Relations Amogn Speech Lenguaje and Reading Disorders

8/12/2019 Relations Amogn Speech Lenguaje and Reading Disorders

1/27

Relations Among Speech,Language, and ReadingDisordersBruce F. Pennington 1 and Dorothy V.M. Bishop21Department of Psychology, University of Denver, Colorado 80208;email: [email protected] of Experimental Psychology, University of Oxford, OX1 3UD,United Kingdom; email: [email protected]

Annu. Rev. Psychol. 2009. 60:283306

First published online as a Review in Advance on July 24, 2008

The Annual Review of Psychology is online at psych.annualreviews.org

This articles doi:10.1146/annurev.psych.60.110707.163548

Copyright c 2009 by Annual Reviews.

All rights reserved0066-4308/09/0110-0283$20.00

Key Wordscomorbidity, speech sound disorder, specic language impairmentdevelopmental dyslexia, genetics

Abstract In this article, we critically review the evidence for overlap among tdevelopmental disorders, namely speech sound disorder (SSD), guage impairment (LI), and reading disability (RD), at three leveanalysis: diagnostic, cognitive, and etiological. We nd that while lap exists at all three levels, it varies by comorbidity subtype, anrelations among these three disorders are complex and not fullyderstood. We evaluate which comorbidity models can be rejectesupported as explanations for why and how these three disorders olap and what new data are needed to better dene their relations.

283

U

R

V


2/27

ContentsINTRODUCTION .. . . . . . . . . . . . . . . . . 284DIAGNOSTIC DEFINITIONS

AND EPIDEMIOLOGY . . . . . . . . . . 284EVIDENCE FOR COMORBIDITY. . 285COGNITIVE MODELS OF

LANGUAGE IMPAIRMENT,READING DISABILITY, ANDSPEECH SOUND DISORDER . . . 288Language Impairment. . . . . . . . . . . . . . 289Reading Disability . . . . . . . . . . . . . . . . . 291Speech Sound Disorder . . . . . . . . . . . . 294

EVIDENCE FOR COGNITIVEO V ERLA P. . . . . . . . . . . . . . . . . . . . . . . . 2 9 4

EVIDENCE FOR ETIOLOGICALO V ERLA P. . . . . . . . . . . . . . . . . . . . . . . . 2 9 5

WHICH COMORBIDITY MODELS

ARE SUPPORTED? . . . . . . . . . . . . . . 298CONCLUDING COMMENTS. . . . . . 301

INTRODUCTION A fundamental question to be addressed by psychologists is how atypical development re-lates to typical development. An adequate the-ory must account for both human universalsand individual differences, preferably with thesame underlying mechanisms. Every exampleof atypical development poses both a challengeand an opportunity for developmental theory.In this review, we focus on a pervasive charac-teristic of atypical development, co-occurrenceor comorbidity among behavioral disorders,specicallyspeech, language, and literacy disor-ders. Understanding this comorbidity has im-plications for the genetics, neuropsychology,prevention, and treatment of these disorders as well as for our understanding of the develop-ment of language. Historically, theorists havefocused on explaining individual disorders (e.g., Morton & Frith 1995) and have viewed theirco-occurrence as a peripheral issue. However,in recent years theorists have increasingly rec-ognizedthatcomorbidityisofinterestinitsownright. We argue that placing comorbidity at the

center of inquiry leads to a new perspecttheoretical models of disorders.

In this review, we rst dene speechguage, and reading disorders and critica view the evidence for their overlap atlevels of analysis: (a) diagnostic, (b) cognor neuropsychological, and (c ) etiologicacause overlap exists at all three levels o ysis, we consider and evaluate which of comorbidity can be rejected by currendence, which are supported, and what neware needed to distinguish among currentlyported models. We conclude with a discuof broader implications and future direcfor research.

DIAGNOSTIC DEFINITIONS AND EPIDEMIOLOGY The essential dening characteristics othree disorders are summarized in Table 1each case, the disorder involves an unexdifculty in oneaspectofdevelopment thanot readily be explained by such factors intelligence or sensorimotor impairmenthree disorders lack a sharp dividing lintween impairment and normality; thus, nosis involves setting an arbitrary thresh what are essentially continua.

It has been customary in language imment (LI) and reading disability (RD) toon specic developmental disorders, i.e., where a signicant discrepancy exists blanguage or literacy and general intelligHowever, such denitions have been criton both logical and practical grounds. Ield of reading disability, it is now broacepted that it is not valid to distinguistween children who have a large discrebetween poor reading and IQ, and thosedo not. Both types of poor reader haveilar underlying decits in phonologicalcessing and both respond to similar kintreatment (see review in Fletcher et al. 1 Analogous arguments have been advancLI (e.g., Bishop & Snowling 2004). Altthe focus here is on children of broadlymal nonverbal ability, we use the term

284 Pennington Bishop

U

R

V


3/27

Table 1 Basic characteristics of language impairment, reading disability, and speech sound disorder

Language impairment (LI) Reading disability (RD) Speech sound disorder (SSD)Synonyms Developmental dysphasia Developmental dyslexia Phonological disorder

Developmental language disorder Articulation disorderDeningcharacteristics

Expressive and/or receptivelanguage development is impairedin the context of otherwise normaldevelopment (i.e., nonverbal IQand self-help skills)

Child has signicant difculty learning to read accurately anduently despite intelligence within normal limits and adequateopportunity to learn

Child substitutes or omits sounfrom words more than dosame-aged peers; speechproduction errors interfere wiintelligibility of speech

Language impairment interferes with activities of daily livingand/or academic achievement

Exclusionary criteria

Severe neglect Acquired brain damage

Inadequate educationalopportunity

Structural or neurologicalabnormality of articulators

Signicant hearing impairment Acquired brain damage Signicant hearing impairmenKnown syndrome, such as autisticdisorder

Signicant hearing impairment Known syndrome, such as autisticdisorder

IQ < 70

Prevalence Depends on cutoff used

Epidemiological study (Tomblinet al. 1997): 7.4% (CI 6.38.5%)of 6-year-olds met psychometriccriterion

Depends on cutoff used; typical

value is around 9%

2%13% (mean = 8.2%)

(Shriberg et al. 1999)

Odds ratio M:F 3 in referred sample (Broomeld &Dodd 2004); 1.5 inepidemiological sample (Tomblinet al. 1997)

1.9 to 3.3 in 4 epidemiologicalstudies (reviewed by Rutter et al.2004)

1.5 to 2.4 (mean = 1.8)(Shriberg et al. 1999)

Risk factors Family history has signicant effect No effect of parental education;slight effect of birth order(later-born at more risk) (Tomblin

et al. 1991)

Parental educationHome literacy environment Bioenvironmental risk factors, suchas lead poisoning and head injuries

No effect of race, SES, or otitimedia history, but signicant effects of gender, family histoand low maternal education

(Campbell et al. 2003)

and LI without the specic prex rather thanadopt a discrepancy-based denition of thesedisorders.

When dealing in particular with LI, thequestion arises as to whether the childs socio-cultural background might affect performanceon language measures and hence liability tobe diagnosed with disorder. In general, that should not be the case, provided one is sen-sitive to the possibility that some nonstandarddialects may have grammatical and vocabulary differences fromthe standard.Tests of languageprocessing, rather than linguistic knowledge,are largely immune to cultural inuence andare sensitive indicators of LI (Campbell et al.1997).

EVIDENCE FOR COMORBIDITY

Any attempt to summarize the evidence forcomorbidity between speech sound disorder(SSD), LI, and RD will be tentative, not least because the prevalence of these disorders is agedependent. Clearly, RD cannot be identieduntil children have been exposed to reading in-struction. On the other hand, SSD is typically most apparent in the preschool years, and it of-ten resolves by the time the child starts to learnto read. LI also declines in prevalence with age, with many children having an early delay in lan-guage development that subsequently resolves(Bishop & Edmundson 1986). Another prob-lem for this eld is that no study has evaluated

www.annualreviews.org Relations Among Speech, Language, and Reading Disorders 285

U

R

V


4/27

prevalence ofSSD,LI, and RDin the samechil-dren; most comorbidity studies examine only two disorders. A third point is that samplingbias in clinical samples may artifactually inatecomorbidity. Berkson (1946) showed that ap-parent comorbidities between otherwise inde-pendent disorders will arise in referred samplesif the probability for referral of either or bothdisorders is less than one. In this case, comor-bid individuals will be overrepresented becausetheir probability of being referred is a com-bined function of the referral rates of each of their disorders. Berksons bias does not imply any overt bias to select comorbid cases; the biasis simply the result of the compounding of in-dependent probabilities. The effect would bemagnied if the concerns of parents or teachersresulted in an additional bias to refer comorbidcases. Epidemiological data are needed for un-biasedestimates of comorbidityrates. However,few investigators have published epidemiolog-ical studies of the comorbidity among pairs of these disorders, and their three-way morbidity is touched upon only in unpublished epidemi-ological data.

Table 2 summarizesexisting data on comor-bidity among SSD, LI, and RD expressed asrelative risks (i.e., rate in the group with therisk diagnosis/rate in the group without therisk diagnosis; for studies that did not have acontrol group, we used the population rate asdenominator). So a relative risk of 2.0 meansthat those in the risk group (e.g., with SSD)are twice as likely to have a comorbid diagno-sis (e.g., LI) as those without the risk diagnosis(e.g., non-SSD controls or the general popu-lation). For a common multifactorial disorder,these two methods of calculating relative risk should yield roughly similar estimates. The ta-ble is divided into data from epidemiological versus referred samples, and it illustrates that more variation exists among individual studies within each type of sample than across sam-ple types. Thus, Berksons and other referralbiases do not appear to play a major role. Gen-uine comorbidity exists among these conditionsbecause the pairwise comorbidities are signi-cantly greaterthanchance in both types of sam-

ples. However, except for the Iowa samplethe risk for later RD in SSD and LI is aentirely restricted to SSD+ LI (RR = 4.6 whereas the rate of later RD in SSD withis negligible (RR = 0.91.6, all ns). Morare needed to specify the risk for later RD without SSD because the two values in the(3.23.6 for Iowa and 0.5 for the Longitu Twin Study) do not agree.

