Vellutino

39
Speci¢c reading disability (dyslexia): what have we learned in the past four decades? Frank R. Vellutino, 1 Jack M. Fletcher, 2 Margaret J. Snowling, 3 and Donna M. Scanlon 1 1 The University at Albany, USA; 2 The University of Texas Health Science Center at Houston, USA; 3 The University of York, UK We summarize some of the most important findings from research evaluating the hypothesized causes of specific reading disability (ÔdyslexiaÕ) over the past four decades. After outlining compon- ents of reading ability, we discuss manifest causes of reading difficulties, in terms of deficiencies in component reading skills that might lead to such difficulties. The evidence suggests that inadequate facility in word identification due, in most cases, to more basic deficits in alphabetic coding is the basic cause of difficulties in learning to read. We next discuss hypothesized deficiencies in reading- related cognitive abilities as underlying causes of deficiencies in component reading skills. The evidence in these areas suggests that, in most cases, phonological skills deficiencies associated with phonological coding deficits are the probable causes of the disorder rather than visual, semantic, or syntactic deficits, although reading difficulties in some children may be associated with general language deficits. Hypothesized deficits in general learning abilities (e.g., attention, association learning, cross-modal transfer etc.) and low-level sensory deficits have weak validity as causal fac- tors in specific reading disability. These inferences are, by and large, supported by research evalu- ating the biological foundations of dyslexia. Finally, evidence is presented in support of the idea that many poor readers are impaired because of inadequate instruction or other experiential factors. This does not mean that biological factors are not relevant, because the brain and environment interact to produce the neural networks that support reading acquisition. We conclude with a discussion of the clinical implications of the research findings, focusing on the need for enhanced instruction. The question of why some children have difficulty learning to read has been the focus of a great deal of research over the past four decades and much has been learned about the probable and improbable causes of such difficulty. Of special interest in this very rich and prolific area of inquiry have been chil- dren who have at least average intelligence, who do not have general learning difficulties, and whose reading problems are not due to extraneous factors such as sensory acuity deficits, socioeconomic dis- advantage, and like factors. Reading problems in such children are manifested in extreme difficulties in acquiring basic reading subskills such as word identification and phonological (letter-sound) de- coding. Such difficulties have been estimated to occur in approximately 10% to 15% of school age children (Benton & Pearl, 1978; Harris & Sipay, 1990; Shaywitz, Escobar, Shaywitz, Fletcher, & Makuch, 1992) and tend to be accompanied by specific deficits in cognitive abilities related to reading and other lit- eracy skills. This symptom pattern is often called ÔdyslexiaÕ, or, alternatively, Ôspecific reading disabil- ityÕ, and the terms are often used interchangeably. We will adopt this convention to refer to children whose reading difficulties occur at the level of basic reading subskills and are not caused by the exclusionary factors just mentioned (Lyon, 1995; Lyon, Fletcher, & Barnes, 2002; Shaywitz, 1996). The primary purpose of the present paper is to provide a selective and relatively up-to-date review of research, conducted over the past four decades, emanating both from historically influential and contemporary conceptualizations of the basic cause(s) of developmental dyslexia. We focus on conceptualizations specifying cognitive and biologi- cal deficits that may underlie this disorder and special emphasis is placed on contemporary and somewhat controversial theories of dyslexia that have gained some degree of prominence in more recent years. We briefly describe and critically an- alyze each of the theories instantiated and sum- marize some of the more important findings amassed by researchers who have evaluated its correlates and extensions, in the interest of distin- guishing between probable and improbable causes of early reading difficulties in children presumed to be dyslexic. Our discussion, throughout, is based on the assumption that causal relationships between given skills and abilities hypothesized to underlie the ability to learn to read, on the one hand (e.g., visual perception, verbal memory), and reading ability on the other (e.g., word identification, reading com- prehension), can only be confidently inferred in instances where results are generated by experi- mental or quasi-experimental research designs. Journal of Child Psychology and Psychiatry 45:1 (2004), pp 2–40 Ó Association for Child Psychology and Psychiatry, 2004. Published by Blackwell Publishing, 9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main Street, Malden, MA 02148, USA

description

Vellutino: artigo científico

Transcript of Vellutino

Page 1: Vellutino

Speci¢c reading disability (dyslexia): what havewelearned in the past four decades?

Frank R. Vellutino,1 Jack M. Fletcher,2 Margaret J. Snowling,3

and Donna M. Scanlon1

1The University at Albany, USA; 2The University of Texas Health Science Center at Houston, USA;3The University of York, UK

We summarize some of the most important findings from research evaluating the hypothesizedcauses of specific reading disability (�dyslexia�) over the past four decades. After outlining compon-ents of reading ability, we discuss manifest causes of reading difficulties, in terms of deficiencies incomponent reading skills that might lead to such difficulties. The evidence suggests that inadequatefacility in word identification due, in most cases, to more basic deficits in alphabetic coding is thebasic cause of difficulties in learning to read. We next discuss hypothesized deficiencies in reading-related cognitive abilities as underlying causes of deficiencies in component reading skills. Theevidence in these areas suggests that, in most cases, phonological skills deficiencies associated withphonological coding deficits are the probable causes of the disorder rather than visual, semantic, orsyntactic deficits, although reading difficulties in some children may be associated with generallanguage deficits. Hypothesized deficits in general learning abilities (e.g., attention, associationlearning, cross-modal transfer etc.) and low-level sensory deficits have weak validity as causal fac-tors in specific reading disability. These inferences are, by and large, supported by research evalu-ating the biological foundations of dyslexia. Finally, evidence is presented in support of the idea thatmany poor readers are impaired because of inadequate instruction or other experiential factors. Thisdoes not mean that biological factors are not relevant, because the brain and environment interact toproduce the neural networks that support reading acquisition. We conclude with a discussion of theclinical implications of the research findings, focusing on the need for enhanced instruction.

The question of why some children have difficultylearning to read has been the focus of a great deal ofresearch over the past four decades and much hasbeen learned about the probable and improbablecauses of such difficulty. Of special interest in thisvery rich and prolific area of inquiry have been chil-dren who have at least average intelligence, who donot have general learning difficulties, and whosereading problems are not due to extraneous factorssuch as sensory acuity deficits, socioeconomic dis-advantage, and like factors. Reading problems insuch children are manifested in extreme difficultiesin acquiring basic reading subskills such as wordidentification and phonological (letter-sound) de-coding. Such difficulties have been estimated tooccur in approximately 10% to 15% of school agechildren (Benton&Pearl, 1978;Harris &Sipay, 1990;Shaywitz, Escobar, Shaywitz, Fletcher, & Makuch,1992) and tend to be accompanied by specific deficitsin cognitive abilities related to reading and other lit-eracy skills. This symptom pattern is often called�dyslexia�, or, alternatively, �specific reading disabil-ity�, and the terms are often used interchangeably.Wewill adopt this convention to refer to children whosereading difficulties occur at the level of basic readingsubskills and are not caused by the exclusionaryfactors just mentioned (Lyon, 1995; Lyon, Fletcher, &Barnes, 2002; Shaywitz, 1996).

The primary purpose of the present paper is toprovide a selective and relatively up-to-date reviewof research, conducted over the past four decades,emanating both from historically influential andcontemporary conceptualizations of the basiccause(s) of developmental dyslexia. We focus onconceptualizations specifying cognitive and biologi-cal deficits that may underlie this disorder andspecial emphasis is placed on contemporary andsomewhat controversial theories of dyslexia thathave gained some degree of prominence in morerecent years. We briefly describe and critically an-alyze each of the theories instantiated and sum-marize some of the more important findingsamassed by researchers who have evaluated itscorrelates and extensions, in the interest of distin-guishing between probable and improbable causesof early reading difficulties in children presumed tobe dyslexic.

Our discussion, throughout, is based on theassumption that causal relationships between givenskills and abilities hypothesized to underlie theability to learn to read, on the one hand (e.g., visualperception, verbal memory), and reading ability onthe other (e.g., word identification, reading com-prehension), can only be confidently inferred ininstances where results are generated by experi-mental or quasi-experimental research designs.

Journal of Child Psychology and Psychiatry 45:1 (2004), pp 2–40

� Association for Child Psychology and Psychiatry, 2004.Published by Blackwell Publishing, 9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main Street, Malden, MA 02148, USA

Page 2: Vellutino

Such research must document improved perform-ance on measures of reading ability as a function ofimproved performance on measures of the skills orabilities hypothesized to underlie reading ability(Shadish, Cook, & Campbell, 2002). Absent suchdocumentation, no causal relationships can be in-ferred, even in instances where poor and normalreaders are shown to be reliably different on givenmeasures, at least not until such correlated rela-tionships are substantiated through controlledexperimentation. It follows that research demon-strating that poor and normal readers do not differreliably on measures of skills and abilities hypo-thesized to underlie reading ability, in most cases,can be taken as evidence that such variables arenot causally related to reading ability.1 It will be-come apparent that there are fewer causal rela-tionships that have been documented in the readingdisability literature that is reviewed herein thanthere are correlated relationships that are not fullyunderstood.

A secondary purpose of our review is to presentresearch evidence documenting the importance ofdistinguishing between early reading difficulties thatmay be caused primarily by cognitive and biologicaldeficits and early reading difficulties that may becaused primarily by experiential and instructionaldeficits. Results from recent intervention studiessuggest that explanations of reading difficulties inmost children must incorporate experiential andinstructional deficits as possible causes of such dif-ficulties, rather than focus exclusively on the types ofcognitive and biological deficits that have predom-inated theory and research in this area of inquirythroughout the previous century. Selected findingsfrom some of this research are discussed, placingspecial emphasis on their implications for practi-tioners.

However, to set the stage for our review, we firstdiscuss the components of literacy in terms of theknowledge, skills, and abilities presumed to underliereading ability. The intent here is to embed our dis-cussion of the different theories of dyslexia withinthe context of a process model that specifies possiblesources of difficulties in learning to read. We thendiscuss the manifest causes of such difficulties, asreflected in deficiencies in basic reading subskillssuch as word identification, phonological awareness,

and phonological decoding. However, because thecausal relationships between deficits in these read-ing subskills and early reading difficulties have beenreasonably well established in previous research, wedo not discuss this research in great detail. Thus, themain body of the text is devoted to discussion of thevarious cognitive deficit theories of dyslexia thathave generated a great deal of empirical research inthe last four decades and we highlight convergingevidence from cross-sectional, longitudinal, andcross-linguistic studies supporting the view thatlinguistic coding deficits are the most probablecauses of reading difficulties in dyslexic children.This view is given additional support from results ofneurobiological, genetic, family risk, and life-spandevelopment studies, which are discussed in sub-sequent sections of the review. We then summarizethe most important findings from a first grade read-ing intervention study that was specifically designedto evaluate the utility of using response to remedialintervention as a primary vehicle for distinguish-ing between cognitive/biological and experiential/instructional deficits as basic causes of early readingdifficulties. We close with a brief discussion of theimplications of the results of this and other studiesreviewed in this paper for practitioners working withreading impaired children. Note, however, that ourreview is not exhaustive and we do not presume todiscuss all areas of research on dyslexia. For exam-ple, there is an extensive research literature onsubtypes of dyslexia that we did not address, largelybecause this research has not been fruitful inenhancing our understanding of dyslexia subtypesat the cognitive level, with few studies finding evid-ence of relations between subtypes and biological orintervention findings (Lyon et al., 2002).

Components of reading ability

Normal reading ability assumes adequate languagecomprehension and fluent word identification. Writ-ten words are encoded (symbolized) representationsof spoken words, and spoken words are encodedrepresentations of environmental experiences andentities. Thus, the ability to learn to read depends onthe acquisition of a variety of different types ofknowledge and skills, which, themselves, depend onnormal development of reading-related linguisticand non-linguistic cognitive abilities. Figure 1 pre-sents a model depicting the cognitive processes anddifferent types of knowledge involved in learning toread. The model depicts processes whereby worldknowledge and domain specific knowledge stored inpermanent memory are transformed into units ofspoken and written language. These include lin-guistic processes and knowledge that allow one toacquire a spoken word vocabulary and languageskills in general, as well as visual, linguistic, andmetalinguistic processes and knowledge that allow

1 Exceptions to this generalization would, of course, include

instances where deficiencies in a basic skill or ability might be

causally related to difficulties in learning to read at beginning

stages of reading development, but not at later stages, because

of experiences that corrected such deficiencies (e.g., reading

instruction) and, thereby, masked the causal relationship. For

example, deficiencies in letter-name knowledge has been

shown to reliably distinguish between poor and normal readers

early in their reading development, but not later (Vellutino

et al., 1996), despite the distinct possibility that deficiencies in

letter-name knowledge might well be a factor contributing to

difficulties in learning to read.

Specific reading disability (dyslexia) 3

Page 3: Vellutino

one to acquire knowledge and skills that are influ-ential determinants of the ability to learn to read.Visual coding processes, broadly defined, refer tosensory and higher-level visualization processes thatfacilitate storage of representations defining the vis-ual attributes of environmental stimuli, includingthe graphic symbols used to represent written words.Linguistic coding processes refer to processes thatfacilitate language acquisition and the use of lan-guage for coding, storing and retrieving information.They include phonological coding – the ability to usespeech codes to represent information in the form ofwords and word parts; semantic and morphologicalcoding – the ability to store information about themeanings of concepts represented by words andword parts (e.g., ing, ed); syntactic coding – theability to store word order rules that set constraintson how words are organized in sentences; andpragmatic coding – the ability to store informationabout conventions governing the use of language asa medium of communication (e.g., modifications inmeaning signaled by changes in volume, pitch, andintensity in spoken language or punctuation marksin written language, use of understandable languageon the part of the speaker or writer, etc.).

Linguistic and visual coding processes togetherfacilitate the establishment of firm associationsbetween the spoken and written counterparts of

printed words, in the interest of helping the childacquire a sight word vocabulary – that is, a corpus ofprinted words the child is able to identify (name) onsight as lexical units (unanalyzed meaning-bearingunits). This associative learning process, itself, de-pends on the child’s understanding of print conceptsand conventions: that written words represent wordsin spoken language, that they are comprised of let-ters, that they are processed from left to right (inwritten English), that they are demarcated byspaces, and so forth. However, because of the heavyload on visual memory imposed by the high degree ofsimilarity characteristic of words derived from analphabet (pot/top; was/saw), sight word learningalso depends on the child’s ability to acquireunderstanding and functional use of the alphabeticprinciple. Understanding the alphabetic principle isimportant for acquiring proficiency in phonological(letter-sound) decoding, which is the primary vehiclebeginning readers use for reducing the load on visualmemory imposed by an alphabetic writing system.

Such proficiency, in turn, will require that the childactively engage in the type of metalinguistic analysis(analysis of language structures) that will facilitateacquisition of sublexical (letter-level) knowledge, inparticular, phonological and orthographic aware-ness, alphabetic knowledge, and general orthogra-phic knowledge. Phonological awareness refers to

Knowledge of Print Concepts and Conventions

Permanent Memory: World Knowledge and Domain Specific

Knowledge

Word Identification, LanguageComprehension and

Reading Comprehension

Metalinguistic Processes and Knowledge

Linguistic Coding Processes and Knowledge:

♦ Phonological ♦ Semantic ♦ Morphological ♦ Syntactic ♦ Pragmatic

Visual Coding Processes and Knowledge

Lexical Knowledge:

♦ Spoken Words

♦ Printed Words

Sublexical Knowledge:♦ Phonological Awareness ♦ Orthographic Awareness ♦ Alphabetic Knowledge ♦ General Orthographic

Knowledge

Written Text

Working Memory Processes

Permanent Memory: World Knowledge and Domain Specific

Knowledge

Figure 1 Cognitive processes and different types of knowledge entailed in learning to read

4 Frank R. Vellutino et al.

Page 4: Vellutino

conceptual understanding and explicit awarenessthat spoken words consist of individual speechsounds (phonemes) and combinations of speechsounds (syllables, onset-rime units). Such knowledgeis believed to be important for learning that letterscarry sound values and for learning tomapalphabeticsymbols to sounds. Orthographic awareness refers tothe child’s sensitivity to constraints on how the lettersin written words are organized (vid is legal, xqr isillegal). Phonological and orthographic awareness arereciprocally related cognitions that ultimately work inconcert to help the child acquire andmake functionaluse of general orthographic knowledge, in the form ofsensitivity to the regularities and redundanciescharacteristic of analphabeticwritingsystem (e.g., �at�in �cat�, �fat� and �rat�; �ing� in �walking� and �running�).These processes are complemented by another type ofmetalinguistic knowledge: syntactic awareness (notshown in Figure 1). Syntactic awareness refers to thechild’s sensitivity to grammatical form in terms of er-rors that violate conventional usage in spoken andwritten language (�Mom brung the cat to the vet�). Itfacilitates detection of reading errors, and, thereby,conjoint use of context-based strategies and phono-logical decoding strategies for word identification andcomprehension during text processing. Together,these three types of knowledge help the child acquireand consolidate both alphabetic knowledge and themore general orthographic knowledge that ultimatelyleads tomastery of the alphabetic code and increasingaccuracy and fluency in word identification andspelling.

Finally, the model in Figure 1 depicts both thepermanent memory and the working memory sys-tems and processes involved in learning to read. Thedouble-directional arrows represent the reciprocaland interactive relationships between the differentcoding and memory systems involved in (a) estab-lishing firm connective bonds between lexical andsublexical components of spoken and printed words,and (b) encoding, storing, and retrieving the differenttypes of information entailed in learning to read.

It should be clear that, given adequate exposure toprint, adequate literacy instruction, and adequatemotivation, the child’s ability to acquire the skillsand subskills necessary to learn to read is dependenton normal development and functioning of thesedifferent coding and memory systems and processes.Thus, difficulties in learning to read could comeabout, either from specific deficiencies in reading-related cognitive abilities resulting from abnormaldevelopment and consequent dysfunction in one ormore of these coding and memory systems and pro-cesses or from a less than adequate mix of reading-related cognitive abilities (the child’s phenotype)resulting from the interaction of the child’s particu-lar genetic endowment (the child’s genotype) and theparticular environmental and instructional experi-ences to which he or she has been exposed. Mosttheories of dyslexia are of the former variety and can

be classified as qualitative difference, �all or nonetype� theories (you either have or do not have dys-lexia) because they are based on the assumption thatreading difficulties are ultimately caused by struc-tural or functional anomalies in the brain. However,some scholars have given serious consideration toquantitative difference, continuous abilities typetheories of dyslexia (e.g., Olson & Gayan, 2001;Pennington & Lefly, 2001; Shaywitz et al., 1992;Snowling, Gallagher, & Frith, 2003; Stanovich,1988; Vellutino et al., 1996). Such theories suggestthat reading ability is a multifactorial trait that isdetermined by the action of a particular assortmentof genes (called quantitative trait loci) that areimportant for acquiring reading skills and subskills,interacting with environmental factors to producequantitative variation in cognitive abilities underly-ing reading ability and reading achievement. Con-tinuous abilities type theories of dyslexia are basedon the assumption that reading disability occursalong a continuum defining levels of reading abilityand that there is a gradation of risk for becomingdyslexic, depending on the particular assortment ofreading-related cognitive abilities with which thechild is endowed and the degree to which that child’shome and school environments capitalize and buildon his or her cognitive strengths and compensate forhis or her cognitive weaknesses. Because such the-ories are not well developed, we only touch uponthem in this review. In the next section, we brieflydiscuss manifest causes of specific reading disabil-ity, in terms of deficiencies in reading subskills thathave been found to be causally related to earlyreading difficulties.

Manifest causes of specific reading disability:deficiencies in reading subskills

Reading may be defined as the process of extractingand constructing meaning from written text for somepurpose. Skilled reading entails on-line comprehen-sion of meaning from running text. It is a complexprocess that depends on adequate development oftwo component processes: word identification andlanguage comprehension. Word identification is alexical retrieval process (see Figure 1) that involvesvisual recognition of a uniquely ordered array ofletters as a familiar word and implicit (or explicit)retrieval of the name and meaning of that word frommemory. Language comprehension involves integ-ration of the meanings of spoken or written words inways that facilitate understanding and integration ofsentences in spoken or written text in the interest ofunderstanding the broader concepts and ideas rep-resented by those sentences. Thus, in order tocomprehend what one reads, one must be able toidentify the words contained in running text withenough accuracy and fluency to allow computationof the meanings embodied in the text within the

Specific reading disability (dyslexia) 5

Page 5: Vellutino

limits of working memory. One must also have ade-quate language comprehension, and, ultimately,adequate world knowledge and adequate domain-specific knowledge. However, research in the studyof reading disability has made it clear that earlyreading difficulties in the population of children de-fined herein, that is, children who might qualify for adiagnosis of dyslexia in accord with the criteriaoutlined earlier, are manifested primarily in inad-equate facility in printed word identification as wellas inadequate facility in related skills such asspelling and phonological (letter-sound) decoding.Such difficulties may or may not be accompanied bysignificant deficits in language comprehension, butthey are not necessarily accompanied by such defi-cits. Thus, specific reading disability (dyslexia) inotherwise normal children has been and continuesto be defined as a basic deficit in learning to decodeprint. There are several pieces of evidence to supportthis definition.

First, we know from studies evaluating the rela-tionship between printed word identification andwritten and oral language comprehension processesthat reading comprehension is impaired in an indi-vidual who has inadequate facility in word identi-fication, in terms of both accuracy and fluency, evenif that individual has adequate language compre-hension skills (Gough & Tunmer, 1986; Perfetti,1985; Snowling, 2000a; Stanovich, 1991; Vellutino,1979, 1987; Vellutino, Scanlon, & Tanzman, 1994;Vellutino, Scanlon, & Chen, 1995a; Vellutino et al.,1996). These same studies demonstrate the con-verse, that is, that children who have inadequatefacility in reading comprehension are typically foundto have inadequate facility in word identification andrelated word-level skills such as spelling and pho-nological decoding. Thus, it would seem that themost basic and most ubiquitous cause of difficultiesin learning to read is inadequate facility in wordidentification, which, itself, appears to be caused bybasic difficulty in learning to decode print.

This possibility is given added credibility by re-sults from a second line of research, that is, regres-sion studies evaluating skills and abilitiesunderlying reading ability (Catts, Hogan, & Fey,2003; Curtis, 1980; Foorman, Francis, Shaywitz,Shaywitz, & Fletcher, 1997; Hoover & Gough, 1990;Vellutino, Scanlon, Small, & Tanzman, 1991; Vellu-tino et al., 1994). These studies have shown thatthere is a developmental asymmetry in the acquisi-tion of skill in comprehending written text such thatfacility in word identification carries much greaterweight as a determinant of reading comprehensionin children at the early stages of reading develop-ment than in children at later stages, whereaslanguage comprehension processes carry muchgreater weight as determinants of reading compre-hension in children at later stages of reading devel-opment than in children at early stages of readingdevelopment. Such research has also shown that

tests evaluating word identification skills were muchbetter predictors of performance on reading com-prehension tests than were tests evaluating lan-guage comprehension skills in beginning and lessskilled readers, whereas the opposite pattern wasevident in more skilled readers. Moreover, testsevaluating sublexical knowledge, such as phonolo-gical awareness, application of letter-sound corres-pondence rules (i.e., phonological decoding), andspelling ability, along with related phonological skillssuch as name retrieval, and verbal memory, weremuch better predictors of facility in word identifica-tion than were tests evaluating vocabulary know-ledge, general knowledge, and syntactic processing,which were found to be better predictors of facility inlanguage comprehension (Vellutino et al., 1991,1994). This pattern of results suggests that adequatefacility in word identification and other word-levelskills is a necessary (though not sufficient) conditionfor reading for meaning and is quite in keeping withthe idea that dyslexia is accurately defined, at thebehavioral level, as a basic impairment in printdecoding.

The regression studies evaluating the componentsof reading ability cited above also suggest that ade-quate facility in word identification, itself, dependsheavily on the beginning reader’s ability to acquirefacility in alphabetic coding. Additional support forthis possibility comes from a large number of studiesproviding independent and convergent evidence thatchildren who have difficulty in mapping alphabeticsymbols to sound also have difficulty learning toread and spell (e.g., Fletcher et al., 1994; Liberman &Shankweiler, 1979, 1991; Olson, Forsberg, Wise, &Rack, 1994; Share & Stanovich, 1995; Shankweileret al., 1979; Siegel & Ryan, 1984; Snowling, 1980;Stanovich & Siegel, 1994; Tunmer, 1989; Vellutino,1979, 1987, 1991; Vellutino & Scanlon, 1987a, b;Vellutino et al., l995a; Vellutino et al., 1991, 1994,1996; Wagner & Torgesen, 1987). Moreover, suchdifficulties have been found to continue well intoadulthood (Bruck, 1990, 1992, 1993; Satz, Buka,Lipsett, & Seidman, 1998; Spreen, 1989).

