Childhood Speech Sound Disorders: From Postbehaviorism to

Introduction

Thanks

I’ll never forget the sunny spring day inMadison when Rhea and Peter spilled thebeans about this book. What an incredible,wonderful surprise. It took two monthsbefore I could get my head around the real-ity of this gracious gesture and hunkerdown to begin my writing assignment. Rheaand Peter asked me to sketch the “arc” ofmy research to date. I’ve tried to capturethat rather wobbly line in the title and con-tent of this chapter, and more personally, inthe next few paragraphs. I’m sorry thereisn’t room to thank people individually byname. I trust that each of you will recognizeyourself and your influence in the follow-ing brief chronology.

My interest in causality research datesback to the masters program in Commu-nicative Disorders at Boston University,where I found an engaging faculty and a ter-rific group of fellow students with diverselife experiences. After an aimless series ofundergraduate majors at Syracuse Univer-sity, followed by a string of forgettable jobs,BU was a stimulating and challenging expe-rience. We talked in and out of class aboutthe big stuff (e.g., ‘the subsoils of humanexistence’), which somehow set me on aquest to try to understand the origins ofspeech disorders of unknown origin. Thedetective work continued at my first clini-cal position in Bridgeport, Connecticut—abusy rehabilitation center where clientsand their families taught me so much morethan I helped them.

At the Lawrence and the Medical Cen-ter campuses of the University of Kansas,

CHAPTER 1

Childhood Speech SoundDisorders: From

Postbehaviorism to thePostgenomic Era

LAWRENCE D. SHRIBERG

01_Flipsen_1-34 10/14/09 7:43 PM Page 1

I had the good fortune to learn “dust-bowlempiricist” research from a scholarly facultyand a knowledgeable, fun-loving, and veryverbal gang of doctoral students. The “post-behaviorism” in the chapter title alludes tothis heady period in our discipline whencarefully planned and reported treatmentresearch offered the possibility to effect significant behavioral and social change.There is much about the earnest goals ofthe current focus on evidence-based prac-tice that is reminiscent of the Zeitgeist ofthis period.

At the Department of CommunicativeDisorders, University of Wisconsin-Madisonand the Communication Processes Unit atthe Waisman Center, I have been privilegedto have long-term interactions with anextraordinary cohort of academic and clin-ical colleagues and forward-thinking admin-istrators. I want to acknowledge the contri-butions of dozens of wonderful alumnaefrom our research group at the PhonologyProject, many doctoral and postdoctoralresearchers who have shared their skillsand enthusiasm with us, and investigators inMadison and elsewhere with whom I havehad the honor of working in past and con-tinuing collaborative projects. One long-term collaborator and good friend I willthank by name is Joan Kwiatkowski, whocontinues to contribute her immense talentto speech research and to set a standard for clinical efficacy in our university andPhonology Project speech clinics. It’s sucha joy to share with these colleagues theexcitement on the other side of the arc—the boundless opportunities for discoveryin the current “postgenomic era.”

To Rhea, Peter, and each of the othergood friends who have written such lucidpapers for this volume—and to SadanandSingh, a long-time friend and tireless advo-cate for our discipline—my humble andheartfelt thanks.

Overview

What follows is the latest version of “theTalk.” I seem to have been updating varia-tions of this presentation for a very longtime. It is an overview of a vision to developand validate an etiologic classification sys-tem for childhood speech sound disordersof currently unknown origin. I first intro-duce a set of working terms and conceptsthat constitute the nosological frameworkfor the system. Then, I discuss epidemio-logic and other research findings viewed assupport for the hypothesis of etiologic sub-types of childhood speech sound disorders.

I hope this review enriches or at leastcomplements each of the thoughtful essaysby my colleagues. I thank them in advancefor playing nicely with me at points in theirdiscussions where they may need to addressthe many gaps in theoretical and empiricalsupport for the proposals offered in the fol-lowing work in progress.

Explanation in Speech SoundDisorders

The American Speech-Language-HearingAssociation’s recent adoption of the termspeech sound disorders (SSD) is a welcomesolution to the constraints associated withthe articulation disorders versus phono-logical disorders dichotomy of the past threedecades. The term SSD provides a theory-neutral cover term for researchers and cli-nicians who may, as I do, view SSD as acomplex neurodevelopmental disorder.Theterm childhood (or in medical contexts,pediatric) speech sound disorders, whichparallels the term childhood language dis-orders, unifies the study of speech sounddisorders of both known (e.g., Down syn-drome, cleft palate) and presently unknownorigin.

2 SPEECH SOUND DISORDERS IN CHILDREN

01_Flipsen_1-34 10/14/09 7:43 PM Page 2

Figure 1–1 is a sketch of four epochs inthe history of causality research in childhoodspeech sound disorders. The first epoch isthe 40-year period from the earliest researchstudies in this country in the 1920s until themany classic studies of the 1950s, in whichepidemiologic and descriptive linguisticmethods were used to identify and classifychildren with speech sound errors. Espe-cially toward the end of this period, distalcauses of speech errors were addressedand research focused primarily on explana-tory theories and constructs from articula-tory phonetics, speech motor control, anddevelopmental psychology.

In the following perhaps 30-year period,both linguistic descriptions and causal stud-ies of SSD changed markedly. Methods inour discipline included a succession ofalternative descriptive, psycholinguistic,and sociolinguistic paradigms from allieddisciplines, with markedly decreased inter-est in the search for distal causes of SSD.Focus clearly shifted to the identificationand delineation of core deficits in proximalprocesses that constrain speech acquisitionand performance.

A third epoch, lasting perhaps 10 years(note the shrinking shelf-life of epochs), wasdubbed the decade of the brain. Advances

in neuroimaging and other assessment tech-nologies enabled renewed interest in bothdistal and proximal causal perspectivesunderlying SSD, especially, in the presentcontext, as it became possible to describeneural correlates of speech sound process-ing more directly.

Our discipline is currently enjoying theopportunities presented in a fourth epoch—the postgenomic era. Following the success-ful conclusion of the Human Genome Pro-ject in 2001, continuous technical advancesmake it possible to study the distal originsof many putative sources of SSD. Overviewsof the current period often allude to the“Omics,” with levels of explanation pro-ceeding downstream from the genome:Genomics > Transcriptomics > Proteomics> Glycomics > Metabalomics > Epigenomics> Phenomics and others. Vernes et al.(2006) report the first example of func-tional genetic analyses of a gene underlyingone subtype of SSD (FOXP2), demonstrat-ing the potential of neurodevelopmentalresearch using systems biology.

We take the perspective that an etio-logic classification system for SSD is neededif this highly prevalent disorder (to be dis-cussed) is to participate in the scientific andclinical advances being achieved in other

CHILDHOOD SPEECH SOUND DISORDERS 3

Figure 1–1. Four epochs of causality research in speech sound disorders.

01_Flipsen_1-34 10/14/09 7:43 PM Page 3

childhood diseases and disorders. The fol-lowing section describes a research frame-work proposed for this challenge.

The Speech DisordersClassification System (SDCS)

Speech Disorders ClassificationSystem–Typology (SDCS-T)

Figure 1–2 is the Speech Disorders Classi-fication System (SDCS), a framework forresearch in SSD that has evolved from rudi-mentary descriptions (Shriberg, 1980, 1982a,1982b; Shriberg & Kwiatkowski, 1982), acall for speech-genetics research (Shriberg,1993), and several preliminary presenta-tions (Shriberg, 1994, 1997; Shriberg,Austin,Lewis, McSweeny, & Wilson, 1997b). Theleft arm of the SDCS, titled SDCS-T, providesa typologic nosology that divides SSD ofunknown origin into two subtypes. Themore clinically significant subtype is termedspeech delay (SD) with delay highlightingthe finding that most children with this sub-type of SSD normalize with treatment. TheSDCS defines SD as a pattern of speechsound deletions and/or substitutions char-acteristic of Ingram’s (1989) PhonologicalStage III that persists past 4 years of age (cf.Shriberg & Kwiatkowski, 1982; Shriberg,Kwiatkowski, & Gruber, 1994). Notice thatwe use the term SD as one of two subtypesof SSD, whereas SSD and SD typically areused synonymously in the literature. Asreviewed shortly, there are few data on therisk and protective factors that predict nor-malization versus persistence of SD at 6 yearsof age (Peterson, Pennington, Shriberg, &Boada, in press). Crucially, although speechsound production errors may normalize withtreatment, SD places a child at increased

risk for literacy delays (Hesketh, 2004; Leitão& Fletcher, 2004; Raitano, Pennington,Tunick, Boada, & Shriberg, 2004; Shriberg &Kwiatkowski, 1988), lowered self-concept(Barrett & Hoops, 1974), and restrictedvocational choices (Felsenfeld, Broen, &McGue, 1994).

The second subtype of SSD of currentlyunknown origin shown in Figure 1–2 istermed speech errors (SE). Children with SEhave histories of speech sound distortionerrors (for English-speaking children typi-cally on sibilants and rhotics) that are notassociated with the risk domains docu-mented for SD and that do not interferewith intelligibility. The prevalence of SEbelow age 9 is estimated at approximately5% (Shriberg & Austin, 1998). Reviews ofthe limited epidemiologic data indicate thatafter 9 years of age, 1 to 2% of adolescentsand adults have one or more of a small setof residual distortion errors from prior SDor SE, errors that may persist for a lifetime(Lewis & Shriberg, 1994).

Speech Disorders ClassificationSystem–Etiology (SDCS-E)

The right arm of Figure 1–2 is termed theSpeech Disorders Classification System–Etiology (SDCS-E). The SDCS-E provides theconceptual framework and working terms forseven etiologic subtypes of SSD. Table 1–1includes additional speculation on geneticversus environmental contributions under-lying each of seven subtypes. The centralclaim is that SD and SE do not arise fromthe same monolithic causal domain andthat each includes etiologic subtypes. Thehypothesis for SD is that it includes fiveindividual and overlapping etiologies, eachwith one or more distal and proximal ori-gins with risk and protective factors in both genetic and environmental domains.


01_Flipsen_1-34 10/14/09 7:43 PM Page 4

5

Fig

ure

1–2

.A

fra

mew

ork

for

caus

ality

res

earc

h in

chi

ldho

od s

peec

h so

und

diso

rder

s.

01_Flipsen_1-34 10/14/09 7:43 PM Page 5

The hypothesis for SE is that it includes twosubtypes, each based on a different groupof environmental risk factors (Shriberg,1994).

