Haskins Laboratories Status Report on Speech Research1991, SR-105/106, 231-242

Subject Definition and Selection Criteria for StutteringResearch in Adult SUbjects*

Peter J. Alfonsot

The principal topics assigned to this authorwere to review subject definition and selectioncriteria reported in experiments investigatingspeech behaviors in adult stutterers, to determinewhether the typical description of these criteria issufficient to meet the demands of scientificinvestigation and replication, and to discuss anumber of topics that should be considered in thedevelopment of subject definition and selectioncriteria. Although the organization of this paper islargely motivated by the assigned topics, I havedeviated somewhat by reviewing in Section I thedefinitions of stutterers and stuttering thattypically appear in the literature. The review isfollowed by a discussion of how the precise andcareful use of these definitions bear directly on thedevelopment of subject definition and selectioncriteria. Section II represents the primary focus ofthe paper and includes: (a) a review of typicalsubject definition and selection criteria, and (b) adiscussion of a number of topics that should beconsidered in the development ofthese criteria.

The argument that appropriate subjectdefinition and selection criteria is an essentialcomponent of good experimental technique isbased on the ubiquitous observations that somany of the overt characteristics of the disorderare highly variable across subjects at all levels ofmeasurement. The argument developed here isthat until the many facets of the heterogeneity ofstuttering are better understood, criteria shoulderr on the side of over-defining rather than under­defining essential details about the stutterer andhis or her behaviors.

Preparation of this paper supported in part by NIH NS.13870 and NS-13617 awarded to Haskins Laboratories. Theauthor wishes to thank three anonymous reviewers for theircomments.

231

Finally, because this paper centers onexperiments dealing with adults, much of what iswritten here presumes that the proceduresemployed in the experiments of interest arecommon to that population and are morephysiologically based and invasive than thosecommonly used with children who stutter.Problems associated with the development ofsubject definition and selection criteria in thelatter population group are considered in detail ina paper entitled "Childhood stuttering: What is itand who does it?" by Edward Conture (1990).

SECTION I: DEFINING STUTIERINGAND STUTIERERS

lAo An overviewDefinitions of stuttering continue to evolve with

our theories and abilities to measure various as­pects of the disorder. Traditionally, most defini­tions are descriptions of behaviors. They are typi­cally presented as a comprehensive list of behav­iors that are common to all stutterers and that dif­ferentiate stuttering from normal speech. An oftencited example of a descriptive definition is givenby Wingate: "The term 'stuttering' means: I. (a)Disruption in the fluency of verbal expression,which is (b) characterized by involuntary, audibleor silent, repetitions or prolongations in the utter­ance of short speech elements, namely: sounds,syllables, and words of one syllable... II.Sometimes the disruptions are (e) accompanied byaccessory activities involving the speech appara­tus, related or unrelated body structures, orstereotyped speech utterances... III. Also, thereare not infrequently (f) indications or report of thepresence of an emotional state, ranging from ageneral condition of 'excitement' or 'tension' tomore specific emotions... (g) The immediate sourceof stuttering is some incoordination expressed in

232 Alfonso

the peripheral speech mechanism; the ultimatecause is presently unknown and may be complexor compound" (Wingate, 1964). Most definitionsinclude at least the following three descriptions ofthe verbal behavior associated with stuttering;involuntary, repetitions, and prolongations. Forexample, stuttering is defined in the InternationalClassification of Diseases as "disorders in therhythm of speech, in which the individual knowsprecisely what he wishes to say, but at the time isunable to say it because of an involuntary,repetitive prolongation or cessation of a sound"(World Health Organization, 1977, p. 202). For agood discussion of the various categories ofdefinitions see Van Riper, 1982, Chapter 2.