Yet, there is also a great deal of evidence thatacquisition of facility in alphabeticmapping depends,inpart, on theacquisitionof phonological awareness –which we defined earlier as conceptual grasp andexplicit awareness that spoken words are comprisedof individual speech sounds (phonemes) and com-binations of speech sounds (syllables, onset rimes).Compared with normally developing readers, poorreaders commonly manifest difficulty acquiringphonological awareness and phonological analysisskills during their childhood years and continue to bedeficient in phonological analysis during their adultyears. Bruck, 1992; Liberman, Shankweiler, Fischer,& Carter, 1974; Liberman & Shankweiler, 1979,1991; Snowling, 2000a; Wagner & Torgesen, 1987).More direct evidence for the possibility that defi-ciencies in phonological awareness and alphabetic

6 Frank R. Vellutino et al.

Page 6: Vellutino

mapping may be causally related to reading difficul-ties comes from naturalistic studies, controlled la-boratory studies, and intervention studies in which itwas found that training that helped children acquirethese skills had a beneficial effect on word identifica-tion, spelling, and reading ability in general (Adams,1990; Blachman, 1994, 2000; Bradley & Bryant,1983; Foorman, Francis, Novy, & Liberman, 1991;Foorman, Francis, Fletcher, Schatschneider, &Mehta, 1998; Hatcher, Hulme & Ellis, 1994; Lund-berg, Frost, & Petersen, 1988; Olson, Wise, & Ring,1999; Scanlon & Vellutino, 1996; Scanlon, Vellutino,Small, & Fanuele, 2000; Torgesen, Rose, Lindamood,Conway, & Garvan, 1999; Vellutino & Scanlon,1987a; Vellutino et al., 1996; Williams, 1980). Thus,although there is abundant evidence that difficulty inlearning to identify printed words is the manifestcause of reading difficulties in beginning readers,there is also abundant evidence that this problem,itself, is causally related to significant difficultiesacquiring phonological analysis skills and masteringthe alphabetic code, regardless of more distinctcauses (intrinsic vs. environmental and instruc-tional). A caveat to the latter generalization concernsthe role of other language-based skills (e.g., semanticand syntactic skills) in learning to read. We discussthese possibilities in greater detail below.

In our analysis of the components of readingability, consolidation of alphabetic knowledge wassaid to be dependent, not only on phonologicalawareness and the ability to map alphabetic sym-bols to sounds, but also on orthographic aware-ness, which we defined as the child’s sensitivity toconstraints on how the letters in written words areorganized. Phonological awareness and orthogra-phic awareness are reciprocally related cognitionsthat facilitate alphabetic mapping and help thechild acquire the more general orthographic know-ledge (e.g., �at� in �cat�, �fat� and �rat�), that helps tomake the writing system more manageable andultimately leads to mastery of the alphabetic code(Ehri, 1999). It follows that children who have dif-ficulty in acquiring phonological awareness andlearning to map alphabetic symbols to sound willalso have difficulty acquiring orthographic aware-ness and general orthographic knowledge. There isabundant evidence that the child who has limitedphonological awareness and limited alphabeticmapping skills also has limited orthographicawareness and limited orthographic knowledge (cf.Siegel, Share, & Geva, 1995). These limitations havebeen observed in both dyslexic children and adults(e.g., Bruck, 1990, 1992; Manis, Custodio, & Szes-zulski, 1993; Snowling, 2000a; Olson et al., 1994;Vellutino et al., 1994, 1995a).

Finally, given the importance of acquiring know-ledge of print concepts and conventions in learningto read along with the importance of acquiringpragmatic knowledge as an important component oflanguage and reading comprehension, it seems

reasonable to inquire whether deficiencies in suchknowledge have been found to be causally related tospecific reading disability. Although there is consid-erable evidence that limited knowledge of print con-cepts and conventions and limited pragmaticknowledge have often been observed in children whoalso experience early reading difficulties and cancertainly contribute to early reading and languagedifficulties (e.g., Adams, 1990; Snow, Burns, &Griffin, 1998), limitations in such knowledge areprobably not basic causes of specific reading dis-ability, in the biological sense, and are, in mostcases, caused by experiential and instructional de-ficits rather than by biologically based cognitive de-ficits. Support for this inference comes from studiesfinding that many children have extreme difficultylearning to read, despite having entered school withage-appropriate pragmatic skills and a reasonablyfirm grasp of print concepts and conventions, havingcome, in most such cases, from enriched homebackgrounds (e.g., Vellutino et al., 1996).

Underlying causes: cognitive deficit theoriesof dyslexia

Deficits in general learning abilities

The study of basic cognitive deficits as underlyingcauses of specific reading disability has a long his-tory (Fletcher, Foorman, Shaywitz, & Shaywitz,1999; Lyon et al., 2002; Snowling, 2000a; Vellutino,1979, 1987; Vellutino & Scanlon, 1982). Etiologicaltheories that have been proffered over the years arelegion, and most have been falsified both empiricallyand logically. Dyslexia has most often been attrib-uted to deficiencies in visual, linguistic, and low-level sensory functions, and we exemplify suchtheories below. However, dyslexia has also been at-tributed to deficiencies in general learning abilitiesthat are involved in all learning enterprises and notjust learning to read. For example, specific readingdisability has been variously attributed to defi-ciencies in selective attention (Douglas, 1972),associative learning (Brewer, 1967; Gascon & Good-glass, 1970), cross-modal transfer (Birch, 1962),serial-order processing (Bakker, 1972), and bothpattern analysis and rule learning (Morrison &Manis, 1982). Such theories can be questioned onlogical grounds alone. As stated elsewhere, �dys-function in one or another of these rather basic andgeneral learning abilities would seem to be ruled outas significant causes of the disorder in a child whohas at least average intelligence and who does nothave general learning difficulties, given that all ofthese cognitive abilities are entailed on virtually alltests of intelligence and are most certainly entailedin all academic learning� (Vellutino et al., 1996,p. 602). More important, however, is the fact thateach of these hypotheses has also been discreditedby empirical research. This research has been

Specific reading disability (dyslexia) 7

Page 7: Vellutino

summarized elsewhere and will not be reviewed here(Vellutino, 1979, 1987; Vellutino & Scanlon, 1982;see also Katz, Shankweiler, & Liberman, 1981; Katz,Healy, & Shankweiler, 1983). Most of the studiesreporting differences between poor and normalreaders on measures of these general learning abil-ities did not control for reader group differences inverbal coding ability and/or working memory pro-cesses that might be affected by verbal coding defi-cits. In subsequent studies that did implement suchcontrols, group differences on measures of theseabilities were generally eliminated.

To cite one example, Birch (1962) hypothesizedthat reading impaired children may be encumberedby a developmental lag in the establishment of cross-modal transfer. He intuited that this lag would im-pair their ability to represent the same information intwo sensory-based systems, as exemplified inlearning to read. Initial support for this theory wasprovided by Birch and Belmont (1964), who observedthat poor readers performed below the level of nor-mal readers in matching auditorily presentedrhythmic patterns with visual representations ofthose patterns. Because this task confounds cross-modal transfer with working memory and verbalcoding ability, Vellutino and his associates (Vellu-tino, 1979, 1987; Vellutino & Scanlon, 1982) con-ducted a series of studies that compared poor andnormal readers on both intramodal (visual–visual;auditory–auditory) and intermodal (visual–auditory)non-verbal learning tasks that minimized the influ-ence of verbal coding ability. They also comparedthese groups on visual–verbal learning tasks, andfound that the poor readers performed below thelevel of the normal readers only on the visual–verballearning tasks. These findings are at variance withBirch’s cross-modal explanation of reading disabil-ity. They are also at variance with etiological theorieswhich suggest that deficits in association learning orattention are root causes of specific reading disabil-ity, given that all of the tasks used in these studiesinvolved paired associates learning and required theutmost in attention, concentration, and workingmemory. The data are more in keeping with verbaldeficit explanations of the disorder.

Finally, Birch’s cross-modal theory of reading dis-ability was derived from the more basic assumptionthat cross-modal transfer is a developmental phe-nomenon that is not established until early childhood(Birch & Lefford, 1963). However, this assumptionhas since been obviated by infant research whichshows that perception of equivalences across mod-alities is present either at birth or shortly after birth(e.g., Gibson, 1969; Bryant, 1974; Meltzoff & Kuhl,1994). Thus, the theory is no longer viable.

Visual deficits

Visual perceptual and visual memory deficits. The-ories implicating deficiencies in the visual system

have been the most ubiquitous and most influentialtheories of dyslexia, from before the turn of the cen-tury (Morgan, 1896; Hinshelwood, 1917) up throughthe 1970s and 1980s, when linguistic deficit ex-planations of the disorder began to compete withvisual deficit explanations (Lyon et al., 2002; Snow-ling, 2000a; Vellutino, 1979, 1987). However, despitetheir popularity, the most prominent visual deficittheories in the early reading disability literature hadlittle empirical support, and confounded the visualand verbal components of reading and spelling.

The demise of these theories was initiated over twodecades ago through a series of related studies thatsystematically evaluated traditional and widelyaccepted etiological conceptualizations such asOrton’s (1925) optical reversibility theory of dyslexia(i.e., perceiving letters and words as reversed forms),Hermann’s (1959) spatial confusion theory (i.e.,inherent spatial disorientation), and a variety ofother theories that implicated deficits in visual pro-cesses such as visualization, visual sequencing, andvisual memory as basic causes of reading difficulties.The studies challenging these theories made use of awide variety of visual processing paradigms thatwere carefully designed to control for or minimize theinfluence of verbal coding (e.g., visual discrimina-tion, spatial orientation, visual memory, and visuallearning paradigms), and they generally replicatedgiven findings with independent samples of poor andnormal readers.

To be brief, in studies conducted comparing poorand normal readers across a broad age range (mostoften grades 2 through 8), few significant differencesbetween these groups were found on measures ofvisual processing ability when the influence of verbalcoding was controlled. For example, in experimentalstudies evaluating such processes (Vellutino, 1979,1987; Vellutino & Scanlon, 1982), it was found thatmemory for visually presented letters and words thatwere visually similar (such as �b�, �d�, �was�, and �saw�)was as good in poor readers as it was in normalreaders when the task required a written responserather than a naming response, which did differen-tiate these two groups. At the same time, there wereno statistically significant differences between poorand normal readers on measures evaluating visualrecognition and visual recall of letters and wordsfrom an alphabetic orthography with which the twogroups were unfamiliar, specifically, written Hebrew.Moreover, these groups did not differ on measuresevaluating orientation and left to right processing ofthe letters in Hebrew words, compared with childrenwho were learning to read and write Hebrew, whoseaccuracy was greater and whose (right to left) pro-cessing strategies were different from the childrenwho were unfamiliar with Hebrew. Research con-ducted later provided additional confirmation thatpoor and normal readers have comparable visualabilities when task requirements for verbal medi-ation were minimized (Fletcher et al., 1999). Thus, it

8 Frank R. Vellutino et al.

Page 8: Vellutino

seems reasonable to conclude that visual deficits ofthe types that had been touted in the early literatureare no more prevalent in poor readers than they arein normal readers. It may also be concluded thatcertain hypothesized and highly popularized deficitssuch as Orton’s optical reversibility and Hermann’sspatial confusion are pseudo-problems that have nopsychological reality.

The final source of evidence is derived from regres-sion studies evaluating the components of readingability (e.g., Vellutino et al., 1991, 1994) discussedearlier. It was found that visual abilities were relat-ively poor predictors of word identification, spelling,pseudoword decoding, and reading comprehension.This was true at all age and grade levels evaluated inthese studies (grades 2 through 7). These results, to-gether with the results just discussed, provide strongevidence that reading is primarily a linguistic skill (seeFrost, 1998 for a recent review).

Low-level visual deficits. Specific reading disabilityhas been attributed to visual tracking problemsassociated with oculomotor deficiencies (Getman,1985), in addition to visual masking effects associ-ated with a hypothesized deficit in the �transientvisual system� (Breitmeyer, 1989; Lovegrove, Martin,& Slaghuis, 1986; Stein, 2001). Reading disabilityhas also been associated with abnormalities in per-ception of visual motion (Eden et al., 1996). At thesame time, transient system and motion perceptiondeficits in disabled readers have both been linked tofunctional anomalies in the magnocellular visualsubsystem.

The visual tracking theory of reading disability hasbeen discredited by well-controlled eye movementstudies finding no differences between poor andnormal readers on visual tracking of non-verbalstimuli (Olson, Kleigl, & Davidson, 1983; Stanley,Smith, & Howell, 1983). However, the transient sys-tem theory has some empirical support (Breitmeyer,1989; Badcock & Lovegrove, 1981; Lehmkuhle,Garzia, Turner, Hash, & Baro, 1993; Lovegrove et al.,1986; Lovegrove, Garzia, & Nicholson, 1990; Martin& Lovegrove, 1984), as does the motion perceptiontheory (Eden et al., 1996). Thus, both warrant furthercomment.

First, note that the visual system is comprised oftwo parallel systems, the magnocellular system andthe parvocellular system, both residing in the layersof the lateral geniculate nucleus of the visual cortex.The magnocellular system – often called the tran-sient system – consists of large neurons that havehigh conduction velocity and demonstrate a highdegree of sensitivity to movement and rapid changesin the visual field. In contrast, the parvocellullarsystem consists of small neurons that are sensitiveto color and fine spatial details. In reading, the par-vocellular system is believed to be operative duringeye fixations and the magnocellular (transient) sys-tem is believed to be operative during saccadic

movements of the eyes. The magnocellular system is(presumably) responsible for suppressing the activ-ity of the parvocellular system when the eyes are inmotion, thereby inhibiting the visual trace that nor-mally persists for a short duration (approximately250 milliseconds) after a visual stimulus has beenterminated. It has been suggested that dyslexicssuffer from a deficit in the inhibitory function of thetransient system, producing a visual trace ofabnormal longevity that creates masking effectsalong with visual acuity problems when such chil-dren are reading connected text. Thus, Lovegroveand his associates have shown that poor and normalreaders process high and low spatial frequency gridsdifferently. They also have different contrast sensit-ivity functions, such that the poor readers requiregreater luminosity than the normal readers for dis-tinguishing low frequency grids (Badcock & Love-grove, 1981; Lovegrove et al., 1986, 1990; Martin &Lovegrove, 1984). That dyslexics are subject to tracepersistence has been inferred from such findings.

However, as pointed out by Hulme (1988), thetrace persistence theory of reading disability predictsthat dyslexics should be impaired only when they arereading connected text and not when they encounterprinted words one at a time under foveal visionconditions. Yet, we know that poor readers find it asdifficult to identify printed words one at a time underfoveal vision conditions as to identify them whenthey are reading connected text. In addition, a sig-nificant number of normally achieving readers werealso found to have transient system deficits in thestudies conducted by Lovegrove and his associates.Moreover, there is no evidence that dyslexicsexperience visual acuity and visual masking prob-lems under normal reading conditions. Furthermore,dyslexic children evaluated typically had phonologi-cal deficits of the types that have been shown to becausally related to reading difficulties (e.g., defici-encies in letter-sound decoding and phonologicalawareness). At the same time, Eden, Stein, Wood,and Wood (1995) found that while visual processescontributed unique variance in predicting readingskills in poor readers, the amount of variance wasquite small compared to the variance contributed bymeasures of phonological skills.

Finally, given that some normal readers have beenfound to manifest abnormalities consistent withtransient system deficits, and absent any clinicalevidence that dyslexics typically experience the typesof visual perceptual anomalies that are said to be aconsequence of transient system deficits (e.g., visualmasking caused by trace persistence), it would ap-pear that transient system deficits have not beenshown to be causally related to reading difficulties.

All things considered, we doubt that visual tracepersistence associated with transient system deficitsis a significant cause of specific readingdisability. Yet,it might be a correlate of the disorder, and perhapsevenabiologicalmarker.Thispossibility is givensome

Specific reading disability (dyslexia) 9

Page 9: Vellutino

support by anatomical and electrophysiologicalstudies demonstrating structural and functionalanomalies in the magnocellular pathways of a smallnumber of those with dyslexia studied (Lehmkuhleet al., 1993; Livingstone, Rosen, Drislane, & Gala-burda, 1991). However, a recent review of the litera-ture by Skottun and Parke (1999) presented evidencefrom several studies of saccadic suppression thatstrongly suggests that it is the magnocellular systemthat is suppressed during saccadic movements of theeyes, not the parvocellular system. Such findings arecontrary to the major premise of the transient systemdeficit theory of dyslexia, so there is considerablereason to question the viability of the theory.

As a correlate, dysfunction in the magnocellularsystems in dyslexia has also been implicated byrecent studies demonstrating that perception ofvisual motion is deficient in this population (Edenet al., 1996; Eden & Zeffiro, 1998). It has been reas-onably well established through study of nonhumanprimates and patients with vascular lesions that acomponent of the visual system located at the tem-poral–occipital–parietal junction (the MT/V5 com-plex) is sensitive to motion and this area of the brainis believed to be dominated by input from the mag-nocellular neurons. Thus, on the strength of thepossibility that dysfunction in the magnocellularsystem might be the result of a partially compens-ated developmental lesion, Eden et al. (1996) usedfunctional magnetic resonance imaging (fMRI) toevaluate visual motion processing in dyslexic andnormal adults and found that for dyslexics, presen-tation of moving stimuli failed to produce the sametask-related pattern of activation in the MT/V5complex as that observed in the normal controls.Similar results have been obtained in contrasts of9- and 10-year-old dyslexic and normal readers(Cornelissen, Hansen, Hutton, Evangeliou, & Stein,1997; Raymond & Sorensen, 1998). The combinedresults lend additional credence to the idea that themagnocellular pathways may be deficient in somedyslexic individuals.

In their interesting review of this literature, Edenand Zeffiro (1998) stop short of inferring a causalrelationship between magnocellular dysfunction andspecific reading disability and put forward anintriguing hypothesis that potentially explains bothlinguistic and visual (magnocellular) deficits ob-served in dyslexics. This hypothesis could alsoaccount for the logical disconnect between the con-sistent observation of low-level visual processingdeficits in these children and the absence of associ-ated clinical symptoms that might be causally rela-ted to their reading difficulties. Specifically, Edenand Zeffiro (1998) underscore the complexity of thesymptom pattern that has been observed in dyslex-ics and point out that low-level visual deficits thathave been observed in some of these individualshave typically been accompanied by deficits inreading-related language and language-based skills,

such as phonological awareness and verbal workingmemory, and sometimes, by low-level auditory defi-cits as well (see discussion below). And, althoughthey acknowledge that these low-level sensory defi-cits have not been shown to be causally related todifficulties in learning to read, they make specialnote of the fact that sensory deficits and the reading-related language deficits that sometimes co-occur indyslexic children and adults have not been shown tobe causally related to each other. To explain thispattern of results, Eden and Zeffiro (1998) suggestthat �these perceptual and cognitive abnormalitiesarise from dysfunction of a neural system common toboth� (p. 279).

In support of this conjecture, the investigators citeresearch conducted by Rumsey et al. (1997), com-paring adult dyslexic and normal reading controls onrhyme detection/judgment and pseudoword readingtasks, in which it was found that the dyslexicsshowed �significantly less task-related signal in-crease in temporoparietal areas bilaterally, consis-tent with a role for the angular gyrus (and nearbytemporal and parietal areas) in reading� (p. 280). Alsoconsistent with this possibility are results they citefrom a study conducted by Horwitz, Rumsey, andDonahue (1998) demonstrating that measures ofactivity in temporal and parietal areas thought to beinvolved in reading were found to be correlated whennormal readers were engaged in a pseudowordreading task and uncorrelated when dyslexic sub-jects performed the same task (see also discussionbelow). Thus, in combining experimental findingsfrom behavioral, electrophysiological, and functionalneuroimaging studies that provide documentation ofphonological processing deficits in dyslexia withexperimental findings from similar studies demon-strating magnocellular system deficits in this group,and coupling these findings with the demonstratedinvolvement of temporoparietal areas in reading,phonological processing, and magnocellular systemfunctioning, Eden and Zeffiro suggest that �the cor-tical regions surrounding the temporoparietal junc-tion emerge as possible candidates for the principalloci of cerebral dysfunction in dyslexia� (p. 281). Inother words, dyslexic persons may have structuraland functional anomalies in adjacent regions of thebrain (respectively) supporting linguistic and visualprocesses of the types found to be deficient in thisgroup. But, whereas the linguistic deficits have beendemonstrated to be causally related to reading dis-ability, the visual deficits have not been demon-strated to be causally related to reading disability,though they may serve as biological markers that aiddifferential diagnosis. This is a highly plausiblehypothesis that merits further consideration.

Language and language-based deficits

Semantic and syntactic deficits. If reading wereprimarily a linguistic skill, as many scholars as-

10 Frank R. Vellutino et al.

Page 10: Vellutino

sume, then it would seem that reading disabilitycould be caused by deficiencies in the semantic,syntactic, or phonological components of language.For example, some have theorized that vocabularydeficits may be a basic cause of difficulties inlearning to read in some impaired readers (e.g.,Dickinson & Tabors, 2001; Snow & Tabors, 1993;Vellutino, 1979, 1987; Vellutino & Scanlon, 1982). Itseems reasonable to infer, on logical grounds, that achild will have less difficulty in learning to readwords that are in his or her speaking vocabularythan in learning to read words that are not in his orher speaking vocabulary. There is some evidence tosupport this possibility. It has been demonstrated, insimulated reading tasks, that normal as well as poorreaders had more difficulty establishing connectivebonds between low meaning words and the charac-ters representing those words than between highmeaning words and their representative characters.In one such study (Vellutino, Scanlon, & Spearing,1995b), low meaning words were words the childrenhad heard before, but did not produce many se-mantic associates on a word association task wherethose words were used to stimulate as many asso-ciations as came to mind. High meaning words werenot only familiar to the children, but produced alarge number of semantic associations on the wordassociation task and were more likely than the lowmeaning words to be in the children’s speakingvocabularies. It was found that the high meaningwords were easier to learn to �read� (on the simulatedreading task) than were the low meaning words.Similar results were obtained in experimental simu-lations of beginning reading that compared poor andnormal readers� ability to learn to �read� nonwordswith their ability to learn to �read� the same nonwordsafter pairing these stimuli with novel cartoon char-acters that imbued them with meaning (Vellutino &Scanlon, 1987a). These findings were apparent, re-gardless of whether the characters were ideographicor alphabetic in nature.

Additional support for the possibility that defi-ciencies in semantic knowledge may be causally re-lated to difficulties in learning to read comes fromstudies finding that vocabulary knowledge in pre-first grade children is a good predictor of early andlater reading achievement (Dickinson & Tabors,2001; Scarborough, 1990, 1991; Snow, Barns,Chandler, Goodman, & Hemphill, 1991; Snowlinget al., 2003). Moreover, deficient vocabulary know-ledge has been found to be a significant cause ofreading difficulties in second language learnershaving limited proficiency in spoken English (Tabors& Snow, 2001). Thus, it follows that a child who hasa limited vocabulary could have difficulties in ac-quiring fluency in printed word identification, even ifhe or she has adequate phonological decoding skills.

Vocabulary knowledge has also been implicated inthe acquisition of reading-related phonologicalskills. To be brief, Metsala and Walley (Metsala,

1999; Metsala & Walley, 1998; Walley, 1993; seealso Goswami, 2001) have proposed that phonolo-gical representations become increasingly less globaland more highly specified with developmentalchanges in vocabulary knowledge, especially in thecase of words having similar phonological properties(e.g., phonological �neighbors� such as �sail�, �mail�,�fail�, �jail�, �bail� etc.). This is because encoding andretrieval of such words for functional use requiresmore in the way of (implicit) segmental analysis thanwords having fewer phonological neighbors. Theseinvestigators have provided some evidence to sup-port this hypothesis. Thus, if it proves to be valid,then it would be expected that deficiencies in vo-cabulary knowledge would be accompanied bypoorly specified phonological representations,which, by some accounts, would impair readingdevelopment (see discussion below).