The convergent results for elevated ralaterRDinSSD + LIcould reect that SSDis an etiological and/or cognitive subtythat etiological and/or cognitive risk factSSD and LI interact synergistically to gincrease risk for later RD. If the latter is timplies that the developing reader maypensate for the cognitive risk factors by SSD or even LI alone by using alterstrategies, but the combination of theirnitive risk factors makes compensationmore difcult.

In studies specically concerned withand LI, comorbidity varies with age. Breld & Dodd (2004) categorized all neferrals to pediatric speech and languageapy services in a British Primary Care and found robust bidirectional comorbidittween SSD and LI (entry 6 in Table 2contrast, Shriberg et al. (1999) found a comorbidity rate in an epidemiological of 6-year-olds in the United States (entin Table 2 ). Although different modes ofplingmayhavecausedthedifferencesinthstudies (clinical sample in the U.K. studan epidemiological sample in the U.S. sit is also likely that the degree of comity varies with age: Only 15% of childthe study by Broomeld & Dodd (2004)over age 6. Bishop & Edmundson (1987)that althoughpreschoolchildren appearedmore vulnerable to speech than language lems, speech problems resolvedmore readtheirclinically identied sample, manych who presented with both speech and lanproblems at age 4 had only residual landifculties when seen 18 months later.

The best evidence for the comorbof LI and RD comes from the same


U

R

V


5/27

Table 2 Comorbidity rates among speech sound disorder (SSD), language impairment (LI), and reading disability (RD)(relative risks)

Samplesizes

SSDin LI

LI in SSD RD in LI

RDin SSD

RD in SSD+ LI

RD in SSDLI

RD in LISSD

Epidemiological 1328 1ar 3.3 2.3 570 1br 6.2 (second),

6.9 (fourth)

527 1cp 1.92.2(second,fourth, eighth)

604 1dr 2.8 (second),3.1 (eighth)

3.9 (second),4.9 (eighth)

2.2 (second),2.3 (eighth)

3.2 (seco3.6 (eig

453 2r 2.2 2.3 1.9 2.6 6.0 1.6+ 0.5+

925955 3r 4.4 (7y), 4.9(9y), 5.1(11y)

1655 4r 6.1 6.1 Referred 277 5p 9.1 1.6 8.1 1.6+

936 6p

5.7 6.9 123 7p 4.0 2.5 7.4 1.1+ 110 8p 5.7 82 9p 2.3

+ = NS.= dened RD as 1 SD below weighted mean on reading comprehension composite.

Key: relative risk = rate in risk group/population rate (p) or rate in risk + group/rate in risk group (r).1: Iowa sample. 1a, Shriberg et al. (1999); 1b, Catts et al. (2002); 1c, Catts et al. (2005); 1d, Tomblin (unpublished).2: Colorado Longitudinal Twin Sample, R.L. Peterson, B.F. Pennington, L.D. Shriberg, & R. Boada (manuscript under review).3: Silva et al. (1987).4: Beitchman et al. (1986).5: Cleveland SSD sample, B.A. Lewis & L.A. Freebairn (unpublished).6: Broomeld & Dodd (2004).7: Denver SSD sample, Raitano et al. (2004), R.L. Peterson, B.F. Pennington, L.D. Shriberg, & R. Boada (manuscript under review).8: McArthur et al. (2000).9: Bishop & Adams (1990).

epidemiological sample that was studied by Shriberg et al. (1999). Catts et al. (2002) fol-lowed up children who had been identied withspecic language impairment (SLI) at 6 years of ageandfoundarelativeriskofabout3.Inalaterfollow-up of the Iowa sample to eighth grade(Catts et al. 2005), the comorbidity rates be-tweenLI and RDwere lower but stillsignicant ( Table 2 ). Interestingly, this study also foundthat the LI without RD group performed sig-nicantly better than either the RD or the co-morbid groups on both phonological awarenessand nonword repetition. Of these two phono-

logical measures, the LI without RD group was worse than controls only on nonword repeti-tion, not on phonological awareness. Whethera similar cognitive prole would be found in LI without SSD is not known.

However, because LI and SSD are com-monly comorbid, it can be difcult to estab-lish whether associations with RD are specicto one of these disorders. Several prospectivestudies of smaller referred samples of children with SSD have found an increased rate of laterRD(Bishop&Adams1990,Catts1993,Nathanet al. 2004, Snowling et al. 2000), although


U

R

V


6/27

Figure 1Connectionist models of speech development. Adapted from Guenther (1995), Joanisse (2000), Westerman & Miranda (2004).

several studies have found that the associationis most robust for children with SSD+ LI and isnot found for children with isolated SSD (Birdet al. 1995, Larrivee & Catts 1999, Leitao &Fletcher 2004, Lewis & Freebairn 1992). New unpublisheddatafromthreeprospectivestudies with large samples of children with SSD haveresolved this issue. As mentioned above, in allthree studies the risk for later RD in SSD ismediated by comorbid LI and this risk is sub-stantial (4.38.9). In contrast, the risk is neg-ligible for later RD in SSD without LI or, inone case, LI without SSD. This pattern for RDrisk in SSS+ /LI was found across referred andepidemiological samples; thus, it is not due to areferral artifact. But more research is needed tounderstand the conicting ndings for RD risk in LI without SSD.

In summary, population samples have estab-lished that SSD and LI are comorbid, althoughtheratesvarywithage,andthatLIandlaterRDarecomorbid,but we lack published population

Figure 2Connectionist model of reading development. From Harm & Seidenberg(1999).

data on the comorbiditybetween SSD andRD.However, SSD+ LI carries most of thefor later RD, and the risk posed by SSDor sometimes LI only, appears to be negliOverlap among pairs of these three disosometimes varies as a function of como with the third disorder. This complex paof comorbidity makes it unlikely that alldisorders are pleiotropic manifestations same underlying cognitive or genetic lia(e.g., generalist genes for a general verbaPlomin& Kovas 2005).This raises thequeof whether the cognitive overlap of thedisorders mirrors the diagnostic overlap.is, is the cognitive prole in SSD+ LI (andRD) distinct from that in either SSD wiLI or LI without SSD?

COGNITIVE MODELS OF LANGUAGE IMPAIRMENT,READING DISABILITY, ANDSPEECH SOUND DISORDER It is useful to consider cognitive modLI, RD, and SSD in the context of tydevelopment and in relation to each Figures 1 and 2 depict typical developof speech, language, and literacy, withten language skills building on earlier deing oral language skills. In developing alanguage, one of the rst tasks an infanmaster is the perception and production ospeech sounds specic to the native lan(Kuhl 2004). Although innate constraintsence some aspects of human language action (e.g., Pinker 1991), mastering a partoral, and especially written, language reextensive learning, much of it implicit scal learning (Saffran et al. 1996). Consequconnectionist models, which implement tical learning of speech, language, and reprovide a useful framework for thinkingrelations among LI, RD, and SSD at thenitive level.

Figure 1 depicts a simplied connectmodel of speech development (adaptedGuenther 1995, Joanisse 2000, Wester& Miranda 2004) and Figure 2 depi


U

R

V


7/27

connectionist model of single-word reading development (Harm & Seidenberg 1999). Twokey components are shared by both models:phonology and semantics. These models illus-trate that a problem in developingphonologicalrepresentations could affect speech, language,and reading development, and indeed, as reviewed below, evidence does exist for phonolog-ical impairment in SSD, LI, and RD. However,these models also illustrate that development in each of these three domains is affected inmultiple ways, and some of these ways have not been considered in cognitive models of thesedisorders.

The development of speech productionhas been depicted in several computationalmodels (Guenther 1995; Joanisse 2000, 2004,2007; Joanisse & Seidenberg 2003; Markey 1994; Menn et al. 1993; Plaut & Kello 1999; Westermann & Miranda 2004). The difcult developmental task these models address is how young children learn the complex mapping be-tween acoustic features and articulatory ges-tures, particularly when important aspects of the articulatory gestures made by adult mod-els are not observable. Babbling almost cer-tainly supplements imitation in learning thesemappings. In these models, hidden units learnthe particular abstract mapping between acous-tic features and articulatory gestures that sig-nal meaning differences for that child, andthe representations of these abstract mappingsare phonological representations. However,these models do notexplain clinical cases wherephonological representations develop without speech (e.g., anarthric children with oral read-ing skills; Bishop 1985).

Lexical semantics is intimately involved inthe determination of which acoustic and artic-ulatory features are counted as relevant for a young childs particular lexicon in a particularlanguage. As the childs vocabulary increases,the nature of her phonological representationsalso changes. Consequently, the development of phonological representations is protracted,and the weighting given to different acousticfeatures in speech perception changes with development (Nittrouer 1999).

These computational models demonstratethat a problem in speech production such asthat found in SSD could have several causes,including a bottom-up problem in processingacoustic features, a motor problem in planningand producing articulatory gestures, a prob-lem learning the mapping between the two, aproblem identifyingwhichphoneticdifferencessignal differences in meaning and which areequivalent (i.e., in learning phonological rep-resentations, a top-down problem in learningsemantic representations that impedes the dif-ferentiation of phonological representations),or some combination of these problems. Simi-larly, these models suggest that there could bebottom-up auditory, representational (phonol-ogy in RD and syntax in LI), top-down seman-tic, learning, and multiple decit theories of RD and LI. As we discuss below, existing cog-nitive models of these disorders have focusedon single cognitive decits and have tended tobe static rather than developmental. That is,they have posited a congenital decit in ei-ther a bottom-up auditory skill or a particu-lar kind of representation (phonology or syn-tax) and have not considered the possibilitiesof how decits might emerge from a develop-mentalprocess or howdecient learning of new mappings between representations could causedisorders.