LB. Needs and future directions

I.B.1. Disfluency, dysfluency, and fluency. Theintent prescribed to the terms "disfluency,""dysfluency," and "fluency" varies considerably asa function of the distinctions among theoreticalmodels of stuttering. This issue is considered insufficient detail elsewhere, for example, Ham(1989), Perkins (1984), and Wingate (l984a.b,1988). There should be no disagreement, however,about the critical necessity to make explicit,especially for research purposes, the descriptionsof the speech behavior(s) under observation. Byway of example, I will throughout this paperretain distinct definitions for the terms"disfluency," "dysfluency," and "fluency," largelyfollowing the rationale reported by Wingate(l984a,b). Disfluency is used here as a generalreferent, pertaining to the usual and normaldisruptions in the patterns of speech movementsthat are perceived as "fluent speech." Dysfluency,on the other hand, is used to mean abnormaldisruption in the normal patterns of speechmovements. Fluency is used in the perceptualsense, to mean the realization of flowing, smooth,and easily produced speech; that is, as anabstraction of the underlying articulatorygestures. Thus, a sample of speech produced by anormal talker that has been judged to be fluent, asdefined here, may include, at the level of speechproduction, disfluent segments but not dysfluentones. Once again, the critical point is that theseterms, and the criteria developed to operationallydefine them, must be made explicit. It will beshown in Section II that the failure to do so canled to conflicting conclusions about a number ofaspects related to stuttering.

I.B.2. Stuttere,r's self-identification andstuttering-identification. In defining stutteringand stutterers, infrequent attention seems to be

given to the stutterer's "self-identification" and"stuttering-identification." In particular, thestutterer's identification of a dysfluency can bevery important since, as discussed below in moredetail, it is often the case that disagreement willoccur in fluent-dysfluent judgments that are basedon data representing different accessible levels ofmeasurement, that is, perceptual, acoustic,movement, and neuromuscular. For example, it isnot unusual that an utterance is judged "fluent" atthe perceptual level by an experienced listener,while analysis at deeper levels of speech,kinematic for instance, indicate inappropriate or"dysfluent" production. Until the distinctionbetween fluent, disfluent, and dysfluent speech isbetter understood, the adult stutterer's judgmentin the classification of him- or herself as astutterer and in the fluency-dysfluency distinctionofhis or her speech should be encouraged.

I.B.3. Voluntary and involuntary speech motoroutput. The distinction between "voluntary" and"involuntary" in defining disfluency and/ordysfluency is not consistently made, although itmay represent a critical distinction betweencertain types of dysfluency exhibited by stutterersand disfluency exhibited by adults who do notstutter. The significance of "involuntary" indefinitions of stuttering and stutterers isdiscussed in detail by Perkins (1983) in responseto a review article on stuttering by Andrews andhis colleagues (Andrews et al., 1983) and need notbe elaborated in great detail here. By way of abrief example, Perkins (1983) writes that thepresumption is that a voluntary disfluency in theadult population results from "linguisticuncertainty." This implies that a voluntarydisfluency, as in the prolongation of the isolatedvowel tal for example, is a voluntary strategyinvoked by the speaker while attempting toresolve a high-level linguistic query, such as inlexical retrieval. On the hand, the mechanismsunderlying involuntary speech acts are far lessagreed upon. Citing Perkins (1983) once again asan example, dysfluency "presumably is a motorspeech blockage." Certainly, one can argue withlinguistic uncertainty and speech motor blockagemodels of the voluntary-involuntary distinction,but until the volition of speech motor output canbe measured with validity, the stutterer'sidentification of hislher fluent and dysfluentspeech should be encouraged. That is, stutterersappear to be in a better position to subjectivelyrate their utterances as involuntary dysfluent,voluntary disfluent, or fluent than are listener­judges.

Subject Definition and Selection Criteria for Stuttering Research in Adult Subjects 233

These two notions, the stutterer's self­identification and the voluntary-involuntarydistinction, are important because they appear tobe essential parameters in distinguishing amongdisfluent, dysfluent, and perceptually fluentspeech. The significance of developing criteria toexplicitly differentiate among involuntarydysfluency, voluntary disfluency, and perceptualfluency will be discussed in greater detail inSection II.B.3.