Finally, given the demonstrated utility of linguis-tic context in facilitating and monitoring wordidentification, especially in poor readers (Perfetti &Roth, 1981; Stanovich, 1980; Tunmer, 1989;Tunmer & Chapman, 1998), it would seem thatsyntactic deficits that impede a child’s ability to uselinguistic context to aid word identification couldcontribute to beginning reading problems in suchchildren. Yet, vocabulary and syntactic knowledgedo not often distinguish between children from thepopulation defined herein and normally achievingreaders, except in contrasts involving older childrencharacterized by long-standing reading disorder(e.g., Fletcher, Satz, & Scholes, 1981; Shankweileret al., 1999; Snowling, 2000a; Stanovich, 1986;Vellutino, Scanlon, & Tanzman, 1988; Vellutinoet al., 1995b, 1996). Such findings suggest thatearly reading difficulties in children from thispopulation may not be caused primarily by vocabu-lary and syntactic deficits and may, more often, bea consequence of prolonged reading problems.These deficits may also be co-morbid, reflecting theco-occurrence of oral language and reading diffi-culties (e.g., Catts et al., 2003). However, existingdata do not preclude the possibility that vocabularyand syntactic deficits could contribute to difficultiesin learning to read in some children. They wouldinevitably be a significant cause of reading com-prehension problems, even in children who haveadequate facility in word identification (Snowling,2000b).

It is also worth noting that most of the studies thathave appeared in the reading disability literaturepurport to use sampling criteria that exclude disad-vantaged children and children who have limitedproficiency with spoken English from their researchsamples. This, of course, may account for theapparent discrepancy between studies evaluatingthe relationship between vocabulary and/or syn-tactic knowledge and early reading difficulties inchildren impaired by specific reading disability ascompared with studies evaluating disadvantaged

Specific reading disability (dyslexia) 11

Page 11: Vellutino

children or children with limited English proficiency.Thus, we suggest that limited vocabulary knowledgeand/or syntactic deficits might be significant sour-ces of difficulties in learning to read in at least somebeginning readers, but likely have little to do with theword recognition and phonological decoding prob-lems that are the primary markers of such difficul-ties in most impaired readers.

Phonological coding deficits. We earlier definedphonological coding as the ability to use speechcodes to represent information in the form of wordsand parts of words. In contrast to the weak supportfor semantic and syntactic deficit explanations ofspecific reading disability in the population of chil-dren typically studied, there is now strong and highlyconvergent evidence in support of weak phonologicalcoding as an underlying cause of the disorder. As weindicated earlier, the most compelling evidence for acausal relationship between phonological skillsdeficiencies, as manifest causes of inadequateachievement in beginning readers, is provided bytraining and intervention studies which have docu-mented that direct instruction designed to facilitatephonological awareness and letter-sound mappinghas a positive effect on word identification, spelling,and reading ability in general. In addition, poorreaders have been consistently found to performbelow the level of normal readers on phonologicalawareness and letter-sound decoding tasks as wealso indicated earlier (Blachman, 2000; Fletcheret al., 1994; Shankweiler et al., 1979; Share & Sta-novich, 1995; Stanovich & Siegel, 1994; Snowling,2000a; Torgesen, Wagner, & Rashotte, 1994; Wag-ner, Torgesen, & Rashotte, 1994; Vellutino, 1979,1987; Vellutino & Scanlon, 1987a, b; Vellutino et al.,1994, 1995a, b, 1996; Wagner & Torgesen, 1987).Such findings have led to a growing consensus thatthe most influential cause of difficulties in learningto read is the failure to acquire phonological aware-ness and skill in alphabetic coding. Difficulties inacquiring phonological awareness and skill in al-phabetic coding are believed to be due, in manycases, to weak phonological coding characterizedby poorly specified phonological representations(Griffiths & Snowling, 2002).

Weak phonological coding is presumed to underlieother problems that may contribute to difficulties inlearning to read. Some scholars suggest, for example,that weak phonological coding can lead to difficultiesin storing and/or retrieving printed words as unitizedand distinctive orthographic representations as wellas to difficulties in processing information in workingmemory (Brady, Shankweiler, & Mann, 1983; Elbro,1997; Gathercole & Baddeley, 1990; Katz, 1986;Shankweiler et al., 1979; Share & Stanovich, 1995;Stanovich & Siegel, 1994; Snowling, 2000a; Torgesenet al., 1994; Torgesen, Wagner, Rashotte, Burgess, &Hecht, 1997; Vellutino, 1979, 1987; Vellutino et al.,1994, 1995a, b, 1996; Wagner et al., 1994; Wagner &

Torgesen, 1987). More specifically, it has been sug-gested that difficulties in name storage and retrievalcould impair the beginning reader’s ability to estab-lish connective bonds between the spoken and gra-phic counterparts of printed words, which, in turn,could impair the reader’s ability to store quality rep-resentations of word spellings, and, thereby, impedehis or her ability to acquire fluency in word identi-fication. Fluency in word identification is, of course, acritically important prerequisite for adequate readingcomprehension (Perfetti, 1985). And, given the dem-onstrated importance of workingmemory in languageprocessing (Baddeley, 1986; Daneman & Carpenter,1980; Ericsson & Kintsch, 1995), working memorydeficits associated with weak phonological codingcould impair reading comprehension, independent ofdysfluency in word identification.

Thus, in accord with these possibilities, poorreaders have been consistently found to performbelow the level of normally achieving readers, notonly on tests evaluating word identification, phono-logical awareness, and letter-sound decoding, butalso on tests evaluating confrontational naming,rapid naming, verbal learning, and verbal memory(Blachman, 1997; Bowers & Wolf, 1993; Katz,1986; Snowling, 2000a; Torgesen et al., 1994;Vellutino, 1979, 1987; Vellutino & Scanlon, 1987a,b; Vellutino et al., 1994, 1995a, b, 1996; Wagner etal., 1994; Wolf, Bowers, & Biddle, 2000a). Alongwith phonological awareness and phonological de-coding deficits, this collection of deficits has beencommonly attributed to weak phonological coding.Thus, weak phonological coding is hypothesized tobe the central cause of specific reading disability inmany if not most impaired readers, as articulated,for example, in what has been called the �phonolo-gical core variable differences� model of specificreading disability (Stanovich, 1988; Stanovich &Siegel, 1994).

Finally, a number of researchers have attemptedto provide more direct evidence for the inference thatweak phonological coding is the central cause ofspecific reading disability and that dyslexic childrenare impaired by poorly specified phonological repre-sentations. The studies conducted have compareddyslexic and normal readers on both speech per-ception and production tasks, using both word andnon-word stimuli varying in degree of phoneticsimilarity, in addition to reading age and chrono-logical age-matched designs to evaluate reader groupdifferences in speech discrimination and encoding.Studies using speech perception tasks typicallyevaluated categorical perception of stop consonantsin these two groups and obtained suggestive, thoughsomewhat inconsistent, evidence that dyslexicreaders perceive phonetic boundaries less sharplythan do normal readers (Adlard & Hazan, 1997;Brandt & Rosen, 1980; Godfrey, Syral-Lasky, Millay,& Knox, 1981; Hurford & Sanders, 1990; Maniset al., 1997; McBride-Chang, 1995; Mody, Studdert-

12 Frank R. Vellutino et al.

Page 12: Vellutino

Kennedy, & Brady, 1997).2 However, a small numberof studies that have used the �gating� paradigm toassess spoken word identification rather than morebasic speech perception processes have providedless consistent evidence of dyslexic difficulties insuch perceptual processing (Griffiths & Snowling,2001; Metsala, 1997).

Studies using speech production tasks (whichentail both speech perception and speech productionmechanisms) have typically evaluated verbal repe-tition of both high and low frequency words, inaddition to nonsense words presented under bothnoise-masked and noise-free stimulus conditions.The researchers have generally found reader groupdifferences on these tasks, especially under noise-masked conditions, and more reliably when thestimuli were nonsense words (Brady et al., 1983;Elbro, 1997; Elbro, Borstrom, & Petersen, 1998;Snowling, Goulandris, Bowlby, & Howell, 1986).

Noteworthy is the observation that reader groupdifferences in these studies tended to be more reli-able in contrasts involving more severely impairedreaders, suggesting that basic deficits in speechperception and production may underlie phonolo-gical skills deficiencies in only some poor readers(Heath, Hogben, & Clark, 1999). This pattern of re-sults suggests that deficiencies in phonological skillssuch as phonological awareness and letter-sounddecoding may, in some children, be caused by fac-tors other than basic deficits in speech perceptionand/or production, such as inadequate instructionand experience. If so, then it would be important todevelop the means for distinguishing between thesetwo groups of impaired readers (see discussion be-low). Nevertheless, the evidence garnered from thesemore direct tests of the weak phonological codingtheory of reading disability, although inconclusive, ishighly suggestive. Additional research of this type iscertainly warranted.

The double deficit hypothesis. Not all scholars ac-cept the view that specific reading disability andreading-related cognitive deficits such as thosementioned in the previous section are caused

exclusively or primarily by limitations in phonolo-gical coding. And, not all scholars accept the viewthat the name retrieval deficits that have often beenobserved in poor readers are due to weak phonolo-gical coding and (by extension) phonological memoryproblems. In particular, Wolf, Bowers, and theircolleagues (Bowers & Wolf, 1993; Wolf & Bowers,1999; Wolf et al., 2000a) postulate the existence ofthree subtypes of reading disability: one caused bydeficiencies in phonological skills such as phonolo-gical awareness and letter-sound decoding, in ac-cord with the phonological deficit model; a secondcaused by slow naming speed that specifically dis-rupts orthographic processing and reading fluency;and a third caused by a combination of both types ofdeficit (the �double deficit�). The latter is consideredto be the most serious form of reading disability,owing to the combined effects of phonological andrapid naming skills deficiencies.

According to this view, naming speed deficits arecaused by disruption of a �precise timing mechanism�that normally influences temporal integration of thephonological and visual counterparts of printedwords, thereby impairing the child’s ability to detectand represent orthographic patterns. It is assertedthat if a word’s letters cannot be identified withsufficient ease and rapidity, they will not be pro-cessed close enough in time to detect orthographicredundancies and regularities. This, in turn willimpair the child’s ability to store distinct and unit-ized representations of word specific spellings. Thus,Bowers and Wolf (1993) hypothesize �that slow letter(or digit) naming may signal disruption of the auto-matic processes which support induction of ortho-graphic patterns, which, in turn, result in quickword recognition� (p. 70). Disruption in this timingmechanism has also been assumed to be manifestedin slow object and color naming, both of which havealso been observed in many poor readers.

Four types of evidence have been offered to sup-port the double deficit theory of reading disability(see Wolf et al., 2000a for a recent review). First,naming speed tasks, especially letter and digitnaming tasks, have consistently been found to ac-count for unique variance in reading performancebeyond that explained by phonological skills such asphonological awareness (Manis, Doi, & Bhadha,2000; Wolf et al., 2000a). Second, a number ofstudies have shown that children grouped into eithersingle deficit or double deficit subgroups, on thebasis of their performance on speed of naming andphonological awareness tasks (respectively), tend toperform below children manifesting neither deficit onindependent measures of reading achievement.Similarly, the double deficit groups typically performbelow the single deficit groups on such measures(Wolf et al., 2000a). Third, phonological awarenessand rapid naming appear to be differentially relatedto reading subskills, insofar as the former has beenshown to be more strongly correlated with accuracy

2 Note that categorical perception in phonological processing

refers to a well established phenomenon in speech perception

whereby listeners perceive a continuum defining a range of

changing acoustic signals produced by the vocal apparatus as

a single phoneme and sharply distinguish that phoneme from

a second phoneme occupying an adjacent position on the same

continuum and differs from the first by a single phonetic fea-

ture. For example, the initial consonants in the syllables /ba/

and /da/ differ by only a small change in the Voice Onset Time

(VOT) – that is, the amount of time between closure of the lips

and vibration of the vocal cords. Thus, using synthetic speech

to produce continuous changes in VOT, researchers have ob-

served that listeners draw sharp categorical boundaries be-

tween these two syllables, despite the fact that the changes in

the acoustic signals, which define these boundaries, are

continuous.

Specific reading disability (dyslexia) 13

Page 13: Vellutino

in word identification and letter-sound decodingwhereas the latter has been shown to be morestrongly correlated with speed of word identificationand speed of letter-sound decoding (Manis et al.,2000; Wolf et al., 2000a). Finally, Morris et al. (1998)used cluster analysis techniques with reading dis-abled populations and provided documentation fornaturally occurring subgroups of disabled readers(as opposed to artificially constituted subgroups) ofthe types specified by double deficit theorists.Moreover, in accord with results obtained in otherstudies, the double deficit group was found to bemore severely impaired in reading than the singledeficit groups.

Although such findings are suggestive, the doubledeficit hypothesis can be challenged on theoretical,interpretive, and methodological grounds. As re-gards its theoretical underpinnings, we suggest thatthe �precise timing mechanism� that presumablyunderlies the formation of orthographic codes lacksthe type of specification that would lend it psycho-logical reality and allow it to be evaluated experi-mentally as a valid hypothetical construct. Indeed,descriptions of its properties and its role in wordrecognition tend to be complex and abstruse (e.g.,Wolf et al., 2000a) and do not readily generatetestable hypotheses that would facilitate suchexperimentation. More important is the fact thatcharacterizations of the central deficit that is said toresult from disruption of this timing mechanism –that is, inadequate temporal integration of letteridentities – are reminiscent of earlier conceptual-izations of the word recognition process which werebased on the assumption that skilled word recog-nition is the end result of serial processing of aword’s component letters (Gough, 1972). If thedouble deficit theory of reading disability is alsobased on this assumption (as it appears to be), thenit is based on a conceptualization of word recogni-tion that has long since been discredited (seeGough, 1984 for a review of this work). There is, infact, good evidence that serial processing of aword’s component letters is an immature mode ofprocessing that is gradually abandoned as the childacquires facility in word recognition (LaBerge &Samuels, 1974). It may be inferred from this evid-ence that rapid serial processing and temporalintegration of letter identities may not be theprimary means by which orthographic codes areformed.

Complicating the picture even further is the factthat more recent formulations of the double deficittheory have viewed the fluency component of readingas a non-phonological deficit that is related primarilyto the fluency of reading connected text. Thisobservation has led to interventions focused onspeed of processing of whole words and text char-acteristics in attempts to address this putativecomponent of reading disability (Wolf, Miller, &Donnelly, 2000b). Although this formulation is con-

sistent with the finding of Morris et al. (1998), thatthe subtype impaired only in rate of processing wasnot impaired in word recognition skills, it appears tobe a significant departure from initial formulations ofthe double deficit theory, which suggest that a timingdeficit resulting in poor integration of orthographicpatterns is one of the two major processing deficitscontributing to word recognition problems in poorreaders (see above discussion).

Another question that has been raised about thedouble deficit theory of reading disability is con-cerned with the consistent finding that measures ofrapid naming skill tend to account for unique vari-ance in reading performance when phonologicalskills such as phonological awareness have beencontrolled (phonological awareness is the phonolo-gical skill almost always evaluated in the doubledeficit literature). Specifically, Torgesen et al. (1997)pointed out that such studies have not typicallycontrolled for autocorrelation effects created by thevariance that rapid naming and phonological skillsshare with reading performance (e.g., Bowers, 1995).Thus, in a longitudinal study addressing this ques-tion, these investigators found that second grademeasures of phonological awareness accounted forunique variance on fourth grade timed and untimedreading and orthographic coding tasks, when per-formance on second grade reading measures wascontrolled. However, measures of speeded letter anddigit naming did not account for unique variance onthe fourth grade reading and orthographic codingtasks, when second grade reading performance wascontrolled. Essentially the same pattern of resultswas obtained with third grade predictors and fifthgrade reading outcome measures, thereby callinginto question conclusions drawn from previousdouble deficit studies that did not control for auto-correlation.

We should also point out that the consistentfinding, that phonological and rapid naming skillseach account for unique variance on measures ofreading performance, need not be interpreted assupport for the double deficit hypothesis and doesnot necessarily undermine phonological accessexplanations of slow naming speed. Given the like-lihood that acquisition of these skills may be influ-enced, not only by common underlying cognitiveabilities such as phonological coding and phonolo-gical access, but also by appreciably different cog-nitive abilities, different educational histories, and/or motivational factors, it is not surprising to findthat they contribute unique as well as shared vari-ance to reading performance. For example, whereasspeeded naming tasks have a strong speech-motorcomponent that may significantly influence per-formance on these tasks, phonological awarenesstasks do not have a strong speech-motor componentand are less apt to be influenced by individual dif-ferences in speech-motor articulation. Conversely,unlike speeded naming tasks, phonological aware-

14 Frank R. Vellutino et al.

Page 14: Vellutino

ness tasks have strong metacognitive and metalin-guistic components that no doubt influence per-formance on such tasks. Moreover, both types oftasks may be differentially affected by instructionalfactors (e.g., instruction that focuses on letterrecognition but not phonological awareness or viceversa) and by personality and motivational charac-teristics of the child (e.g., willingness to offer a re-sponse in the face of uncertainty and/or inclinationto comply with the request to name things as quicklyas possible). Such differences could also account, inpart, for the consistent finding that phonologicalawareness and rapid naming tasks tend to be dif-ferentially related to different types of reading per-formance, such as accuracy versus speed of wordidentification and letter-sound decoding. Yet, suchvariability does not rule out the possibility thatindividual differences in phonological coding andphonological access may be the primary factorinfluencing performance on both types of tasks.

Perhaps the most serious challenge to the doubledeficit theory of reading disability inheres in certainmethodological problems associated with constitut-ing hypothesized subgroups of impaired readers onthe basis of scores on phonological awareness andrapid naming tasks. To be brief, Schatschneider,Carlson, Francis, Foorman, and Fletcher (2002) havedemonstrated that because these two types ofmeasures tend to be significantly correlated, andbecause the relationship between phonologicalawareness and reading performance tends to becurvilinear, it is likely to be the case that doubledeficit reader groups will have phonological aware-ness scores that are substantially lower than those ofsingle deficit reader groups found to be deficient onlyin phonological awareness. Compton, DeFries, andOlson (2001) have obtained similar results. More-over, Compton et al. (2001) also found that when thedouble and single deficit groups were matched onthe phonological awareness and rapid naming tasksused to define these (respective) groups, many of thedifferences on the reading measures disappeared.Furthermore, they were unable to find matches forthe most severely impaired readers. It is also worthnoting that, in many of these studies, the namingspeed deficit groups were found to have at leastaverage-level word recognition skills, unlike thephonological deficit groups (e.g., see Manis et al.,2000). Such findings suggest that the larger differ-ences typically observed between single and doubledeficit groups on reading tasks are due primarily todeficiencies in phonological awareness and relatedphonological skills, rather than to the combined ef-fects of phonological and rapid naming deficits,thereby compromising a basic assumption of thedouble deficit theory of reading disability. They alsofavor phonological deficit explanations of specificreading disability, and further question the roleplayed by rapid naming skills in reading develop-ment, as propounded by double deficit theorists. We

should also point out that relationships betweenrapid naming tests and reading tests tend to bestronger for letter/digit tasks than for color/objecttasks. Such relationships are also stronger for flu-ency than for accuracy tasks. Thus, rapid letter/digitnaming may be a simple, early measure of readingspeed as opposed to a component process(Schatschneider et al., 2002).

Finally, Kail, Hall, and Caskey (1999) evaluatedthe unique contributions made by general process-ing speed (timed matching and target search tasks),naming speed (letter and digit tasks), and printexposure (author recognition) to reading perform-ance (word recognition and reading comprehensiontasks) in randomly selected groups of childrenacross a broad age span (ages 7 to 13 years). Usingstep-wise regression techniques, these investigatorsfound that, whereas print exposure predicted per-formance on the reading measures, it did not predictperformance on the rapid naming measures, aswould be anticipated by double deficit conceptuali-zations of reading disability. In contrast, generalprocessing speed did predict speed of naming, butnot reading performance when age was controlled. Atthe same time, none of these variables predictedperformance on the reading comprehension measurewhen word recognition was controlled. The investig-ators concluded from these findings that the rapidnaming–reading link may stem from individual dif-ferences in �global processing speed�, rather thanfrom individual differences in reading-specific skillssuch as speed of letter naming, implying that speedof naming deficits may not be causally related tophonologically based reading disability.

However, because Kail et al. (1999) did not employtimed reading tasks, relationships with fluency arepossible. Moreover, rapid naming deficits have beenobserved in populations of children with learningimpairments that are not associated with readingdisabilities, including attention deficit hyperactivitydisorder (Tannock, Martinussen, & Friitjers, 2000).Such findings have implications for phonologicaldeficit as well as for double deficit conceptualizationsof the relationship between naming speed andreading development and certainly question the no-tion that rapid naming deficits are specifically re-lated to reading disability (Waber, Wolff, Forbes, &Weiler, 2000).

Low-level auditory deficits

We should also mention another theory of readingdisability that has attracted widespread attention inrecent years – one that implicates deficiencies in low-level auditory processing as the basic cause of thephonological deficits typically observed in dyslexics.The reference here is to Tallal’s (1980) temporal or-der perception theory of dyslexia (see Farmer &Klein, 1995 and Tallal, Miller, Jenkins, & Merzenich,1997, for recent reviews). This causal hypothesis

Specific reading disability (dyslexia) 15

Page 15: Vellutino

was based on earlier research conducted by Tallaland her associates, which demonstrated that chil-dren suffering from specific language impairmentshad difficulty making temporal order judgments(TOJ) with high and low tones presented at long(400 ms) versus short (50 ms) interstimulus inter-vals (ISIs; Tallal & Percy, 1973, 1975). In extendingthis research, Tallal (1980) found that dyslexicsperformed below normal readers on essentially thesame TOJ tasks at short ISIs, but as well as thenormal readers at long ISIs. Despite the fact thatonly 9 of the 20 impaired readers in the sample haddifficulty with the TOJ task (a task with which manyof the normal controls also had difficulty) and largelybecause of a high correlation between performanceon the TOJ task and performance on a nonsenseword decoding task (rho ¼ .81), Tallal inferred thatdyslexic readers suffer from a basic, non-linguisticdeficit in temporal resolution of rapidly changingauditory stimuli that impairs speech perception. Shealso inferred that this disorder underlies the pho-nological deficits typically observed. However, thisinference was speculative because Tallal did not usespeech stimuli to assess temporal order judgment atvarying inter-stimulus intervals. In addition, manychildren had raw scores of 0 on the reading task,which would inflate nonparametric correlations. Yet,in a later study, Reed (1989) attempted to replicateTallal’s findings with dyslexic and normal readingchildren, using both stop consonants and steady-state vowels along with brief tones, and found thosewith dyslexia had more difficulty than the normals inmaking temporal order judgments with the stopconsonants and tones, but performed as well as thenormal readers with the steady state vowels. Thosewith dyslexia also performed below the normalreaders on a phoneme discrimination task.

Although these findings would appear to offersupport for Tallal’s interpretation of reader groupdifferences on TOJ tasks, they do not confirm thatthe poor readers� difficulties on both the verbal andnon-verbal TOJ tasks arise from the same underly-ing perceptual mechanism. Thus, in a series ofexperiments that more carefully controlled the typesof verbal and non-verbal stimuli presented to par-ticipants, Mody et al. (1997) provided documentationthat the difficulties of poor readers on TOJ tasksusing speech stimuli were due to speech discrimina-tion deficits rather than temporal order judgmentdeficits (see also Studdert-Kennedy & Mody, 1995).In the first experiment in this study (Experiment 1a),the investigators verified that poor readers had moredifficulty than normal readers with TOJ tasks in-volving phonetically and acoustically similar speechstimuli (e.g., /ba/–/da/) at short interstimulusintervals (ISI). However, when each of these stimuliwas paired with a consonant–vowel stimulus thatwas acoustically and phonetically very different (e.g.,/ba/–/sa/, Experiment 1b), the groups performed atcomparable levels on the TOJ tasks. In a third

experiment (Experiment 2), the poor and normalreader groups were given TOJ tasks using non-speech stimuli that were acoustically matched to theonset transitions of the speech stimuli used in theseexperiments. No statistically significant group dif-ferences emerged at any of the ISIs used in theexperiment. In fact, the normal readers were slightlyworse on these tasks than the poor readers. Theinvestigators concluded from these results that ageneral auditory deficit of the type posited by Tallal isan unlikely source of the phonological deficits typ-ically observed in poor readers.

In virtually all of the studies providing support forTallal’s temporal order perception theory of dylexia,questions can be raised about the adequacy of thecriteria used to identify children as dyslexic. As ourreview of research on the core deficits showed, it isessential to define children as dyslexic on the basisof difficulties identifying words in isolation. In addi-tion, it has been shown that 30%–70% of childrenwith dyslexia also have attention deficit–hyperactiv-ity disorder (ADHD), depending on the setting andhow ADHD is defined (Fletcher, Shaywitz, & Shay-witz, 1999). Two recent studies provided evaluationsof the auditory temporal processing hypothesis inwell-defined samples of children with dyslexia thatalso controlled for the presence of ADHD.Waber et al.(2001) selected children with dyslexia and no ADHDfrom a larger group of children originally referred forclinical evaluations of learning problems. Auditoryprocessing tests involving speech and nonspeechstimuli were used that varied in the interstimulusinterval, thus permitting evaluation of Tallal’s (1980)hypothesis that rate of processing was impaired inchildren with dyslexia. The results revealed signific-ant differences between good and poor readers in thediscrimination of speech and nonspeech stimuli, butno effect of interstimulus interval. Thus, group dif-ferences in perceptual ability were apparent that didnot appear to be related to rapid temporal process-ing.