Language Impairment A broad distinction can be drawn betweentwo classes of LI model: those that regard thelanguage difculties as secondary to more gen-eral nonlinguistic decits, and those that pos-tulate a specically linguistic decit. The best-known example of the rst type of model isthe rapid temporal processing (RTP) theory of Tallal and colleagues, which maintains that lan-guage learning is handicapped because of poortemporal resolutionof perceptual systems. Thisbottom-up auditory model of LI has also beenapplied to RD and SSD.

The rst evidence for the RTP theory camefrom a study where children were required tomatch the order of two tones (Tallal & Piercy


U

R

V


8/27

1973).When tones were rapid or brief,children with LI had problems in correctly identifyingthem, even though they were readily discrim-inable at slow presentation rates. The theory has continued to develop over the years, and Tallal (2004) proposed a neural basis in the formof spike-timing-dependent learning. Tallal ar-gued that although the underlying mechanismaffected all auditory stimuli, its effects wereparticularly detrimental to language learningbecause development of neural representationsof phonemes depends on ne-grained tempo-ral analysis. Children who have poor tempo-ral resolution will chunk incoming speech inblocks of hundreds of milliseconds rather thantens of milliseconds, and this will affect speechperception and hence on aspects of languagelearning.

Another theoretical account that stressesnonlinguistic temporal processing has beenproposed by Miller et al. (2001), who showedthat children with LI had slower reaction timesthan did control children matched on nonver-bal IQ on a range of cognitive tasks, includingsome, such as mental rotation, that involved nolanguage. Unlike the RTP theory, this account focuses on slowing of cognition rather thanperception.

A more specialized theory is the phono-logical short-term memory decit account of SLI by Gathercole & Baddeley (1990a). Theseauthors noted that many children with SLIare poor at repeating polysyllabic nonwords, adecit that has been conrmed in many sub-sequent studies (Graf Estes et al. 2007). Thisdecit has been interpreted as indicating a lim-itation in a phonological short-term memory system that is important for learning new vo-cabulary (Gathercole & Baddeley 1990b) andsyntax. This theory, like the more specically linguistic theories, places the core decit in asystem that is specialized for language process-ing, but the system is for memory and learningrather than for linguistic representations per se.

More recently, Ullman & Pierpont (2005)proposed a theorythat encompasses both short-term memory and syntactic decits under theumbrella of procedural learning, which is

contrasted with a declarative learning sthat is involved in learning new verbal inftion. They argue that LI is not a specicallguistic disorder but is rather theconsequenan impaired system that will also affect leof other procedural operations, such as mskills.

Many authors, such as Bates (2004)argued that domain-general decits in ctive and perceptual systems are sufciaccount for LI. This position differs radfrom linguistic accounts of LI, which mathat humanshave evolved specialized langlearning mechanisms and that LI results these fail to develop on the normal sch Thus, language-specic representational ries of LI coexist with similar linguistic thof RD and SSD that focus on phonologicalresentations. A range of theories of this typLI focus on the syntactic difculties thacore feature of many children with LI.Chil with LI tend to have problems in usinginections that mark tense, so they migyesterday I walk to school rather thanterday I walked to school. Different lingaccounts of the specic nature of such lems all maintain that the decit is locaa domain-specic system that handles synoperations and is not a secondary conseqof a more general cognitive processing (see, e.g., Rice & Wexler 1996, van de1994).

Although these theories focus on difperceptual, cognitive, and linguistic dethey arenonetheless hard to choosebetweeseveral reasons. First, the theories do noessarily predict pure decits in just one for instance, the RTP theory predictschildren with SLI will have phonologicsyntacticproblems,but the theoryregardsas secondary to the basic perceptual dEven where a domain-specic linguistic is postulated, it could beargued that othergeneral decits may coexist, perhaps bof pleiotropic effects of genes. Second, iten easy to explain away a failure to nddicted decit on the grounds that the chilgrown out of the decit (which neverthele


U

R

V


9/27

affected language acquisition) or that thedecit applies only to a subgroup of children withSLI.

Studies that examine decits predicted by different theories in the same children aidin disentangling different theoretical accounts. The results of such studies can be illuminating.Bishop et al. (1999) studied a sample of twinchildren, many of whom met criteria for LI. These children were given a battery of tests,including a measure of auditory processing, de-rived from the RTP theory, and a measureof phonological short-term memory, nonwordrepetition. Children with LI did worse thancontrols did on both measures, but some chil-dren have normal language despite poor scoreson the testsof RTPornonword repetition.Theintercorrelation between the measures, thoughsignicant, was low (around 0.3). Furthermore,genetic analysis suggested different etiologiesfor the auditory decit, which appeared envi-ronmental in origin, and the nonword repeti-tion decit, which was heritable. One might wonder whether these decits identify different subgroups of children with LI, but the resultsindicated rather that the two decits interactedand that children with a double decit were themost severely affected. A similar pattern of re-sults was obtained in a later twin study in whichchildren were assessed on a test of nonwordrepetition and on a test of productive verb mor-phology (Bishop et al. 2006). Both measures re- vealed decits in children with LI, and in thiscase, both were heritable, yet the intercorrela-tion between these measures was low (thoughsignicant) andtherewasnoevidence that com-mon genes were implicated. Once again, chil-dren with both decits had the most severeproblems, and some children with normal lan-guage scored in the impaired range on one of the measures of underlying decit.

These two studies raise some general pointsthat also apply to other developmental disor-ders:

a) Any theory that postulatesa singleunder-lying decit is inadequate to account forthe disorder: several distinct decits seem

implicated, none of which is necessary orsufcient on its own for causing LI.

b) Although the different decits can be dis-sociated and appear to have distinct eti-ologies, they tend to co-occur at above-chance levels.

c) It is possible to have a single decite.g., in auditory processing or phonolog-ical short-term memorywithout neces-sarily showing LI.

d) Children with LI typically show morethan one decit.

Reading Disability A cognitive model of RD has greater consen-sus than does a model of LI. Figure 3 depictsthe processes involved in extracting meaning

from written text. This gure shows that read-ing comprehension can be rst broken downinto cognitive components and then into developmental precursors of these cognitive com-ponents. One key component is uent printed word recognition, which is highly predictiveof reading comprehension, especially in theearly years of reading instruction (Curtis 1980). The other key component is listening compre-hension, that is, oral language comprehension(Hoover & Gough 1990).

The terms developmental dyslexia orreading disability have traditionally been re-served for children who have difculties withbasic printed word recognition. It is possible fora child to have reading comprehension prob-lems despite adequate printed word recogni-tion, but this is not counted as dyslexia. Instead,individuals with such problems are described aspoor comprehenders, and the cognitive causesof their reading comprehension problems aredistinct from those that interfere with word

recognition (Nation 2005).Printed word recognition canbe brokenintotwo component written language skills, phono-logical and orthographic coding (Figure 3 ).Phonological coding refers to the ability to useknowledge of rule-like letter-sound correspon-dences to pronounce words that have never


U

R

V


10/27

Figure 3 The processes involved in extracting meaning from written text.

been seen before (usually measured by pseu-doword reading); orthographic coding refers tothe use of word-specic patterns to aid in wordrecognition and pronunciation. Words that donot followtypical letter-sound correspondences(e.g., have or yacht) must rely, at least in part,on orthographic coding to be recognized, as dohomophones (e.g., rows versus rose).

A large body of work has shown that most children with RD have disproportionate prob-lems with pseudoword reading and that thisdecit in phonological coding is related to poorphonological awareness. Phonological aware-ness is measured by tasks that require ma-nipulation of the sound structure of spoken words (e.g., what is cat without the /k/?).Despite agreement about the importance of phoneme awareness decits in RD, there isdisagreement about whether these difcultiesare themselves causedby lower-levelprocessingdecits. Phoneme awareness is a complex meta-linguistic skill that clearly involves multiplecomponents. One argument is that phonemeawareness decits arise from impaired phono-logical representations (Fowler 1991a, Swan& Goswami 1997a). Another argument postu-

lates that the central decit is not speclanguage, but rather is a consequence osame RTP decit proposed as an explanfor SLI (Tallal 1980). However, evidenthe RTP hypothesis for RD is patchy a(see McArthur & Bishop 2001 for a revie

A parsimonious explanation for currenis that decits in phonological representlead to both phoneme awareness andphonical codingdifculties in RD.Thephonolorepresentations hypothesis is appealing beit helps explain why RD is associated no with decits in phoneme awareness, bu with impairments on a wide variety of plogical tasks, including phonological m(Byrne & Shea 1979, Shankweiler et al.and picture naming (Fowler & Swainson Swan & Goswami 1997b).