I.B.4. Core versus secondary stutteringbehaviors. Since many developmental models ofstuttering consider some form of repetitions torepresent the "core" of stuttering (e.g., Bloodstein,1987; Stromsta, 1986; Van Riper, 1982) theinclusion of core behaviors in a definition appearswarranted) Beyond "involuntary repetitions," andperhaps somewhat secondarily the duration andfrequency of prolongations and silent pauses,there appears to be less agreement as to what issufficient to delimit stuttering, except perhaps forthe frequent remark that defining stuttering ismuch more complicated than some would think.Of course, there is a wide variety of so-calledsecondary behaviors or accessory featuresassociated with stuttering (e.g., Wingate, 1964),the specification of which would be importantespecially in the consideration of severity.

I.B.5. Inclusive definitions of stuttering.Although there currently seems to be a betterappreciation for psychological effects onphysiological behaviors (e.g., Smith & Weber,1988; Zimmerman, 1980c) definitions for the mostpart are rarely inclusive of external and internalinfluences on the disorder. Rather, they are eitherpredominantly psychologically or physiologicallybased. Comprehensive definitions of stutteringsimilar to the "integrated theory" notion proposedmost recently by Smith and Weber (1988) need tobe better developed.

Definitions of stuttering do not always includean account of the suspected etiology of thedisorder. For example, definitions could includethe notion that involuntary "core" behaviors occuras a consequence of deficits, at various levels, intemporal programming (e.g., Caruso, Abbs, &Gracco, 1988; Kent, 1984), spatial or movementprogramming (e.g., Zimmerman, 1980c), or bothtemporal and spatial programming (e.g., Alfonsoet al. (l986a; 1987a,b,c).

With a proper definition of stuttering, theresearcher is better able to define stutteringsubjects. Section II.A. shows, however, thatresearchers generally provide little definition ofthe stuttering populations that serve in their

experiments, one consequence of which is that itoften makes it difficult to make appropriatecomparisons among experiments.

SECfION II: SUBJECf DEFINITIONAND SELECfION CRITERIA

ll.A. An OverviewA review of the literature indicates that little

detail is given in journal articles regarding eithersubject definition or selection criteria. It is moreoften the case that stuttering severity is reported,although the means by which the severityestimate is determined is highly variable acrossexperiments. What follows are examples of subjectselection criteria (and subject definition, if given)that have been reported in recent or frequentlycited research papers. The aim is to demonstratethe variability among published descriptions ofexperimental subjects and the criteria employed toselect and define them. The following citations arein alphabetical order.

Many reports of experiments in thecontemporary literature provide little informationat all. For example, Freeman and Ushijima (1978)state only that their subjects were mild-moderateor severe. No other details of subject selection ordefinition criteria are given. Guitar et al. (1988)reports the gender of the subjects, that they wereall native speakers of the same language, and hadnever received treatment for stuttering. The firstauthor subjectively judged the severity of thesubjects. The criteria for estimating severity werenot given. Although the motivation for the Martinand Haroldson (1988) study was to experimentallyincrease stuttering frequency, so that thedefinition of stuttering, the frequency ofstuttering, and the severity criteria, are crucial inthis type of study, very little detail is given. Theprocedure employed is difficult to ascertain. "Inthe control room, the experimenter monitored allsessions auditorily and depressed a handswitcheach time the subject stuttered. Stuttering wasdefined in terms of moments or instances ofstuttering and not in terms of specific disfluencytypes." McClean (1987) reports that "informalassessment of conversations with the stutterers (7adult male) suggested that as a group theirstuttering severity ranged from mild to severe."No other details of subject selection or definitioncriteria are given. Zimmerman (1980a,b) andZimmerman and Hanley (1983) used subjects whowere enrolled in speech and hearing clinics at thetime of data collection. The specific type of therapyis not mentioned. No' details of the severity

234 Alfonso

criteria are given: "They ranged in severity frommild to severe as judged by a certified speech­language pathologist."