In the second study, Breier, Fletcher, Foorman,and Gray (2002) administered temporal order judg-ment and discrimination tasks that also varied indemands for processing of acoustic stimuli acrossinterstimulus intervals. The authors created 4groups of children with sample sizes of about 40 pergroup: children with reading disability and noADHD, children with reading disability and ADHD,children with ADHD and no reading disability, andtypically achieving children with no ADHD. The re-sults were not consistent with the auditory temporalprocessing hypothesis. As in Waber et al. (2001),children with dyslexia did not show a specific sen-sitivity to variations in interstimulus intervals, al-though they tended to perform more poorly thanthose children without reading difficulties. In addi-tion, phonological processing measures were onlycorrelated with the processing of speech stimuli.However, in contrast to Waber et al. (2001), the

16 Frank R. Vellutino et al.

Page 16: Vellutino

difficulties were more apparent on speech thannonspeech stimuli. It may be that Waber et al. (2001)found differences on nonspeech stimuli because theyincluded children with both reading and oral lan-guage difficulties in their sample. Breier et al. (2002)specifically excluded children with oral languagedifficulties. Moreover, Heath et al. (1999) also foundnonspeech auditory processing deficits in only thosepoor readers with concomitant oral language diffi-culties and not those poor readers without oral lan-guage difficulties. Both Waber et al. (2001) andBreier et al. (2002) concluded that children withdyslexia have difficulties with speech perception thatproduce deficits on temporal processing tests, but donot have a pervasive deficit in auditory temporalprocessing. The speech perception difficulties arerelated to and possibly contribute to phonologicalprocessing difficulties that in turn, affect readingskills.

Results from several other studies testing variousaspects of Tallal’s theory have led to similar conclu-sions (Best & Avery, 1999; Bishop, Carlyon, Deeks,& Bishop, 1999; Bradlow et al., 1999; McNally,Hansen, Cornelissen, & Stein, 1997; Nittrouer,1999), thus questioning the viability of the theory.The most significant problem is that sensory deficitsat the auditory level do not explain the word recog-nition difficulties in a parsimonious manner. Thelink with phonological processing is tenuous, thoughthere is evidence that speech perception difficultiesare related to phonological processing capabilities, afinding in many studies of children with dyslexia (seeMody et al., 1997). Dyslexia implies more thanreading difficulties and children with dyslexia differfrom typically achieving children on a variety ofdimensions, many of which relate to co-morbiditiesassociated with dyslexia. However, these differencesdo not explain the reading problem and they tend tobe small relative to those associated with phonolo-gical processing. Yet, they could be related to theunderlying neurobiological problems that causedyslexia in some children, as suggested by Eden andZeffiro (1998) and other scholars (Fletcher et al.,1999).

Dyslexia in different languages

Our discussion so far has focused on the cognitivecharacteristics of dyslexia and its manifestationsfrom the perspective of children learning to read inEnglish. However, it needs to be borne in mind thatEnglish has an opaque (or deep) orthography inwhich the relationships between letters and soundsare inconsistent and many exceptions are permitted.As such, English presents a significantly greaterchallenge to the beginning reader than other moreregular alphabetic systems that contain consistentmappings between letters and sounds and are des-cribed as transparent (or shallow) orthographies.Indeed, several studies now show that the develop-

mental progress of children learning to read intransparent orthographies such as German or Ital-ian is generally faster than that of children learningwritten English (Harris & Hatano, 1999). Such chil-dren also show correspondingly faster developmentof phoneme awareness (Cossu, 1999). It follows fromthese findings that the utility of phonological abilitiesas predictors of reading development varies acrossdifferent languages. For instance, while rhymingskill predicts learning to read in English (Bradley &Bryant, 1983), it is a poor predictor of subsequentreading achievement in German (Wimmer, Landerl,& Schneider, 1994) and Dutch (de Jong & van derLeij, 1999), where rapid naming ability accounts forlarger proportions of unique variance in readingability.

Although there has been a burgeoning of researchon dyslexia in different languages in recent years(Goulandris, 2003), cross-linguistic studies that di-rectly compare dyslexia in English (about which weknow a great deal) and dyslexia in a different lan-guage are still comparatively rare. Nonetheless, theprevailing view is that the core phonological deficitsof dyslexia are harder to detect in children who havelearned to read in transparent orthographies such asGerman or Italian. In such languages, impairmentscan be identified most clearly on tasks that requireimplicit phonological processing, such as thoseevaluating verbal short-term memory, rapid namingand visual–verbal paired associate learning (Wim-mer, Mayringer, & Landerl, 1998), rather than ontests evaluating explicit phonological processingsuch as phonological awareness and phonological(letter-sound) decoding. The reading problems as-sociated with dyslexia also differ in regular ortho-graphies as compared to less regular orthographiessuch as written English. For example, Wimmer andhis colleagues have conducted studies showing thatGerman-speaking dyslexic children can read longunfamiliar words and also nonwords as well as theirnormal reading peers (Frith, Wimmer, & Landerl,1998). However, the fluency of their reading is af-fected; they read single words more slowly thancontrols and sometimes, reading comprehensiondifficulties follow as the consequence of a �bottleneck�in the reading process (Wimmer et al., 1998). Thisdisparity notwithstanding, a common finding is thatspelling presents a major stumbling block for chil-dren with dyslexia across alphabetic writing sys-tems. Dyslexic spelling deficits are very welldocumented in English (Treiman, 1997) and workwith French-speaking children, for example, showsthat children with dyslexia lag behind in theirspelling development as well as in their use ofphonological spelling strategies (Caravolas, 2003).

A more stark comparison is afforded betweenlearning to read in alphabetic languages and learningto read in non-alphabetic languages such as Chineseor Japanese. The Chinese writing system differsfrom an alphabetic system in that it contains a large

Specific reading disability (dyslexia) 17

Page 17: Vellutino

numberof visual symbolsor characters that representunits of meaning (morphemes) rather than phonemesas in an alphabet. The task of learning to read istherefore a considerable feat for Chinese childrenwhomust learn literally hundreds of visually complexcharacters that containphonetic and radical compon-ents. It is not surprising, therefore, that visual skillspredict readability inChinesebetter than theypredictreading ability in alphabetic writing systems. Moresurprising, at first glance, is that phonological skillsarealsopredictors of individualdifferences inChinesechildren’s reading skills (Ho & Bryant, 1997), eventhough readers of Chinese do not develop phonemeawareness to the level observed among readers ofalphabetic languages (Read, Zhang, Nie, & Ding,1986).

The most comprehensive study of dyslexia inChinese to date was reported by Ho, Chan, Tsang,and Lee (2002), who classified data from 30 childrenwith dyslexia on tests evaluating visual processing,phonological processing, and rapid naming, as wellas on tests evaluating their knowledge of the ortho-graphic regularities of characters and radicals. Re-sults from this study indicate that a rapid namingdeficit was the most dominant type of cognitive def-icit in Chinese dyslexic children, affecting some 60%of the cases. However, over half of the dyslexic chil-dren exhibited deficits in three or more cognitivedomains and there was a significant associationbetween the number of cognitive deficits and thedegree of reading and spelling impairment. Thisstudy was small in scale and needs to be replicatedon a larger sample. However, its findings providetentative support for the idea that dyslexia in chil-dren learning to read written Chinese is associatedwith multiple deficits, rather than with a core phono-logical deficit as has been proposed for dyslexia inchildren learning to read written English.

The cross-linguistic study of dyslexia is at a rel-atively early stage. Moreover, the extant data aredifficult to interpret. A fundamental difficulty is inequating diagnostic criteria across languages. InEnglish, the diagnosis of dyslexia proceeds on thebasis of a discrepancy between reading accuracy andage (or in some cases IQ). The same criteria cannotbe used in transparent writing systems whereaccuracy levels are typically high (Paulesu et al.,2001). Rather, criteria tend to center on speed andfluency rather than error-rate. Thus, it is entirelypossible that different individuals will fulfill criteriafor dyslexia in one language but not in another. In-deed, Wydell and Butterworth (1999) reported a caseof a bilingual boy who was able to read normally inJapanese, but manifested many of the reading be-haviors characteristic of dyslexicia when attemptingto read English.

Taken together, the evidence underlines the factthat although dyslexia, quite likely, has a neurobio-logical basis, it is not just a product of nature but,rather, a complex condition that depends on the

dynamic interaction between certain innate suscep-tibilities as well as the home and school environ-ments on one hand, and the cultures in whichchildren learn to read on the other. It is not an all ornone condition. Just as some individuals reach thethreshold for a diagnosis of dyslexia and others witha similar cognitive profile do not succumb to readingfailure, certain transparent orthographies, like Ger-man and Italian, benefit learning to the point wheredyslexia may be �hidden� in the majority of cases,while other, less transparent orthographies, likeEnglish, may aggravate the problem.

Underlying causes: biological foundationsof dyslexia

Neurobiological factors

The World Federation of Neurology definition ofdyslexia provides a major role for neurobiologicalfactors by indicating that dyslexia is �dependentupon fundamental cognitive disabilities, which arefrequently of constitutional origin� (Critchley, 1970,p. 11). However, this statement for many years waslittle more than an assumption based on severalsources of indirect information. One source was thelinguistic and behavioral characteristics of adultswith documented brain injury, where fractionationsof reading from other skills in otherwise normalindividuals could be observed. The second involvedassociations of nonspecific indices of neurologicaldysfunction with dyslexia, including perceptual-motor problems, �soft� neurological signs, and motorclumsiness, along with a potpourri of findings onelectrophysiological measures (Dykman, Ackerman,Clements, & Peters, 1971; Taylor & Fletcher, 1983).Finally, it was commonly assumed that if dyslexiawas not due to intellectual, sensory, socio-cultural,or instructional factors, it must be due to con-stitutional factors. Thus, the constitutional natureof dyslexia was inferred on the basis of what it wasnot, rather than on the basis of direct evidence ofcentral nervous system involvement (Rutter, 1978).During the past 15 years, it has become possible tomore directly evaluate the hypothesis that dyslexiais caused by constitutional factors that are intrinsicto the child. In the next section, we review studiesof a) brain structure, b) brain function, and c)genetics.

Brain structure

Post mortem studies. The most significant researchon brain structure utilizes post mortem studies oranatomical magnetic resonance imaging (aMRI).Computed tomography studies were also utilized, butwill not be reviewed here as MRI superseded thesemodalities (see Hynd & Semrud-Clikman, 1989).

Post mortem evaluations of the brains of peoplewith dyslexia are rare since it is not a cause of

18 Frank R. Vellutino et al.

Page 18: Vellutino

death. Nonetheless, the findings, while representinga cumulative total of 10 cases, are instructive in thatthey indicate that the underlying neuroanatomicalbasis of dyslexia is complex and not reducible to asingle finding or area of the brain. One set of findingsfocus on a structure on the plane of the temporallobe known as the planum temporale. This structurehas been reported to be often larger in the lefthemispheres than the right hemispheres of neuro-logically normal adults (Geschwind & Levitsky,1968). In post mortem studies of dyslexics, it hasbeen reported that there are unexpected symmetriesin the left versus the right hemispheres of thosewhose brains were studied (Galaburda, Sherman,Rosen, Aboitiz, & Geschwind, 1985; Humphreys,Kaufmann, & Galaburda, 1990). This area of theleft hemisphere supports language functions, so thesymmetry is viewed as a partial cause of languagedeficiencies that, in turn, lead to reading problems.

A second set of findings involves small, focalanomalies that appear when microscopic evalua-tions of the brain are performed. These anomalieswere found to be more common in the left hemi-spheres of people with a history of dyslexia, thoughthey were also found to be more frequent in otherparts of the brain. Microscopic examinations ofsubcortical structures have also shown differencesin the thalamus that may be related to visual pro-cessing (Livingstone et al., 1991).

These studies, while implicating difficulties at thelevel of brain structure, have some limitations.Obviously, the educational histories, reading char-acteristics, and other factors that influence brainorganization (e.g., handedness) are difficult toevaluate on a post hoc basis. There are no specificcontrol groups, and all the historical information isretrospective, often derived from a period where thediagnosis of specific reading problems was not ad-vanced. Given these difficulties and the limitednumber of available brains, investigators haveturned to methods based on neuroimaging.

Anatomical magnetic resonance imaging. Nonin-vasive methods based on aMRI have revolutionizedthe search for neural correlates of dyslexia. Theresolution is excellent and sufficient so that precisemeasurements of brain structure can be completed.A variety of structures have been evaluated,including the planum temporale, temporal lobes,and corpus callosum. Unfortunately, while the aMRIdata are easily acquired, the measurement of brainstructures is arduous and often requires consider-able human effort, which restricts sample size. Thereare also variations in how structures are measuredand in the definition of landmarks. Thus, it is notsurprising that the results are mixed and tend tovary across laboratories.

For the planum temporale, different studies reportsymmetry (Hynd, Semrud-Clikeman, Lorys, Novey, &Eliopulos, 1990; Larsen, Høien, Lundberg, & Ode-

garrd, 1990), reversals in the expected pattern ofasymmetry (Hynd et al., 1990), and no relationshipsof size or symmetry of the planum temporale indyslexia (Rumsey et al., 1997; Schultz et al., 1994).Leonard et al. (1996) reported that higher degrees ofasymmetry of the temporal lobes, favoring the lefthemisphere, were associated with better readingperformance, regardless of whether the child wasdisabled in reading. This finding implies lack ofspecificity with regard to children with reading dis-ability, but Leonard et al. (2001) failed to replicatethe finding. Other studies report that the temporo-parietal brain areas are smaller (Duara et al., 1991;Kushch et al., 1993), or not different in those with orwithout reading difficulties (Hynd et al., 1990;Jernigan, Hesselink, Sowell, & Tallal, 1991). Con-sistent with this theme, studies that measure thecorpus callosum find differences in its size (Duaraet al., 1991; Hynd et al., 1990) as well as no differ-ences in its size (Larsen et al., 1990; Schultz et al.,1994) between groups with and without dyslexia.

The differences across studies clearly relate tomany factors, including small samples and variationin imaging methods. Subject-level factors are clearlyimportant (Filipek, 1996; Shaywitz et al., 2000).Schultz et al. (1994) found statistically significantdifferences on several measures between childrenwith dyslexia and age matched controls, includingthe planum temporale and other left hemispherestructures. However, when age, gender, and hand-edness were covaried, the only reliable finding was asmall reduction in the size of the left temporal lobesin the group with dyslexia. Similarly, Penningtonet al. (1999) found reductions in the size of theinsula and anterior superior cortex in both hemi-spheres of a group of twins with dyslexia. An area ofthe brain defined as posterior to the splenium of thecorpus callosum, comprising the posterior temporal,parietal, and occipital areas, was larger in bothhemispheres of the group with dyslexia. These dif-ferences persisted when age, gender, and IQ werecontrolled, but were relatively small.

Although the findings are consistent, there is evid-ence suggestive of subtle differences in severalbrain structures between groups with and withoutdyslexia. But these studies have diminished in fre-quency because of their difficulty and relatively lowyield, especially relative to functional neuroimagingmethods (see below). However, new modalities forstructural neuroimaging may lead to resurgence. Toillustrate, Klingberg et al. (2000) used diffusiontensor imaging to evaluate the integrity of the ce-rebral white matter in the left hemisphere languageregions, showing less development of white matter ina group with dyslexia. These results suggested re-duced myelination of these language-mediatingareas. In the future, more application of thesemethods are likely, especially in conjunction withmulti-modal studies that also employ functionalbrain imaging with the same person.

Specific reading disability (dyslexia) 19

Page 19: Vellutino

Brain function

Most current studies of neural factors in dyslexia usefunctional neuroimaging modalities that assess theresponse of the brain to cognitive challenges. Func-tional neuroimaging in dyslexia utilizes five differentmodalities, including variations in electrophysiolo-gical methods, positron emission tomography (PET),functional magnetic resonance imaging (fMRI), mag-netic source imaging (MSI), and magnetic resonancespectroscopy (MRS). A review of these methods is be-yond the scope of this paper (seePapanicolaou, 1998).The methods all measure changes in the brain thatoccur during cognitive processing as a basis formapping where and/or when in the brain thesechanges occurred. Thus, metabolic changes reflectedby glucose utilization or shifts in blood flow from onepart of the brain to another part of the brain occurduring cognitive processing and can be captured byPET or fMRI. In contrast, when neurons discharge,changes in the brain electrical activity occur that canbe captured at the scalp by electrophysiologicalmethods. Changes also occur in the magnetic fieldssurrounding these electrical discharges at the neur-onal level, which can be detected with MSI.

Finally, MRS detects changes in brain chemistrythat occur in relation to changes in state (Hunter &Wang, 2001). Methods that are sensitive to metabolicactivity capture changes that occur after the cognit-ive activity has occurred and are not sensitive towhen the change occurred. Methods such as MSI(and EEG) take place in real time and provide infor-mation on the time course of neural events. Theirspatial resolution is weak, so the maps are co-registered on an MRI scan.

Previous research has used all five modalities.However, since the resolution of maps from electro-physiological procedures does not have very goodspatial resolution, these studies will not be furtherdiscussed. There are a few MRS studies, which willbe discussed within the context of intervention. Thefindings from PET, fMRI, and MSI converge. Theyshow that tasks requiring reading and phonologicalprocessing are associated with increased activationin the basal surface of the temporal lobe, the pos-terior portion of the superior and middle temporalgyri, extending into temporoparietal areas (supra-marginal and angular gyri), and the inferior frontallobe. These activations are often bilateral in PET andfMRI (Eden & Zeffiro, 1998; Rumsey et al., 1997;Shaywitz et al., 2000), but tend to be more lateralizedin the MSI studies (Simos, Breier, Fletcher, Berg-man, & Papanicolau, 2000a; Simos et al., 2000b).

In MSI studies, children who are typically achiev-ing show a pattern in which the occipital areas of thebrain that support primary visual processing areinitially activated. Then regions in the basal tem-poral areas in both hemispheres are activated.Shortly thereafter, there is virtually simultaneousactivation of three areas in the left temporal and

parietal areas of the left hemisphere, roughly cor-responding to the superior temporal gyrus, Wer-nicke’s area, and the angular gyrus. In contrast,children with dyslexia activate the same areas of theright hemisphere, with a similar time course. Otherstudies using PET and fMRI also show evidence ofright hemisphere activation in the posterior temporalparietal regions (Shaywitz et al., 2002). This couldreflect compensatory processes or could indicatethat other nonlinguistic factors are related to readingdisability (Grigorenko, 2001; Joseph, Noble, & Eden,2001; Wood & Grigorenko, 2001). These differencesin activation patterns across modalities as well asother inconsistencies across studies are apparent(Poeppel, 1996). However, a network of areas isconsistently implicated in studies of people withdyslexia. Particularly important is the angular gyrus,which studies of acquired reading disorders com-monly implicate as a feature of the pattern of lesionsthat cause pure alexia. Horwitz et al. (1998) andPugh et al. (2000) conducted statistical analyses ofPET and fMRI results and showed that the angulargyrus in the left hemisphere was poorly connectedwith other areas involved in the mediation of readingin dyslexic compared with proficient readers.

Most recently, imaging studies have been con-ducted before and after attempts to remediate read-ing impairment (Richards et al., 2000; Simos et al.,2002a). Richards et al. (2000) used MRS to evaluatechanges in brain chemistry as part of a comprehen-sive, phonologically oriented three-week (30-hour)intervention. Before intervention, MRS showed ahigher metabolic rate of lactate in the left anteriorquadrant of the left hemisphere when children withdyslexia completed a reading task. After interven-tion, measures of lactate metabolism taken duringreading were not different in children with dyslexiacompared with controls.

In a similar study using MSI, Simos et al. (2002b)evaluated neural responses to an intense phonolo-gical intervention in 8 children with severe dyslexia.The children were 7–17 years, with 6 of 8 readingbelow the fourth percentile and the other two readingbelow the 19th percentile. After about 80 hours ofinstruction over 8 weeks, their scores on measuresof word reading accuracy were in the average range.In each child, there was a significant increase in theactivation of neural circuits in the left hemispherethat tend to be activated in good readers, but werenot activated in the dyslexics at baseline (see exam-ple in Figure 2). A comparison group of good readersassessed 8 weeks apart showed no significantchanges in left hemisphere activation. Although thechanges were statistically significant, the samplesize was small and replication is needed. In addition,latency data showed delays in the evoked fieldsassociated with the left hemisphere response. Suchdelays may relate to the failure of these types ofinterventions to produce dramatic gains in readingfluency with older children (Torgesen et al., 2001).

20 Frank R. Vellutino et al.

Page 20: Vellutino

Results from these two studies suggest thatinstruction may be necessary to establish the neuralnetworks that support word recognition abilities indyslexia. The resultsmay also extendbeyonddyslexiaand imply that environmental factors are essential inestablishingthesenetworksinallchildren.Simosetal.(2002b) identified children who showed risk charac-teristics for reading difficulties at the end of kinder-garten due to poor development of phonologicalawareness skills. Application of MSI during a taskassessing the child’s ability to identify letter soundsshowedbrainactivationprofilesthatwerequitesimilarto those identified in older children and adults withdyslexia. Children who were not at risk showed thepatterns of left hemisphere activation seen in olderproficient readers. These children are being followedand the children at risk for reading difficulties receiveintensive intervention in Grade 1. Follow-up imagingwill be completed at the end of grade 1 to determinewhether the patterns of the at-risk children shift to-wards those of the not-at-risk children. But the direc-tion of the research suggests that the brain andenvironment,either through instructionorsomeotherform of early literacy support, interact to produce theneural networks that must be in place to mediate theunique component of reading – word recognition.

Genetic studies

Studies of the heritability of dyslexia also show evid-ence for a prominent role of environmental factors

as well as converging evidence that reading skillshave a strongly genetic influence (see recent reviewsby Grigorenko, 2001; Pennington, 1999; Olson,Forsberg, Gayan, & DeFries, 1999; Olson & Gayan,2001). In family segregation studies, the risk ofdyslexia is 8 times higher in children where there is aparental history of reading difficulties. Differentstudies report that 25%–60% of the parents of dys-lexic children also display reading difficulties. Intwin studies, concordance rates are almost alwaysabove 80% for monozygotic twins and usually below50% for dizygotic twins. As monozygotic and dizy-gotic twin pairs share the same environment, differ-ences in concordance rates presumably relate to theheritability. Statistical approaches to family andtwin studies also produce evidence for the herita-bility of reading disability. From these studies, about50–60% of the variance in reading achievement andreading-related abilities can be explained by geneticfactors. By the same token, the studies also showthat environmental factors account for significantvariance in reading skills, but the contribution ofgenetic factors is consistently higher (Olson et al.,1999; Grigorenko, 2001). Finally, linkage studies offamilies with many individuals with dyslexia havebeen used to identify specific genes involved inreading disability. Five different laboratories haveidentified an area on chromosome 6 (Grigorenko,2001). Replicated findings have also occurred forchromosome 15 (Grigorenko et al., 1997; Smith,Pennington, Kimberling, & Ing, 1990). Potential

BEFORE

AFTER

Left Hemisphere Right Hemisphere

Figure 2 Brain activation maps from a child with severe dyslexia before and after an 8 week intense intervention inwhich word reading skills moved into the average range. The upper panel shows the typical brain activation map frommagnetic source imaging studies of dyslexia, with predominant activity in temporal and parietal areas of the righthemisphere, but little activation in homologous areas of the left hemisphere. In the lower panel there is a significantincrease in the activation of these left temporoparietal areas associated with the significant improvement in wordreading accuracy that parallels the patterns observed in proficient readers (based on Simos et al., 2002a).

Specific reading disability (dyslexia) 21

Page 21: Vellutino

markers on chromosome 1 and 2 have been reportedbut not replicated (Grigorenko, 2001).

Fisher and DeFries (2002) provide a detailed re-view of genetic studies of dyslexia. They place thestudy of the heritability of dyslexia squarely withinthe study of other complex human problems, andlink it closely with studies of the heritability ofreading skills. Their extensive review of geneticmethodologies and their application to dyslexia andreading should be consulted for a more extensiveevaluation of an area of research with a long history(see also Grigorenko, 2001).