An important caveat is that the relship between phoneme awareness and ing is bidirectional, so that over time,reading also causes poor phoneme aness (Morais et al. 1979, Perfetti et al. Wagner et al. 1994). Another caveat is thevidence for the emergence of phoneme awness being a necessary precursor for re


U

R

V


11/27

development is not airtight (Castles &Coltheart 2004) because longitudinal studiessupporting this claimhave notcompletelyelim-inated the confound of preschoolers already having some reading skill at time one. A recent study found that childrenwith chance-levelper-formance on phoneme awareness tasks couldnonetheless use letter names to learn lettersounds and thus begin to decode printed words(Treiman et al. 2008). So explicit phonemeawareness may not be necessary for learning toread, but it seems clear that appropriatelystruc-tured phonological representations are needed. The exact meaning of appropriately struc-tured is still unclear in the context of read-ing development. As discussed above, phono-logical representationsdevelopandcandevelopatypically in a variety of ways. The nature of the underlying phonological decit in RD isthe subject of severalhypotheses, including that children with RD (a) lack segmental phono-logical representations (Boada & Pennington2006, Fowler 1991), (b) have problems detect-ing suprasegmental information in phonologi-cal representations(Goswamiet al.2002),(c )re-tainallophonicrepresentations (Serniclaeset al.2004), or (d ) have less-distinct phonologicalrepresentations (Elbro et al. 1998). More re-search is needed to test these hypotheses not only in RD, but also across all three disordersconsideredhere(e.g.,Corriveauetal.2007haverecently extended Goswamis suprasegmentedtheory to LI). These three disorders possibly share decits on broad phonological measures,such as phonological awareness or nonwordrepetition, but they differ in how their phono-logical representations are decient.

Because RD is often comorbid with LI, thequestion is raised as to the extent to which RDisassociated with broader language decits, e.g.,as measured by tests of vocabulary and syn-tax. Although the phonological decit hypoth-esis stresses childrens difculties in learningletter-sound mappings, poor general languageskills could also handicap reading acquisitionbecause one can use linguistic context to in-fer meaning of a novel word (e.g., Cain et al.2004). On IQ tests, children with RD tend to

underperform relative to their typically devel-oping counterparts not only on phonologicaltasks, such as digit span, but also on all ver-bal subtests (DAngiulli & Siegel 2003). Someof this performance decit likely resulted fromRD, since children with reading difcultieshave impoverished opportunities to learn fromprint (e.g., Stanovich 1986; cf. Scarborough &Parker 2003), but some may well reect subtle, wide-ranging language impairments. The evi-dence on this point from predyslexic childrenis particularly compelling, because decits on a wide range of language skills are evident beforethey learn to read (Pennington & Ley 2001,Scarborough 1990).

The presence of comorbid language prob-lems in RD raises doubts that a phonolog-ical decit is sufcient to cause RD. More-over, as is discussed below, children with SSDhave phonological decits similar to thosefound in RD, but they usually do not developRD unless they have comorbid LI. It appearsthat normal performance on rapid serial nam-ing (RSN) tasks is a protective factor (R.L.Peterson,B.F. Pennington, L.D.Shriberg,& R.Boada, manuscript under review; Raitano et al.2004). RSN is impaired in both RD and at-tention decit/hyperactivity disorder (ADHD)(Shanahan et al. 2008), so RSN appears to be acognitive risk factor shared by RD and ADHD. Moreover, because nonlinguistic processing-speed measures such as perceptual speed tasks(Wechsler Symbol Search) played a role simi-lar to that of RSN as a cognitive risk factor forbothRDandADHD,itdoesnotappearthattheRSN problem is just a by-product of phonolog-ical or name-retrieval problems. We recently tested the hypothesis that processing or per-ceptual speed is a shared risk factor for RDandADHD using structural equation modeling(L. McGrath, B.K. Pennington, R.K. Olson, &E.G. Willcutt, manuscript under review). Wefound that processing speed (i.e., latent traitscomposed of RSNandnonlinguisticperceptualspeed tasks) was a unique predictor of both RDand ADHD symptoms and reduced the corre-lation between them to a nonsignicant value.Phoneme awareness and language skill were


U

R

V


12/27

unique predictors of RD symptoms, and inhi-bition was a unique predictor of ADHD symp-toms. These results support a multiple-decit model of both RD and ADHD. The total vari-ance explained in RD symptoms by phonemeawareness, language skill, and processing speed was more than 80%. Thus, the best current understanding of the neuropsychology of RDindicates that at least three cognitive risk fac-tors are involved, which is consistent with amultiple-decit model.

At least one of these underlying decits, de-cient phonological representations, overlaps with decits that are found in studies of LI.However, rather surprisingly, a decit in RSNis not characteristic of children with LI un-less they also have reading impairment (D.V.M.Bishop, D. McDonald, & S. Bird, manuscript under review). Furthermore, although children with RD tend to have lower scores on languagetests, they typically do not show the kinds of grammatical limitation seen in LI (Bishop &Snowling 2004). Overall, multiple underlyingdecits appear to exist in RD, as in LI, with themost serious problems being found in children who have two or more of the disorders. Of par-ticular interest is the indication that at least oneunderlying decit, poor phonological process-ing, is common to both RD and LI.

The procedural learning account of LI isrelatively new, and few studies have tested itspredictions. Nevertheless, it is noteworthy that it overlaps with the automatization decit ac-count of dyslexia (Nicolson & Fawcett 1990).Both theories maintain that specic brain cir-cuitry involving the cerebellum is involved inthe poor learning of reading or rule-governedaspects of language, especially phonology andsyntax, and in both cases it is argued that associ-ated motor impairments are another symptomof this neurobiological decit (Nicolson et al.2001, Ullman & Pierpont 2005). The autom-atization decit account of dyslexia has beenchallenged as a general account of this disorderby ndings that motor impairments are seen inonly a subset of cases (e.g., Ramus 2003). Nev-ertheless, as we have argued for LI, this doesnot necessarily mean that these decits are ir-

relevant to the causation of the disordermay have theireffect only when in combin with other decits.

Speech Sound Disorder SSD was originally considered a disorgenerating oral-motor programs, and chi

with speech sound impairments were shave functional articulation disorder (B1997). However, a careful analysis of erroterns has rendered a pure motor decit unlas a full explanation for the disorder. For ple, children with SSD sometimes prodsound correctly in one context but incorin another. If children were unable to exparticular motor programs, then we mighpect that most of their errors would takform of phonetic distortions arising fro

approximation of that motor program. Hever, the most common errors in childrenSSD are substitutions of phonemes, notortions (Leonard 1995). Moreover, a grobody of research demonstrates that indials with SSD often show decits on a of phonological tasks, including speechception, phoneme awareness, and phonolomemory (Bird & Bishop 1992, Kenney2006, Leitao et al. 1997, Raitano et al. 2 Though it remains possible that a subgro

children have SSD primarily because of impairments, it nowseemslikelythat themity ofchildren with SSD have a type of landisorder that primarily affects phonologic velopment. Interestingly, RSN is not impin SSD (Raitano et al. 2004), and SSD chcan have persisting phoneme awarenesslems but normal reading development Peterson, B.F. Pennington, L.D. ShriberR. Boada, manuscript under review). Thutact RSN appears to be a protective fac

these children.

EVIDENCE FOR COGNITIVEOVERLAP This brief review of cognitive models ofders indicates that some close similaritiist in the theories that have been advan


U

R

V


13/27

account for LI, RD, and SSD. For all threedisorders, phonological decits, possibly due toauditory perceptual problems, have been pro-posed as a core underlying cause. Although thisoverlap may help explain why the disorders areoften comorbid, it leaves us with the puzzleof phenotypic variation between disorders. Inshort, if the same theoretical account applies toall disorders, why do they involve different be-havioral decits? And why, for instance, do wend children with SSD who have poor phono-logical skills yet do not have reading problems? An answer is suggested by our analysis of RDand LI as disorders that involve multiple cog-nitive decits. A phonological decit may be akey feature of all three disorders, yet its spe-cic manifestation will depend on the pres-ence of other decits. This kind of model isimplicit in the analysis by Bishop & Snowling(2004), who argue that LI is not just a moresevere form of RDrather, RD and LI bothusually involve poor phonological processing,but LI is seen when this decit is accompa-nied by broader difculties affecting aspects of language such as syntax. The nding (by R.L.Peterson, B.F. Pennington, L.D. Shriberg, &R. Boada, manuscript under review) that chil-dren with SSD often read well despite poorphonological skills indicates that phonologicaldecit alone will not usually lead to later read-ing problemsit does so when it is accompa-nied by poor RSN. A closely similar conclusion was reached (by D.V.M. Bishop, D. McDon-ald, & S. Bird, manuscript under review) in astudy of comorbidity between RD and LI. Thestudy found that children who had LI without RD performed normally on tests of RSN. Boththese studies suggest that although phonologi-cal decit is a risk factor for RD, good RSN canact as a protective factor. This evidence for in-teraction between decits has implications forhow we model comorbidity between disorders(discussed below).

In summary, although cognitive overlap ex-ists among these three disorders (i.e., phono-logical decits), the cognitive prole varies as afunction of comorbidity. Moreover, these cog-nitiveproledifferences appearto maponto the

comorbidity patterns reviewed above. That is,the presence or absence of RSN decits in SSDand LI relates to their comorbidity with laterRD. But more systematic research is needed totest how cognitive proles vary by comorbidity subtypes. This research will require large sam-ples that have been followed longitudinally.

EVIDENCE FOR ETIOLOGICALOVERLAPStrongevidence demonstrates that LI, RD,andSSD are genetically inuenced. That evidenceis summarized in Table 3 , which shows that each disorder is familial, moderately herita-ble, and has several replicated linkages to spe-cic chromosome locations (for reviews, seeFisher & Francks 2006, Lewis et al. 2000,

McGrath et al. 2006, Paracchini et al. 2007). Table 3 also contains a footnote that explainsthe nomenclature for chromosome locationsand loci associated with disorders. The RD locihave replicated across languages and cultures,including Swahili-speaking children in Tanza-nia (Grigorenko et al. 2007). For both RD andSSD, several candidate genes have been identi-ed, and several of these are candidates for bothdisorders.