Other reports of experiments provide somedetail regarding subject selection criteria, thoughthe type and amount vary in considerable degree.For example, Alfonso and his colleagues (Alfonsoet aI., 1986a,b; 1987a,b,c; Kalinowski & Alfonso,1987; Story & Alfonso, 1988; Watson & Alfonso,1982, 1983, 1987) used subjective and objectivecriteria to identify stutterers and group them onthe basis of stuttering severity. Objectiveevaluations of stuttering frequency and type werecompleted using a combination of proceduresdescribed in the Stuttering Interview (SI) (Ryan,1974) and the Stuttering Severity Instrument(SS1) (Riley, 1972). Subjective judgments ofstuttering severity were obtained from certifiedspeech-language pathologists and from theexperimental subjects. Additional criteria aredeveloped if severity ratings differ markedlybetween reading and conversational speechsamples, among objective and subjective criteria(e.g., Watson & Alfonso, 1987), or if inter-testreliability, as a function of time periods or theidentity of the judges, was low (Kalinowski &Alfonso, 1987). Caruso et aI., (1988) assessedstutterers' dysfluency during a conversationalspeech sample using two objective measures: 1),mean stuttering frequency (MSF), and 2), mean

, stuttering duration (MSD). Stuttering was definedas sound I syllable repetitions and soundprolongations. Conture et aI. (1977) used MSFduring conversation and oral reading. Stutteringseverity was determined by use of the Iowa scale(Johnson, Darley, & Spriestersbach, 1963) basedon the MSF. All of the subjects for Conture et aI.,(1985) were receiving therapy at a universityspeech and hearing clinic. MSF for sound/syllablerepetition, sound prolongation, or within-wordpause was used as a measure of severity. Metz etaL (1983) and Sacco et aL (1987) selected subjectsenrolled in a residential stuttering treatmentprogram. They used MSF based on sound andsyllable repetitions, sound prolongations, and/orbroken words produced during an unidentifiedreading sample. Severity was calculated using theSSI (Riley, 1972; 1980). The Shapiro (1980)experiment required that the locus of stuttering,type of dysfluency, and stuttering severity bereliably determined. Subjects read the RainbowPassage five times and performed the Job andTAT Tasks (Johnson et aI., 1963). Subjects wereaccepted if the interjudge agreement across fourjudges meet criteria. Operational definitions of

severity of stuttering were obtained using theStuttering Severity Scale (Johnson et al., 1963),and estimates of the specific type and locus ofstuttering were subjectively made from videotapeviewing.

ILB. Needs and future directions

The section above indicates that it is usually thecase that one would find little detail in a journalarticle regarding the subject definition andselection criteria employed in the experiment.Stuttering severity estimates are given more oftenthan subject definition and selection criteria,although the means by which the severityestimate is arrived at is highly variable acrossexperiments. In what follows, certain issuesrelevant to the development of subject definitionand selection criteria are discussed in detail. Theissues represent broad areas of concern, andshould not be construed as suggested minimalcriteria. Rather, criteria should be developed as afunction of the nature of the experiment at hand.The goal that all researchers should share,however, is that adequate descriptions should begiven in sufficient detail so that experiments canbe replicated and/or results can be appropriatelyinterpreted and compared among experiments.

II.B.1. What identification measures should beused? The common use of a standardizedperceptual test such as the SSI would lead to anobvious advantage of direct comparison of subjectsacross different experiments. Other identificationmeasures should be considered, and include: (a)familial history of stuttering,2 (b) type(s) andduration(s) of therapies received, the distinctivecharacteristics of a therapy program, a descriptionof clinical goals (e.g., slowed speech rate, gentleonsets) of these therapies, and how well they aremaintained in habitual speech, (c) estimates ofcovert stuttering severity, for example, the stut­terer's judgment of the frequency and severity ofdysfluency, secondary characteristics (e.g., Riley,1972; Van Riper, 1982), descriptions of contextualconditions in which fluency and dysfluency areenhanced, and an estimate of the success to whichthe subject is able to use fluency enhancers topromote fluent speech, (d) estimates of overt stut­tering severity, for example, the frequency andduration of repetitions, prolongations, and silentpauses, and (e) a description of the subject's fluentspeech, for example, rate and naturalness. Speechsamples gathered as part of the evaluation shouldbe obtained during both extemporaneous and readspeech, and the differentiating stutteringbehaviors should be noted. Speech samples

Subject Definition and Selection Criteria for Stuttering Research in Adult Subjects 235

gathered as part of the evaluation andexperimental run should routinely be video-taped.

