed












4 Discussion




sylvianevaldois
Texte surligneacute


Reading accuracy









Spelling

















sylvianevaldois
Texte surligneacute














Acknowledgments




Appendix

















Total 248

References



















































Spelling

R2-change p B SE B












Reading speed












4 Discussion




sylvianevaldois
Texte surligneacute


Reading accuracy









Spelling

















sylvianevaldois
Texte surligneacute














Acknowledgments




Appendix

















Total 248

References



















































Reading accuracy









Spelling

















sylvianevaldois
Texte surligneacute














Acknowledgments




Appendix

















Total 248

References































































Acknowledgments




Appendix

















Total 248

References




















































Appendix

















Total 248

References


















































Cognitive mechanisms underlying reading and spelling development in five European orthographies

Documents

Transcript of Cognitive mechanisms underlying reading and spelling development in five European orthographies