Twin studies have also been used to exam-

ine relations among these disorders. Most twinstudies have found high h2g (0.6 or above)1

for LI. However, a recent analysis by Bishop& Hayiou-Thomas (2008) found that this de-pended on whether children with SSD were in-cluded in the sample. For 4-year-olds who hadLI without SSD, genes did not seem implicatedin the etiology. Few twin studies have looked at both LI and RD in the same children, althoughin two separate samples Bishop and colleaguesreported that RD was heritable only when

1 The term h 2g refers to the heritability of the extremegroups decit. Unlike h2, which estimates what proportionof the phenotypic variance across the whole distribution isattributable to genetic inuence, h2g estimates the magni-tude of genetic inuences on the low (or high) tail of thedistribution. Unless they are completely categorical, h2g isthe appropriate heritability for disorders.


U

R

V


14/27

Table 3 Summary of genetic studies of speech sound disorder (SSD), language impairment (LI),and reading disability (RD)

SSD LI RDFamiliality (relative risk) 6a 24a 48d

Heritability 0.801.00a, b 0.360.96b, c 0.58Chromosome regions e, g 1p34-36 (DYX8) 13q21 (SLI3) 1p34-36 (DYX

3p12-q13 (DYX5) 16q24 (SLI1) 2p15-16 (DYX

6p22 (DYX2) 19q13 (SLI2) 3p12-q13 (DY15q21 (DYX1) 6p22 (DYX2)15q21 (DYX1)18p11 (DYX6) Xq27.3 (DYX9

Candidate genese, f FOXP2 ROBO1ROBO1 DCDC2DCDC2 KIAA0319KIAA0319 DYX1C1DYXC1 MRPL19

C20RF3

Estimates based on twin concordance data (double difference between monozygotic and dizygotic) or from groupheritability computed using DeFries-Fulker method (DeFries & Fulker 1985).Nonsex chromosomes are numbered according to their size, so chromosome 1 is the largest and chromosome 22

possibly 21) is the smallest. Each chromosome has two arms, one short (p) and one long (q). Morphologically denregions within each arm are denoted by a number, counting outward from the centromere that lies between the two(e.g., p1, p2, p3, and q1, q2, q3), and these regions are subdivided into bands (p11) and subbands (p11.1) and sub-s(p11.11). So the term 1p34-p36 means a location on the short arm of chromosome 1 including regions 34, 35, andnames of loci associated with a disorder are capitalized and numbered according to order of discovery (DYX1 meandyslexia locus discovered, DYX2 means the second, and so on).aLewis et al. (2006).b Viding et al. (2004).cBishop & Hayiou-Thomas (2008).dPennington & Olson (2005).e McGrath et al. (2006).f Anthoni et al. (2007).gNewbury & Monaco (2008).

accompanied by poor nonword repetition(Bishop 2001, Bishop et al. 2004).

A multivariate analysis of reading skill intwins (Tiu et al. 2004) supported the multiple-decit model (L. McGrath, B.K. Pennington,R.K. Olson, & E.G. Willcutt, manuscript un-der review) discussed above, in which phoneme

awareness, RSN, and language skill indepen-dently contributed to predicting reading skill. Tiu et al. (2004) found that phoneme aware-ness, RSN, and full-scale IQ each made inde-pendent phenotypic contributions to readingskill and, in the etiological model based on thetwin design, each construct had both shared

and independent genetic relations to reaskills. More work of this kind is neededmultiple-decit models of each disordehowboth familialityand heritability vary bmorbidity subtypes.

Family and twin studies can providdence for genetic inuences on disorder

their relations, but to identify the gene volved we need to use methods of moleculnetic analysis. Linkage analysis identiesmosomal regions that are likely to harbor ginvolved in etiology of disorder. This mcapitalizes on the fact that genes close toon a chromosome tend to be inherited toge


U

R

V


15/27


16/27


17/27

genetic andenvironmental

risk factors

liabilitycutoff

observedclinicalcondition

RDLI or SSD

Figure 4Continuous liability distribution. RD, readingdisability; LI, language impairment; SSD, speechsound disorder.

genes (Plomin & Kovas 2005) may exist for ageneral verbal trait, the low extreme of whichunderlies these disorders. In its simplest form,this corresponds to Neale & Kendlers (1995)alternate forms model,which hypothesized that individuals who cross a particular thresholdonaliability distribution have the probability of p of having a disorderA and the probability r of hav-inga disorderB.This means that both disordersshare a single liability, yet whether the personmanifests disorder A or B depends on chanceor risk factors that vary across individuals. Thismodel raises the question of why the same li-ability should manifest differently in different people: It seems unlikely that this would be en-tirely due to chance; rather, it seems plausiblethat different environmental or genetic risks inthe individualchild interact with theunderlyingliability to determine the outcome.

Age is one systematic factor that has beenproposed as inuencing outcome. Scarborough& Dobrich (1990) suggested that the same risk factors that led to LI in a preschool child couldlead to RD in a school-aged child. They talkedof illusory recovery, in which the problems of a child with LI appeared to resolve only to bereplaced later by literacy problems. Subsequent research, however, suggests that LI does not disappear in children with LI who then develop

RD,althoughLI maybecomeless overt (Bishop& Adams 1990, Catts et al. 1999). A variant of this model treats RD as a condition that is bothlater in onset and less severe than SSD or LI;according to such a severity model, a child witha moderate liability may present only with RDin middle childhood, whereas one with a higherliability will be identied with SSD and/or LIin preschool and with RD a few years later.

Although a model with a single-liability di-mension is parsimonious, most experts in thiseld would regard it as unlikely on the groundsthat all three disordersSSD, LI, and RDappear heterogeneous. Although agreement islacking about the best way of subtyping thesedisorders, quite marked phenotypic and etio-logic differences can exist within each category,as we have shown. Furthermore, some children with severe LI do not have RD, which is in-consistent with a model that treats LI as indica-tive of a more severe liability. In addition, somechildren with SSD do not have RD, which alsorejects this severity hypothesis for the relationbetween SSD and RD.

These results suggest that a better kind of model may be one that allows for separate lia-bilities for the different disorders, but where at least one of the liabilities can lead to a comor-bid form. This corresponds to a multiformity model (Neal & Kendler 1995). Penningtonet al. (1993) considered this kind of model when investigating comorbidity between RDand ADHD, and suggested that RD might leadto the phenotypic manifestation of ADHD inthe absence of etiological inuences typically associate with ADHD in isolation. One canreadily imagine that a child might appear to beinattentive or hyperactive in the classroom be-cause of the frustration elicited by difculties with reading rather than as a consequence of the neurocognitive difculties that are typically associated with ADHD in the absence of RD. The multiformity model usually describes eachdisorder as having its own distinct etiology, but there arecases where theetiology fordisorder A can lead to comorbid A + B. However, if rates of comorbidity are high, the multiformity modelcan also be seen as an extension of the alternate


U

R

V


18/27

forms model, whereby A and B usually have thesame underlying cause (but with different lia-bility thresholds), but subtypes of either A or Bexist with separate etiological pathways.

A more extreme view of subtypes is totreat the comorbid disorder as etiologically dis-tinct from either disorder occurringalone.Thismodel of three independent disorders might postulate, for instance, that SSD with LI is eti-ologically distinct from pure SSD or pure LI,requiring three separate liability distributionsto account for the patterns of co-occurrencebetween these two disorders. The comorbidity and genetic results presented here are consis-tent with this model, but the cognitive resultsare not consistent because SSD+ LI overlapscognitively with both SD and LI.

Neale & Kendler (1995)also postulated cor-related liabilities models. In these models, eachdisorder has its own liability, a continuous rela-tion exists between the liability to one disorderand the liability to the second disorder. An in-crease in liability for one disorder is correlated with the increase in liability for the second dis-order. The relationship between the liability of the two disorders occurs via a signicant corre-lation between the risk factors (correlated lia-bilities) or a direct causal relationship betweenthemanifest phenotypes of the twodisorders (A causes B, B causes A, or reciprocal causation).If the relationship is between manifest pheno-types, this actually becomes the multiformity model. In contrast, if the liabilities themselvesarecorrelated, evenat subthreshold levels, thensubclinical comorbidity will occur even if onediagnostic phenotype is not expressed.

Our review of the etiology of LI, RD, andSSD indicates that some support exists forthe correlated liabilities model of the relationbetween SSD and RD because both disordersare linked to some of the same QTLs. But thisnding is also somewhat puzzling because SSD without LI does not pose much risk for laterRD. So this nding needs to be tested in largersamples to determine whether the linkage of SSD to RD loci is mainly due to SSD+ LI. An-other puzzle is the robust comorbidity betweenSSD and LI and between LI (even without

SSD) and RD in some studies, but so fQTLs have been discovered that are shareither case.Clearly, more work is needed thow well the correlated liabilities modeshared genetic and possibly environmentfactors) accounts for the comorbidity aSSD, LI, and RD.

Although the Neale & Kendler (1995) mels are a major contribution to the comorliterature because they are the most comset of models yet proposed and because thspecied quantitatively, they nonethelesssome limitations. Specically, they do nclude either a neural or a cognitive leveare not explicitly developmental, and thedeal with pairwise comorbidities. The maiof testing between them involves considfamily or twin data to see how far a givender, or a comorbid form, breeds true, but with ideal (simulated) data sets, some models can be difcult to distinguish ecally (Rhee et al. 2004).