However, perceptual testing alone poses a num­ber of problems. The first is that the magnitude ofstuttering severity is highly variable withinsubjects so that a classification of "severe" could begiven to a subject tested in the morning while aclassification of "moderate" could be given to thesame subject tested in the afternoon (This prob­lem is discussed in greater detail in SectionILB.4.) A second problem, if one accepts the notionof physiological loci, is that available standardizedtests do not adequately differentiate between, forexample, labial stutterers and laryngealstutterers. The latter problem is related to a moreserious shortcoming of subject selection andclassification schemes made on the basis of stan­dardized perceptual testing alone, in that no phys­iological measurement of speech is attempted. Theomission of physiological description is a seriousomission because: (a) a physiologically basedclassification may be more stable across time thana perceptual method, especially as the linguisticstructure of dysfluent speech is better understoodand the corresponding linguistic-physiologicaldata base is increased, and (b) a dysfluent utter­ance may not be identified at the perceptual level.That is, while a segment of a stutterer's speechmay be judged "fluent" by a group of listeners, theacoustic, and/or movement, and/or electromyo­graphic signals underlying the perceptual seg­ment may appear inappropriate or "dysfluent"(e.g., Alfonso et aI., 1984; Baer & Alfonso, 1984;Shapiro, 1980).

A relatively noninvasive physiological-basedbattery could be devised. The following are exam­ples of possible criteria that could be included in asubject selection and definition battery. Of course,the selection and applicability of physiologicalcriteria are dependent upon the nature of the pro­posed experiment, the point, however, is that fre­quent and systematic use of certain physiologicalmeasures would enhance the validity of data com­parison across different experiments. Noninvasiverespiratory criteria, based on data obtained by aRespitrace inductive plethysmograph for example,could include the magnitude of the lung volumeexchange for speech inspiration and expiration, orflow rates during fluent speech, since both havebeen shown to differentiate stutterers from non­stuttering adults, and, perhaps distinguish mildfrom severe stutterers (e.g., Lewis, 1975; Peters &Boves, 1987, 1988; Watson & Alfonso, 1987). Anextensive body of literature on the differencesbetween stutterers' and control subjects' acoustic

reaction-time responses (e.g., Watson & Alfonso,1983, 1987) suggest that predominantly laryngeal(and secondarily respiratory and supralaryngeal)noninvasive criteria could include acoustic reac­tion-time latencies for steady-state vowelresponses. More direct noninvasive laryngeal esti­mates could be made using an electroglottographwith the reaction-time paradigm. Vocal-tract cri­teria based on direct measures of supralaryngealarticulator movements would require more tech­nologically sophisticated noninvasive paradigmsthan those discussed above, and could include, forexample, strain-gauge or opto-electronic move­ment transducers to measure lip and jaw dis­placement. These techniques are becoming morecommon in many laboratories. It would be impor­tant that supralaryngeal physiological criteria forthe selection and definition of stuttering subjectsbe centered on organizational principles of speechmotor control, that is, centered on good represen­tatives of what are believed to be speech motorcontrol parameters. For example, noninvasive lipand jaw displacement amplitude data could beanalyzed to assess motor equivalence covariationand sequential ordering in labial gestures (Alfonsoet aI., 1986a, 1987b,c; Caruso et aI., 1988). Thisissue is discussed in greater detail in SectionII.B.3. It is less important to know and base groupcomparisons on displacement amplitudes andvelocities of individual lip and jaw movements,and at what rates they move, than it is to knowabout the organizational principles underlying lip­jaw movements because (a) efficient, rapid, andfluent speech requires a relatively high degree ofspatial and temporal coordination among thesupralaryngeal speech structures, and (b)interspeaker variability at the motor control levelis inherently less variable than at the phoneticlevel. Of course, data obtained by a combination ofthese techniques could be used to measure inter­system physiological parameters, for example,respiratory-laryngeal timing (e.g., Peters & Boves,1988; Watson & Alfonso, 1987).