A set of working terms (and theirabbreviations) is used to reference childrenwhose speech delay may be due to one ormore of the five proposed distal-proximalorigins shown in Figure 1–2 and Table 1–1.The five etiologic subtypes of SD are thoseassociated with (a) cognitive-linguistic pro-cessing constraints that may be, in part,genet-ically transmitted (SD-GEN); (b) auditory-perceptual processing constraints that arethe consequence of the fluctuant conduc-tive hearing loss associated with early recur-

rent otitis media with effusion (SD-OME);(c) affective, temperamental processing con-straints associated with developmental psy-chosocial involvement (SD-DPI); (d) speechmotor planning/programming constraintsconsistent with apraxia of speech (SD-AOS);and (e) speech motor execution constraintsconsistent with several forms of dysarthria(SD-DYS). The term motor speech disorder(MSD) is used for children suspected tohave either or both of the latter two senso-rimotor speech disorders. It is important tounderscore the epidemiologic observationthat a significant proportion of childrenwith SD have involvement in two or moreof the five distal and proximal domains.


Table 1–1. Seven Subtypes of Speech Sound Disorders in the Speech DisordersClassification System-Etiology (SDCS-E)

Working Term

1 Speech Delay–Genetic

2 Speech Delay–Otitis Media with Effusion

3 Speech Delay–Developmental Psychological Involvement

4 Speech Delay–Apraxia of Speech

5 Speech Delay–Dysarthria

6 Speech Errors–Sibilants

7 Speech Errors–Rhotics

Undifferentiated Speech Delay

Undifferentiated Speech Sound Disorders

Abbreviation

SD-GEN

SD-OME

SD-DPI

SD-AOS

SD-DYS

SE-S

SE-R

USD

USSD

Primary Origin

Polygenic/Environmental



Monogenic?Oligogenic?

Monogenic?Oligogenic?

Environmental

Environmental

Any of 1–5

Any 1–7

Processes Affected

Cognitive-Linguistic

Auditory-Perceptual

Affective-Temperamental

Speech Motor Control

Speech Motor Control

PhonologicalAttunement

PhonologicalAttunement

Any of 1–5

Any 1–7

01_Flipsen_1-34 10/14/09 7:43 PM Page 6

The two subtypes of speech errors (SE)included in the SDCS-E provide classifica-tions for English speakers who have tran-sient or persistent distortions of sibilants(SE-S) and/or rhotics (SE-R). Changing viewsof handicap and competing service deliveryneeds have greatly affected research andapplied interests in children and adults withSE. Although SE was well studied in the firsttwo epochs delineated in Figure 1–1, persist-ent speech sound distortions such as den-talized /s/, lateralized /s/, derhotacized /r/, orvelarized / l/ are currently viewed as havingnegligible or minor social consequences.The causal origins of such allophones andtheir natural histories, however, remain ofconsiderable theoretical interest. We haveproposed a variant of attunement theory(phonological attunement) to account forsociodemographic differences observed be-tween children with SD and SE and betweenchildren with each of the two proposedsubtypes of SE (Shriberg, 1975, 1994).

The dashes in the lower rows of theright arm in Figure 1–2 are placeholders for the research and applied goals of theSDCS-E. Our aims have been to providediagnostic markers that discriminate eachof the five etiologic subtypes of SD, and to

develop binary and quantitative phenotypesand endophenotypes for use in moleculargenetic studies. As discussed presently, en-couraging progress has been made towardfilling in the blanks.

Finally, the two cover terms at the bot-tom of Table 1–1 are needed to differentiateamong research samples. Undifferentiatedspeech sound disorders (USSD) is a usefulclass term for speakers who may have eitherSD or SE.Undifferentiated speech delay (USD)is a useful class term for speakers who haveor have had SD, but have not been differen-tiated on the basis of the proposed etiologicsubtypes (e.g., SD-GEN, SD-OME, SD-DPI,SD-AOS, SD-DYS).

Epidemiology of Speech Sound Disorders

Prevalence and Persistence ofSpeech Delay

Figure 1–3 includes prevalence estimatesfor USD (i.e., speakers who meet criteriafor SD, but are undifferentiated relative tosubtypes of SD). Methodological details for


Figure 1–3. Prevalence estimates for Undifferentiated Speech Delay (USD) at six yearsof age (adapted from Shriberg, Tomblin, & McSweeny, 1999).

01_Flipsen_1-34 10/14/09 7:43 PM Page 7

the population-based sample from whichestimates were obtained are provided inShriberg, Tomblin, and McSweeny (1999).The prevalence estimate of 3.8% at 6 yearsof age in the left panel indicates that SD ishighly prevalent, with implications for bothgenetic and environmental explanatoryaccounts and service delivery needs. Becauseseveral of our prospective studies have indi-cated that approximately 75% of childrenhave normalized their SD by 6 years of age,we estimate that SD occurs in 15 to 16% ofchildren at 3 years of age. A study by Camp-bell et al. (2003) using the SDCS-T to clas-sify speech disorders cross-validated thatprojection, reporting a prevalence of SD atage 3 of 15.2%. This latter prevalence esti-mate is the area approximately equivalentto one-standard deviation below the normalcurve, suggesting that speech acquisition isa normally distributed trait with SD reflect-ing scores below the 16th percentile. Froma genetic perspective, as posited in Table1–1, the mode of genetic transmission for anormally distributed trait is consistent withpolygenic, rather than monogenic or oligo-genic transmission, in addition to risk asso-ciated with environmental sources. Theprevalence estimate for males (4.5%) com-pared to females (3.1%), a ratio of 1.5:1, andthe differing prevalence estimates associatedwith the three geographic strata shown inthe right panel in Figure 1–3, further sup-port the need for explanatory models thatinclude both genetic and environmental riskand protective factors. Notice that the preva-lence estimate for children from urban strata(4.9%) is more than twice that for childrenfrom rural strata (2.3%). Campbell et al.(2003) reported three genetic-environmentalrisk factors that best predicted SD: male,lower educational level of the mother, anda history of SD in other family members.

The three panels in Figure 1–4 providepreliminary estimates of short- and long-termnormalization of SD, with implications for

etiologic subtypes of SD (Lewis & Shriberg,1994). From 9 to 12 years of age, the periodwhen even severe SD should normalize,about 30% of children shown in the twoindependent subsamples retained distortionsof sibilants and rhotics.That figure droppedto approximately 9% from 12 to 18 years.Rhotic errors persisted after 18 years in 9%of children with prior SD, possibly persist-ing throughout these individuals’ lifetimes.

The normalization/persistence data inFigure 1–4 raise questions of considerableinterest for explanatory accounts of SSD.Clinical experience indicates that dental-ized and lateralized sibilant errors persist inadults. Yet the preliminary estimates in Fig-ure 1–4, which are based on children whohad SD, not SE, indicate persistence prima-rily of rhotic distortions. If reliable, whatexplanatory mechanisms might account forthis difference in the persistence of sub-classes of residual errors? Epidemiologicdata using population sampling designscould provide such information, particularlyas there are now cohorts of adults withuntreated SE due to school districts’ con-temporary definitions of handicap. Laterdiscussion of research findings in SD-GENwill consider related issues.

Emerging Epidemiologic Data for Subtypes of Speech Sound Disorders

We have been continuously updating epi-demiologic and speech findings for thehypothesis of etiologic subtypes of SSD.Theentries in Table 1–2 are current estimatesfor five such variables, based on publishedand unpublished clinical samples from col-laborative studies. The question marks indi-cate cells for which there presently are noavailable preliminary estimates, mainly dueto the lack of emerging diagnostic markersfor some SDCS-E subgroups.


01_Flipsen_1-34 10/14/09 7:43 PM Page 8

9

Fig

ure

1–4

.P

oint

pre

vale

nce

estim

ates

for

resi

dual

spe

ech

erro

rs in

chi

ldre

n w

ith p

rior

SD

(R

E-S

D).

01_Flipsen_1-34 10/14/09 7:43 PM Page 9

Clinical Prevalence

The clinical prevalence estimates in Table 1–2are the percentages of children with speechdelay in study samples who met diagnosticcriteria for the five proposed etiologic sub-types.These estimates are based on referralsto one university-affiliated speech clinic(Hauner, Shriberg, Kwiatkowski, & Allen,2005; Shriberg & Kwiatkowski, 1994). UsingSDCS inclusionary markers for each proposedsubtype (see later discussion), estimates indi-cate that SD-GEN (56%), SD-OME (30%), andSD-DPI (12%) account for 98% of childrenreferred for assessment/treatment of SD, withthe remaining 2% possibly having apraxia ofspeech and/or dysarthria. Again, these esti-mates are based on referral rather than pop-ulation samples, and are not adjusted foroverlapping categories. If cross-validated inother clinics, and perhaps in a large epidemi-ologic study of speech sound disorders, theyare presumably informative for research, clin-ical training, and treatment. Notably, theysuggest that the majority of children with SDwould best profit from treatment procedures

that address delays in cognitive-linguisticprocesses (i.e., the proximal causes of SDfor SD-GEN; see Figure 1–2 and Table 1–1).They also suggest that a significant percent-age of children might require treatmentprocedures that address alternative or addi-tional speech processing needs.

Sex Ratios

The 1.5:1 boys-to-girls prevalence estimatenoted previously for USD (see Figure 1–3)may not obtain for each of the proposedsubtypes of SD. It holds for the estimated68% of children with SD posited to have SD-GEN (56%) or SD-DPI (12%), but not forSD-OME, in which the sex ratio is estimatedat 1:1. Boys-to-girls ratios for SD-AOS havebeen estimated to even more greatly favorboys (Hall, Jordan, & Robin, 1993), but arecent review of 55 cases reported to havegenetically-based (i.e., nonidiopathic) SD-AOSoccurring in complex neurodevelopmentaldisorders indicated approximately equalpercentages of boys and girls (Shriberg,in press). Finally, as indicated in Table 1–2,


Table 1–2. Epidemiologic Estimates for Seven Proposed Etiological Subtypes of SSD

Speech Delay (SD) Speech Errors (SE)

Variables SD-GEN SD-OME SD-DPI SD-AOS SD-DYS SE-S SE-R

Clinical 56% 30% 12% <1% ? ? ?Prevalence

Sex M > F M = F M > F M > > F ? M < F M > F

Delayed NO NO NO YES ? NO NOOnset ofSpeech

Language YES YES YES YES ? NO NODisorder

Normalization EARLY ? LATE LATE ? Variable Variableof Speech

01_Flipsen_1-34 10/14/09 7:43 PM Page 10

preliminary findings of more boys with SE-R (rhotics) errors and more girls with SE-S(sibilant) errors are viewed as support foran attunement theory of SE discussed else-where (Shriberg, 1975, 1994).

Onset of Speech

As indicated in Table 1–2, a significant delayin the onset of speech may be one featurethat differentiates SD-AOS from the otherproposed subtypes. However severe orunintelligible their speech, children in eachof the four other proposed subtypes of SDbegin speaking at the expected time. Theassumption is that for children who are truepositives for SD-AOS, the praxis deficit makesspeaking inordinately difficult.

Comorbidity of Language Disorder

Entries in the fourth row of Table 1–2 indi-cate that children in at least four of the fivesubgroups of SD are at increased risk for lan-guage disorder. As with most other domainsshown in Table 1–2, there are few data onlanguage impairment in children posited tohave a clinical or subclinical type of dys-arthria (SD-DYS). Language impairment is aprimary feature differentiating children withSD from those with SE. Of considerableinterest in recent genetic research is whysome children with SD are spared languageimpairment, and more generally, what arethe important genotype-phenotype relation-ships across verbal disorders (McGrath et al.,2008; McGrath et al., 2007; Miscimarra et al.,2007; Peterson et al., in press).