The specic developmental disorder we consider here, however, provide annative route to testing between models vided one is willing to make the assumptiodisorders caused by different liabilities have different cognitive proles. For insto test the three independent disorders mto explain comorbidity between RD and Lmight predict that the cognitive prole imorbid RD+ LI would differ from that inRD or pure LI. Because the phenotype ofof the disorders considered here can be variable, and well-articulated models elutheunderlyingcognitivedecits associatethese disorders, this seems a promisinproach. Contrary to the three independenorders model, cognitive decits characteof SSD, LI, and RD overlap to some exte

In summary, the comorbidity obsamong LI, RD, and SSD in not readilplained by any of the Neale and Kendlerels but rather may require a model in wmultiple cognitive decits interact. Modthis type have been proposed by BishSnowling (2004), Bishop (2006), and Penton (2006), and a specic version to acco


U

R

V


19/27

comorbidity between LI and RD was formally specied and tested by D.V.M. Bishop, D. Mc-Donald, & S. Bird (manuscript under review).

The multiple overlapping risk factorsmodel of developmental disorders proposedby D.V.M. Bishop, D. McDonald, & S. Bird(manuscript under review) differs from theNeale and Kendler models in that it does not have a separate liability for each disorder, nordoes it attempt to account for two disordersin terms of a single liability. Rather, it postu-lates several independent liability distributions,each of which determines a specic underlyingdecit, and the disorder that is observed de-pends on the combination of liabilities that aresuprathreshold.Some liabilities, such as that forphonological processing decit, are implicatedin SSD, LI, and RD. Others, such as liability for RSN decit, appear to be specic to RD(although they might turn out to be implicatedin other neurodevelopmental disorders such as ADHD; see Shanahan et al. 2008).

The multiple overlapping risk factors modelcan account for several features of the data reviewed here. In particular, it predicts that weshould nd some risk factors that are generalto the three disorders of SSD, LI, and RD, andothers that are specic. Furthermore, it pre-dicts that a decit in one underlying cognitiveskill will not lead to overt disorder unless otherdecits are also present.

CONCLUDING COMMENTSLI, RD, and SSD are conditions that have tra-ditionally been viewed as separate, and indeed,

they can occur in pure form. This has led re-searchers to look for a single underlying causeforeach disorder, both at thecognitive level andat the etiological level. Insofar as single causeexplanations do not work, the alternative ap-proach has been to look for subtypes that havea single cause. The research reviewed here sug-gests that such an approach is doomed to fail-ure. LI, RD, and SSD are complex multifac-torial disorders, not only in terms of their ge-netic and environmental etiology, but also interms of their cognitive underpinnings. Eachdisorder appears to arise as the consequence of a specic constellation of underlying decits.Each individual decit may be common in thegeneral population and may only assume clin-ical signicance when combined with anotherdecit. Some decits, especially those affectingphonological processing, appear to play a part in all three disorders; others are specic to oneof the disorders.

We have shown that some evidence existsfor similar patterns of relations among thethree disorders at the three levels of analy-sis considered here: diagnostic, cognitive, andetiological. But more systematic research isneeded to test these patterns and dene therelations among these three communicationdisorders.

In the past, researchers often either ignoredcomorbidity or strenuously attempted to avoidit by studying pure groups. We argue that thisis the wrong approach, and that to understandthese disorders fully, we need to consider therelationships between them, both cognitive andetiological.

ACKNOWLEDGMENTS The preparation of this article was supported by NIH grants HD27802 and HD049027 (to BFP)and a Wellcome Trust Principal Research Fellowship (to DVMB). Barbara Lewis, Richard Olson,and Bruce Tomblin very generously provided unpublished data from their longitudinal samplesfor Table 2. Irina Kaminer and Jenni Rosenberg helped prepare this table and Suzanne Millerhelped prepare the manuscript.

LITERATURE CITED Anthoni H, Zucchelli M, Matsson H, Muller-Myhsok B, Fransson I, et al. 2007. A locus on 2p12 containing

the coregulated MRPL19 and C2ORF3 genes is associated to dyslexia. Hum. Mol. Genet. 16:66777


U

R

V


20/27

Bates EA. 2004. Explaining and interpreting decits in language development across clinical groups:do we go from here? Brain Lang. 88:24853

Beitchman JH, Nair R, Clegg M, Ferguson B, Patel PG. 1986. Prevalence of psychiatric disorders in c with speech and language disorders. J. Am. Acad. Child Psychiatry 25(4):52835

Berkson J. 1946. Limitations of the application of fourfold table analysis to hospital data. Biometrics 2:4Bird J, Bishop DV. 1992. Perception and awareness of phonemes in phonologically impaired childre

J. Disord. Commun. 27:289311Bird J, Bishop DV, Freeman NH. 1995. Phonological awareness and literacy development in childre

expressive phonological impairments. J. Speech Hear. Res. 38:44662Bishop DV. 1997. Uncommon Understanding: Development and Disorders of Language ComprehensionHove, UK: Psychol. Press

Bishop DV. 2001. Genetic inuences on language impairment and literacy problems in children: sadifferent? J. Child Psychol. Psychiatry 42:18998

Bishop DV. 2006. Developmental cognitive genetics: how psychology can inform genetics and vicQ. J. Exp. Psychol. 59:115368

Bishop DV, Adams C. 1990. A prospective study of the relationship between specic language impaphonological disorders and reading retardation. J. Child Psychol. Psychiatry 31:102750

Bishop DV, Adams CV, Norbury CF. 2004. Using nonword repetition to distinguish genetic and envmental inuences on early literacy development: a study of 6-year-old twins. Genes Brain Behav. 5:15

Bishop DV, Adams CV, Norbury CF. 2006. Distinct genetic inuences on grammar and phonological sterm memory decits: evidence from 6-year-old twins. Genes Brain Behav. 5:15869

Bishop DV, Bishop SJ, Bright P, James C, Delaney T, Tallal P. 1999. Different origin of auditoryphonological processing problems in children with language impairment: evidence from a twin J. Speech Lang. Hear. Res. 42:15568

Bishop DV, Edmundson A. 1986. Is otitis media a major cause of specic developmental language disBr. J. Disord. Commun. 21:32138

Bishop DV, Edmundson A. 1987. Language-impaired 4-year-olds: distinguishing transient from persimpairment. J. Speech Hear. Dis. 52:15673

Bishop DV, Haylou-Thomas ME. 2008. Heritability of specic language impairment depends on diagcriteria. Genes Brain Behav. In press

BishopDV,Snowling MJ.2004. Developmentaldyslexia andspecic language impairment: same or diff Psychol. Bull. 130:85886

Bishop DVM.1985.Spellingability in congenitaldysarthria: evidenceagainst articulatorycoding in transbetween graphemes and phonemes. Cogn. Neuropsychol. 2:22951

Bishop DVM,Hayiou-ThomasME.2008.Heritability of specic language impairment depends on diagncriteria. Genes Brain Behav. 7(3):36572

Boada R, Pennington BF. 2006. Decient implicit phonological representations in children with dy J. Exp. Child Psychol. 95:15393

Bronfenbrenner U, Ceci SJ. 1994. Nature-nurture reconceptualized in developmental perspective: a bilogical model. Psychol. Rev. 101:56886

Broomeld J, Dodd B. 2004. Children with speech and language disability: caseload characteristics I Lang. Comm. Dis. 39(3):30324

Byrne B, Shea P. 1979. Semantic and phonetic memory codes in beginning readers. Mem. Cogn. 7:333Cain K, Oakhill J, Lemmon K. 2004. Individual differences in the inference of word meanings from c

J. Educ. Psychol. 96:67181Campbell T, Dollaghan C, Needleman H, Janosky J. 1997. Reducing bias in language assessment: proc

dependent measures. J. Speech Lang. Hear. Res. 40:51925Campbell T, Dollaghan C, Rockette H, Paradise JL, Feldman HM, et al. 2003. Risk factors for speech

of unknown origin in 3-year-old children. Child. Dev. 74(2):34657Castles A, Coltheart M. 2004. Is there a causal link from phonological awareness to success in lea

read? Cognition 91:77111Catts HW. 1993. The relationship between speech-language impairments and reading disabilities. J. S

Hear. Res. 36:94858


U

R

V


21/27

Catts HW, Adlof SM, Hogan TP, Weismer SE. 2005. Are specic language impairment and dyslexia distinct disorders? J. Speech Lang. Hear. Res. 48:137896

Catts HW, Fey ME, Tomblin JB, Zhang X. 1999. Language basis of reading and reading disabilities. Sci. Stud. Reading 3:33161

Catts HW, Fey ME,Tomblin JB,Zhang X. 2002. A longitudinal investigation of reading outcomes in children with language impairments. J. Speech Lang. Hear. Res. 45:114257

CorriveauK, Pasquini E,Goswami U.2007. Basic auditoryprocessing skillsandspecic language impairment:a new look at an old hypothesis. J. Speech Lang. Hear. Res. 50:64766

Curtis ME. 1980. Development of components of reading skill. J. Educ. Psychol. 72:65669DAngiulli A, Siegel LS. 2003. Cognitive functioning as measured by the WISC-R: Do children with learningdisabilities have distinctive patterns of performance? J. Learn. Disabil. 36:4858

DeFries JC, Fulker DW. 1985. Multiple regression analysis of twin data. Behav. Genet. 15:46773Elbro C, Borstrom I, Petersen D. 1998. Predicting dyslexia from kindergarten: the importance of distinctness

of phonological representations of lexical items. Read. Res. Q. 33:3660Fisher SE. 2007. Molecular windows into speech and language disorders. Folia Phoniatrica et Logopaedica

59:13040Fisher SE, Francks C. 2006. Genes, cognition and dyslexia: learning to read the genome. Trends Cogn. Sci.