11.B.2. What behaviors during the moment ofstuttering should be included in the definition?The type, duration, and severity3 of the involun­tary dysfluency should be identified during themoment of stuttering. Because the relationshipbetween linguistic structure and speech motorspecification in dysfluent speech is not fullyunderstood, a description of the linguistic contextin which the dysfluency occurred should be given.At the very least, the intended fluent phonetictarget should be identified, for example, stressedsyllable initial voiceless aspirated stop. Other

236 Alfonso

linguistic descriptors that could be specifieddepending on the focus of the experiment includeword position and content versus function word. Adetailed consideration of psycholinguistic vari­ables associated with stuttering is given inWingate (1988). A number of secondary character­istics, including anxiety and emotional stress,should also be considered.

Although Shapiro (1980) concluded on analysisof EMG data that the perceptions of the judges aswell as the subjects themselves regarding theidentification of labial, laryngeal, and lingualpredominate locus were erroneous, it might stillbe clinically and theoretically useful, particularlyin regard to the subject's perception of hislherproduction of the dysfluency, to include in thedefinition a statement about clinician and/or clientestimate of locus. We need more physiologicaldata, in parallel with perceptual and acoustic, todetermine with certainty whether or notstuttering results from a physiological disruptionat one location, the larynx for example, or if theentire speech motor system fails simultaneously.Shapiro concludes that the stuttering may not beable to identify the location of the disruption,however, we don't have enough data to know thisfor certain, nor do we have enough data to knowwhether a breakdown at the larynx, for example,occurs first and leads to failures in othercomponents of the speech system in response tothe laryngeal failure.

Finally, for those experiments based onphysiological data it may be useful to segment theevents surrounding a dysfluency. By way ofexample, Alfonso and Seider (1986) examinedacoustic, respiratory and laryngeal kinematicdata, and laryngeal electromyographic (EMG)data during an interval beginning with thetermination of fluency, followed by an inaudibledysfluent period, followed by an audible dysfluentperiod, and ending with a fluent period. LaryngealEMG and movement data showed that thelaryngeal configuration was clearly differentduring the inaudible than during the audibleperiods of the dysfluent episode. The configurationappeared most in~ppropriate during the initialinaudible period and less so during the followingaudible period. Segmentation also allows for thecomparison of speech motor events during thedysfluent periods with events immediatelyfollowing, that is, during fluent production of thephonetic target. Thus, statements about theconfiguration of the vocal tract during moments ofstuttering can be made in reference to the vocaltract configuration during more fluent episodes

immediately preceding the dysfluency, and inreference to the vocal tract configurationimmediately following the dysfluency andassociated with the intended fluent phonetictarget.

I1.B.3. Is there such a thing as "normaldisfluency" and can it be reliably differentiatedfrom stuttering? Perceptual fluency spans a widecontinuum, the endpoints of which could beidentified as "severely fluent" to "severelydysfluent." There is large variability within the"normal" subsection of the continuum, rangingfrom something like "severely fluent" to "normaldisfluent" and apparently even more variabilitywithin the "abnormal" subsection of thecontinuum, ranging from something like"abnormal fluent" to "severely dysfluent." Becausethe "normal" and "abnormal" subsections of thecontinuum appear to overlap perceptually, itseems that defining abnormal fluency, particularlywhen produced by severe stutterers, may be morestraightforward than defining normal disfluency.That is, it may be more experimentally viable andmore fruitful in the long run to modify the abovequestion and to ask: is the perceptual "fluent"speech of severe stutterers similar to the fluentspeech of control subjects? The latter form of thequestion may be more experimentally viablebecause: (a) the resolution involves extremecontrasts, the continuum endpoints, representedby a severe stutterer contrasted with a fluentcontrol subject in the example here. The extremecontrast should be easier to differentiate thanmore subtle contrasts; for example, certainphysiological characteristics of the perceptuallyfluent speech of a severe stutterer compared withthe corresponding characteristics of a normalsubject's fluent speech should be easier todifferentiate than a mild stutterer's fluent speechcompared with a normal subject's fluent speech,and (b) it would be easier to define a severelydysfluent stutterer than it would be to define aseverely fluent control subject. The latter form ofthe question may be more fruitful because a betterunderstanding of stutterers' perceptually fluentspeech would represent a relatively direct andimmediate increase in our understanding of