Normalization/Persistence

Entries in the fifth row of Table 1–2 are esti-mates of normalization versus persistenceof SD, with implications for explanatory

theories and clinical decision making. Pres-ently there are no data on the recoveryrates for children differentiated into the fiveproposed SDCS classifications; retrospectiveand prospective studies are in process.Withthe 75% short-term normalization rates forUSD referenced previously, and SD-GEN esti-mated to comprise 56% of clinical referrals,a large percentage of children who normal-ize may have this subtype of SD. In contrast,the increased severity of involvement in chil-dren meeting criteria for SD-DPI (discussedlater), and the persistent segmental andespecially suprasegmental deficits reportedfor children with SD-AOS, warrant classify-ing children in these groups as having latenormalization or persistent disorder. Again,data on the natural histories of SE wouldinform theories of both SE and SD.

Emerging Research Using theSDCS Framework: Overview

To this point, we have proposed that an etiologic classification system for SSD isneeded for this highly prevalent disorder to participate in the translational scientificadvances emerging for other childhood dis-orders. What types of evidence in additionto epidemiologic information have beenreported supporting the hypothesis of etio-logic subtypes of speech sound disorders?The following sections provide brief researchoverviews and highlight selected method-ological and substantive findings. Due tospace limitations, the focus is on two of theseven proposed subtypes—SD-GEN and SD-OME—with only brief comment on SD-DPIand SD-AOS. As indicated, there is sparse lit-erature on SD-DYS, and we comment on SEonly in the context of the following discus-sion of SD-GEN.


01_Flipsen_1-34 10/14/09 7:43 PM Page 11

Emerging Research Using theSDCS Framework: SpeechDelay–Genetic (SD-GEN)

Literature Review

Significant recent progress has been madein the molecular genetics of SD. Behavioralgenetic research in the previous decade pro-vided strong support for the likelihood of agenetic form of SD based on findings fromfamilial aggregation studies and twin studies(Stromswold, 1998, 2001). Compelling find-ings from a twin-adoption study (Felsenfeld& Plomin, 1997) provided strong supportfor genetic, rather than shared or nonsharedenvironmental, sources of risk for SD in fam-ilies with histories of verbal trait disorders.

The primary molecular genetic researchfindings to 2008, using a variety of posi-tional cloning and other methods, have beenthe identification of significant regions ofinterest for what we term SD-GEN on fourchromosomes. Each of these regions also is reported to be a susceptibility locus forlanguage and/or reading disorders. In Lewiset al. (2006), we provide a glossary of com-mon terms in genetics research and areview of findings, focusing on the limitedresearch in speech-genetics compared to themore extensive genetics literature in otherverbal trait disorders. A summary table listsfindings from 34 linkage studies of language,reading, and spelling impairment that reportsusceptibility loci on 11 of the 22 auto-somes (1–3, 6, 7, 13, 15, 16, 18, 19, 21). Incontrast, our review of linkage findings forSD in Lewis et al. includes only four studiesreporting significant or suggestive linkageof SD to regions on chromosomes 1 (1p36:Smith, Pennington, Boada, & Shriberg, 2005),3 (3p12-q13: Stein et al., 2004),6 (6p22:Smithet al., 2005), and 15 (15q21: Smith et al.,2005; 15q14: Stein et al., 2006). Candidate

genes for SD-GEN, based on their associationwith reading, language, or speech disordersfor these proposed studies include loci on chromosomes 3 (ROBO1), 6 (DCDC2,KIAA0319, THEM2), and 15 (DYX1C1). Asdiscussed later, separate literatures continueto address the origins of vocal communica-tion catalyzed by the finding that disruptionsin the FOXP2 gene are associated withreported apraxia of speech. Chapter 3 byBarbara Lewis may also be of interest here.

The Mendelian Inheritance in Man data-base has created a new entry, Speech SoundDisorder (SSD: MIM %608445), to archivegenetic findings for SSD. Figure 1–5 is asample of the rich information available inMIM, one of many databases and bioinfor-matic resources used in genetic research.Interested readers are invited to visit theonline version of this database (OMIM:www.ncbi.nlm.nih.gov/omim/) and searchon “SSD” for more information on thisentry. In this view of a susceptibility regionof interest for Speech Sound Disorders onchromosome 3 (11 rows from top), thereare two nearby regions of interest andgenes associated with other verbal trait dis-orders: the DYX5 gene above, and (notshown in this display), the ROBO1 gene,both of which have been linked to dyslexia.

As indicated previously, a number ofrecent analyses have explored genotype-phenotype associations within and acrossverbal domains (McGrath et al., 2007; Misci-marra et al., 2007; Peterson et al., in press).For example, using linkage analyses proce-dures, Miscimarra et al. (2007) demon-strated that the DYX8 region (1p34-p36)may include genes with pleiotropic (multi-ple) effects, including SD, language impair-ment, and reading disorders. Such findingssupporting common genetic influencesacross verbal domains are consistent withthe polygenic-environmental causal modelposited for SD-GEN in Table 1–1 and the


01_Flipsen_1-34 10/14/09 7:43 PM Page 12

contemporary perspective of “generalistgenes” (Butcher, Kennedy, & Plomin, 2006).

Phenotypes for SD-GEN

A primary goal of our collaborative geneticresearch in SD-GEN has been to developquantitative phenotypes that are sensitive toand specific for this proposed subtype of SD

(Shriberg, 1993). The goal in family-basedgenetic designs is to assess relevant familymembers of the proband using measures thatprovide one or more quantitative indices foreach sign of the disorder. Obtaining suchinformation from direct behavioral testingof individuals who are at risk for the disorder,are currently expressing the disorder, or whohave normalized prior disorder is prefer-able to using case records data (Barry,Yasin,


Figure 1–5. Entry for Speech Sound Disorders (%608445) in Mendelian Inheritance in Man(MIM). This figure is included only to illustrate a display from one of the most frequentlyconsulted bioinformatic sources. Definitions for column headings and other terms are avail-able at the online site: www.ncbi.nlm.nih.gov/omim/

01_Flipsen_1-34 10/14/09 7:43 PM Page 13

& Bishop, 2007; Lewis et al., 2007; PlanteShenkman, & Clark, 1996). Phenotypes mustbe sensitive to all levels of severity of expres-sion of the disorder, as adjusted for possibleage, gender, and other sociodemographicvariables, but they may or may not be specificfor it (i.e., they may be narrower or morebroad relative to the domain of interest).

Several phenotypes have been used forfamily members of different ages in collabo-rative speech-genetics studies. For probandsand young siblings assessed during thedevelopmental period (years 3–6) in whichmisarticulation profiles are especially sensi-tive and specific for SD, two sets of pheno-types have been productive in the moleculargenetic studies with colleagues just re-viewed: the binary SDCS-T phenotype (Nor-mal Speech Acquisition [NSA] versus SpeechDelay [SD]; Shriberg et al., 1997b) and twocontinuous phenotypes (ZPCC and ZPCCR;z-scores for Percentage of Consonants Cor-rect [PCC] and Percentage of ConsonantsCorrect-Revised [PCCR]; Shriberg, Austin,Lewis, McSweeny, & Wilson, 1997a). For ages6 to 9 years, when speech production errorsmay be nearly normalized with treatment,two continuous phenotypes developed tobe maximally sensitive to persistent SDerrors have also been productive: Speech1and Speech2. The latter two phenotypes(log of the odds and probability of havingprior SD, respectively) were obtained froma 7-variable composite derived from a mul-tivariate logistic regression using 17 out of120 speech metrics that had the highestdiagnostic accuracy for a sample of 759speakers with prior SD (Smith et al., 2005).

Endophenotypes for Verbal Trait Disorders

Significant linkage findings in verbal traitdisorders have most often been made to

several widely used nonword repetitiontasks (Newbury, Bishop, & Monaco, 2005).A significant confound with such tasks inspeech-genetics research is that nonwordrepetition errors cannot be disambiguatedfrom misarticulations.We have reported psy-chometric and substantive findings for an18-item syllable repetition task that requiresrespondents to repeat nonwords comprisedof only four Early-8 consonants (/b/, /d/,/m/, /n/) and one nonscored vowel (/ɑ/) intwo-, three-, and four-syllable words (e.g.,/bɑmɑdɑ/; Shriberg et al., 2008). Re-sponses to this task, titled the Syllable Rep-etition Task (SRT), and to a comparisonmeasure, the Nonword Repetition Task(NRT; Dollaghan & Campbell, 1998), wereobtained from 158 children assessed in a col-laborative physiology study, including 95preschool participants with SD. Findingsfrom three substudies support the con-struct and concurrent validity of the SRTand its internal reliability. All children hadthe four consonants in their inventories.Effect sizes estimating its ability to discrim-inate comorbid expressive language disor-der from no language disorder were signif-icant and comparable in magnitudes tothose obtained with the NRT.

Additional analyses indicated that theSRT can aid in dissecting the speech pro-cessing constraints underlying poor perform-ance on nonword repetition tasks. Briefly,we used differences in item length (2-, 3-, 4-syllable nonwords), error type (within ver-sus across class manner substitutions), andarticulatory difficulty (i.e., two-syllableitems with homorganic [same place] versushomotypic [same manner] consonants) toassess the level of support for errors reflect-ing processing constraints at three phasesprior to articulatory execution: auditory-perceptual encoding, storage-retrieval, andarticulatory planning-programming. Find-ings provided strong support for auditory-


01_Flipsen_1-34 10/14/09 7:43 PM Page 14

perceptual encoding deficits underlyingrepetition errors of children with typical anddelayed speech, mixed support for memorialprocesses, and no support for articulatoryplanning/programming deficits.On memorialprocessing, although longer SRT words gen-erally were more difficult to imitate correctly,twenty-four percent of the 158 participants(87% of whom had SD) had repetition errorson at least half of the 16 consonant targetsin the two-syllable nonwords (e.g.,/dɑmɑ/). We interpreted these findings asarguing against a simple memory capacityconstraint explanation for poor nonwordrepetition of these short CVCV words. Onauditory-perceptual processing, the substitu-tion errors of children with typical speech-language development more often retainedthe correct manner feature of the targetconsonant (e.g., /d/ for /b/, rather thaneither of the two nasal consonants for /b/)compared to the percentage of within-classsubstitution errors made by participantswith speech-language impairment. We sug-gest that the SRT may be a useful endophe-notypic metric for speech-genetics research,as well as for research with speakers withdisorders of any type who have limited pho-netic inventories or misarticulations.