10:25057Fletcher JM,Foorman BR,Shaywitz SE,ShaywitzBA.1999. Conceptual andmethodological issuesin dyslexia

research: a lesson for developmental disorders. In Neurodevelopmental Disorders , ed. H Tager-Flusberg,pp. 271305. Cambridge, MA: MIT Press

Fowler A. 1991. How early phonological development might set the stage for phoneme awareness. In Phono-logical Processes in Literacy: A Tribute to Isabelle Y. Liberman, ed. SA Brady, DP Shankweiler, pp. 97117.Hillsdale, NJ: Erlbaum

Fowler AE, Swainson B. 2004. Relationships of naming skills to reading, memory, and receptive vocabulary:evidence for imprecise phonological representations of words by poor readers. Ann. Dyslexia 54:24780

Galaburda AM, Sherman GF, Rosen GD, Aboitiz F, Geschwind N. 1985. Developmental dyslexia: fourconsecutive cases with cortical anomalies. Ann. Neurol. 18:22233

Gathercole SE, Baddeley AD. 1990a. Phonological memory decits in language disordered children: Is therea causal connection? J. Mem. Lang. 29(3):336

Gathercole SE, Baddeley AD. 1990b. The role of phonological memory in vocabulary acquisition: a study of young children learning new names. Br. J. Dev. Psychol. 81:43954

Goswami U, Thomson J, Richardson U, Stainthorp R, Hughes D, et al. 2002. Amplitude envelope onsets anddevelopmental dyslexia: a new hypothesis. Proc. Natl. Acad. Sci. USA 99:1091116

Graf Estes K,Evans JL,Else-Quest NM.2007. Differences in thenonword repetition performance of children with and without specic language impairment: a meta-analysis. J. Speech Lang. Hear. Res. 50:17795

Grigorenko EL, Naples A, Chang J, Romano C, Ngorosho D, et al. 2007. Back to Africa: tracing dyslexiagenes in east Africa. Read. Writ. 20:2749

Guenther FH. 1995. Speech sound acquisition, coarticulation, and rate effects in a neural network model of speech production. Psychol. Rev. 102:594621

Harm MW, Seidenberg MS. 1999. Phonology, reading acquisition, and dyslexia: insights from connectionist models. Psychol. Rev. 106:491528

Hoover WA, Gough PB. 1990. The simple view of reading. Read. Writ. 2:12760 Joanisse MF. 2000. Connectionist phonology. Ph.D. thesis. Univ. South. Calif., Los Angeles Joanisse MF. 2004. Specic language impairments in children: phonology, semantics, and the English past

tense. Curr. Dir. Psychol. Sci. 13:15660 Joanisse MF. 2007. Phonological decits and developmental language impairments: evidence from connec-

tionist models. In Neuroconstructivism: Perspectives and Prospects , ed.D Mareschal, S Sirois, G Westermann, MH Johnson, pp. 20529. Oxford, UK: Oxford Univ. Press

Joanisse MF, Seidenberg MS. 2003. Phonology and syntax in specic language impairment: evidence from aconnectionist model. Brain Lang. 86:4056

Kenney MK, Barac-Cikoja D, Finnegan K, Jeffries N, Ludlow CL. 2006. Speech perception and short-termmemory decits in persistent developmental speech disorder. Brain Lang. 96:17890


U

R

V


22/27

Klein DN, Riso LP. 1993. Psychiatric disorders: problems of boundaries and comorbidity. In Basic Is Psychopathology, ed. CG Costello, pp. 1966. New York: Guilford

Kuhl PK. 2004. Early language acquisition: cracking the speech code. Nat. Rev. Neurosci. 5:83143Larrivee LS, Catts HW. 1999. Early reading achievement in children with expressive phonological dis

Am. J. Speech Lang. Pathol. 8:11828Leitao S, Fletcher J. 2004. Literacy outcomes for students with speech impairment: long-term follo

Int. J. Lang. Commun. Dis. 39:245Leitao S, HogbenJ, Fletcher J. 1997. Phonologicalprocessing skills in speechandlanguage impaired chi

Eur. J. Disord. Commun. 32:91111Leonard LB. 1995. Phonological impairment. In The Handbook of Child Language, ed. P Fletch

MacWhinney, pp. 573602. Oxford, UK: BlackwellLewis BA, Freebairn LA.1992.Residual effects of preschoolphonology disordersin grade school, adoles

and adulthood. J. Speech Hear. Res. 35:81931Lewis BA, Freebairn LA, Hansen AJ, Miscimarra L, Iyengar SK, Taylor HG. 2007. Speech and langua

of parents of children with speech sound disorders. Am. J. Speech Lang. Pathol. 16:10818Lewis BA, Freebairn LA, Hansen AJ, Stein CM, Shriberg LD, et al. 2006. Dimensions of early speech

disorders: a factor analytic study. J. Commun. Dis. 39:13957Lewis BA, Freebairn LA, Taylor HG. 2000. Follow-up of children with early expressive phonology di

J. Learn. Disabil. 33:433 Markey KL. 1994. The sensorimotor foundations of phonology: a computational model of early childh

and phonetic development . Ph.D. thesis. Univ. Colorado, Boulder McArthur GM, Bishop DV. 2001. Auditory perceptual processing in people with reading and oral lan

impairments: current issues and recommendations. Dyslexia 7(3):15070 McArthur GM, Hogben JH, Edwards VT, Heath SM, Mengler ED. 2000. On the specics of sp

reading disability and specic language impairment. J. Child Psychol. Psychiatry 41:86974 McGrath LM, Hutaff-Lee C, Scott A, Boada R, Shriberg LD, Pennington BF. 2007. Children wit

morbid speech sound disorder and specic language impairment have increased rates of attdecit/hyperactivity disorder. J. Abnorm. Child Psychol. 36:15163

McGrath LM, Pennington BF, Willcutt EG, Boada R, Shriberg LD, Smith SD. 2007. Gene envment interactions in speech sound disorder predict language and preliteracy outcomes. Dev. Psycho19:104772

McGrath LM, Smith SD, Pennington BF. 2006. Breakthroughs in the search for dyslexia candidate Trends Mol. Med. 12:33341 Menn L, Markey KL, Mozer M, Lewis C. 1993. Connectionist modeling and the microstructure of p

logical development: a progress report. In Developmental Neurocognition: Speech and Face Proce First Year of Life, ed. B. de Boysson-Bardies, pp. 42133. Dordrecht, The Netherlands: Kluwer Ac

Miller CA, Kail R, Leonard LB, Tomblin JB. 2001. Speed of processing in children with specic limpairment. J. Speech Lang. Hear. Res. 44(2):41633

Monaco AP. 2007. Multivariate linkage analysis of specic language impairment (SLI). Ann. Hum. G71:66073

Morais J, Cary L, Alegria J, Bertelson P. 1979. Does awareness of speech as a sequence of phospontaneously? Cognition 7:32331

Morton J, Frith U. 1995. Causal modeling: structural approach to developmental psychopathology. In D

opmental Psychopathology, ed. D Cicchetti, DJ Cohen, pp. 35790. New York: Wiley Nathan L, Stackhouse J, Goulandris N, Snowling MJ. 2004. The development of early literacy skills

children with speechdifculties: a test of theCritical Age Hypothesis. J. Speech Lang. Hear. Res.4791

Nation K. 2005. Childrens reading comprehension difculties. In The Science of Reading , ed. MJ SnowC Hulme, pp. 24865. Oxford, UK: Blackwell Sci.

Neale MC,Kendler KS.1995. Models of comorbidity formultifactorial disorders. Am. J. Hum. Genet.5753


U

R

V


23/27

Newbury DF, Monaco AP. 2008. Theapplication of molecular genetics to thestudy of language impairments.In Understanding Developmental Language Disorders , ed. CF Norbury, JB Tomblin, DVM Bishop. Hove,UK: Psychol. Press

Nicolson RI, Fawcett AJ. 1990. Automaticity: a new framework for dyslexia research? Cognition 35:15982Nicolson RI, Fawcett AJ, Dean P. 2001. Developmental dyslexia: the cerebellar decit hypothesis. Trends

Neurosci. 24:50811Nittrouer S. 1999. Do temporal processing decits cause phonological processing problems? J. Speech Lang.

Hear. Res. 42:92542

Paracchini S, Scerri T, Monaco AP. 2007. The genetic lexicon of dyslexia. Annu. Rev. Genom. Hum. Genet.8:5779

Pennington BF. 2006. From single to multiple decit models of developmental disorders. Cognition 101:385413

Pennington BF, Groisser D, Welsh MC. 1993. Contrasting cognitive decits in attention decit hyperactivity disorder versus reading disability. Dev. Psychol. 29:51123

Pennington BF, Ley DL. 2001. Early reading development in children at family risk for dyslexia. Child Dev.72:81633

Pennington BF, Olson RK. 2005. Genetics of dyslexia. In The Science of Reading: A Handbook, ed. M Snowling,C Hulme, pp. 45372. Oxford, UK: Blackwell Sci.

Pennington BF, Willcutt EG, Rhee SH. 2005. Analyzing comorbidity. In Advances in Child Development and Behavior , ed. RV Kail, pp. 263304. Oxford, UK: Elsevier

Perfetti CA, Beck I, Bell LC, Hughes C. 1987. Phonemic knowledge and learning to read are reciprocal: alongitudinal study of rst grade children. Merrill-Palmer Q. 33:283319

Pinker S. 1991. Rules of language. Science 253:53035Plaut DC, Kello CT. 1999. The emergence of phonology from the interplay of speech comprehension and

production: a distributed connectionist approach. In The Emergence of Language, ed. B MacWhinney,pp. 381415. Hillsdale, NJ: Erlbaum

Plomin R, Kovas Y. 2005. Generalist genes and learning disabilities. Psychol. Bull. 131:592617Raitano NA, Pennington BF, Tunick RA, Boada R, Shriberg LD. 2004. Pre-literacy skills of subgroups of

children with speech sound disorders. J. Child Psychol. Psychiatry 45:82135Ramus F. 2003. Developmental dyslexia: specic phonological decit or general sensorimotor dysfunction?