Fisher and DeFries (2002) emphasize that theheritability of dyslexia has biological and environ-mental influences, with Olson and Gayan (2001)providing an extensive discussion of shared andnon-shared environmental factors. Studies of twinsare especially useful in teasing apart the contribu-tion of shared environmental and genetic factors.Olson and Gayan (2001) observe, for example, thatwhile twin pairs (identical and fraternal) may sharesimilar environments, the differences in the genes offraternal twins are more likely to lead them to selectdifferent environments. As groups, identical andfraternal twins are treated differently in schools socomparisons of identical and fraternal twins helpestablish not only the differential influence of geneticfactors, but also the differential impact of environ-mental factors. Olson and Gayan (2001) also sum-marize behavioral-genetic studies involving a varietyof reading skills and reading-related subskills, suchas phonological and orthographic awareness. Forexample, word recognition shows a genetic influenceof .45 and a shared environmental influence of .15.Related subskills, however, have higher geneticinfluences (e.g., phonological awareness ).56; ortho-graphic awareness ).58), lower shared environ-mental influences (phonological awareness ).24;orthographic awareness ).20) and higher non-shared environmental influences (phonologicalawareness ).20; orthographic awareness ).22).Thus, individual differences in reading are partly dueto shared genetic influences and partly due to sharedand non-shared environmental influences. Olsonand Gayan (2001) also note that these influencesappear linked to early literacy development. It is alsoclear that pre-schoolers with early language diffi-culties are at greater risk for reading problems(Snowling, Bishop, & Stothard, 2000) and that chil-dren with oral language problems also show evid-ence that these difficulties have a genetic component(see discussion below). Studies that link the herita-bility of early language difficulties and the extensiveresearch on the heritability of dyslexia are emergingand will be a significant contribution to our under-standing of both (Olson & Gayan, 2001).

Like the initial studies of intervention and neuro-imaging, genetic studies suggest that environmentalfactors are important determinants of individualreading profiles for many children with dyslexia.

Both types of studies also show that intrinsic biolo-gical factors are important determinants of suchprofiles. Neither type of study fully explains whysome children develop dyslexia and others do not doso, although the idea that people may vary in theintensity and quality of instruction necessary toestablish word recognition skills is clearly apparent.Keep in mind that parents with reading problemsread less frequently to their children than parentswho do not have reading problems and are likely tohave fewer books in the home (Wadsworth, Olsen,Pennington, & DeFries, 2000). The genetic studiesalso do not indicate that there are specific genesthat produce dyslexia. As Gilger (2002) pointed out,the genetic influences appear to affect all levels ofreading ability. Thus, many people are not born withdyslexia, but, rather, have a susceptibility thatrequires more intense instruction.

Dyslexia across the life-span

Additional support for the possibility that readingdifficulties, in some children, are caused, in part, bygenetically based cognitive deficits is provided byseveral family risk studies that have recently ap-peared in the reading disability literature. A strategythat has been increasingly used to identify the earlyprecursors of dyslexia is to follow the developmentalprogress of children at high risk of dyslexia becauseof a genetic liability associated with the occurrence ofdyslexia in at least one family member. This ap-proach capitalizes on the fact that dyslexia runs infamilies and that there is approximately a 50% riskthat a child will develop reading difficulties, giventhat they have a parent with dyslexia.

The first prospective study of children at familyrisk of dyslexia was reported by Scarborough (1990),who followed the progress of 32 two-year-olds fromfamilies with a history of reading disability andcompared them with children from control familieshaving similar socioeconomic backgrounds. At8 years, 65% of the high-risk sample (20 children)was classified as reading-disabled. Examination ofthe data from earlier assessments of these childrenshowed that, at 30 months, those who later becamedyslexic used as wide a range of vocabulary in theirconversation as controls and children from high-riskfamilies who were normal readers. However, theyused a more restricted range of syntactic devices andmade more speech production errors. At 36 and42 months, the high-risk children’s vocabularydevelopment was less well developed than that ofcontrols and syntactic difficulties persisted (Scar-borough, 1991). At 5 years of age, the children wholater became dyslexic had poor letter knowledge,poorly developed phonological awareness, andexpressive naming difficulties.

Thus, contrary to the findings of studies of school-age dyslexic samples that implicate a specificphonologicaldeficit (Snowling,2000a),Scarborough’s

22 Frank R. Vellutino et al.

Page 22: Vellutino

study found that dyslexia was characterized by orallanguage difficulties that transcended phonology inthe pre-school years. Converging evidence has sincebeen reported by several studies of children at highrisk of dyslexia before formal schooling begins(Byrne, Fielding–Barnsley, Ashley, & Larsen, 1997;Gallagher, Frith, & Snowling, 2000; Locke et al.,1997; Lyytinen, Poikkeus, Laakso, Eklund, & Lyyti-nen, 2001; Lefly & Pennington 1996). However, itneeds to be borne in mind that this method of re-cruitment is very different from that used in moreconventional group studies that have selected chil-dren who fulfill �discrepancy� and other more con-ventional criteria for reading disability. Furthermore,the finding of more general language delays inaffected children applies to group data and, withingroups of affected cases, some children may havemore selective impairments or pre-school languagedelays that may resolve into more specific phonolo-gical deficits. An alternative interpretation of thesefindings, however, is that poor phonological pro-cessing carries the risk for reading disability in high-risk samples, but the impact of the risk may bemodified or mitigated by variations in other cognitiveand language skills.

In relation to this alternative version of thephonological deficit hypothesis, three studies pro-vide support for the view that the risk of dyslexia iscontinuous rather than �all-or-none�, in the sense inwhich we discussed this dichotomy earlier. Penn-ington and Lefly (2001) followed the progress of 67children at high risk of dyslexia and 57 controlsconsidered to be at low risk, from before entry intokindergarten through second grade. At the end of thestudy, children who showed a discrepancy betweenIQ and composite reading and spelling scores wereclassified as reading-disabled. Using this criterion,34% of the high-risk group were diagnosed as�reading-disabled� in second grade, compared to only6% of the low-risk (control) group. Consistent withthe phonological deficit hypothesis, children whobecame reading-disabled showed deficits on tests ofspeech perception, verbal short-term memory, rapidserial naming, and phonological awareness at alltesting points, relative to both the control childrenand the high-risk unimpaired children. Importantly,however, children at high risk of reading disabilitywho were considered to be normal readers at the endof second grade scored significantly lower thanchildren in the low-risk control group on mostmeasures of reading and spelling. They also hadmore difficulty on tests of implicit phonological pro-cessing (particularly verbal short-term memory andrapid serial naming measured around second grade),although they were unimpaired on tests evaluatingexplicit phonological awareness. So it seems that theunimpaired children shared at least some of thecognitive characteristics of reading difficulties, de-spite the fact that their reading outcomes werewithin the normal range. More research is needed to

determine how the high-risk unimpaired childrenmanaged to �compensate� on tests of explicit phono-logical awareness in the face of weak implicitphonological processing skills.

In a similar vein, Snowling et al. (2003) followedthe progress of 56 children from families in which atleast one parent was dyslexic. These children wereassessed periodically from the age of 3 years9 months to age 8 years and their performance onmeasures of reading skills and reading-related cog-nitive abilities was compared to that of normalreading control children from similarly advantagedhome backgrounds, but with no history of readingfailure in family members. Sixty-six percent of thechildren had reading skills more than one standarddeviation below the mean of the normal readingcontrols (the rate of discrepancy-defined dyslexiawas lower at 32%). In line with previous studies,retrospective analyses revealed that the children whowent on to develop reading disabilities had slowervocabulary development and poorer narrative skillsin the pre-school years than high-risk children whowent on to develop normal reading ability. They alsohad more difficulty on tests of verbal short-termmemory and phonological processing (nonwordrepetition). At 6 years, the high-risk impaired groupcontinued to have difficulty on tasks evaluatingvocabulary development and language processingskills, and they also performed poorly on tests evalu-ating explicit phonological awareness. On all of thesetests, the unimpaired group performed within thenormal range. Nonetheless, they knew fewer lettersthan controls at age 3 years 9 months (althoughmore than the reading impaired children), and at6 years, their reading, spelling and reading compre-hension skills were less advanced than those of thenormal reading controls. Importantly, on two meas-ures evaluating phonological reading and spellingstrategies, one involving nonword decoding andthe other involving the phonetic accuracy of earlyspelling, the high-risk unimpaired group performedas poorly as the high-risk impaired group and muchless well than expected for their age. Since thesechildren were not poor readers at 8 years, it appearsthat they were able to compensate for deficits inphonological reading and spelling strategies at6 years, perhaps by using intact language skills to�bootstrap� inefficient decoding mechanisms (cf. Na-tion & Snowling, 1998). Indeed, it seems that bothaffected and unaffected members of dyslexic familiesshare the risk of reading failure that is characterizedby poor �phonic� skills (e.g., poor nonword reading).However, while some succumb to reading impair-ments, others do not.

Finally, further evidence for a �broader phenotype�of dyslexia comes from a study of Danish childrenconducted by Elbro et al. (1998), who reported thatnon-dyslexic children in dyslexic families had defi-cits, relative to controls, on tests evaluating mor-phological awareness and articulatory accuracy.

Specific reading disability (dyslexia) 23

Page 23: Vellutino

Surprisingly, the difficulties these children wereexperiencing did not extend to letter knowledge,phoneme awareness, verbal STM or the distinctnessof phonological representations. However, it isimportant to note that the high-risk children in theElbro et al. study were being educated very differ-ently than the high-risk children in the studies wehave discussed thus far, most of whom were edu-cated either in the United States or in the UnitedKingdom. For example, the children in the Elbroet al. study were in kindergarten classes for longerperiods of time and they were learning to read in adifferent orthography, albeit an opaque one. Itseems likely, therefore, that environmental factorsplay a role in the determination of the �dyslexiaphenotype�.

Taken together, the findings from the family riskstudies suggest that dyslexia is a multifactorial traitin which basic constitutional (genetic) vulnerabilities(notably in phonological skills) interact with othercognitive skills and environmental factors to producean increased risk of dyslexia in a continuous way.Arguably, when the level of risk reaches a certainthreshold, the classic dyslexia profile emerges, butthe evidence suggests that there are varying degreesof �sub-clinical� impairment, particularly in dyslexicfamilies.

Additional support for this conceptualizationcomes from studies evaluating the manifestationsand correlates of dyslexia at the other end of the agespectrum, that is, studies of adults with a history ofearly reading difficulties. The hypothesis that thecore phonological deficit in dyslexia is manifested indysfunctional phonological processes receives sup-port from the finding that reading difficulties andrelated phonological deficits persist across develop-ment, even in people with dyslexia who have com-pensated for their reading difficulties (Bruck, 1990,1992, 1993; Hatcher, Snowling, & Griffiths, 2002;Pennington, Orden, Smith, Green, & Haith, 1990).Such findings apply equally to adults with dyslexiawho read in transparent languages (Paulesu et al.,2001).

For example, Bruck (1990, 1993) compared col-lege students who had childhood diagnoses of dys-lexia with age-matched and reading-level matched(6th graders) control students on measures of wordrecognition, spelling, and reading comprehensionskills, and found that, despite having readingcomprehension skills that approximated those ofthe age-matched control students, the dyslexicstudents manifested inaccurate and dysfluent wordrecognition and spelling skills, relative to the age-matched controls. Moreover, the dyslexics usedstrategies and processes, for both word recognitionand spelling, that were immature and more likethose of the reading-level matched controls thanlike those of the age-matched controls. Like thereading-level (sixth grade) controls, the adult withdyslexia relied more than the age-matched controls

on spelling-sound information, syllabic information,and use of context for word recognition, and theyrelied more heavily on spelling-sound informationthan on visual information, in both word recogni-tion and spelling of irregular words such as put andstrange words such as yacht. This was found to bethe case, despite the fact that the dyslexics haddeficient letter-sound knowledge (as manifested inpoor nonword decoding and poor nonword spellingskills). Evidence that the persistent word recogni-tion and spelling problems of the dyslexic collegestudents were due, in part, to a core phonologicaldeficit is reported in another paper by Bruck (1992),who found that these same students performedsignificantly below both the age-matched and thereading-level matched controls on a variety of pho-nological awareness measures.

Bruck’s (1990) finding that adults with dyslexiahad reasonably good reading comprehension skills,despite dysfluent word recognition skills, is some-what at variance with conventional views of readingcomprehension that are based on the assumptionthat fluent word recognition is a prerequisite foradequate reading comprehension. To further evalu-ate this disparity, Bruck divided the dyslexic stu-dents into poor and good comprehender groups andfound that the good comprehenders had higherchildhood IQs and better developed vocabularyknowledge than the poor comprehenders, but nobetter word recognition skills. Bruck inferred fromthese results that the superior cognitive abilities ofthe good comprehenders allowed them to compen-sate more effectively than the poor comprehendersfor their deficient word recognition skills. Thisinterpretation is quite in keeping with the implica-tion from family risk studies that the ability to ac-quire functional reading skills is determined, in part,by the particular mix of cognitive abilities with whichthe individual is endowed interacting with uniqueenvironmental circumstances that serve to facilitateor impede the acquisition of such skills.

Studies of adult dyslexia conducted by Bruck(1990, 1992, 1993) and others (see above refer-ences) provide documentation that the majorsymptom pattern defining dyslexia, in terms of ba-sic deficits in word recognition and related literacyand language-based skills, persist well into adult-hood. To compare competing theories of the etiologyof dyslexia and to further examine the core deficit inadults, Ramus et al. (2003) conducted a multiple-case study involving 16 university students identi-fied as dyslexics and 16 controls. The studentsidentified as dyslexics had reading and spellingskills within the normal range, even though theirliteracy skills were poorer than expected, given theirgeneral cognitive ability (IQ). Each student com-pleted a number of different tests assessingauditory, phonological, visual, and motor domainsof processing. Phonological deficits on tasks evalu-ating rapid automatized naming, spoonerisms, and

24 Frank R. Vellutino et al.

Page 24: Vellutino

nonword repetition were observed in all 16 of thestudents with dyslexia. In addition, 10 of the dys-lexic students showed auditory impairments, 4 hadproblems with motor skills, and 2 showed deficitson visuo-magnocellular tasks.

Auditory deficits were much more common thanother forms of sensory impairment in this group,lending face valid credence to Tallal’s (1980) hypo-thesis that language learning and reading difficul-ties can be traced to auditory processing problems.However, the nature of the auditory deficit variedacross these students – some but not all showedproblems on rapid auditory processing tasks, andsome had problems with basic auditory processing.Others only had problems on speech perceptiontasks, and six manifested no deficits in any of thesecognitive domains. To explain the heterogeneity inthese data, the investigators suggested that phono-logical skills may be involved in �top-down control�of auditory skills, but auditory processing skills arenot universally impaired in dyslexia. We would add,in keeping with a suggestion made by Eden andZeffiro (1998), that the low-level visual and auditorydeficits that were observed in some of the dyslexicsin this study were unlikely causes of their lowerthan expected reading levels, but could have beenclassified as biological markers signifying functionalanomalies, both in areas of the brain that supportthese sensory processes and in adjacent areas thatsupport reading and language processes.

In sum, results from the life-span dyslexia studiesexemplified in the foregoing sections complementresults from the genetic studies discussed earlier.Together, these studies provide documentation thatreading difficulties in otherwise normal individuals,in many cases, are caused primarily by biologicallybased cognitive deficits that can be identified quiteearly in life and that persist well into adulthood.They also provide documentation that there is agradation of risk for becoming dyslexic, dependingon the particular assortment of (reading-related)cognitive abilities with which the individual is en-dowed and the degree to which the individual’senvironmental experiences allow him or her to cap-italize on cognitive strengths in ways that will com-pensate for cognitive weaknesses. Thus, the majorconclusions that can be reasonably drawn from thelife-span dyslexia studies we have discussed accordquite well with the major conclusions that can bereasonably drawn from the genetic studies we havediscussed.

Cognitive and biological versus experiential andinstructional causes of early reading difficulties

Specific reading disability, as an etiological con-struct, incorporates the assumption that childrenwho qualify for this diagnosis on the basis of psy-chometric and exclusionary criteria such as those

outlined earlier suffer from basic cognitive deficits ofbiological origin. Empirical support for this poss-ibility is seminal and suggestive, if not always con-clusive. As we have discussed, such support comesfrom results obtained in genetic, neuroanatomical,and psycho-physiological studies, which, collec-tively, provide some reason to believe that poor andnormal readers have structurally and functionallydifferent architecture for processing spoken andwritten language (Grigorenko, 2001; Lyon et al.,2001). However, as pointed out by Clay (1987), vir-tually all reading disability research has been com-promised by the failure to control for the child’seducational history, given that the adverse effects ofinadequate pre-reading experience and/or inad-equate instruction can often lead to reading skillsdeficiencies that mimic the effects of basic cognitivedeficits. Consider, for example, that the acquisitionof skills such as phonological awareness and letter-sound decoding can be greatly influenced by the typeof reading instruction to which a child has been ex-posed. Yet, it is commonly assumed that difficultiesin acquiring one or both of these skills are a mani-festation of basic deficits in phonological coding.Indeed, it is possible (and we think highly likely) thatmany of the children placed in single deficit (or evendouble deficit) sub-categories in recent studiesevaluating the double deficit theory of reading dis-ability (e.g., Wolf, Bower, & Biddle, 2000a) werechildren whose reading difficulties were causedprimarily by inadequate instruction. Similarly, IQscores are typically used to help diagnose specificreading disability. Most intelligence tests, however,contain tasks and items that depend heavily on theacquisition of knowledge and skills, like vocabularyknowledge and domain-general knowledge, that areacquired in part through reading. As a consequence,intelligence in children having long-standing readingdifficulties may be underestimated, particularly asthey grow older (Ackerman, Weir, Holloway, & Dyk-man, 1995; Shaywitz et al., 1995; Stanovich, 1986;Vellutino et al., 1988, 1995b). Thus, there is apressing need for research that facilitates identi-fication of markers that would aid in distinguishingbetween children whose reading problems arecaused primarily by cognitive deficits of biologicalorigin and those whose reading problems are causedprimarily by limitations in pre-reading experienceand/or inadequate instruction.

One such study was recently reported by Vellu-tino, Scanlon, and their associates (Vellutino et al.,1996). This was a longitudinal study that incor-porated an intervention component (daily one-to-one tutoring) to distinguish between cognitivelyversus experientially impaired readers. Thus, chil-dren classified in first grade as impaired or nor-mally developing readers were tracked from thetime they entered kindergarten through the end offourth grade, that is, before and after they wereclassified, and before and after intervention. Inter-

Specific reading disability (dyslexia) 25

Page 25: Vellutino

vention was initiated in mid-first grade, and wasterminated either at the end of first grade or in themiddle of second grade, depending on the child’sprogress. Given results obtained in previous inter-vention studies (Clay, 1985; Iversen & Tunmer,1993; Pinnell, 1989; Wasik & Slavin, 1993), it wasexpected that most, but not all of the tutored chil-dren would be successfully remediated, and theinvestigators were especially interested in compar-ing the entry-level skills and cognitive profiles ofchildren who were the most difficult to remediate,with those of children who were readily remediated,relative to normally achieving readers. It was alsoexpected that the entry-level skills of the impairedreaders would be uniformly deficient. And from theconvergent evidence implicating phonological cod-ing deficits as a basic cause of reading difficulties insome impaired readers, it was anticipated that thechildren who were the most difficult to remediatewould perform below the children who were readilyremediated on measures evaluating cognitive abil-ities such as phonological awareness, verbal mem-ory, and rapid naming. These groups were notexpected to differ on measures evaluating semantic,

syntactic, and visual skills. These predictions gen-erally were confirmed.

First, it was found that relative to normallyachieving readers, entry-level literacy skills such asletter naming and phonological awareness weredeficient in the group of kindergarten childrenidentified as impaired readers in first grade. Second,67.1% of the tutored children were brought to withinthe average range of reading ability in only onesemester of remediation, and the majority main-tained this level of functioning through the end offourth grade (see Figures 3 and 4). Third, the chil-dren who were found to be the most difficult to re-mediate performed well below the normal readers,and quite often below the children who were readilyremediated on kindergarten, first, and third gradetests evaluating phonological abilities such as thosementioned previously. In addition, the children whowere readily remediated often performed as well asthe normal readers on such tests. However, therewere no significant differences among any of thegroups on the semantic, syntactic, and visualmeasures, although mean effect sizes tended to benegative, thereby reflecting a general trend, on the

Normal and Tutored Poor Readers

010

2030

4050

6070

8090

100

0 5 10 15 20 25 30 35 40 45 50 55

VLG (n = 19)

LG (n = 15)

GG (n = 17)

VGG (n = 18)

AvIQNorm (n = 21)

AbAvIQNorm (n = 30)

KindergartenWinter Spring

Grade Grade 1 Grade 1 2Fall

Grade 2Winter

Grade 2Spring

Grade 3Spring

Grade 4Spring

Time Intervals Between Tests in Months

VLG = Very Limited Growth LG = Limited GrowthGG = Good GrowthVGG =Very Good GrowthWRMT = Woodcock Reading Mastery TestRevised

Wor

d Id

entif

icat

ion

Raw

Sco

res

on W

RM

T-R

Figure 3 Growth curves for mean raw scores on the WRMT-R word identification subtest for normal and tutored poorreaders

26 Frank R. Vellutino et al.

Page 26: Vellutino

part of the impaired readers, to perform below thelevel of the normal readers on these measures as wellas on the phonological measures. Because the nor-mal readers in this study generally scored abovenational norms on the reading measures, the neg-ative effect sizes on the semantic, syntactic, andvisual measures were taken as an indication thatthey were characterized by a more �optimal mix� ofreading-related cognitive abilities than were the im-paired readers, rather than an indication that thecognitive abilities evaluated by these measures wereseriously deficient in the impaired readers (Vellutinoet al., 1996). This analysis is more in keeping withwhat we have called �gradation of risk�, continuousability type conceptualizations of dyslexia (Olson &Gayan, 2001; Pennington & Lefly, 2001; Snowlinget al., 2003; Stanovich, 1988; Vellutino et al., 1996)than with what we have called �all or none� typeconceptualizations of dyslexia, as discussed in ear-lier sections of this paper.

Finally, contraindicating the use of IQ scores toidentify disabled readers or to predict readingachievement in beginning readers, the investigatorsfound that the tutored groups did not differ on any

of the intelligence tests, nor did they differ from anaverage IQ normal reader group on these tests.Moreover, the average IQ normal reader group didnot differ from an above average IQ normal readergroup on any of the reading tests, except for tests ofreading comprehension, on which the above aver-age IQ group performed at significantly higherlevels than the average IQ group (and, of course, athigher levels than the tutored groups). In addition,IQ-achievement discrepancy scores were not signific-antly correlated with initial growth in reading per-formance following one semester of one-to-onetutoring (Vellutino, Scanlon, & Lyon, 2000; see alsoShare, McGee, & Silva, 1989). Presently, there is alarge body of research showing that children with IQdiscrepant and IQ non-discrepant reading scorescannot be adequately differentiated, vis-a-visresponse to instruction or prognosis (see Francis,Fowler, & Shaywitz, 1994; Fletcher et al., 1994,2002; Lyon et al., 2001, 2002; Vellutino, Scanlon, &Lyon, 2000). There are also two recent meta-analysesshowing null to small differences between the cogni-tive skills of these two populations (Hoskyn & Swan-son, 2000; Steubing et al., 2002).

Subtest for Normal and Tutored Poor Readers

0

5

10

15

20

25

30

35

40

0 5 10 15 20 25 30 35 40 45 50 55

Time Intervals Between Tests (Months)

Wor

d A

ttack

Raw

Sco

re o

n W

MR

T-R

VLG (n=19)

LG (n=15)

GG (n=17)

VGG (n=18)

AvIQNorm(n=21)

AbAVIQNorm (n=30)

Kindergarten

VLG = Very LimitedGrowthLG = Limited GrowthGG = Good GrowthVGG = Very Good GrowthWRMT = WoodcockReading Mastery TestRevised

Winter SpringGrade 1 Grade 1 Grade 2

FallGrade 2Winter

Grade 2Spring

Grade 3Spring

Grade 4Spring

Figure 4 Growth curves for mean raw scores on the WRMT-R word attack subtest for normal and tutored poorreaders

Specific reading disability (dyslexia) 27

Page 27: Vellutino

These results are quite in keeping with Clay’s(1987) contention that reading difficulties in begin-ning readers are, in most cases, caused primarily byexperiential and/or instructional deficits. Indeed,the impaired reader sample initially identified in firstgrade, using exclusionary criteria such as thosetypically used to identify �disabled readers� in suchresearch, represented approximately 9% of the(available) population from which these childrenwere drawn. Yet, the impaired readers who contin-ued to qualify for this diagnosis after only onesemester of remediation represented only 1.5% of thepopulation from which these children were drawn,which is a far cry from the 10% to 15% figures thathave emerged as estimates of the incidence of read-ing disability in the relevant literature (e.g., Shaywitzet al., 1992; Harris & Sipay, 1990).