Diagnostic Markers for SD-GEN

Recall that as indicated at the bottom of theright arm of the SDCS-E (see Figure 1–2), aprimary research need is for diagnosticmarkers to identify children with one ormore of the five proposed subtypes of SD.Diagnostic markers have to be sensitive toSD at all levels of severity of expression.Crucially, unlike phenotype markers, theymust be specific for each subtype. We havemade some progress identifying diagnosticmarkers for SD-GEN for use in geneticresearch.

A Diagnostic Marker for Childrenwith SD-GEN

In Shriberg et al. (2005), we reported diag-nostic marker findings for children at riskfor SD-GEN. We divided 72 preschool chil-dren expressing SD into two groups compa-rable in age, gender, language status, andspeech severity, but differing in family his-tory of speech-language-reading disorder.Group 1 had two or more nuclear familymembers with histories of or active speech-language-reading disorder, whereas Group 2had no nuclear family members that metthis criterion. Although the two groupswere assembled to have similar levels ofseverity of involvement (Group 1: PCC =70.0% [10.3]; Group 2: PCC = 71.6% [10.2]),there were statistically significant between-group differences in their absolute and rel-ative percentages of omission errors. Group 1children, who presumably had higher liabil-ities for SD-GEN than Group 2 children, hadsignificantly higher percentages of relativeomission errors, particularly on the Late-8(i.e., most difficult) speech sounds. The sig-nificant effect size for this latter compari-son was 0.86 (C.I. = 0.35–1.37).

In addition to its diagnostic signifi-cance, we interpret these findings as sup-port for the core claim (see Figure 1–2 andTable 1–1) that SD-GEN is due to geneticallytransmitted cognitive-linguistic constraintsaffecting phonological processing. We pro-pose that speech sound omissions (com-pared to speech sound substitutions anddistortions) reflect significant deficits inencoding and/or storage-retrieval processes.Literature support associating omissionswith cognitive constraints includes thewidely attested findings of proportionallymore speech sound omissions in cognitivedisability (e.g., Shriberg & Widder, 1990)and in the higher percentage of omissionsin children with comorbid SD and language


01_Flipsen_1-34 10/14/09 7:43 PM Page 15

impairment. Speech sound omissions werealso more prevalent in children with low-ered performance on the SRT, the nonsenseword task described above, which appearsto be sensitive to individual differences incognitive-linguistic aspects of speech pro-cessing. The omission marker has beencross-validated in an unpublished study thatincluded 95 participants with SD dividedby familial aggregation status.

Acoustic Markers to Recover thePhenotype in Residual SpeechErrors

What speech signs or signatures might beused to recover SD versus SE origins in olderspeakers who have normalized or nearlynormalized sibilant or rhotic distortions?Recall in Table 1–1 that SD-GEN is positedto have genetic origins, whereas the distaland proximal origins of SE are posited to bemoderated by environmental risk and pro-tective factors.

Figure 1–6 provides summary data fromtwo study series attempting to determine ifthere were acoustic signatures of former SDversus former SE in speakers with residualdistortion errors on rhotics (i.e., RE-R) orsibilants (i.e., RE-S).We tested several groupsof adolescents including speakers whosespeech histories were documented fromour clinic records or school records as hav-ing either prior SD or prior SE and controlchildren from the same classrooms. Bothnarrow phonetic transcription and acousticmethods were used to classify the speechtokens from word lists of speakers with nor-mal or normalized /�/ and those with prior/s/ errors.

As shown in Figure 1–6, /�/ tokenswere quantified acoustically by subtractingF2 from F3 and standardizing the result(z-score) using age × sex reference data

(Flipsen, Shriberg, Weismer, Karlsson, &McSweeny, 2001). A zF3–F2 value of 3.0provided excellent discrimination betweentokens transcribed perceptually as correct/�/ productions (<3.0) and derhotacized/�/ productions (>3.0) yielding sensitivity/specificity estimates of 95% and 94%, re-spectively. Moreover, a zF3–F2 cutpoint of6.0 provided good discrimination betweenthe perceptually derhotacized /�/ tokens of speakers with RE-SD (<6.0) versus RE-SE(>6.0) speech histories (sensitivity/speci-ficity estimates of 85% and 79%, respec-tively). Notice that the /�/ tokens from thetypical speech control group and the nor-malized speech group had z-score valuesbelow 3.0. Crucially, the /�/ tokens ofspeakers with RE-SD were mostly below6.0, whereas the /�/ tokens of speakerswith RE-SE were mostly above 6.0.Thus, thepersistent /�/ distortions of speakers withprior SE were acoustically most discrepantfrom typically produced /�/ productions.

As shown in Figure 1–6, trends were inthe same direction for /s/ productions ofadolescents with former SD versus thosewith former SE measured acoustically usingthe first spectral moment (M1; Milenkovic,1996). All tokens from the adolescentspeakers were transcribed as perceptuallycorrect, supporting the previous epidemio-logic data indicating fewer residual sibilantthan rhotic distortion errors in adolescents.Nevertheless, compared to the perceptuallycorrect /s/ tokens from control speakers,the perceptually correct /s/ tokens from thespeakers who had prior SE (see Figure 1–6,lower right panel) had notably higher zMoment 1 values compared to those fromthe speakers who had prior SD.

These /�/ and /s/ findings suggest thepossibility of acoustic signatures that mayprovide the specificity needed to classifyfamily members correctly for speech-genetics


01_Flipsen_1-34 10/14/09 7:43 PM Page 16

studies. Studies in process are also pursuingimplications of the findings in Figure 1–6 intheir own right as discussed elsewhere (Karls-

son, Shriberg, Flipsen, & McSweeny, 2002;Shriberg, Flipsen, Karlsson, & McSweeny,2001). There are a number of cognitive and


Figure 1–6. Recovery of the speech phenotypes in /�/ and /s/ productions in older speakers.The panels on the left side include the number of tokens at each standardized value (zF3-F2)produced by participants in each of four speaker groups.The panels on the right side includethe number of tokens at each standardized value (z Moment 1) produced by participantsin each of three speaker groups.

01_Flipsen_1-34 10/14/09 7:43 PM Page 17

sensorimotor developmental frameworksthat would predict that early distortions aremore resistant to change than early errorsof omission or substitution. Although con-sidered more detrimental for intelligibilitythan distortion errors, early omissions of orsubstitutions for Late-8 sounds—as occursin SD—may actually have a better prognosisfor complete normalization than the earlyspeech sound distortions that occur in SE.

Emerging Research Using theSDCS Framework: Speech

Delay–Otitis Media with Effusion (SD-OME)

Literature Review

The hypothesis that early recurrent OMEplaces a child at increased risk for SSD isbased primarily on the assumption that thefluctuant conductive hearing loss that mayaccompany OME can affect the develop-ment of veridical and stable phonologicalrepresentations (Shriberg, 1987). As indi-cated in Table 1–1, such proximal processesare presumed to have their origins in bothpolygenetic and environmental risk fac-tors. In a review of 27 OME-speech studies,we concluded that support for correlativeassociations among early OME, hearing loss,and speech delay was equivocal, and thatsupport for causal associations remainedundocumented (Shriberg, Flipsen, et al.,2000).This is essentially the position of sev-eral large-scale prospective studies thathave concluded that the mild hearing lossthat may occur during episodes of early fre-quent OME is not a risk factor for early or later impairments in speech, language,academics, or social function. Variants ofthis conclusion have been expressed in a

comprehensive literature review (Roberts,Hunter, et al., 2004), a meta-analysis (Roberts,Rosenfeld, & Zeisel, 2004), an updated set ofclinical practice guidelines (Rosenfeld et al.,2004), and, most recently, in a long-term follow-up study of cohorts in the Pittsburghotitis media project (Paradise et al., 2007).

A consistent trend in smaller scale stud-ies of outcomes for children with significanthistories of early frequent OME and hearingloss, however, is for study participants tohave significantly lowered performance oncognitive and auditory perception tasks.Three research examples illustrate this trend.Nittrouer and Burton (2005) reported that5-year-old children with histories of OMEand low socioeconomic status scored lowerthan a control group on tasks involvingspeech perception, verbal working memory,and sentence comprehension. Gravel et al.(2006) reported that OME and early hearingloss was significantly associated with severalmeasures of auditory processing in childrenassessed at school age, including measuresof extended high-frequency hearing and mea-sures assessing brainstem auditory pathways.Majerus et al. (2005) in a study of 8-year-oldchildren with histories of early recurrentOME reported normal performance on short-term memory and new word learning tasks,but small, statistically significant performancedecrements on several phonological process-ing tasks. These latter authors suggestedthat one negative outcome of OME appearsto be “. . . subtle impairments at the level ofperceptual-phonologic analysis . . . ” (p. 473).

Is Mild Hearing Loss a Risk Factorfor Speech Delay?

How can we reconcile the negative riskfindings for mild hearing loss reported inthe large-scale, prospective studies of OMEwith the positive risk findings reported


01_Flipsen_1-34 10/14/09 7:43 PM Page 18

from small-scale retrospective or ambispec-tive studies using convenience samples?Review of the hearing loss information inour prior and continuing research collabo-rations prompted us to examine the OMEliterature for the rationales used to subgroupchildren by hearing loss histories in bothlarge-group and small-group studies, includ-ing the classic paper by Fria, Cantekin, andEichler (1985) and the detailed informationin Gravel and Wallace (2000).We wonderedif the independent variable of mild conduc-tive hearing loss based on the pure-toneaverage (PTA) metric may be the source ofimportant methodological differences amongOME outcome studies. Some differenceswe found in the computation of PTA acrossOME outcome studies include: (a) the num-ber of frequencies tested (3 or 4), (b) thefrequencies used to compute the PTA (.5K,1K, 2K, or 4K), (c) whether the PTA is usedas the index if there is greater than a 20 dbHL difference in the better ear between anytwo frequencies, (d) differences in the cut-off levels used to convert PTA values to theordinal, adjectival classifications for hearingloss (i.e., mild versus moderate conductivehearing loss), and (e) other testing differ-ences, including the number of hearing eval-uations available, the age(s) at which one ormore PTAs were obtained, methods used toaverage multiple PTAs or to select the worstPTA as the primary independent variable.

In recent collaborative research, weexplored the implications of prior fluctuanthearing loss on speech outcomes in sub-samples of 60 children in each of two pro-spectively assessed otitis media projects:the Chapel Hill (Roberts, Burchinal, Koch,Footo, & Henderson, 1988) and the Pitts-burgh (Paradise et al., 2000) studies. Based onprior collaborative research with the Dallasotitis media study (Shriberg, Friel-Patti,Flipsen, & Brown, 2000) and preliminary

analyses of pure tone averages available inthe Chapel Hill and Pittsburgh projects, wedefined children as having typical hearingor (negligible hearing loss) if all of their PTAsduring the first three years of life rangedfrom 0 to 24 dB HL. Participants who hadat least one PTA from 35 to 45 db HL dur-ing this period were classified as havingmild-moderate hearing loss. Participants ineach subsample who did not meet eithercriteria (i.e., 25-34 db HL and above 45 db HL)were excluded from further analyses.