Curr. Opin. Neurobiol. 13(2):21218Rhee SH, Hewitt JK, Lessem JM, Stallings MC, Corley RP, Neale MC. 2004. The validity of the Neale and

Kendler model-tting approach in examining the etiology of comorbidity. Behav. Genet. 34:25165Rice ML, Wexler K. 1996. Toward tense as a clinical marker of specic language impairment in English-

speaking children. J. Speech Hear. Res. 39:123957Rutter M. 2006. Genes and Behavior: Nature-Nurture Interplay Explained . Oxford, UK: Blackwell Sci.Rutter M, Caspi A, Fergusson DM, Horwood LJ, Goodman R, et al. 2004. Gender differences in reading

difculties: ndings from four epidemiological studies. JAMA 291:200712Saffran JR, Aslin RN, Newport EL. 1996. Statistical learning by 8-month-old infants. Science 274:192628Scarborough HS. 1990. Very early language decits in dyslexic children. Child Dev. 61:172843Scarborough HS, Dobrich W. 1990. Development of children with early language delay. J. Speech Hear. Res.

33(1):7083Scarborough HS, Parker JD. 2003. Matthew effects in children with learning disabilities: development of

reading, IQ, and psychosocial problems from grade 2 to grade 8. Ann. Dyslexia 53:4771

Serniclaes W, Van Heghe S, Mousty P, Carre R, Sprenger-Charolles L. 2004. Allophonic mode of speechperception in dyslexia. J. Exp. Child Psychol. 87:33661

Shanahan M, Yerys B, Scott A, WillcuttE, DeFries JC,et al. 2006. Processing speed decits in attentiondecit hyperactivity disorder and reading disability. J. Abnorm. Child Psychol. 34(5):585602

Shankweiler DP, Liberman IY, Mark LS, Fowler CA, Fischer FW. 1979. The speech code and learning toread. J. Exp. Psychol.: Hum. Learn. Memory 5:53145

Shriberg LD, Tomblin JB, McSweeny JL. 1999. Prevalence of speech delay in 6-year-old children and comor-bidity with language impairment. J. Speech Lang. Hear. Res. 42:146181


U

R

V


24/27

Silva PA, Williams S, McGee R. 1987. A longitudinal study of children with developmental language age three: later intelligence, reading and behaviour problems. Dev. Med. Child Neurol. 29(5):6304

SmithSD,PenningtonBF, Boada R,ShribergLD.2005.Linkage of speechsounddisorder toreading disabloci. J. Child Psychol. Psychiatry 46:104556

Snowling M, Bishop DV, Stothard SE. 2000. Is preschool language impairment a risk factor for dysadolescence? J. Child Psychol. Psychiatry 41:587600

StanovichKE.1986.Matthew effects in reading: someconsequencesof individual differences in the acquof literacy. Read. Res. Q. 21:360406

Stein CM, Millard C, Kluge A, Miscimarra LE, Cartier KC, et al. 2006. Speech sound disorder inuena locus in 15q14 region. Behav. Genet. 36:85868

Stein CM, Schick JH, Taylor H, Shriberg LD, Millard C, et al. 2004. Pleiotropic effects of a chromoslocus on speech-sound disorder and reading. Am. J. Hum. Genet. 74:28397

Swan D, Goswami U. 1997a. Phonological awareness decits in developmental dyslexia and the phonrepresentations hypothesis. J. Exp. Child Psychol. 66:1841

Swan D, Goswami U. 1997b. Picture naming decits in developmental dyslexia: the phonological retations hypothesis. Brain Lang. 56:33453

Tallal P. 1980. Auditory temporal perception, phonics, and reading disabilities in children. Brain Lang. 998

Tallal P. 2004. Improving language and literacy is a matter of time. Nat. Rev. Neurosci. 5:72128 Tallal P, Piercy M. 1973. Developmental aphasia: impaired rate of nonverbal processing as a func

sensory modality. Neuropsychologia 11:38998 Tiu RD Jr, Wadsworth SJ, Olson RK, DeFries JC. 2004. Causal models of reading disability: a twinTwin Res. 7:27583

Tomblin JB, Hardy JC, Hein HA. 1991. Predicting poor-communication status in preschool children risk factors present at birth. J. Speech Hear. Res. 34(5):1096105

Tomblin JB, Records NL, Buckwalter P, Zhang X, Smith E, OBrien M. 1997. Prevalence of specic laimpairment in kindergarten children. J. Speech Lang. Hear. Res. 40(6):124560

TreimanR, PenningtonBF, ShribergLD, Boada R. 2008. Which childrenbenet fromletternames in learnletter sounds? Cognition 106(3):132238

Ullman MT, Pierpont EI. 2005. Specic language impairment is not specic to language: the procedecit hypothesis. Cortex 41:399433

van der Lely HK. 1994. Canonical linking rules: forward versus reverse linking in normally develop

specically language-impaired children. Cognition 51:2972 Viding E, Spinath FM, Price TS, Bishop DV, Dale PS, Plomin R. 2004. Genetic and environmental ence on language impairment in 4-year-old same-sex and opposite-sex twins. J. Child Psychol. Ps45:31525

Wagner RK, Torgesen JK, Rashotte CA. 1994. Development of reading-related phonological procabilities: new evidence of bidirectional causality from a latent variable longitudinal study. Dev. P30:7387

Westermann G, Miranda ER. 2004. A new model of sensorimotor coupling in the development of sBrain Lang. 89:393400


U

R

V


25/27

Annual Review ofPsychology

Volume 60, 2009Contents

Prefatory

Emotion Theory and Research: Highlights, Unanswered Questions,and Emerging IssuesCarroll E. Izard 1

Concepts and Categories

Concepts and Categories: A Cognitive Neuropsychological PerspectiveBradford Z. Mahon and Alfonso Caramazza 27

Judgment and Decision Making

Mindful Judgment and Decision Making Elke U. Weber and Eric J. Johnson 53

Comparative Psychology

Comparative Social Cognition Nathan J. Emery and Nicola S. Clayton 87

Development: Learning, Cognition, and Perception

Learning from Others: Childrens Construction of ConceptsSusan A. Gelman 115

Early and Middle Childhood

Social Withdrawal in Childhood Kenneth H. Rubin, Robert J. Coplan, and Julie C. Bowker 141

Adulthood and Aging

The Adaptive Brain: Aging and Neurocognitive ScaffoldingDenise C. Park and Patricia Reuter-Lorenz 173

Substance Abuse Disorders

A Tale of Two Systems: Co-Occurring Mental Health and Substance Abuse Disorders Treatment for Adolescents Elizabeth H. Hawkins 197

vii

U

R

V


26/27

Therapy for Specic Problems

Therapy for Specic Problems: Youth Tobacco CessationSusan J. Curry, Robin J. Mermelstein, and Amy K. Sporer

Adult Clinical Neuropsychology

Neuropsychological Assessment of Dementia

David P. Salmon and Mark W. Bondi

Child Clinical Neuropsychology

Relations Among Speech, Language, and Reading DisordersBruce F. Pennington and Dorothy V.M. Bishop

Attitude Structure

Political Ideology: Its Structure, Functions, and Elective Afnities John T. Jost, Christopher M. Federico, and Jaime L. Napier

Intergroup relations, stigma, stereotyping, prejudice, discrimination

Prejudice Reduction: What Works? A Review and Assessment of Research and Practice Elizabeth Levy Paluck and Donald P. Green

Cultural Inuences

Personality: The Universal and the Culturally SpecicSteven J. Heine and Emma E. Buchtel

Community Psychology

Community Psychology: Individuals and Interventions in Community

Context Edison J. Trickett

Leadership

Leadership: Current Theories, Research, and Future DirectionsBruce J. Avolio, Fred O. Walumbwa, and Todd J. Weber

Training and Development

Benets of Training and Development for Individuals and Teams,Organizations, and Society Herman Aguinis and Kurt Kraiger

Marketing and Consumer Behavior

Conceptual ConsumptionDan Ariely and Michael I. Norton

vii i Contents

U

R

V


27/27

Psychobiological Mechanisms

Health Psychology: Developing Biologically Plausible Models Linkingthe Social World and Physical HealthGregory E. Miller, Edith Chen, and Steve Cole 501

Health and Social Systems

The Case for Cultural Competency in Psychotherapeutic InterventionsStanley Sue, Nolan Zane, Gordon C. Nagayama Hall, and Lauren K. Berger 525

Research Methodology

Missing Data Analysis: Making It Work in the Real World John W. Graham 549

Psychometrics: Analysis of Latent Variables and Hypothetical Constructs

Latent Variable Modeling of Differences and Changes withLongitudinal Data John J. McArdle 577

Evaluation

The Renaissance of Field Experimentation in Evaluating InterventionsWilliam R. Shadish and Thomas D. Cook 607

Timely Topics

Adolescent Romantic RelationshipsW. Andrew Collins, Deborah P. Welsh, and Wyndol Furman 631

Imitation, Empathy, and Mirror Neurons Marco Iacoboni 653

Predicting Workplace Aggression and Violence Julian Barling, Kathryne E. Dupr e, and E. Kevin Kelloway 671

The Social Brain: Neural Basis of Social Knowledge Ralph Adolphs 693

Workplace Victimization: Aggression from the Targets Perspective Karl Aquino and Stefan Thau 717

Indexes

Cumulative Index of Contributing Authors, Volumes 5060 743Cumulative Index of Chapter Titles, Volumes 5060 748

Errata

An online log of corrections to Annual Review of Psychology articles may be found at h // h l i / h l

U

R

V

Relations Amogn Speech Lenguaje and Reading Disorders

Documents

Transcript of Relations Amogn Speech Lenguaje and Reading Disorders