Additional support for the contention that readingdifficulties in many children are caused primarilyby experiential and/or instructional deficits comesfrom other intervention studies which show thatmost impaired readers can acquire at least grade-level reading skills if they are identified early andare provided with comprehensive and intensivereading instruction tailored to their individualneeds (Clay, 1985; Iversen & Tunmer, 1993; Pinn-ell, 1989; Scanlon et al., 2000; Torgesen et al.,1999; Wasik & Slavin, 1993). Similarly, recentclassroom observation and classroom interventionstudies (Foorman et al., 1991; Foorman et al.,1998; Scanlon & Vellutino, 1996) have shown thatcomprehensive and well-balanced reading instruc-tion that facilitates the acquisition of phonologicalawareness and letter-sound decoding skills alongwith other word-level skills, in addition to orallanguage and reading comprehension skills, canprevent long-term reading difficulties in childrenwho would otherwise qualify for a diagnosis ofreading disability. Nevertheless, the observation ofreading-related cognitive deficits in the poor readerswho were found to be difficult to remediate, relativeto both the normal readers and the poor readerswho were found to be readily remediated, suggeststhat reading difficulties in some impaired readersmay be caused, in part, by basic cognitive deficits ofbiological origin. Thus, additional research designedto trace the ultimate origins of such deficits is cer-tainly warranted.

Implications for practitioners

The knowledge acquired from the research reviewedin this paper has several important implications forpractitioners involved in diagnosing specific readingdisability (dyslexia), especially those who work inschool systems and serve as consultants to teachersand other school officials responsible for educationaland remedial planning. First, the research questionsthe utility of psychometric assessment as the sole or

even primary vehicle for determining the origin ofreading difficulties for purposes of educational andremedial planning. In many instances, this enter-prise is motivated by what some have called �a searchfor pathology� (Ysseldyke & Christensons, 1988) –that is, assessment to determine what we have calledunderlying (cognitive and biological) rather thanmanifest causes of a child’s reading difficulties interms of poorly developed reading subskills. Thisenterprise is typically implemented for purposes ofclassification associated with official mandatesrequiring such classification, such as categoricalfunding for �learning disabled� children versus �so-cioeconomically disadvantaged� children. Accord-ingly, the child is given an individually administeredbattery of tests that typically include an intelligencetest, one or more measures of reading achievement,and a variety of measures to evaluate reading-relatedcognitive abilities. Other �exclusionary� criteria aretypically employed to rule out extraneous factors ascauses of the child’s reading difficulties, for example,uncorrected sensory deficits, emotional disorder,frequent absences from school, and/or socioeco-nomic disadvantage. The end result of this process isthat the child’s reading difficulties, in most cases,are attributed to a basic cognitive deficit of one des-cription or another, which, in turn, tends to beattributed to some sort of neuro-developmentalanomaly.

Yet, as pointed out by Clay (1987), psychometric/exclusionary approaches to assessment do not con-trol for the child’s educational history and early lit-eracy experiences. And as we have seen fromintervention studies such as those just discussed,there is strong evidence that most early reading dif-ficulties are caused primarily by experiential andinstructional deficits, rather than basic cognitivedeficits associated with neuro-developmentalanomaly. Moreover, despite the progress made inidentifying what we have called manifest andunderlying causes of difficulties in learning to read,it should also be clear from the research reviewedherein that there is no uniform consensus as to whatmight be called the �ultimate� or neuro-biologicalcause(s) of specific reading disability and contro-versy abounds in this area of inquiry. Although someetiological theories enjoy greater currency than oth-ers, none of these theories yet provides a clear-cut,definitive, and unequivocal set of diagnostic criteriathat would pinpoint the ultimate (neuro-biological)origin of the child’s reading difficulties in the sameway that a biopsy pinpoints the type and locus of aphysical disease such as cancer. As a consequence,the approach to diagnosing the underlying cause(s)of a child’s reading difficulties tends to vary with thetraining and/or conceptual biases of the clinician.The reports written by these clinicians often havelittle prescriptive value vis-a-vis educational orremedial planning. Thus, given the state of the art, itcould be argued with some legitimacy that psycho-

28 Frank R. Vellutino et al.

Page 28: Vellutino

metric assessment performed by practitioners for thesole purpose of �diagnosing� reading disability is aquestionable enterprise that might better be aban-doned. This, of course, would mean that officialmandates, which require traditional diagnosticassessments for purposes of official classification,should also be abandoned.

A related problem underscored by current re-search – one that also has important implications forpractitioners – has to do with current psychometric-exclusionary approaches to diagnosing specificreading disability that adopt the IQ–achievementdiscrepancy as their central defining criterion. Inmost venues and locales, specific reading disabilityis defined on the basis of a significant discrepancybetween a child’s expected reading achievement, aspredicted by that child’s IQ, and his or her actualachievement, as defined by a score on a standardizedtest of word identification and/or other word-levelskills (see Lyon et al., 2002 and Vellutino et al., 2000for historical reviews of this practice). Aside from theobvious fact that exclusionary factors discussedabove provide no clear-cut criteria for identifying ei-ther manifest or underlying causes of a child’sreading difficulties, it should be apparent, from theresults obtained in the intervention study conductedby Vellutino et al. (1996), that the IQ–achievementdiscrepancy criterion is not a sufficiently precisemetric to warrant its continued use to define specificreading disability. In this study, intelligence testscores did not distinguish between poor and nor-mally achieving readers or between poor readers whowere difficult to remediate and those who werereadily remediated. They also failed to predict word-reading performance in normally achieving readers.Moreover, IQ–achievement discrepancy scores didnot predict initial growth in reading in poor readerswho received intensive remediation.

Altogether, the evidence supporting the use of IQ–achievement discrepancy classifications models isquite limited (Fletcher et al., 2002; Hoskyn &Swanson, 2000; Lyon et al., 2001, 2002; Stuebinget al., 2002; Vellutino et al., 2000; see also Shareet al., 1989; Siegel, 1988, 1989). Thus, it is apparentthat a child may need little more than average oreven low average intelligence to learn to decipherprint. Some scholars even assert that it has norelevance at all (e.g., Siegel, 1988, 1989). Althoughthis may be an extreme view, the fact remains that inany child learning to read in an alphabetically basedorthography, regardless of level of intelligence, theability to learn to decode print will be determinedprimarily by phonological skills such as phonologicalawareness, facility in alphabetic mapping, nameencoding and retrieval, and verbal memory. Indeed,how else would we account for hyperlexia (i.e.,atypically strong decoding ability), which occurssometimes even in children with mental deficiency.Thus, if our analysis proves to be correct, not onlywould the IQ–achievement discrepancy to define

reading disability be invalidated, there would be norole for IQ tests in this enterprise. Accordingly,practitioners would be well advised to abandon theuse of such measures for children with word-levelreading disabilities, which, of course, constitute thecore deficit in dyslexic individuals. This suggestiondoes not discount the utility of using intelligencetests (among other relevant measures) to aid indiagnosing the origin of reading comprehension dif-ficulties in children with adequate word-level skills,especially intelligence tests that evaluate verbalabilities and other cognitive abilities that are entailedin comprehending linguistic text, for example verbalreasoning, inferencing, and logical deduction. Suchhigher-level intellectual skills are not entailed, how-ever, in learning to decode print, in keeping with ourcontention that intelligence tests have little utility fordiagnosing specific reading disability.

Still another problem with currently employedpsychometric approaches to assessing the origin of achild’s reading difficulties is that they typically pro-vide no direction for educational or remedial plan-ning (Lyon & Moats, 1988, 1993). This problemoccurs, not only because such approaches tend tofocus on cognitive and biological rather than man-ifest causes of a child’s reading difficulties, but alsobecause the clinicians performing such assessmentstend to have limited background and expertise fordiagnosing and remediating deficiencies in founda-tional reading subskills such as phonologicalawareness, word identification, alphabetic mapping,and language comprehension (Fish & Margolis,1988). Consequently, their role in assisting educa-tors in educational and remedial planning is limitedto the administration of norm-referenced tests,which typically provide little that is of practical valuefor purposes of remedial instruction. And because ofthe lack of expertise in educational and remedialplanning characteristic of clinicians working inschool settings (and elsewhere), such assessmentstend to be �static� rather than �dynamic� insofar asthey evaluate existing abilities in terms of a child’srelative position in a normative population and pro-vide little or no information about the type ofinstruction that might be most effectively tailored tothe child’s individual needs (Reschly, Tilly, &Grimes, 1999).

This state of affairs underscores perhaps the mostgeneral implication of the reading disability researchreviewed herein: specifically, the need for a radicalchange in the perceived and implemented role ofclinical assessment in diagnosing and remediatingreading difficulties. Rather than select psychometrictests for purposes of evaluating cognitive abilitiesthat underlie reading ability, in the interest of de-tecting underlying (cognitive and biological) causesof a child’s reading difficulties for purposes of cat-egorical labeling (e.g., �specific reading disability�,�attention deficit disorder�, etc.), the clinician wouldmore profitably select psychometric tests that have

Specific reading disability (dyslexia) 29

Page 29: Vellutino

demonstrated validity for assessing strengths andweaknesses in reading subskills (what we havetermed �manifest causes� of reading difficulties) forpurposes of developing an appropriate educationalplan tailored to the child’s individualized needs.Moreover, such tests should be based on criterion-referenced standards derived from a thoroughunderstanding of the components of reading ability,rather than norm-reference standards based exclus-ively on placement in a normative distribution.

Results from current intervention studies suggestthat the most informative and most effective ap-proach to distinguishing between cognitive/biolo-gical and experiential/instructional causes of earlyreading difficulties would be to implement an initialperiod of remedial intervention, in lieu of assessmentof reading-related cognitive abilities as a �first cut�approach to diagnosis. Accordingly, psychometricassessment of strengths and weaknesses in a child’sreading subskills would not be used to classify thatchild as �reading-disabled� at this stage of the dia-gnostic process, but, rather to provide the educatorwith guidance for purposes of initiating remedialinstruction. This approach would not only evaluatethe child’s existing skills and abilities to insureindividualized and well-balanced instruction, butwould also evaluate the child’s initial response toremediation to aid in determining whether his or herreading difficulties are caused primarily by instruc-tional and experiential deficits. Assessment ofstrengths and weaknesses in reading-related cognit-ive abilities could thereafter be implemented tocross-validate initial impressions derived from thechild’s initial response to remediation. Recall that inthe Vellutino et al. (1996) intervention study dis-cussed earlier, the cognitive profiles of children whowere found to be readily remediated were closer tothose of normally achieving readers than were thecognitive profiles of those who were found to bedifficult to remediate. Thus, assessment of reading-related cognitive abilities for cross-validationpurposes would appear to be useful.

However, the primary purpose of these comple-mentary approaches to assessment would be to de-velop a long-range remedial plan that would facilitateacquisition of functional reading skills, rather thancategorical labeling. But, if such labeling is neces-sary, then the implication of the approach toassessment we are suggesting is that the assignmentof a categorical label be deferred until an attempt ismade to remediate the child’s reading difficulties(see Abbott, Reed, Abbott, & Berninger, 1997 andReschly et al., 1999 for additional support for thisapproach to diagnostic assessment). The net effect ofthis approach would be to shift our conceptualiza-tion of dyslexia away from scores on a set of teststoward response to intervention as the primarymeans for defining the disorder. From a policystandpoint, identification of children who might beeligible for special educational services would focus

on the child who has not demonstrated the acceler-ated growth in reading skills that characterizes manypoor readers who receive intensive intervention.Without this component, it becomes impossible todistinguish those with a true disability from thosewho are instructional casualties.

Finally, it should be clear that clinical practition-ers working with children who have reading diffi-culties should reset their priorities and shift thefocus of their clinical activities so as to place muchgreater emphasis on instructional and remedialactivities and much less emphasis on psychometricactivities in their work with reading impaired chil-dren. We suspect that this shift would not only en-hance their acumen in clinical diagnosis, but, moreimportantly, would facilitate the development ofeffective educational programs for correcting readingdifficulties in these children. Such a shift in focuswill require that the practitioner update and/or up-grade her or his knowledge base so as to learn moreabout the reading process and reading development,and focus on instructional factors that would facilit-ate or impair such development. In other words, themost effective practitioner would not simply be apsychometrician, as is true of many practitionersworking with reading impaired individuals, butwould also be a recognized expert in the psychologyof reading and reading development, the psychologyof reading disability, and the psychology of readinginstruction.

Summary and conclusions

We have learned much about manifest and under-lying causes of reading difficulties in otherwise nor-mal children over the past four decades. It is clearfrom the relevant research that reading is primarily alinguistic skill, contrary to the once popular notionthat it is primarily a visual skill. And, because of thestructural properties of an alphabetic system, it isalso clear that linguistic abilities are themselvesdifferentially weighted in reading development suchthat phonological skills carry greater weight asdeterminants of beginning reading ability than dosemantic and syntactic skills, whereas semantic andsyntactic skills carry greater weight than do phono-logical skills in more advanced readers. It follows, asthe evidence confirms, that inadequate facility inword identification constitutes the manifest andmost ubiquitous cause of reading difficulties. More-over, there is reliable and highly convergent evidencethat word identification problems, themselves, arecausally related to deficiencies in phonologicalawareness, alphabetic mapping, and phonologicaldecoding that lead to difficulties in establishingconnective bonds between a word’s spoken andwritten counterparts. However, causal relationshipsbetween word identification problems and deficien-cies in such phonological skills are more prominent

30 Frank R. Vellutino et al.

Page 30: Vellutino

in dyslexics learning to read in opaque orthographiessuch as written English than in dyslexics learning toread in more transparent orthographies such asGerman or Italian. In transparent orthographiessuch as these, word identification problems aremore often associated with deficiencies in implicitphonological processes (e.g., verbal memory, nameretrieval, etc.) that impair fluency in word iden-tification and text processing, and, thereby, readingcomprehension.

As regards underlying causes, the researchstrongly suggests that reading difficulties in mostchildren are caused by deficits in phonologicalcoding. Such deficits are believed to account for thereliable and robust differences observed betweenpoor and normal readers on measures of phono-logically based skills such as phonological aware-ness, alphabetic mapping, phonological decoding,verbal memory, and name encoding and retrieval.However, some researchers suggest that dysfluencyin name retrieval is caused by a hypothesized tim-ing deficit that impairs temporal integration of aword’s component letters, but the evidence to sup-port this hypothesis is tenuous and the issue hasbecome controversial. Semantic and syntactic defi-cits, in most cases, do not appear to be a primarycause of reading difficulties in otherwise normalchildren, but they are a likely consequence of long-standing reading disorder or a co-morbid oral lan-guage disorder. They may, however, be a primarycause of early reading difficulties in some children,especially those from disadvantaged or bilingualpopulations.

Reading disability research has also establishedthat reading difficulties are not caused by visualdeficits of the types most often proposed over theyears. Contrary to popular belief, impaired readersdo not see letters and words in reverse, nor do theysuffer from inherent spatial confusion or other visualanomalies of the types proposed in the early liter-ature. More recent research provides suggestiveevidence that some poor readers may suffer fromlow-level sensory deficits in both the visual andauditory spheres, but the evidence is inconclusive,and in, some instances, equivocal and controversial.Moreover, no causal relationships have been estab-lished between such deficits and difficulties inlearning to read.

Similarly, there is no reason to believe that defici-encies in general learning abilities such as attention,association learning, cross-modal transfer, serialmemory, pattern analysis, and rule learning arebasic causes of reading difficulties in impairedreaders who do not have general learning difficulties.Etiological theories, which implicate deficits in suchabilities as causally related to reading difficulties,can be ruled out on logical grounds alone and theyhave not fared well in empirical research.3

Finally, recent intervention studies have clearlydemonstrated that reading difficulties in most

beginning readers are not invariably caused by basiccognitive deficits of biological origin, from which itcan be concluded that current estimates of the incid-ence of true reading disabilities are greatly inflated.However, recent studies of neurological and geneticcorrelates of dyslexia, along with recent life-spandevelopment and intervention studies, providestrong reason to believe that a very small percentageof impaired readers may well be afflicted by basiccognitive deficits of biological origin, especiallyphonological deficits that lie at the root of theirdifficulties in learning to read. These and otherfindings we have discussed have obvious implica-tions for the diagnosis and remediation of readingdisability, the most general implication being theneed for practitioners to shift the focus of theirclinical activities away from emphasis on psycho-metric assessment to detect cognitive and biologicalcauses of a child’s reading difficulties for purposes ofcategorical labeling in favor of assessment thatwould eventuate in educational and remedial activ-ities tailored to the child’s individual needs. It wassuggested that a �first cut� approach to suchassessment should entail well-balanced and indi-vidualized remedial intervention that would buildupon the child’s existing knowledge base. The evi-dence suggests that a child’s response to this type ofintervention would provide guidance as to his orher long-term instructional needs, regardless of theorigin of his or her reading difficulties.

Acknowledgments

Much of the work discussed in this paper was sup-ported by grants from the National Institute of ChildHealth and Human Development to the co-authors.The data for the intervention study reported inVellutino et al. (1996) were collected as part of aproject conducted under the auspices of a specialCenter grant (P50HD25806) awarded to the KennedyKrieger Institute by NICHD. Martha Bridge Dencklawas the principal investigator overseeing the variousresearch projects initiated under this grant. Severalgrants involving Fletcher supported this review,including P50 HD21888, R01 HD35938, and NSF9979968.

3 These assertions in no way discount the possibility that

reading difficulties could conceivably arise because of deficits

in general learning abilities such as selective attention, asso-

ciative learning, and cross-modal transfer. Indeed, several

studies have demonstrated that some children who have dif-

ficulties learning to read suffer from both phonological and

attention deficits (e.g., Fletcher, Shaywitz, & Shaywitz, 1999),

and such co-morbidity would be expected in children who have

multiple handicaps. These appear largely a matter of the child

having two problems as opposed to the hypothesis that

the reading difficulties are a probable consequence of such

co-morbidity.

Specific reading disability (dyslexia) 31

Page 31: Vellutino

Correspondence to

Frank Vellutino, Child Research and Study Center,Department of Educational Psychology and Statis-tics, State University of New York at Albany, Albany,New York 12222, USA; Email: [email protected]

ReferencesAbbott, S.P., Reed, E., Abbott, R.D., & Berninger, V.W.(1997). Yearlong balanced reading/writing tutorial: Adesign experiment used for dynamic assessment.Learning Disability Quarterly, 20, 249–263.

Ackerman, P.T., Weir, N.L., Holloway, C.A., & Dykman,R.A. (1995). Adolescents earlier diagnosed as dyslexicshow major IQ declines on the WISC-III. Reading andWriting, 7, 163–170.

Adams, M.J. (1990). Beginning to read: Thinking andlearning about print. Cambridge, MA: MIT Press.

Adlard, A., & Hazan, V. (1997). Speech perception inchildren with specific reading difficulties (dyslexia).Quarterly Journal of Experimental Psychology, 51,152–177.

Badcock, D., & Lovegrove, W. (1981). The effects ofcontrast stimulus, duration and spatial frequency invisual persistence in normal and specifically disabledreaders. Journal of Experimental Psychology: HumanPerception and Performance, 1, 495–505.

Baddeley, A.D. (1986). Working memory. London:Oxford University Press.

Bakker, D.J. (1972). Temporal order in disturbed read-ing – developmental and neuropsychological aspectsin normal and reading-retarded children. Rotterdam:Rotterdam University Press.

Benton, A.L., & Pearl, D. (Eds.). (1978). Dyslexia: Anappraisal of current knowledge. New York: Oxford.

Best, C.T., & Avery, R.A. (1999). Left hemisphereadvantage for click consonants is determined bylinguistic significance and experience. PsychologicalScience, 10, 65–70.

Birch, H.G. (1962). Dyslexia and maturation of visualfunction. In J. Money (Ed.), Reading disability: Pro-gress and research needs in dyslexia (pp. 161–169).Baltimore: Johns Hopkins Press.

Birch, H.G., & Belmont, L. (1964). Auditory–visualintegration in normal and retarded readers. AmericanJournal of Orthopsychiatry, 34, 852–861.

Birch, H.G., & Lefford, A. (1963). Intersensory develop-ment in children. Monographs of the Society forResearch in Child Development, 28, 5 (Whole No. 89).

Bishop, D.V.M., Carlyon, R.P., Deeks, J.M., & Bishop,S.J. (1999).Auditory temporal processing impairment:Neither necessary nor sufficient for causing languageimpairment in children. Journal of Speech, Language,and Hearing Research, 42, 1295–1310.

Blachman, B.A. (1994). Early literacy acquisition – therole of phonological awareness. In G. Wallach & K.Butler (Eds.), Language learning disabilities in school-age children and adolescents: Some underlying prin-ciples and applications (pp. 253–274). Columbus,OH: Merrill.

Blachman, B.A. (Ed.). (1997). Foundations of readingacquisition and dyslexia. Mahwah, NJ: LawrenceErlbaum.

Blachman, B.A. (2000). Phonological awareness. InM.L. Kamil, P.B. Mosenthal, P.D. Pearson, & R. Barr(Eds.), Handbook of reading research (vol. III, pp.483–502). Mahwah, NJ: Lawrence Erlbaum.

Bowers, P.G. (1995). Tracing symbol naming speed’sunique contributions to reading disabilities over time.Reading and Writing: An Interdisciplinary Journal, 7,189–216.

Bowers, G., & Wolf, M. (1993). Theoretical links amongnaming speed, precise timing mechanisms, andorthographic skill in dyslexia. Reading and Writing,5, 69–85.

Bradley, L., & Bryant, P.E. (1983). Categorizing soundsand learning to read – a causal connection. Nature,303, 419–421.

Bradlow, A.R., Kraus, N., Nicol, T., McGee, T.J., Cun-ningham, J., Zecker, S.G., & Carrell, T.D. (1999).Effects of lengthened formant transition duration ondiscrimination and neural representation of syntheticCV syllables by normal and learning-disabled chil-dren. Journal of the Acoustical Society of America,106, 2086–2096.

Brady, S.A., Shankweiler, D., & Mann, V. (1983).Speech perception and memory coding in relation toreading ability. Journal of Experimental Child Psy-chology, 35, 345–367.

Brandt, J., & Rosen, J.J. (1980). Auditory phonemicperception in dyslexia: Categorical identification anddiscrimination of stop consonants. Brain and Lan-guage, 9, 324–337.

Breier, J.I., Fletcher, J.M., Foorman, B.R., & Gray,L.C. (2002). Perception of speech and nonspeechstimuli by children with and without readingdisability and attention deficit hyperactivity disor-der. Journal of Experimental Child Psychology, 82,226–250.

Breitmeyer, B.G. (1989). A visual based deficit inspecific reading disability. Irish Journal of Psycho-logy, 10, 534–541.

Brewer, W.F. (1967). Paired-associate learning of dys-lexic children. Unpublished doctoral dissertation,University of Iowa.

Bruck, M. (1990). Word recognition skills of adults withchildhood diagnoses of dyslexia. Developmental Psy-chology, 26, 439–454.

Bruck, M. (1992). Persistence of dyslexics� phonologicalawareness deficits. Development Psychology, 28,874–886.

Bruck, M. (1993). Component spelling skills of collegestudents with childhood diagnoses of dyslexia. Learn-ing Disabilities Quarterly, 16, 171–184.

Bryant, P.E. (1974). Perception and understanding inyoung children. New York: Basic Books Inc.

Byrne, B., Fielding-Barnsley, R., Ashley, L., & Larsen,K. (1997). Assessing the child’s contribution to read-ing acquisition: What we know and what we don’tknow. In B. Blachman (Ed.), Cognitive and linguisticfoundations of reading acquisition. Hillsdale, NJ:Lawrence Erlbaum Associates.

Caravolas, M. (2003). Similarities and differencesbetween English and French poor spellers. InN. Goulandris (Ed.), Dyslexia in different languages.London: Whurr.

Catts, H.F., Hogan, T.P., & Fey, M. (2003). Subgroupingpoor readers on the basis of individual differences in

32 Frank R. Vellutino et al.

Page 32: Vellutino

reading-related abilities. Journal of Learning Disabil-ities, 36, 151–164.

Clay, M.M. (1985). The early detection of reading difficul-ties (3rd edn). Auckland, New Zealand: Heinemann.