Figure 1–7A provides speech profiles(PCCR calculated on the Goldman-FristoeTest of Articulation-2 [GFTA-2]; Goldman & Fristoe, 2000) at 3 years of age for thechildren from the Chapel Hill study. Group 1(n = 9; filled circles) participants had typi-cal hearing levels (0-24 dB HL) as assessedfrom 6 to 36 months and Group 2 (n = 8;open circles) participants had PTAs meet-ing criteria for mild-moderate hearing loss(35-45 dB HL) on at least one audiologicevaluation. Figure 1–7B provides compara-ble GFTA-2 data from the Pittsburgh database(Group 1: n = 10; Group 2: n = 6). Over half(57%) of the Group 2 participants in the twodatasets had hearing loss from 35 to 40 dbHL (i.e., within the standard 40 db HL upperlimit for mild hearing loss). Two trends inthese independent samples are interpretedas support for the hypothesis that mild-moderate hearing loss associated with OMEplaces a child at increased risk for SD.

First, as shown for both datasets, notablyin Figure 1–7A, Group 2 participants hadmarkedly lower average percentages of con-sonants correct (i.e., ignoring distortions)compared to participants in Group 1. Thestatistically significant mean between-groupdifferences for each developmental soundclass in the top section of Figure 1–7A (per-centaged separately for Singletons [S], Clus-ters [C], and Total [T]) have a “box” drawn


01_Flipsen_1-34 10/14/09 7:43 PM Page 19

20

Figure 1–7. Early fluctuant mild-moderate hearing loss as a risk factor for speech delayat three years of age. The data in (A) are from the Chapel Hill database (Roberts, Burchi-nal, Koch, Footo, & Henderson, 1988); the data in (B) are from the Pittsburgh database(Paradise et al., 2000).

01_Flipsen_1-34 10/14/09 7:44 PM Page 20

around them. Conventional inferential sta-tistical symbols and underscored capital let-ters (S: Small; M: Moderate; L: Large; V: VeryLarge; and E: Extremely Large) indicate themagnitudes of significant t-tests and effectsizes, respectively.The significant effect sizefor the total PCCR was 1.21, especially largefor small sample comparisons. Trends forthe Pittsburgh data (see Figure 1–7B) weresimilar, including a significant effect size of1.29 for total PCCR. Notice that the meansfor both groups in the two studies are com-parable, with most Group 1 children scoringabove 85% (typical for 3-year-old childrenon this speech measure), whereas Group 2children scored approximately 15–20%lower, in the range reported for childrenwith SD (Shriberg et al., 1997a).

Significant between-group differencesin the percentages of correct /s/ and /z/productions were also evident in both data-sets in Figure 1–7. Group 2 participantsaveraged approximately 20% lower in theatypical 45% to 70% correct range. Themagnitudes of the significant effect sizes forthe four sibilant comparisons ranged from1.11 to 1.57. We have reported perceptual(Shriberg & Smith, 1983;Thielke & Shriberg,1990) and acoustic (Shriberg et al., 2003)data indicating that differences in sibilantproduction may be one of several possiblediagnostic markers of SD-OME.Although con-sidered preliminary due to cell sizes, these areour first data linking mild-moderate hearingloss to later deficits in sibilant production.We interpret findings as reflecting Group 2participants’ reduced attention to the sa-lience of fricative energy in the 4kHz regionand above. Again, although the large-scalestudies have concluded that mild hearingloss is not a risk factor for SD, over half ofthe Group 2 children (8/14 = 57%) in thesesmall subsamples had fluctuant hearing lossof 35 to 40 dB HL and all had fluctuantlosses below 45 db HL.

Summary and ResearchDirections

Our current research perspective on theimportant public health issue in early recur-rent OME (i.e., “watchful waiting” versusinsertion of tympanostomy tubes) focuseson the need to determine the level and pro-file of hearing loss that places a child at riskfor SD. Auditory perceptual constraints onphonological representations are clearly therelevant primitives in the causal pathwaysfrom hearing loss to speech productionerrors (Clarkson, Eimas, & Marean, 1989),and more generally, in current models ofthe development (Guenther, 2006) and per-sistence (Kenney, Barac-Cikoja, Finnegan,Jeffries, & Ludlow, 2006) of speech disor-der. If replicated in larger study samples, thefindings reviewed may explain why it hasbeen so difficult to document the validity ofthis proposed subtype of SD. In an invitedcommentary on the influential Paradise andcolleagues 2007 paper,Berman (2007) noted:“Since a hearing loss of 40 dB or higher wasuncommon among patients in the study byParadise and colleagues, it could not addressthe question of whether this level of hear-ing loss also leads to impairments” (p. 301).If replicated in prospective studies, the pre-liminary findings in Figure 1–7 would suggestthat an early mild-moderate hearing loss of35 to 45 dB HL is a risk factor for SD.

Emerging Research Using theSDCS Framework: Speech Delay–

Developmental PsychosocialInvolvement (SD-DPI)

Literature Review

As indicated previously, due to space con-straints we have elected to focus on SD-GEN


01_Flipsen_1-34 10/14/09 7:44 PM Page 21

and SD-OME, with only brief comments onSD-DPI and SD-AOS. The hypothesis of asubtype of SSD in which psychosocial pro-cesses are the primary domain in explana-tory accounts and central for treatmentplanning has been difficult to test. As indi-cated in Table 1–1, such proximal processesare presumed to have their origins in bothpolygenetic and environmental risk factors.The working term for this proposed etio-logic subtype, Speech Delay–DevelopmentalPsychosocial Involvement (SD-DPI), wascoined expressly to avoid the concept ofpsychopathology or emotional disorder.Rather, we have borrowed from the person-ality and temperament literatures, whichinclude dimensions such as mood, negativeemotionality, approach-withdrawal, dis-tractibility, attention span, task persistence,and adaptability. It has seemed that suchconstructs have been useful to describeindependent variables that are risk factorsfor successful treatment outcomes in ourclinical studies of speech delay (Kwiatkow-ski & Shriberg, 1993, 1998). Contemporarystudies have assessed individual differencesin temperament as a risk factor or correlateof delayed language development (e.g.,Caulfield, Fischel, DeBaryshe, & Whithurst,1989; Paul & Kellogg, 1997) and stuttering(Anderson, Pellowski, Conture, & Kelly,2003; Embrechts, Ebben, Franke, & van dePoel, 2000; Lewis & Goldberg, 1997), but assuggested in Figure 1–1, there have beenfew studies in recent decades on personal-ity or temperament differences and speechsound disorder.

Diagnostic Markers for SD-DPI

Using parental report and clinical records,we found that 29 of 245 children (12%) seenfor speech evaluation and treatment in our

university speech clinic during an 18-yearperiod met a set of temperament-based cri-teria for SD-DPI (Hauner et al., 2005). Thispercentage of children was larger than weexpected; these data are the source of theclinical prevalence entry for SD-DPI inTable 1–2.The SD-DPI groups included chil-dren who met criteria for either approach-related negative affect or withdrawal-relatednegative affect (Goldsmith, Lemery, & Essex,2004).We assembled a comparison group of87 children with speech delay from this data-base, matched to the SD-DPI group in age,gender, and SDCS classification (i.e., SD or anintermediate classification termed NSA/SD).

Speech analyses using a suite of de-scriptive measures indicated that childrenmeeting criteria for SD-DPI had significantlymore severe speech involvement comparedto controls with SD. Their modal profilewas an across the board delay of about anadditional one year, compared to the controlgroup of children with USD not includingSD-DPI. They scored lower than the com-parison group in each of the three develop-mental speech classes (Early-8, Middle-8,Late-8) and had significantly lower totalPCC scores (p <.01; effect size = .57). Untilwe computed the clinical prevalence esti-mate for SD-DPI of 12%, we had assumedthat such children would comprise a muchsmaller percentage of children with SD, andthat they would likely have less severe speechdelay, at least as compared to children sus-pected to have SD-GEN and SD-OME.

Although we have not, to date, identifieda unique speech or prosody-voice markerfor SD-DPI, the unexpected severity featureof SD-DPI is of significant interest for theoryand practice. Clearly, more research isneeded to cross-validate these initial find-ings and to design controlled studies ofpsychosocial processes as possible risk andprotective variables in speech delay.


01_Flipsen_1-34 10/14/09 7:44 PM Page 22

Emerging Research Using theSDCS Framework: SpeechDelay–Apraxia of Speech

(SD-AOS)

Literature Review

Research and particularly clinical concern forapraxia of speech in children has increasedsignificantly in the past two decades (Shri-berg & Campbell, 2003). Whereas Develop-mental Verbal Dyspraxia (DVD) remainsthe classification term in medical contextsand most other countries, a position state-ment by the American Speech-Language-Hearing Association (2007) endorsed theterm Childhood Apraxia of Speech (CAS).The working term SD-AOS references thesame group of children as CAS and theseterms will be used synonymously in the fol-lowing comments.

The primary constraint in CAS researchand applied clinical decision making is thelack of a set of inclusionary/exclusionarycriteria to classify speakers as positive forthis disorder (American Speech-Language-Hearing Association, 2007). From a researchperspective, differentiating SD-AOS fromdysarthria (SD-DYS) is a major researchneed discussed elsewhere (Shriberg, inpress). Our perspective on this issue is thatthe problem is due to the focus on CAS asan idiopathic speech disorder, neglectingthe potential informativeness of research inCAS as a secondary disorder in complexneurodevelopmental contexts.

Our proposed solution to the circular-ity (Guyette & Diedrich, 1981) or tautology(McNeil, Robin, & Schmidt, 1997) problemin CAS research and the escalating highrates of false positives in contemporary clin-ical practice (Shriberg & McSweeny, 2002),is the four-phase research design illustrated

in Figure 1–8. In the first (“1”) phase, wehave begun to describe the core features ofapraxia of speech as it occurs in adult AOS,in CAS following pediatric neurologic disor-ders, and in CAS in complex neurodevelop-mental contexts (Potter, Lazarus, Johnson,Steiner, & Shriberg, 2008; Shriberg et al.,2006; Shriberg, Jakielski, & El-Shanti, 2008;Shriberg, Jakielski, & Tilkens, 2009; Shriberg& Potter, 2008). For a related discussion, alsosee Chapter 7 by Velleman and colleaguesin this volume. The assumption is that a dis-order of speech praxis should have featuresthat are common to both acquired anddevelopmental subtypes, with developmen-tal issues likely moderating severity ofexpression of the core features. Critical tothis approach is use of a well-developedspeech assessment protocol that includesthe same perceptual and acoustic measuresto assess all forms of CAS across the life-span. We have developed perceptual- andacoustic-based analytics to test alternativeperspectives on the precision and stabilityof spatiotemporal signs in the speech andprosody-voice profiles of speakers suspectedto have apraxia of speech. Findings from thefirst phase are used in the second phase(see Figure 1–8) to identify and develop thecriteria that qualify participants as havingidiopathic CAS. Findings from these fourforms of CAS are expected to inform thirdphase studies of the genetic and neural substrates underlying the pathobiology ofapraxia of speech (for overviews of FOXP2research, see Fisher, 2005, 2006; Fisher, Lai,& Monaco, 2003; for an example of systemsbiology to early 2008 see Groszer et al.,2008). The fourth phase focuses on appliedmethods for assessment, treatment, andprevention. Shriberg (in press) includesrationale for the framework and a review of55 cases of CAS occurring in genetic-basedneurodevelopmental contexts.


01_Flipsen_1-34 10/14/09 7:44 PM Page 23

Some Translational Needs in SSD

This progress report has described a diagnos-tic classification system for childhood speechsound disorders and a sample of findings gath-ered within this framework. To this point,there has been little discussion of clinicalissues, specifically,on translating research find-ings to service delivery contexts. I’ll concludewith some personal perspectives on the why,what, and how of these translational needs.

Why Etiologic Classification of SSD?

This chapter’s focus on identifying the eti-ologic causes and clinical signs of subtypes

of SSD is essentially similar to the classifica-tion perspectives in Duffy’s (2005) classictext on motor speech disorders in adults.In his core chapter on differential diagnosis(Chapter 15), Duffy begins with the follow-ing quote from Sackett, Haynes, Guyatt, andTugwell (1991):

The act of clinical diagnosis is classifica-tion for a purpose: an effort to recognizethe class or group to which a patient’sillness belongs so that, based on our priorexperience with that class, the subsequentclinical acts we can afford to carry out,and the patient is willing to follow, willmaximize the patient’s health. (p. 409)

I obviously agree with Sackett and col-leagues and with Duffy on the value of diag-


Figure 1–8. A neurodevelopmental framework for research in childhood apraxia of speech.Reprinted with permission from Shriberg, L. D. (in press), A Neurodevelopmental Frame-work for Research in Childhood Apraxia of Speech. In B. Maassen and P. van Lieshout(Eds.), Speech Motor Control: New Development in Basic and Applied Research. OxfordUniversity Press.

01_Flipsen_1-34 10/14/09 7:44 PM Page 24

nostic classification. As suggested at theoutset of this chapter, diagnostic classifica-tion is required for children with SSD tofully participate in clinical advances in thepostgenomic era, including molecular med-icine and other new forms of personalizedintervention in pre-emptive, prognostic, andtargeted treatment applications. For theseclinical goals, a systems biology approachseems to me to be the forward-lookingframework for classification, rather than, forexample, classification typologies basedsolely on linguistic descriptions or com-mon speech error patterns. On this point,the long-awaited revision of the Diagnosticand Statistical Manual of the American Psy-chiatric Association,Version V promised for2011, reportedly is being reorganized toreflect common genomic and other patho-biological backgrounds as the principleclassification axes (Helmuth, 2003).

What Topic Areas in SSD Are Mostin Need of Programmatic Studies?

Two figures in this chapter have includedplaceholder boxes for topic areas in SSDthat are in need of programmatic study.Thefirst boxes were included in the right armof the SDCS (see Figure 1–2): “Genetic andEnvironmental Risk and Protective Factors”and “Neurodevelopmental Substrates.” Thesecond and overlapping boxed topics wereechoed as later phase research needs inchildhood motor speech disorders (see Fig-ure 1–8): “Genetic Substrates and NeuralSubstrates” and the translational goals of“Assessment, Treatment, and Prevention.”Research in all diseases and disorders pur-sues these public health topics, of course,but only recently have interdisciplinary proj-ects begun to study the genetic, epigenetic,and neurolinguistic substrates of SSD. I wouldsubmit that it is time to recast the long-termdichotomy between SSD of unknown (for-

merly “functional”) origin and those ofknown origin, bringing both together in aconsolidated research framework.

Convergence on common genetic,neurodevelopmental, and environmentalcontributions to typical and atypical speechacquisition should lead to the developmentof common assessment, treatment, and pre-vention frameworks. Using an example fromresearch in childhood apraxia of speech,Table 1–3 is a list of complex neurodevelop-mental disorders and genetic disruptions inwhich speakers have been reported to haveCAS. As shown in Figure 1–8, we have sug-gested that such disorders provide a richsource of information on the substrates ofCAS, and more generally, of SSD.The descrip-tions of children’s speech in these studiesare notably sparse. Most reports continue


Table 1–3. Some ComplexNeurodevelopmental Disorders ReportingCAS as a Secondary Sign. Studies are inprocess for most of the entries in thistable.

Autism

Chromosome Translocations

Coffin-Siris syndrome (7q32–34 deletion)

Down syndrome (Trisomy 21)

Rolandic Epilepsy

Fragile X syndrome (FMR1)

Joubert syndrome (CEP290; AHI1)

Galactosemia

Rett syndrome (MeCP2)

Russell-Silver syndrome (FOXP2)

Velocardiofacial syndrome (22q11.2 deletion)

Williams-Beuren locus duplication (7q11.23)

01_Flipsen_1-34 10/14/09 7:44 PM Page 25

to appear in medical journals which, under-standably, do not include speech phenotypedetails even if available (although usefulphenotypic details are increasingly appear-ing in on-line supplements). Again, I thinkit is no longer tenable to compartmentalizeSSD research into those of known andunknown origin in the postgenomic era,when advances in understanding and treat-ment are likely to emerge from and informone another.

How Can We Best ImproveService Delivery to Children With SSD?

The “how” questions of translational re-search necessarily involve new skills andtechnologies, operationalizing them for usein the field. Among the many instrumentalapproaches that are candidates for theassessment and treatment of children withSSD, skilled use of acoustic techniques (alsosee Chapters 7 and 8 in this volume), cou-pled with competence in narrow phonetictranscription seem to offer the highest pos-sibility for widespread clinical accessibility.As described previously, the assessmentprotocol we use for diagnostic classificationrequires a laptop computer and masters-level skills in phonetic transcription andacoustic analysis.

A key to the success of the workdescribed in this chapter is the identification

and verification of risk factors and diagnos-tic markers for subtypes of SSD. Table 1–4is a list of 38 risk factors and diagnosticmarkers that have been assembled through2008 for the five proposed subtypes of SD.Some are potential entries for the placemarkers at the bottom right side of Figure1–2. The entries in Table 1–4 are prelimi-nary; projects in process are completing val-idation and cross-validation studies of theseand other diagnostic markers with compar-ison acoustic methods. Taken individually,few of the markers have demonstrated suf-ficient diagnostic accuracy, which wouldnominally require that they identify at least90% of the true positives and true negativesfor each subtype. We anticipate using anumber of statistical techniques to maxi-mize their combined power to accuratelyclassify children’s most likely subtype orcomposite subtype of SD.

Information such as in Table 1–4, whichcan be obtained using perceptual and acous-tic procedures, should be viewed as reflect-ing only the current report on this continuingconversation on the causal origins of chil-dren’s speech sound disorders. Once again,I am profoundly grateful to the many peo-ple who have shaped and contributed tothe thoughts and findings in this report andto the children and their families who con-tinue to participate in studies with us. Moregenerally, I salute the international commu-nity of investigators and clinicians who workdaily to help children communicate.


01_Flipsen_1-34 10/14/09 7:44 PM Page 26

27

Table 1–4. Sample Perceptual, Acoustic, and Case History Risk Factors and Markers forFive Subtypes of Speech Delay Emerging from Studies Using the SDCS Framework*

SD- SD- SD- SD- SD-Risk Factors and Diagnostic Markers GEN OME DPI AOS DYS

Familial aggregation of any verbal trait +(pedigree interview)

Lower language tests scores +

Lower nonword repetition task scores +(NRT, SRT)

Higher % relative omissions errors (ROI) +

Lower % relative sibilant distortion errors (RDI) +

Six or more episodes of OME from 0–2 yrs +(medical records)

Mean 3–4 freq. thresholds of at least 35 dB +on any evaluation (audiological records)

Higher % backing of fricatives +(Siblilant Report)

Smaller Intelligibility-Speech Gap (I-S Gap) +

Higher % epenthetic vowels on glides +(Diacritic Modification Index)

Higher % glottal stop substitutions (DMI) +

Higher % nasal-nasal substitutions +(Speech Report 1)

Higher % /h/ insertions/substitutions +(Speech Report 1)

Higher % initial consonant deletions +(Speech Report 1)

Lower first spectral moment on sibilants (<M1) +

History of clinically significant psychosocial +issues (case records)

Lower scores on psychological/social skills +tests (case records)

Lower aggregate speech competence +(PCCR, PVCR)

Late (>2 years) onset of speech (case and +clinical records)

Late (>6 years) normalization of SD +(clinical records)

continues

01_Flipsen_1-34 10/14/09 7:44 PM Page 27

28

Table 1–4. continued

SD- SD- SD- SD- SD-Risk Factors and Diagnostic Markers GEN OME DPI AOS DYS

Higher % vowel errors +

Higher % inconsistent errors on four stability +indices

Higher % utterances with inappropriate +lexical stress (LSI)

Higher % utterances with inappropriate +sentential stress (SSI)

Lower % on Pairwise Variability Index (PVI) +

Higher % on Transition Disruption Index (TDR) +

Higher % on Syllable Segregation Index (SSI) +

Lower ratio on Speech-Pause Index (SPI) +

Clinically significant sensorimotor task scores +(e.g., tapping rate)

Clinically significant oral function task scores +(e.g., DDK)

Higher % imprecision on Diacritic Modification +Index (DMI)

Nasal emissions (oral examination; DMI) +

Smaller planar area for vowels (Vowel Space +Index: VSI)

Slow articulation and speech rates; +by syllable, by phoneme (SRI)

Lower Speech Intensity Index (SII) +

Breathy, strain/strangle; tremulous laryngeal +codes (PVSP)

Rough voice (jitter, shimmer, +harmonics-to-noise ratio) (J, S, HNR)

Nasal, denasal, or nasopharyngeal resonance +quality (RQA)

TOTALS 5 10 3 10 10

*Titles, descriptions, and references for measures and working terms are not included here. Informa-tion about most terms can be retrieved using the search function at the Phonology Project Web site:http://www.waisman.wisc.edu/phonology/index.htm .

01_Flipsen_1-34 10/14/09 7:44 PM Page 28

References

American Speech-Language-Hearing Association.(2007). Childhood apraxia of speech [Posi-tion statement]. Available from www.asha.org/policy .

Anderson, J. D., Pellowski, M. W., Conture, E. G.,& Kelly, E. M. (2003). Temperamental charac-teristics of young children who stutter. Jour-nal of Speech, Language, and HearingResearch, 46, 1221–1233.

Barrett, C. M., & Hoops, H. R. (1974). The rela-tionship between self-concept and the remis-sions of articulatory errors. Language, Speech,and Hearing Services in Schools, 2, 67–70.

Barry, J. G., Yasin, I., & Bishop, D. V. M. (2007).Heritable risk factors associated with languageimpairments. Genes, Brain, and Behavior, 6,66–76.