Clay,M.M. (1987). Learning to be learning disabled.NewZealand Journal of Educational Studies, 22, 155–173.

Compton, D.L., DeFries, & Olson, R.K. (2001). Are RANand phonological awareness deficits additive inchildren with reading disabilities? Dyslexia, 7, 125–149

Cornelissen, P.L., Hansen, P.C., Hutton, J.L., Evangel-inou, V., & Stein, J.F. (1997). Magnocellular visualfunction and children’s single word reading. VisionResearch, 38, 471–482.

Cossu, G. (1999). Biological constraints on literacyacquisition. Reading and Writing, 11, 213–237.

Critchley, M. (1970). The dyslexic child. Springfield, IL:Charles C. Thomas.

Curtis, M.E. (1980). Development of components ofreading skill. Journal of Educational Psychology, 72,656–669.

Daneman, M., & Carpenter, P.A. (1980). Individualdifferences in working memory and reading. Journalof Verbal Learning and Verbal Behavior, 19, 450–466.

de Jong, P.F., & van der Leij, A. (1999). Specificcontributions of phonological abilities to early readingacquisition: Results from a Dutch latent variablelongitudinal study. Journal of Educational Psycho-logy, 91, 450–476.

Dickinson, D.K., & Tabors, P.O. (Eds.). (2001). Buildingliteracy with language: Young children learning athome and school. Baltimore, MD: Brookes.

Douglas, V.I. (1972). Stop, look and listen: The problemof sustained attention and impulse control in hyper-active and normal children. Canadian Journal ofBehavioral Science, 4, 259–281.

Duara, R., Kushch, A., Gross-Glenn, K., Barker, W.,Jallad, B., Pascal, S., Loewenstein, D.A., Sheldon, J.,Rabin, M., Levin, B., & Lubs, H. (1991). Neuroana-tomic differences between dyslexic and normal read-ers on magnetic resonance imaging scans. Archives ofNeurology, 48, 410–416.

Dykman, R.A., Ackerman, P.T., Clements, S.D., &Peters, J.E. (1971). Specific learning disabilities: Anattentional deficit syndrome. In H.R. Myklebust (Ed.),Progress in learning disabilities (vol. II, pp. 56–93).New York: Grune & Stratton.

Eden, G.F., Stern, J.F., Wood, M.H., & Wood, F.B.(1995). Verbal and visual problems in dyslexia.Journal of Learning Disabilities, 28, 272–290.

Eden, G.F., Van Meter, J.W., Rumsey, J.M., Maisog,J.M., Woods, R.P., & Zeffiro, T.A. (1996). Abnormalprocessing of visual motion in dyslexia revealed byfunctional brain imaging. Nature, 382, 66–69.

Eden, G.F., & Zeffiro, T.A. (1998). Neural systemsaffected in developmental dyslexia revealed by func-tional neuroimaging. Neuron, 21, 279–282.

Ehri, L. (1999). Phases of development in learning toread words. In J. Oakhill & R. Beard (Eds.), Readingdevelopment and the teaching of reading: A psycholo-gical perspective (pp. 79–108). Oxford, UK: BlackwellPublishers.

Elbro, C. (1997). Early linguistic abilities and readingdevelopment: A review and a hypothesis aboutunderlying differences in distinctiveness of phonolo-

gical representations of lexical items. Reading andWriting, 8, 453–485

Elbro, C., Borstrom, I., & Petersen, D.K. (1998).Predicting dyslexia from kindergarten: The impor-tance of distinctness of phonological representationsof lexical items. Reading Research Quarterly, 33,36–60.

Ericsson, K.A., & Kintsch, W. (1995). Long-term work-ing memory. Psychological Review, 102, 211–245.

Farmer, M.E., & Klein, R. (1995). The evidence for atemporal processing deficit linked to dyslexia: Areview. Psychonomic Bulletin and Review, 2, 460–493.

Filipek, P. (1996). Structural variations in measures inthe developmental disorders. In R. Thatcher, G. Lyon,J. Rumsey, & N. Krasnegor (Eds.), Developmentalneuroimaging: Mapping the development of brain andbehavior (pp. 169–186). San Diego, CA: AcademicPress .

Fish, M.C., & Margolis, H. (1988). Training and practiceof school psychologists reading assessment and inter-vention. Journal of School Psychology, 26, 399–404.

Fisher, S.E., & DeFries, J.C. (2002). Developmentaldyslexia: Genetic dissection of a complex cognitivetrait. Neuroscience, 3, 767–780.

Fletcher J.M., Foorman, B.R., Shaywitz, S.E., & Shay-witz, B.A. (1999). Conceptual and methodologicalissues in dyslexia research: A lesson for develop-mental disorders. In H. Tager-Flugsberg (Ed.), Neuro-developmental disorders (pp. 271–306). Cambridge,MA: MIT Press.

Fletcher, J.M., Lyon, G.R., Barnes, M., Stuebing, K.K.,Francis, D.J., Olson, R.K., Shaywitz, S.E., & Shay-witz, B.A. (2002). Classification of learning disabil-ities: An evidenced-based evaluation. In R. Bradley,L. Danielson, & D. Hallahan (Eds.), Identification oflearning disabilities: Research to practice (pp. 185–250). Mahwah, NJ: Lawrence Erlbaum.

Fletcher, J.M., Satz, P., & Scholes, R.J. (1981). Devel-opmental change in the linguistic correlates of read-ing achievement. Brain and Language, 13, 78–90.

Fletcher J.M., Shaywitz, S.E., Shankweiler, D.P., Katz,L., Liberman, I.Y., Stuebing, K.K., Fletcher, J.M.,Shaywitz, S.E., & Shaywitz, B.A. (1999). Co-morbidityof learning and attention disorders: Separate butequal. Pediatric Clinics of North America, 46, 885–897.

Fletcher J.M., Shaywitz, S.E., Shankweiler, D.P., Katz,L., Liberman, I.Y., Stuebing, K.K., Francis, D.J.,Fowler, A.E., & Shaywitz, B.A. (1994). Cognitiveprofiles of reading disability: Comparisons of discre-pancy and low achievement definitions. Journal ofEducational Psychology, 86, 6–23.

Foorman, B.R., Francis, D.J., Fletcher, J.M.,Schatschneider, C., & Mehta, P. (1998). The role ofinstruction in learning to read: Preventing readingfailure in at-risk children. Journal of EducationalPsychology, 90, 37–55.

Foorman, B.R., Francis, D.J., Novy, D.M., & Liberman,D. (1991). How letter-sound instruction mediatesprogress in first-grade reading and spelling. Journalof Educational Psychology, 83, 456–469.

Foorman, B.R., Francis, D.R., Shaywitz, S.E., Shaywitz,B.A.,&Fletcher, J.M. (1997). The case for early readingintervention. In B.A. Blachman (Ed.), Foundations ofreading acquisition and dyslexia. Implications for earlyintervention (pp. 243–264). Mahwah, NJ: Erlbaum.

Specific reading disability (dyslexia) 33

Page 33: Vellutino

Frith, U., Wimmer, H., & Landerl, K. (1998). Differencesin phonological recoding in German and Englishspeaking children. Scientific Studies of Reading, 2,31–54.

Frost, R. (1998). Toward a strong phonological theory ofvisual word recognition: True issues and false trails.Psychological Bulletin, 123, 71–99.

Galaburda, A.M., Sherman, G.P., Rosen, G.D., Aboitiz,F., & Geschwind, N. (1985). Developmental dyslexia:Four consecutive patients with cortical anomalies.Annals of Neurology, 18, 222–233.

Gallagher, A., Frith, U., & Snowling, M.J. (2000).Precursors of literacy-delay among children at geneticrisk of dyslexia. Journal of Child Psychology andPsychiatry, 41, 203–213.

Gascon, G., & Goodglass, H. (1970) Reading retardationand the information content of stimuli in pairedassociate learning. Cortex, 6, 417–429.

Gathercole, S.E., & Baddeley, A.D. (1990). Phonologicalmemory deficits in language disordered children: Isthere a connection? Journal of Memory and Lan-guage, 29, 336–360.

Geschwind, N., & Levitsky, W. (1968). Human brain:Left–right asymmetries in temporal speech region.Science, 161, 186–187.

Getman, G.N. (1985). A commentary on vision. Journalof Learning Disabilities, 18, 505–511.

Gibson, E.J. (1969). Principles of perceptual learningand development. New York: Appleton-Century-Crofts.

Gilger, J.W. (2002). Current issues in the neurology andgenetics of learning-related traits and disorders:Introduction to the special issue. Journal of LearningDisabilities, 34, 490–491.

Godfrey, J.J., Syral-Lasky, A.K., Millay, K., & Knox,C.M. (1981). Performance of dyslexic children onspeech perception tests. Journal of ExperimentalChild Psychology, 32, 401–424.

Goswami, U. (2001). Early phonological developmentand the acquisition of literacy. In S.B. Neuman & D.K.Dickinson (Eds.), Handbook of early literacy research(pp. 111–125). New York, NY: The Guilford Press.

Gough, P.B. (1972). One second of reading. In J.F.Kavanagh & I.G. Mattingly (Eds.), Language by earand by eye (pp. 331–358). Cambridge: MIT Press.

Gough,P.B. (1984).Wordrecognition. InP.D.Pearson,R.Barr, M.L. Kamil, & P. Mosenthal (Eds.), Handbook ofreading research (pp. 225–253). New York: Longman.

Gough, P.B., & Tunmer, W.E. (1986). Decoding, read-ing, and reading disability. Remedial and SpecialEducation, 7, 6–10.

Goulandris, N. (2003). Dyslexia in different languages:Cross-linguistic comparisons. London: Whurr Pub-lishers.

Griffiths, Y.M., & Snowling, M.J. (2001). Auditory wordidentification and phonological skills in dyslexic andaverage readers. Applied Psycholinguistics, 22, 419–439.

Griffiths, Y.M., & Snowling, M.J. (2002). Predictors ofexception word and nonword reading in dyslexicchildren: The severity hypothesis. Journal of Educa-tional Psychology, 94, 34–43.

Grigorenko, E.L. (2001). Developmental dyslexia: Anupdate on genes, brains, and environment. Journal ofChild Psychology and Psychiatry, 42, 91–125.

Grigorenko, E.L., Wood, F.B., Meyer, M.S., Hart, L.A.,Speed, W.C., Shuster, A., & Pauls, D.L. (1997).Susceptibility loci for distinct components of devel-opmental dyslexia on chromosome 6 and 15. Amer-ican Journal of Human Genetics, 60, 27–39.

Harris, M., & Hatano, G. (Eds.). (1999). Learning to readand write: A cross-linguistic perspective. Cambridge:Cambridge University Press.

Harris, A.S., & Sipay, E.R. (1990). How to increasereading ability (9th edn). New York: Longman.

Hatcher, J., Snowling, M.J., & Griffiths, Y.M. (2002).Cognitive assessment of dyslexic students in highereducation. British Journal of Educational Psychology,72, 119–133.

Hatcher, P., Hulme, C., & Ellis, A.W. (1994). Amelior-ating early reading failure by integrating theteaching of reading and phonological skills: Thephonological linkage hypothesis. Child Development,65, 41–57.

Heath, S.M., Hogben, J.H., & Clark, C.D. (1999).Auditory temporal processing in disabled readerswith and without oral language delay. Journal ofChild Psychology and Psychiatry, 40, 637–647.

Herman, K. (1959). Reading disability. Copenhagen:Munksgaard.

Hinshelwood, J. (1917). Congenital word blindness.London: H.K. Lewis.

Ho, C.S.H., & Bryant, P. (1997). Phonological skills areimportant in learning to read Chinese. DevelopmentalPsychology, 33, 946–951.

Ho, C.S.H., Chan, D.W.O., Tsang, S.M., & Lee, S.H.(2002). The cognitive profile and multiple-deficithypothesis in Chinese developmental dyslexia. Devel-opmental Psychology, 38, 543–553.

Hoover, W. & Gough, P.B. (1990). The simple view ofreading. Reading and Writing: An InterdisciplinaryJournal, 2, 127–160.

Horwitz, B., Rumsey, J.M., & Donohue, B.C. (1998).Functional connectivity of the angular gyrus innormal reading and dyslexia. Proceedings of theNational Academy of Science, 95, 8939–8944.

Hoskyn, M., & Swanson, H.L. (2000). Cognitive proces-sing of low achievers and children with readingdisabilities: A selective meta-analytic review of thepublished literature. The School Psychology Review,29, 102–119.

Hulme, C. (1988). The implausibility of low-level visualdeficits as a cause of children’s reading difficulties.Cognitive Neuropsychology, 5, 369–374.

Hurford, D.P., & Sanders, R.E. (1990). Assessment andremediation of a phoneme discrimination deficit inreading disabled second and fourth graders. Journalof Experimental Child Psychology and Psychiatry, 50,396–415.

Hynd, G.W., & Semrud-Clikman, M. (1989). Dyslexiaand brain morphology. Psychological Bulletin, 106,447–482.

Hynd, G.W., Semrud-Clikeman, M., Lorys, A.R., No-vey, E.S., & Eliopulos, D. (1990). Brain morphologyin developmental dyslexia and attention deficit/hyperactivity disorder. Archives of Neurology, 47,919–926.

Humphreys, P., Kaufmann, W.E., & Galaburda, A.M.(1990). Developmental dyslexia in women. Annals ofNeurology, 28, 727–738.

34 Frank R. Vellutino et al.

Page 34: Vellutino

Hunter, J.V., & Wang, Z.J. (2001). MR spectroscopy inpediatric neuroradiology. MRI Clinics of North Amer-ica, 9, 165–189.

Iversen, S., & Tunmer, W. (1993). Phonological proces-sing skills and the reading recovery program. Journalof Educational Psychology, 85, 112–126.

Jernigan, T.L., Hesselink, J.R., Sowell, E., & Tallal, P.(1991). Cerebral structure on magnetic resonanceimaging in language and learning-impaired children.Archives of Neurology, 48, 539–545.

Joseph, J., Noble, K., & Eden, G. (2001). The neuro-biological basis of reading. Journal of LearningDisabilities, 34, 566–579.

Kail, R., Hall, L.K., & Caskey, B.J. (1999). Processingspeed, exposure to print, and naming speed. AppliedPsycholinguistics, 20, 303–314.

Katz, R.B. (1986). Phonological deficiencies in childrenwith reading disability: Evidence from an objectnaming task. Cognition, 22, 225–257.

Katz, R.B., Healy, A.F., & Shankweiler, D. (1983).Phonetic coding and order memory in relation toreading proficiency: A comparison of short-termmemory for temporal and spatial order information.Applied Psycholinguistics, 4, 229–250.

Katz, R.B., Shankweiler, D., & Liberman, I.Y. (1981).Memory for item order and phonetic recoding in thebeginning reader. Journal of Experimental ChildPsychology, 32, 474–484.

Klingberg, T., Hedehus, M., Temple, E., Salz, T.,Gabrieli, J., Moseley, M., & Poldrack, R. (2000).Microstructure of temporo-parietal white matter asa basis for reading ability: Evidence from diffusiontensor magnetic resonance imaging. Neuron, 25,493–500.

Kushch, A., Gross-Glenn, K., Jallad, B., Lubs, H.,Rabin, M., Feldman, E., & Duara, R. (1993). Temporallobe surface area measurements on MRI in normaland dyslexic readers. Neuropsychologia, 31, 811–821.

LaBerge, D., & Samuels, J. (1974). Toward a theory ofautomatic information processing in reading. Cognit-ive Psychology, 6, 293–323.

Larsen, J.P., Høien, T., Lundberg, I., & Odegarrd, H.(1990). MRI evaluation of the size and symmetry ofthe planum temporale in adolescents with develop-mental dyslexia. Brain and Language, 39, 289–301.

Lefly, D.L., & Pennington, B.F. (1996). Longitudinalstudy of children at high family risk for dyslexia: Thefirst two years. In M.L. Rice (Ed.), Toward a genetics oflanguage. Hillsdale, NJ: Lawrence Erlbaum.

Lehmkuhle, S., Garzia, R.P., Turner, L., Hash, T, &Baro, J.A. (1993). A defective visual pathway inchildren with reading disability. New England Journalof Medicine, 328, 989–996.

Leonard, C.M., Eckert, M.A., Lombardino, L.J., Oak-land, T., Franzier, J., Mohr, C.M., King, W.M., &Freeman, A. (2001). Anatomical risk factors forphonological dyslexia. Cerebral Cortex, 11, 148–157.

Leonard, C.M., Lombardino, L.J., Mercado, L.R.,Browd, S.R., Breier, J.I., & Agee, O.F. (1996).Cerebral asymmetry and cognitive development inchildren: A magnetic resonance imaging study. Psy-chological Science, 7, 89–95.

Liberman, I.Y., & Shankweiler, D. (1979). Speech, thealphabet and teaching to read. In L. Resnick & P.

Weaver (Eds.), Theory and practice of early reading(vol. II, pp. 109–132). Hillsdale, NJ: Erlbaum.

Liberman, I.Y., & Shankweiler, D. (1991). Phonologyand beginning reading – a tutorial. In L. Rieben & C.Perfetti (Eds.), Learning to read: Basic research andits implications (pp. 3–17). Hillsdale, NJ: Erlbaum.

Liberman, I.Y., Shankweiler, D., Fischer, F.W., &Carter, B. (1974). Explicit syllable and phonemesegmentation in the young child. Journal of Experi-mental Child Psychology, 18, 201–212.

Livingstone, M.S., Rosen, G.D., Drislane, F.W., &Galaburda, A.M. (1991). Physiological and anatomi-cal evidence for a magnocellular defect in develop-mental dyslexia. Proceedings of the National Academyof Science, 88, 7943–7947.

Locke, J. L., Hodgson, J., Macaruso, P., Roberts, J.,Lambrecht-Smith, S., & Guttentag, C. (1997). Thedevelopment of developmental dyslexia. In C. Hulme& M. Snowling (Eds.), Dyslexia: Biology, cognition andintervention. London: Whurr.

Lovegrove, W., Garzia, R.P., & Nicholson, S.B. (1990)Experimental evidence for a transient system deficitin specific reading disability. Journal of the AmericanOptometric Association, 61, 137–146.

Lovegrove, W., Martin, F., & Slaghuis, W. (1986). Atheoretical and experimental case for a visual deficitin specific reading disability. Cognitive Neuropsycho-logy, 3, 225–267.

Lundberg, I., Frost, J., & Petersen, O.P. (1988). Effectsof an extensive program for stimulating phonologicalawareness in preschool children. Reading ResearchQuarterly, 23, 263–285.

Lyon, G.R. (1995). Toward a definition of dyslexia.Annals of Dyslexia, 45, 3–27.

Lyon, G.R., Fletcher, J.M., & Barnes, M. (2002).Learning disabilities. In E.J. Marsh & R.A. Barkley(Eds.), Child psychopathology (2nd edn, pp. 520–588). New York: Guilford.

Lyon, G.R., Fletcher, J.M., Shaywitz, S.E., Shaywitz,B.A., Torgesen, J.K., Wood, F.B., Schulte, A., &Olson, R. (2001). Rethinking learning disabilities. InC.E. Finn, R.A.J. Rotherham, & C.R. Hokanson(Eds.). Rethinking special education for a new century(pp. 259–287). Washington, DC: Thomas B. FordhamFoundation and Progressive Policy Institute.

Lyon, G.R., & Moats, L.C. (1988). Critical issues in theinstruction of the learning disabled. Journal ofConsulting and Clinical Psychology, 56, 830–835.

Lyon, G.R., & Moats, L.C. (1993). An examination ofresearch in learning disabilities: Past practices andfuture directions. In G.R. Lyon, D.B. Gray, J.F.Kavanagh, & N.A. Krasnegor (Eds.), Better under-standing learning disabilities: New views from re-search and their implications for education and publicpolicies (pp. 1–14). Baltimore, MD: Brookes.

Lyytinen, P., Poikkeus, A.M., Laakso, M.L., Eklund, K.,& Lyytinen, H.(2001). Language development andsymbolic play in children with and without familialrisk for dyslexia. Journal of Speech, Language, andHearing Research, 44, 873–885.

Manis, F.R., Custodio, R., & Szeszulski, P.A. (1993).Development of phonological and orthographicskills: A 2-year longitudinal study of dyslexicchildren. Journal of Experimental Child Psychology,56, 64–86.

Specific reading disability (dyslexia) 35

Page 35: Vellutino

Manis, F.R., Doi, L.M., & Bhadha, B. (2000). Namingspeed, phonological awareness, and orthographicknowledge in second graders. Journal of LearningDisabilities, 33, 325–333.

Manis, F.R., McBride-Chang, C., Seidenberg, M.S.,Keating, P., Doi, L.M., Munson, B., & Petersen, A.(1997). Are speech perception deficits associated withdevelopmental dyslexia? Journal of ExperimentalChild Psychology, 66, 211–235.

Martin, F., & Lovegrove, W. (1984). The effects of fieldsize and luminance on contrast sensitivity differencesbetween specifically reading disabled and normalchildren. Neuropsychologia, 22, 73–77.

McBride-Chang, C. (1995). What is phonological aware-ness? Journal ofEducational Psychology,87, 179–192.

McNally, K.I., Hansen, P.C., Cornelissen, P.L., & Stein,J.F. (1997). Effect of time and frequency manipula-tion on syllable perception in developmental dys-lexics. Journal of Speech, Language, and HearingResearch, 40, 912–924.

Meltzoff, A.N., & Kuhl, P.K. (1994). Faces and speech:Intermodal processing of biologically relevant signalsin infants and adults. In D.J. Lewkowicz & R. Lickliter(Eds.), The development of intersensory perception:Comparative perspectives (pp. 335–369). Hillsdale,NJ: Erlbaum.

Metsala, J.L. (1997). Spoken word recognition in read-ing disabled children. Journal of Educational Psycho-logy, 89, 159–169.

Metsala, J.L. (1999). Young children’s phonologicalawareness and nonword repetition as a function ofvocabulary development. Journal of Educational Psy-chology, 91, 3–19.

Metsala, J.L., & Walley, A.C. (1998). Spoken vocabularygrowth and the segmental restructuring of lexicalrepresentations: Precursors to phonemic awarenessand early reading ability. In J.L. Metsala & L.C. Ehri(Eds.), Word recognition in beginning literacy (pp. 89–120). Hillsdale, NJ: Erlbaum.

Mody, M., Studdert-Kennedy, M., & Brady, S. (1997).Speech perception deficits in poor readers: Auditoryprocessing or phonological coding? Journal of Experi-mental Child Psychology, 64, 199–231.

Morgan, W.P. (1896). A case of congenital word blind-ness. British Medical Journal, 11, 378.

Morris, R.D., Steubing, K.K., Fletcher, J.M., Shaywitz,S.E., Lyon, G.R., Shankweiler, D.P., Katz, L., Francis,D.J., & Shaywitz, B.A. (1998). Subtypes of readingdisability: Variability around a phonological core.Journal of Educational Psychology, 90, 347–373.

Morrison, F.J., & Manis, F.R. (1982). Cognitive pro-cesses and reading disability: A critique and propo-sal. In C.J. Brainerd & M. Pressley (Eds.), Verbalprocesses in children (pp. 59–93). New York: Springer-Verlag.

Nation, K., & Snowling, M.J. (1998). Individual differ-ences in contextual facilitation: Evidence from dys-lexia and poor reading comprehension. ChildDevelopment, 69, 996–1011.

Nittrouer, S. (1999). Do temporal processing deficitscause phonological processing problems? Journal ofSpeech, Language, and Hearing Research, 42, 925–942.

Olson, R.K., & Gayan, J. (2001). Brains, genes, andenvironment in reading development. In S.B. Neu-

man & D.K. Dickinson (Eds.), Handbook of earlyliteracy research (pp. 81–94). New York,NY: GuilfordPress.

Olson, R.K., Forsberg, H., Gayan, J., & DeFries, J.C.(1999). A behavioral-genetic analysis of readingdisabilities and component processes. In R.M. Klein& P.A. McMullen (Eds.), Converging methods forunderstanding reading and dyslexia (pp. 133–153).Cambridge, MA: MIT Press.

Olson, R.K., Forsberg, H., Wise, B., & Rack, J. (1994).Measurement of word recognition, orthographic, andphonological skills. In G.R. Lyon (Ed.), Frames ofreference for the assessment of learning disabilities:-New views on measurement issues (pp. 243–277).Baltimore, MD: Brookes.

Olson, R.K., Kleigl, R., & Davidson, B.J. (1983).Dyslexic and normal readers� eye movements. Journalof Experimental Psychology: Human Perception andPerformance, 9, 816–825.