Berman, S. (2007). The end of an era in otitisresearch. New England Journal of Medicine,356, 300–302.

Butcher, L. M., Kennedy, J. K., & Plomin, R.(2006). Generalist genes and cognitive neuro-science. Current Opinion in Neurobiology,16, 145–151.

Campbell, T. F., Dollaghan, C. A., Rockette, H. E.,Paradise, J. L., Feldman, H. M., Shriberg, L.,et al. (2003). Risk factors for speech delay ofunknown origin in 3-year-old children. ChildDevelopment, 74, 346–357.

Caulfield, M., Fischel, J., DeBaryshe, B., & White-hurst, G. (1989). Behavioral correlates ofdevelopmental expressive language disorders.Journal of Abnormal Child Psychology, 17,187–201.

Clarkson, R. L., Eimas, P. D., & Marean, G. C. (1989).Speech perception in children with historiesof recurrent otitis media. Journal of theAcoustical Society of America, 85, 926–933.

Dollaghan, C., & Campbell, T. F. (1998). Non-word repetition and child language impair-ment. Journal of Speech, Language, andHearing Research, 41, 1136–1146.

Duffy, J. R. (2005). Motor speech disorders: Sub-strates, differential diagnosis, and manage-ment (2nd ed.). St. Louis, MO: Mosby.

Embrechts, M., Ebben, H., Franke, P., & van dePoel, C. (2000). Temperament: A comparisonbetween children who stutter and childrenwho do not stutter. In H. G. Bosshardt, J. S.Yaruss, & H. F. M. Peters (Eds.), Proceedingsof the Third World Congress on Fluency Dis-orders: Theory, research, treatment, and self-help (pp. 557–562). Nijmegen, The Nether-lands: University of Nijmegen.

Felsenfeld, S., Broen, P. A., & McGue, M. (1994).A 28-year follow-up of adults with a history ofmoderate phonological disorder: Educationaland occupational results. Journal of Speechand Hearing Research, 37, 1341–1353.

Felsenfeld, S., & Plomin, R. (1997). Epidemiolog-ical and offspring analyses of developmentalspeech disorders using data from the Col-orado Adoption Project. Journal of Speech,Language, and Hearing Research, 40,778–791.

Fisher, S. E. (2005). Dissection of molecularmechanisms underlying speech and languagedisorders. Applied Psycholinguistics, 26,111–128.

Fisher, S. E. (2006). Tangled webs: Tracing theconnections between genes and cognition.Cognition, 101, 270–297.

Fisher, S. E., Lai, C. S. L., & Monaco, A. P. (2003).Deciphering the genetic basis of speech andlanguage disorders. Annual Review of Neu-roscience, 26, 57–80.

Flipsen, P., Jr., Shriberg, L. D., Weismer, G., Karls-son, H. B., & McSweeny, J. L. (2001). Acousticphenotypes for speech-genetics studies: Ref-erence data for residual /�/ distortions. Clin-ical Linguistics and Phonetics, 15, 603–630.

Fria, T. J., Cantekin, E. I., & Eichler, J. A. (1985).Hearing acuity of children with otitis mediawith effusion. Archives of Otolaryngology,111, 10–16.

Goldman, R., & Fristoe, M. (2000). Goldman-Fristoe Test of Articulation (2nd ed.). CirclePines, MN: American Guidance Service.

Goldsmith, H. H., Lemery, K. S., & Essex, M. J.(2004). Temperament as a liability factor forbehavioral disorders of childhood. In L. F.DiLalla (Ed.), Behavior genetics principles:Perspectives in development, personality,


01_Flipsen_1-34 10/14/09 7:44 PM Page 29

and psychopathology. Washington, DC:American Psychological Association.

Gravel, J. S., Roberts, J. E., Roush, J., Grose, J.,Besing, J., Burchinal, M., et al. (2006). Earlyotitis media with effusion, hearing loss, andauditory processes at school age. Ear andHearing, 27, 353–368.

Gravel, J. S., & Wallace, I. F. (2000). Effect of oti-tis media with effusion on hearing in the first3 years of life. Journal of Speech, Language,and Hearing Research, 43, 631–644.

Groszer, M., Keays, D. A., Deacon, R. M. J., deBono, J. P., Prasad-Mulcare, S., Gaub, S., et al.(2008). Impaired synaptic plasticity and motorlearning in mice with a point mutation impli-cated in human speech deficits. Current Biol-ogy, 18, 354–362.

Guenther, F. H. (2006). Cortical interactionsunderlying the production of speech sounds.Journal of Communication Disorders, 39,350–365.

Guyette,T.W., & Diedrich,W. M. (1981).A criticalreview of developmental apraxia of speech.In N. J. Lass (Ed.), Speech and language:Advances in basic research and practice(Vol. 5, pp. 1–49). New York: Academic Press.

Hall, P. K., Jordan, L. S., & Robin, D. A. (1993).Developmental apraxia of speech: Theoryand clinical practice. Austin, TX: Pro-Ed.

Hauner, K. K. Y., Shriberg, L. D., Kwiatkowski, J.,& Allen, C. T. (2005). A subtype of speechdelay associated with developmental psycho-social involvement. Journal of Speech, Lan-guage, and Hearing Research, 48, 635–650.

Helmuth, L. (2003). In sickness or in health? Science, 302, 808–810.

Hesketh,A. (2004). Early literacy achievement ofchildren with a history of speech problems.International Journal of Language andCommunication Disorders, 39, 453–468.

Ingram, D. (1989). Phonological disability inchildren: Studies in disorders of communi-cation (2nd ed.). San Diego, CA: Singular.

Karlsson, H. B., Shriberg, L. D., Flipsen, P., Jr., &McSweeny, J. L. (2002). Acoustic phenotypesfor speech-genetics studies:Toward an acous-tic marker for residual /s/ distortions. ClinicalLinguistics and Phonetics, 16, 403–424.

Kenney, M. K., Barac-Cikoja, D., Finnegan, K., Jef-fries, N., & Ludlow, C. L. (2006). Speech per-ception and short-term memory deficits inpersistent developmental speech disorder.Brain and Language, 96, 178–190.

Kwiatkowski, J., & Shriberg, L. D. (1993). Speechnormalization in developmental phonologicaldisorders: A retrospective study of capability-focus theory. Language, Speech, and HearingServices in Schools, 24, 10–18.

Kwiatkowski, J., & Shriberg, L. D. (1998). Thecapability-focus treatment framework for childspeech disorders.American Journal of Speech-Language Pathology: A Journal of ClinicalPractice, 7, 27–38.

Leitão, S., & Fletcher, J. (2004). Literacy out-comes for students with speech impairment:Long-term follow-up. International Journalof Language and Communication Disor-ders, 39, 245–256.

Lewis, B. A., Freebairn, L. A., Hansen, A. J., Misci-marra, L., Iyengar, S. K., & Taylor, H. G. (2007).Speech and language skills of parents of chil-dren with speech sound disorders. AmericanJournal of Speech-Language Pathology, 16,108–118.

Lewis, B. A., & Shriberg, L. D. (1994, November).Life span interrelationships among speech,prosody-voice, and nontraditional phono-logical measures. Mini-seminar presented at the Annual Convention of the AmericanSpeech-Language-Hearing Association, NewOrleans, LA.

Lewis, B. A., Shriberg, L. D., Freebairn, L. A., Han-sen,A. J., Stein, C. M.,Taylor, H. G., et al. (2006).The genetic bases of speech sound disorders:Evidence from spoken and written language.Journal of Speech, Language, and HearingResearch, 49, 1294–1312.

Lewis, K. E., & Goldberg, L. L. (1997). Measure-ments of temperament in the identificationof children who stutter. European Journal ofDisorders of Communication, 32, 441–448.

Majerus, S., Amand, P., Boniver, V., Demanez, J-P.,Demanez, L., & Van der Linden, M. (2005).A quantitative and qualitative assessment ofverbal short-term memory and phonologicalprocessing in 8-year-olds with a history of


01_Flipsen_1-34 10/14/09 7:44 PM Page 30

repetitive otitis media. Journal of Communi-cation Disorders, 38, 473–498.

McGrath, L. M., Hutaff-Lee, C., Scott, A., Boada,R., Shriberg, L. D., & Pennington, B. F. (2008).Children with comorbid speech sound disor-der and specific language impairment are atincreased risk for attention-deficit/hyperac-tivity disorder. Journal of Abnormal ChildPsychology, 36, 151–163.

McGrath, L. M., Pennington, B. F., Willcutt, E. G.,Boada, R., Shriberg, L., & Smith, S. D. (2007).Gene x environment interactions in speechsound disorder predict language and pre-literacy outcomes. Development and Psy-chopathology, 19, 1047–1072.

McNeil, M. R., Robin, D. A., & Schmidt, R. A.(1997).Apraxia of speech: Definition, differen-tiation, and treatment. In M. R. McNeil (Ed.),Clinical management of sensorimotor speechdisorders (pp. 311–344). New York: Thieme.

Milenkovic, P. (1996). CSpeech (Version 4)[Computer software]. Madison: University ofWisconsin-Madison, Department of ElectricalEngineering.

Miscimarra, L., Stein, C., Millard, C., Kluge, A.,Cartier, K., Freebairn, L., et al. (2007). Furtherevidence of pleiotropy influencing speechand language: Analysis of the DYX8 region.Human Heredity, 63, 47–58.

Newbury, D. F., Bishop, D. V., & Monaco, A. P.(2005). Genetic influence on language impair-ment and phonological short-term memory.Trends in Cognitive Sciences, 9, 528–534.

Nittrouer, S., & Burton, L. T. (2005). The role ofearly language experience in the developmentof speech perception and phonological pro-cessing abilities: Evidence from 5-year-oldswith histories of otitis media with effusion andlow socioeconomic status. Journal of Com-munication Disorders, 38, 29–63.

Paradise, J. L., Dollaghan, C. A., Campbell, T. F.,Feldman, H. M., Bernard, B. S., Colborn, D. K.,et al. (2000). Language, speech sound produc-tion, and cognition in three-year-old childrenin relation to otitis media in their first threeyears of life. Pediatrics, 105, 1119–1130.

Paradise, J. L., Feldman, H. M., Campbell, T. F.,Dollaghan, C. A., Rockette, H. E., Pitcairn,

D. L., et al. (2007). Tympanostomy tubes anddevelopmental outcomes at 9 to 11 years ofage. New England Journal of Medicine, 356,248–261.

Paul, R., & Kellogg, L. (1997). Temperament inlate talkers. Journal of Child Psychology andPsychiatry, 38, 803–810.

Peterson, R. L., Pennington, B. F., Shriberg, L. D.,& Boada, R. (in press). What influences liter-acy outcome in children with speech sounddisorders. Journal of Speech, Language, andHearing Disorders.

Plante, E., Shenkman, K., & Clark, M. M. (1996).Classification of adults for family studies ofdevelopmental language disorders. Journal ofSpeech and Hearing Research, 39, 661–667.