Olson, R.K., Wise, B., & Ring, J. (1999). Trainingphonological awareness with and without explicitattention to articulation. Journal of ExperimentalChild Psychology, 72, 271–304.

Orton, S.T. (1925). �Word-blindness� in school chil-dren. Archives of Neurology and Psychiatry, 14, 581–615.

Papanicolaou, A.C. (1998). Fundamentals of functionalbrain imaging. Lisse, The Netherlands: Swets &Zeitlinger.

Paulesu, E., Demonet, J.F., Fazio, F., McCrory, E.,Chanoine, V., Brunswick, N., Cappa, S.F., Cossu, G.,Habib, M., Frith, C.D., & Frith, U. (2001). Dyslexia;cultural diversity and biological unity. Science, 291,2165–2167.

Pennington, B.F. (1999). Dyslexia as a neurodevelop-mental disorder. In H. Tager-Flusberg (Ed.), Neuro-developmental disorders (pp. 307–330). Cambridge,MA: MIT Press.

Pennington, B.F., Filipek, P.A., Churchwell, J., Ken-nedy, D.N., Lefley, D., Simon, J.H., Filley, C.M.Galaburda, A., Alarcon, M., & DeFries, J.C. (1999).Brain morphometry in reading disabled twins. Neuro-logy, 53, 723–729.

Pennington, B.F., & Lefly, D.L. (2001). Early readingdevelopment in children at family risk for dyslexia.Child Development, 72, 816–833.

Pennington, B.F., Orden, G.C.V., Smith, S.D., Green,P.A., & Haith, M.M. (1990). Phonological processingskills and deficits in adult dyslexics. Child Develop-ment, 61, 1753–1778.

Perfetti, C.A. (1985). Reading ability. New York: OxfordUniversity Press.

Perfetti, C.A., & Roth, S.F. (1981). Some of the inter-active processes in reading and their role in readingskill. In A.M. Lesgold & C.A. Perfetti (Eds.), Interactiveprocesses in reading. Hillsdale, NJ: Erlbaum.

Pinnell, G.S. (1989). Reading recovery: Helping at riskchildren learn to read. Elementary School Journal, 90,161–184.

Poeppel, D. (1996). A critical review of PET studies ofphonological processing. Brain and Language, 55,317–351.

Pugh, K.R., Mencl, W.E., Shaywitz, B.A., Shaywitz, S.E.,Fulbright, R.K., Skudlarski, P., Constable, R.T.,Marchione, K.E., Jenner, A.R., Fletcher, J.M.,

36 Frank R. Vellutino et al.

Page 36: Vellutino

Liberman, A.M., Shankweiler, D.P., Katz, L., Lacadie,C., & Gore, J.C. (2000). The angular gyrus indevelopmental dyslexia: Task specific differences infunctional connectivity in posterior cortex. Psycholo-gical Science, 11, 51–56.

Ramus, F., Rosen, S., Dakin, S.C., Day, B.L., Castellote,J.M., White, S., & Frith, U. (2003). Theories ofdevelopmental dyslexia: Insights from a multiple casestudy of dyslexic adults. Brain, 126, 1–25.

Raymond, J.E., & Sorensen, R.E. (1998). Visual motionperception in children with dyslexia: Normal detec-tion but abnormal integration. Visual Cognition, 5,389–404.

Read, C., Zhang, Y., Nie, H., & Ding, B. (1986). Theability to manipulate speech sounds depends onknowing alphabetic writing. Cognition, 55, 151–218.

Reed, M.A. (1989). Speech perception and the discrim-ination of brief auditory cues in reading disabledchildren. Journal of Experimental Child Psychology,48, 270–292.

Reschly, D.J., Tilly, W.D., & Grimes, J.P. (1999). Specialeducation in transition: Functional assessment andnoncategorical programming. Longmont, CO: SoprisWest.

Richards, T.L., Cornia, D., Serafini, S., Steury, K.,Echelard, D.R., Dager, S.R., Marro, K., Abbott, R.D.,Maravilla, K.R., & Berninger, V.W. (2000). The effectsof a phonologically driven treatment for dyslexia onlactate levels as measures by proton MRSI. AmericanJournal of Neuroradiology, 21, 916–922.

Rumsey, J.M., Nace, K., Donohue, B., Wise, D., Maisog,J.M., & Andreason, P. (1997). A positron emissiontomographic study of impaired word recognition andphonological processing in dyslexic men. Archives ofNeurology, 54, 562–573.

Rutter, M. (1978). Dyslexia. In A.L. Benton & D. Pearl(Eds.), Dyslexia: An appraisal of current knowledge(pp. 3–28) New York: Oxford.

Satz, P., Buka, S., Lipsitt, L., & Seidman, L. (1998). Thelong-term prognosis of learning disabled children: Areview of studies (1954–1993). In B.K. Shapiro, P.J.Accardo, & A.J. Capute (Eds.), Specific readingdisability: A view of the spectrum (pp. 223–250).New York: York Press.

Scanlon, D.M., Vellutino, F.R., Small, S.G., & Fanuele,D.P. (2000). Severe reading difficulties: Can they beprevented? A comparison of prevention and interven-tion approaches. Paper presented at the AnnualConference of The American Educational ResearchAssociation. New Orleans, April.

Scanlon, D.M., & Vellutino, F.R. (1996) Prerequisiteskills, early instruction, and success in first gradereading: Selected results from a longitudinal study.Mental Retardation and Development Disabilities, 2,54–63.

Scarborough, H.S. (1990). Very early language deficitsin dyslexic children. Child Development, 61, 1728–1743.

Scarborough, H.S. (1991). Early syntactic developmentof dyslexic children. Annals of Dyslexia, 41, 207–220.

Schatschneider, C., Carlson, C.D., Francis, D.J., Foor-man, B.R., & Fletcher, J.M. (2002). Relationships ofrapid automatized naming and phonological aware-

ness in early reading development: Implications forthe double deficit hypothesis. Journal of LearningDisabilities, 35, 245–256.

Schultz, R.T., Cho, N.K., Staib, L.H., Kier, L.E., Flet-cher, J.M., Shaywitz, S.E., Shankweiler, D.P., Katz,L., Gore, J.C., Duncan, J.S., & Shaywitz, B.A. (1994).Brain morphology in normal and dyslexic children;the influence of sex and age. Annals of Neurology, 35,732–742.

Shadish, W.R., Cook, T.D., & Campbell, D.T. (2002).Experimental and quasi-experimental designs forgeneral causal inference. Boston, MA: HoughtonMifflin.

Shankweiler, D.P., Liberman, I.Y., Mark, L.S., Fowler,C.A., & Fischer, F.W. (1979). The speech code andlearning to read. Journal of Experimental Psychology:Human Learning and Memory, 5, 531–545.

Shankweiler, D.P., Lundquist, E., Katz, L., Steubing, K.,Fletcher, J., Brady, S., Fowler, A., Dreyer, L., Marchi-one, K., Shaywitz, S., & Shaywitz, B. (1999). Com-prehension and decoding: Patterns of association inchildren with reading difficulties. Scientific Studies ofReading, 3, 69–94.

Share, D.L., McGee, R., & Silva, P.A. (1989). IQ andreading progress: A test of the capacity notion of IQ.Journal of the American Academy of Child andAdolescent Psychiatry, 28, 97–100.

Share, D.L., & Stanovich, K.E. (1995). Cognitive pro-cesses in early reading development: Accommodatingindividual differences into a model of acquisition.Issues in Education: Contributions from EducationalPsychology, 1, 1–57.

Shaywitz, B.A., Holford, T.R., Holohan, J.M., Fletcher,J.M., Francis, D.J., Steubing, K.K., & Shaywitz, S.E.(1995). A Matthew effect for IQ but not for reading:Results from a longitudinal study of reading. ReadingResearch Quarterly, 30, 894–906.

Shaywitz, B.A., Pugh, K.R., Jenner, A.R., Fulbright,R.K., Fletcher, J.M., Gore, J.C., & Shaywitz, S.E.(2000). The neurobiology of reading and readingdisability (dyslexia). In M.L. Kamil, P.B. Mosenthal,P.D. Pearson, & R. Barr (Eds.), Handbook of readingresearch (vol. III, pp. 229–249). Mahwah, NJ: Law-rence Erlbaum.

Shaywitz, B.A., Shaywitz, S.E., Pugh, K.R., Mencl, W.E.,Fulbright, R.K., Constable, R.T., Skudlarski, P.,Jenner, A., Fletcher, J.M., Marchione, K.E., Shank-weiler, D., Katz, L., Lacadie, C., Lyon, G.R., & Gore,J.C. (2002). Disruption of the neural circuitry forreading in children with developmental dyslexia.Biological Psychiatry, 52, 101–110.

Shaywitz, S.E. (1996). Dyslexia. Scientific American,November, 98–104.

Shaywitz, S.E., Escobar, M.D., Shaywitz, B.A., Fletcher,J.M., & Makuch, R.W. (1992). Evidence that dyslexiamay represent the lower tail of a normal distributionof reading ability. New England Journal of Medicine,326, 145–150.

Siegel, L.S. (1988). Evidence that IQ scores are irrele-vant to the Definition and Analysis of ReadingDisability. Canadian Journal of Psychology, 42(2),201–215.

Siegel, L.S. (1989). IQ is irrelevant to the definition oflearning disabilities. Journal of Learning Disabilities,22, 469–478.

Specific reading disability (dyslexia) 37

Page 37: Vellutino

Siegel, L.S., & Ryan, E.B. (1984). Reading disability as alanguage disorder. Remedial and Special Education,5, 28–33.

Siegel, L.S., Share, D.L., & Geva, E. (1995). Evidence forsuperior orthographic skills in dyslexics. Psycholo-gical Science, 6, 250–254.

Simos, P.G., Breier, J.I., Fletcher, J.M., Bergman, E., &Papanicolau, A.C. (2000a). Cerebral mechanismsinvolved in word reading in dyslexic children. Cereb-ral Cortex, 10, 809–816.

Simos, P.G., Breier, J.I., Fletcher, J.M., Foorman, B.R.,Bergman, E., Fishbeck, K., & Papanicolau, A.C.(2000b). Brain activation profiles in dyslexic childrenduring nonword reading: A magnetic source imagingstudy. Neuroscience Reports, 290, 61–65.

Simos, P.G., Fletcher, J.M., Bergman, E., Breier, J.I.,Foorman, B.R., Castillo, E.M., Fitzgerald, M., &Papanicolau, A.C. (2002a). Dyslexia-specific brainactivation profile becomes normal following success-ful remedial training. Neurology, 58, 1203–1213.

Simos, P.G., Fletcher, J.M., Foorman, B.R., Francis,D.J., Castillo, E.M., Davis, M.F., Mathes, P.G., Den-ton, C.A., & Papanicolaou, A.C. (2002b). Brainactivation profiles during the early stages of readingacquisition. Journal of Child Neurology, 17, 159–163.

Skottun, B.C., & Parke, L.A. (1999). The possiblerelationship between visual deficits and dyslexia:Examination of a critical assumption. Journal ofLearning Disabilities, 32, 2–5.

Smith, S.D., Pennington, B.F., Kimberling, W.J., & Ing,P.S. (1990). Familial dyslexia: Use of genetic linkagedata to define subtypes. Journal of the AmericanAcademy of Child and Adolescent Psychiatry, 29,338–348.

Snow, C.E., Barnes, W., Chandler, J. Goodman, I., &Hemphill, L. (1991). Unfulfilled expectations: Homeand school influences on literacy. Cambridge, MA:Harvard.

Snow, C.E., Burns, M., & Griffin, P. (Eds.) (1998).Preventing reading difficulties in young children.Washington, DC: National Academy Press.

Snow, C.E., & Tabors, P.O. (1993). Language skillsthat relate to literacy development. In B. Spodek &O. Saracho (Eds.), Yearbook in early childhoodeducation, 4. New York: Teachers College Press.

Snowling, M.J. (1980). The development of grapheme–phoneme correspondence in normal and dyslexicreaders. Journal of Experimental Child Psychology,29, 294–305.

Snowling, M.J. (2000a). Dyslexia. Oxford: Blackwell.Snowling, M.J. (2000b). Language and literacy skills:Who is at risk and why? In D.V.M. Bishop & L.B.Leonard (Eds.), Speech & language impairments inchildren: Causes, characteristics, intervention andoutcome. Hove, UK: Psychology Press.

Snowling, M.J., Bishop, D.V.M., & Stothard, S.E.(2000). Is pre-school language impairment a riskfactor for dyslexia in adolescence? Journal of ChildPsychology and Psychiatry, 41, 587–600.

Snowling, M.J., Gallagher, A., & Frith, U. (2003). Familyrisk of dyslexia is continuous: Individual differencesin the precursors of reading skill. Child Development,74, 358–373.

Snowling, M.J., Goulandris, N., Bowlby, M., & Howell,P. (1986). Segmentation and speech perception in

relation to reading skill: A developmental analysis.Journal of Experimental Child Psychology, 41, 489–507.

Spreen, O. (1989). Learning disability, neurology, andlong-term outcome: Some implications for the indi-vidual and for society. Journal of Clinical and Experi-mental Neuropsychology, 11, 389–408.

Stanley, G., Smith, G.A., & Howell, E.A. (1983). Eyemovements and sequential tracking in dyslexic andcontrol children. British Journal of Psychology, 74,181–187.

Stanovich, K.E. (1980). Toward an interactive-compen-satory model of individual differences in the develop-ment of reading fluency. Reading Research Quarterly,16, 32–71.

Stanovich, K.E. (1986). Matthew effects in reading:Some consequences of individual differences in theacquisition of literacy. Reading Research Quarterly,21, 360–407.

Stanovich, K.E. (1988). Explaining the differencesbetween the dyslexic and the garden-variety poorreader: The phonological core variable differencemodel. Journal of Learning Disabilities, 21, 590–604.

Stanovich, K.E. (1991). Word recognition: Changingperspectives. In R. Barr, M.L. Kamil, P. Mosenthal, &P.D. Pearson (Eds.), Handbook of reading research(vol. 2, pp. 418–452). New York: Longman.

Stanovich, K.E., & Siegel, L.S. (1994). Phenotypicperformance profile of children with reading disabil-ities: A regression-based test of the phonological-corevariable-difference model. Journal of EducationalPsychology, 86, 24–53.

Stein, J. (2001). The sensory basis of reading problems.Developmental Neuropsychology, 20, 509–534.

Steubing, K.K., Fletcher, J.M., LeDoux, J.M., Lyon,G.R., Shaywitz, S.E., & Shaywitz, B.A. (2002)Validity of IQ–discrepancy classifications of readingdisabilities: A meta-analysis. American EducationalResearch Journal, 39, 469–518. Washington, DC:AERA.

Studdert-Kennedy, M., & Mody, M. (1995). Auditorytemporal perception deficits in the reading impaired:A critical review of the evidence. Psychonomic Bulletinand Review, 2, 508–514.

Tabors, P.O., & Snow, C.E. (2001). Young bilingualchildren and early literacy development. In S.B.Neuman & D.K. Dickinson (Eds.) Handbook of earlyliteracy research (pp. 159–178). New York, NY: TheGuilford Press.

Tallal, P. (1980). Auditory temporal perception, phon-ics, and reading disabilities in children. Brain andLanguage, 9, 182–198.

Tallal, P., Miller, S.L., Jenkins, W.M., & Merzenich,M.M. (1997). The role of temporal processing indevelopmental language-based learning disorders:Research and clinical implications. In B.A. Blachman(Ed.), Foundations of reading acquisition and dyslex-ia: Implications for early intervention (pp. 49–66).Mahwah, NJ: Erlbaum.

Tallal, P., & Percy, M. (1973). Developmental aphasia:Impaired rate of nonverbal processing as a function ofsensory modality. Neuropsychologia, 11, 389–398.

Tallal, P., & Percy, M. (1975). Developmental aphasia:The perception of brief vowels and extended stopconsonants. Neuropsychologia, 13, 69–74.

38 Frank R. Vellutino et al.

Page 38: Vellutino

Tannock, R., Martinussen, R., & Friitjers, J. (2000).Naming speed performance and stimulant effectsindicate effortful, semantic processing deficits inattention-deficit/hyperactivity disorder. Journal ofAbnormal Child Psychology, 28, 237–252.

Taylor, H.G., & Fletcher, J.M. (1983). Biologicalfoundations of specific developmental disorders:Methods, findings, and future directions. Journal ofChild Clinical Psychology, 12, 46–65.

Torgesen, J.K., Alexander, A.W., Wagner, R.K., Rash-otte, C.A., Voeller, K.K.S., & Conway, T. (2001).Intensive remedial instruction for children withsevere reading disabilities: Immediate and long-termoutcomes from two instructional approaches. Journalof Learning Disabilities, 34, 33–58.

Torgesen, J.K., Rose, E., Lindamood, P., Conway, T., &Garvan C. (1999). Preventing reading failure in youngchildren with phonological processing disabilities:Group and individual responses to instruction.Journal of Educational Psychology, 91, 579–594.

Torgesen, J.K., Wagner, R.K., & Rashotte, C.A. (1994).Longitudinal studies of phonological processing andreading. Journal of Learning Disabilities, 27, 276–286.

Torgesen, J.K., Wagner, R.K., Rashotte, C.A., Burgess,S., & Hecht, S. (1997). Contributions of phonologicalawareness and rapid automatic naming ability to thegrowth of word-reading skills in second to fifth gradechildren. Scientific Studies of Reading, 1, 161–185.

Treiman, R. (1997). Spelling in normal children anddyslexics. In B. Blachman (Ed.), Foundations ofreading acquisition and dyslexia: Implications forearly intervention. Mahwah, NJ: Lawrence ErlbaumAssociates.

Tunmer, W.E. (1989). The role of language relatedfactors in reading disability. In D. Shankweiler &I.Y. Liberman (Eds.), Phonology and reading disabil-ity: Solving the reading puzzle (pp. 91–131). AnnArbor, MI: University of Michigan Press.

Tunmer, W.E., & Chapman, J.W. (1998) Languageprediction skill, phonological recoding ability, andbeginning reading. In C.H. Hulme & R.M. Joshi(Eds.), Reading and spelling: Development and dis-order (pp. 33–67). Mahwah, NJ: Erlbaum.

Vellutino, F.R. (1979). Dyslexia: Theory and research.Cambridge, MA: MIT Press.

Vellutino, F.R. (1987). Dyslexia. Scientific American,March, 1, 34–41.

Vellutino, F.R. (1991). Introduction to three studies onreading acquisition: Convergent findings on theoret-ical foundations of code-oriented versus whole-lan-guage approaches to reading instruction. Journal ofEducational Psychology, 83, 437–443.

Vellutino, F.R., & Scanlon, D.M. (1982). Verbal proces-sing in poor and normal readers. In C.J. Brainerd &M. Pressley (Eds.), Verbal processes in children(pp. 189–264). New York: Springer-Verlag.

Vellutino, F.R., & Scanlon, D.M. (1987a). Phonologicalcoding, phonological awareness, and reading ability:Evidence from a longitudinal and experimental study.Merrill-Palmer Quarterly, 33, 321–363.

Vellutino, F.R., & Scanlon, D.M. (1987b). Linguisticcoding and reading ability. In S. Rosenberg (Ed.),Advances in applied psycholinguistics (vol. 2, pp. 1–69). New York: Cambridge University Press.

Vellutino, F.R., Scanlon, D.M., & Chen, R. (1995a). Theincreasingly inextricable relationship between ortho-graphic and phonological coding in learning to read:Some reservations about current methods of oper-ationalizing orthographic coding. In V. Berninger(Ed.), The varieties of orthographic knowledge II:Relationships to phonology reading, and writing(pp. 47–111). Dordrecht: The Netherlands.

Vellutino, F.R., Scanlon, D.M., & Lyon, G.R. (2000).Differentiating between difficult-to-remediate andreadily remediated poor readers: More evidenceagainst the IQ–achievement discrepancy definitionof reading disability. Journal of Learning Disabilities,33, 223–238.

Vellutino, F.R., Scanlon, D.M., Sipay, E.R., Small, S.G.,Pratt, A., Chen, R.S., & Denckla, M.B. (1996).Cognitive profiles of difficult to remediate and readilyremediated poor readers: Early intervention as avehicle for distinguishing between cognitive and ex-periential deficits as basic causes of specific readingdisability. Journal of Educational Psychology, 88,601–638.

Vellutino, F.R., Scanlon, D.M., Small, S.G., & Tanzman,M.S. (1991). The linguistic basis of reading ability:Converting written to oral language. Text, 11, 99–133.

Vellutino, F.R., Scanlon, D.M., & Spearing, D. (1995b).Semantic and phonological coding in poor andnormal readers. Journal of Experimental Child Psy-chology, 59, 76–123.

Vellutino, F.R., Scanlon, D.M., & Tanzman, M.S.(1988). Lexical memory in poor and normal readers:Developmental differences in the use of categorycues. Canadian Journal of Psychology, 42, 216–241.

Vellutino, F.R., Scanlon, D.M., & Tanzman, M.S.(1994). Components of reading ability: Issues andproblems in operationalizing word identification,phonological coding, and orthographic coding. InG.R. Lyon (Ed.), Frames of reference for the assess-ment of learning disabilities: New views on measure-ment issues (pp. 279–324). Baltimore, MD: PaulH. Brookes.

Waber, D.P., Weiler, M.D., Wolff, P.H., Bellinger, D.,Marcus, D.J., Ariel, R., Forbes, P., & Wypij, D. (2001).Processing of rapid auditory stimuli in school-agechildren referred for evaluation of learning disorders.Child Development, 72, 37–49.

Waber, D.P., Wolff, P.H., Forbes, P.W., & Weiler, M.D.(2000). Rapid automatized naming in childrenreferred for evaluation of heterogeneous learningproblems: How specific are naming speed deficits toreading disability? Child Neuropsychology, 6, 251–261.

Wadsworth, S.J., Olson, R.K., Pennington, B.F., &DeFries, J.C. (2000). Differential genetic etiology ofreading disability as a function of IQ. Journal ofLearning Disabilities, 33, 192–199.

Wagner, R.K., & Torgesen, J.K. (1987). The nature ofphonological processing and its causal role in theacquisition of reading skills. Psychological Bulletin,101, 192–212.

Wagner, R.K., Torgesen, J.K., & Rashotte, C.A. (1994).Development of reading-related phonological proces-sing abilities: Evidence of bi-directional causalityfrom a latent variable longitudinal study. Develop-mental Psychology, 30, 73–87.

Specific reading disability (dyslexia) 39

Page 39: Vellutino

Walley, A.C. (1993). The role of vocabulary developmentin children’s spoken word recognition and segmenta-tion ability. Developmental Review, 13, 286–350.

Wasik, B.A., & Slavin, R.R. (1993). Preventing earlyreading failurewithone-to-one tutoring:Areviewoffiveprograms.Reading ResearchQuarterly, 28, 179–200.

Williams, J.P. (1980). Teaching decoding with anemphasis on phoneme analysis and phoneme blend-ing. Journal of Educational Psychology, 72, 1–15.

Wimmer, H., Landerl, K., & Schneider, W. (1994). Therole of rhyme awareness in learning to read a regularorthography. British Journal of Developmental Psy-chology, 12, 469–484.

Wimmer, H., Mayringer, H., & Landerl, K. (1998). Poorreading: A deficit in skill-automatization or a phonolo-gical deficit? Scientific Studies of Reading, 2, 321–340.

Wolf, M., & Bowers, P.G. (1999). The �double deficithypothesis� for the developmental dyslexias. Journalof Educational Psychology, 91, 1–24.

Wolf, M., Bowers, P.G., & Biddle, K. (2000a). Naming-speed processes, timing, and reading: a conceptual

review. Journal of Learning Disabilities, 33, 387–407.

Wolf, M., Miller, L., & Donnelly, K. (2000b). Retrieval,automaticity, vocabulary elaboration, orthography(RAVE-O): A comprehensive, fluency-based readingintervention program. Journal of Learning Disabil-ities, 33, 375–386.

Wood, F.B., & Grigorenko, E.L. (2001). Emergingissues in the genetics of dyslexia: A methodologicalpreview. Journal of Learning Disabilities, 34, 503–512.

Wydell, T.N., & Butterworth, B.L. (1999). A case studyof an English–Japanese bilingual with monolingualdyslexia. Cognition, 70, 273–305.

Ysseldyke, J., & Christensons, S.L. (1988). Linkingassessment to intervention. In J.L. Graden, J.E. Zins,& M.J. Curtis (Eds.), Alternative educational deliverysystems (pp. 91–109). Washington, DC: NationalAssociation of School Psychologists.

Manuscript accepted 18 August 2003

40 Frank R. Vellutino et al.