Potter, N. L., Lazarus, J. C., Johnson, J. M., Steiner,R. D., & Shriberg, L. D. (2008). Language andmotor characteristics of children with galac-tosemia and speech sound disorders. Manu-script submitted for publication.

Raitano, N. A., Pennington, B. F., Tunick, R. A.,Boada, R., & Shriberg, L. D. (2004). Pre-literacyskills of subgroups of children with speechsound disorders. Journal of Child Psychologyand Psychiatry, 45, 821–835.

Roberts, J. E., Burchinal, M. R., Koch, M. A.,Footo, M. M., & Henderson, F. W. (1988). Oti-tis media in early childhood and its relation-ship to later phonological development.Journal of Speech and Hearing Disorders,53, 424–432.

Roberts, J., Hunter, L., Gravel, J., Rosenfeld, R.,Berman, S., Haggard, M., et al. (2004). Otitismedia,hearing loss, and language learning:Con-troversies and current research.Developmentaland Behavioral Pediatrics, 25, 110–122.

Roberts, J. E., Rosenfeld, R. M., & Zeisel, S. A.(2004). Otitis media and speech and language:A meta-analysis of prospective studies. Pedi-atrics, 113, 237–247.

Rosenfeld, R. M., Culpepper, L., Doyle, K. J.,Grundfast, K. M., Hoberman, A., Kenna, M. A.,et al. (2004). Clinical practice guideline: Oti-tis media with effusion. Otolaryngology-Headand Neck Surgery: Official Journal of Amer-ican Academy of Otolaryngology-Head andNeck Surgery, 130(Suppl. 5), S95–S118.


01_Flipsen_1-34 10/14/09 7:44 PM Page 31

Sackett, D. L., Haynes, R. B., Guyatt, G. H., & Tug-well, P. (1991). Clinical epidemiology: A basicscience for clinical medicine (2nd ed.).Boston: Little Brown.

Shriberg, L. D. (1975, November). Preliminariesto a social learning theory view of deviantchild phonology. Paper presented at theannual meeting of the American Speech andHearing Association, Washington, DC.

Shriberg, L. D. (1980). Developmental phonolog-ical disorders. In T. J. Hixon, L. D. Shriberg, &J. S. Saxman (Eds.), Introduction to commu-nicative disorders (pp. 262–309). EnglewoodCliffs, NJ: Prentice-Hall.

Shriberg, L. D. (1982a). Diagnostic assessment of developmental phonologic disorders. InM. Crary (Ed.), Phonological intervention:Concepts and procedures. San Diego, CA:College-Hill.

Shriberg, L. D. (1982b). Toward classification ofdevelopmental phonological disorders. In N.J. Lass (Ed.), Speech and language: Advancesin basic research and practice (Vol. 8, pp.2–18). New York: Academic Press.

Shriberg, L. D. (1987). In search of the otitismedia-speech connection. Journal of theNational Student Speech Language HearingAssociation, 15, 56–67.

Shriberg, L. D. (1993). Four new speech andprosody-voice measures for genetics researchand other studies in developmental phono-logical disorders. Journal of Speech andHearing Research, 36, 105–140.

Shriberg, L. D. (1994). Five subtypes of devel-opmental phonological disorders. Clinics inCommunication Disorders, 4(1), 38–53.

Shriberg, L. D. (1997). Developmental phonolog-ical disorder(s): One or many? In B. W. Hod-son & M. L. Edwards (Eds.), Perspectives inapplied phonology (pp. 105–127). Gaithers-burg, MD: Aspen.

Shriberg, L. D. (in press). A neurodevelopmentalframework for research in childhood apraxiaof speech. In B. Maassen & P. van Lieshout,(Eds.), Speech motor control: New develop-ments in basic and applied research. OxfordUniversity Press.

Shriberg, L. D., & Austin, D. (1998). Comorbidityof speech-language disorder: Implications for

a phenotype marker for speech delay. In R.Paul (Ed.), The speech-language connection(pp. 73–117). Baltimore: Paul H. Brookes.

Shriberg, L. D.,Austin, D., Lewis, B.A., McSweeny,J. L., & Wilson, D. L. (1997a).The Percentage ofConsonants Correct (PCC) metric: Extensionsand reliability data. Journal of Speech, Lan-guage, and Hearing Research, 40, 708–722.

Shriberg, L. D.,Austin, D., Lewis, B.A., McSweeny,J. L., & Wilson, D. L. (1997b). The Speech Dis-orders Classification System (SDCS): Exten-sions and lifespan reference data. Journal ofSpeech, Language, and Hearing Research,40, 723–740.

Shriberg, L. D., Ballard, K. J.,Tomblin, J. B., Duffy,J. R., Odell, K. H., & Williams, C. A. (2006).Speech, prosody, and voice characteristics ofa mother and daughter with a 7;13 transloca-tion affecting FOXP2. Journal of Speech, Lan-guage, and Hearing Research, 49, 500–525.

Shriberg, L. D., & Campbell, T. F. (Eds.). (2003).Proceedings of the 2002 childhood apraxiaof speech research symposium. Carlsbad, CA:The Hendrix Foundation.

Shriberg, L. D., Flipsen, P., Jr., Karlsson, H. B., &McSweeny, J. L. (2001). Acoustic phenotypesfor speech-genetics studies: An acousticmarker for residual /�/ distortions. ClinicalLinguistics and Phonetics, 15, 631–650.

Shriberg, L. D., Flipsen, P., Jr., Thielke, H., Kwiat-kowski, J., Kertoy, M., Katcher, M., et al. (2000).Risk for speech disorder associated with earlyrecurrent otitis media with effusion:Two retro-spective studies. Journal of Speech, Language,and Hearing Research, 43, 79–99.

Shriberg, L. D., Friel-Patti, S., Flipsen, P., Jr., &Brown, R. L. (2000). Otitis media, fluctuanthearing loss, and speech-language outcomes:A preliminary structural equation model.Journal of Speech, Language, and HearingResearch, 43, 100–120.

Shriberg, L. D., Jakielski, K. J., & El-Shanti, H.(2008). Breakpoint localization using array-CGH in three siblings with an unbalanced4q;16q translocation and childhood apraxiaof speech. American Journal of MedicalGenetics, Part A, 146A, 2227–2233.

Shriberg, L. D., Jakielski, K. J., & Tilkens, C. M.(2009). Acoustic and perceptual profiles of


01_Flipsen_1-34 10/14/09 7:44 PM Page 32

Childhood Apraxia of Speech in three sib-lings with an unbalanced 4q;16q transloca-tion. Manuscript in preparation.

Shriberg, L. D., Kent, R. D., Karlsson, H. B., Mc-Sweeny, J. L., Nadler, C. J., & Brown, R. L.(2003). A diagnostic marker for speech delayassociated with otitis media with effusion:Backing of obstruents. Clinical Linguisticsand Phonetics, 17, 529–547.

Shriberg, L. D., & Kwiatkowski, J. (1982). Phono-logical disorders I: A diagnostic classificationsystem. Journal of Speech and Hearing Dis-orders, 47, 226–241.

Shriberg, L. D., & Kwiatkowski, J. (1988).A follow-up study of children with phono-logic disorders of unknown origin. Journalof Speech and Hearing Disorders, 53,144–155.

Shriberg, L. D., & Kwiatkowski, J. (1994). Devel-opmental phonological disorders I: A clinicalprofile. Journal of Speech and HearingResearch, 37, 1100–1126.

Shriberg, L. D., Kwiatkowski, J., & Gruber, F. A.(1994). Developmental phonological disor-ders II: Short-term speech-sound normaliza-tion. Journal of Speech and Hearing Research,37, 1127–1150.

Shriberg, L. D., Lewis, B. L., Tomblin, J. B., Mc-Sweeny, J. L., Karlsson, H. B., & Scheer, A. R.(2005). Toward diagnostic and phenotypemarkers for genetically transmitted speechdelay. Journal of Speech, Language, andHearing Research, 48, 834–852.

Shriberg, L. D., Lohmeier, H. L., Campbell, T. F.,Dollaghan, C. A., Green, J. R., & Moore, C. A.(in press). A nonword repetition task forspeakers with misarticulations: The SyllableRepetition Task. Journal of Speech, Language,& Hearing Research.

Shriberg, L. D., & McSweeny, J. L. (2002). Classi-fication and misclassification of childhoodapraxia of speech (Tech. Rep. No. 11). Pho-nology Project, Waisman Center, Universityof Wisconsin-Madison.

Shriberg, L. D., & Potter, N. L. (June, 2008). Speechcharacteristics of children with galacto-semia and persistent speech disorder. Paper

presented at the 12th meeting of the Interna-tional Clinical Phonetics and Linguistics Asso-ciation, Istanbul, Turkey.

Shriberg, L. D., & Smith, A. J. (1983). Phonologi-cal correlates of middle-ear involvement inspeech-delayed children: A methodologicalnote. Journal of Speech and Hearing Re-search, 26, 293–297.

Shriberg, L. D., Tomblin, J. B., & McSweeny, J. L.(1999). Prevalence of speech delay in 6-year-old children and comorbidity with languageimpairment. Journal of Speech, Language,and Hearing Research, 42, 1461–1481.

Shriberg, L. D., & Widder, C. J. (1990). Speechand prosody characteristics of adults withmental retardation. Journal of Speech andHearing Research, 33, 627–653.

Smith,S.D.,Pennington,B.F.,Boada,R.,& Shriberg,L. D. (2005). Linkage of speech sound disorderto reading disability loci. Journal of ChildPsychology and Psychiatry, 46, 1057–1066.

Stein, C. M., Millard, C., Kluge, A., Miscimarra, L.E., Cartier, K. C., Freebairn, L. A., et al. (2006).Speech sound disorder influenced by a locus in15q14 region. Behavior Genetics,36, 858–868.

Stein, C. M., Schick, J. H., Taylor, H. G., Shriberg,L. D., Millard, C., Kundtz-Kluge,A., et al. (2004).Pleiotropic effects of a chromosome 3 locus onspeech-sound disorder and reading. AmericanJournal of Human Genetics, 74, 283–297.

Stromswold, K. (1998). Genetics of spoken lan-guage disorders.Human Biology,70, 297–324.

Stromswold, K. (2001). The heritability of lan-guage: A review and meta-analysis of twin,adoption, and linkage studies. Language, 77,647–723.

Thielke, H. M., & Shriberg, L. D. (1990). Effects ofrecurrent otitis media on language, speech,and educational achievement in MenomineeIndian children. Journal of Native AmericanEducation, 29, 25–35.

Vernes, S. C., Nicod, J., Elahi, F. M., Coventry,J. A., Kenny, N., Coupe, A. M., et al. (2006).Functional genetic analysis of mutationsimplicated in a human speech and languagedisorder. Human Molecular Genetics, 15,3154–3167.


01_Flipsen_1-34 10/14/09 7:44 PM Page 33

Documents

Childhood Speech Sound Disorders: From Postbehaviorism to