.stutterers' speech motor control. Of course, in thelong run we will need to understand better thefluency variability in the normal population aswell as the stuttering population. Thus, thefollowing discussion is a modification of theoriginally assigned question and asks: (a) cannormal disfluency be reliably differentiated fromstutterers' dysfluency, and (b) can stutterers'

Subject Definition and Selection Criteria for Stuttering Research in Adult Subjects 237

perceptually fluent speech be reliably characteristics of stuttering discussed in Section I,differentiated from normal fluent speech? for example, voluntary disfluency, involuntary

Considering normal disfluency first, certainly dysfluency, and perceptual fluency. That is, theadults who do not stutter do repeat and prolong a results of two experiments may generatevariety of speech segments, usually words and conflicting conclusions whether or not stutterers'phrases, and interject pauses between words and perceptually fluent speech is similar to normallyphrases. Generally, the speech segments in which fluent speech simply because the criteria (whichrepetitions occur differentiate the groups; mayor may not be reported) for distinguishing thenonstutterers predominantly repeat whole words stutterers' perceptually fluent speech from theirand phrases whereas stutterers predominantly dysfluent speech differed across the tworepeat sounds and syllables. The duration of experiments.pauses and prolongations may also distinguish Second, the majority of experiments are basednormal disfluency from abnormal dysfluency. on perceptual data alone, some are based onHowever, the frequency of repetitions and the perceptual and speech acoustic data, and far lessduration of prolongations, regardless of the speech include kinematic and neuromuscular data insegment in which these occur, are far less in parallel with perceptual and acoustic. One sourcemagnitude in the nonstuttering population than of the conflict was discussed above in Sectionin the stuttering population. That is, stutterers do II.B.1, that is, while a segment of a stutterer'smore of everything; repetitions, prolongations, and speech may appear normal or "fluent" at thepauses. Normal disfluencies are not usually perceptual level, it may appear abnormal oraccompanied by secondary characteristics and "dysfluent" at the acoustic, and/or movement andmay be less susceptible to higher linguistic muscular levels (Alfonso et aI., 1984; Baer &influences, for example, word order and word type Alfonso, 1984; Shapiro, 1980). Thus, a comparison(see Starkweather, 1987, Chapter 5 for a more based on perceptual data alone could indicate nodetailed discussion of normal disfluency). An difference between the groups, while a comparisonimportant assumption underlying the distinction of the same utterances using the same perceptualbetween normal and abnormal speech is that criteria for fluency and dysfluency but based onnormal speech movements are voluntary whereas physiological data could find significant groupstutterers' dysfluencies are not. Although it may differences. Clearly, perceptual data alone are toobe difficult, and perhaps impossible, to ascertain far removed from the source to be able to makethe volition of speech motor output, determining detailed analyses of the fluent-dysfluentthe extent to which repetitions, prolongations, and distinction, and as such would mark only the mostpauses are under voluntary motor control may be obvious instances of stuttering. The lack ofthe ultimate test of the normal--dis!dflnume:>YITTlc'Vy"vlTle:>yr'l'lsTTu~s----,pmllrlyVls'tiinotrlon:gri-icf']aillf--rld~atJta~a'l1dndhr""e'1'ls~sii-rnTtgrlt.hhiJisrii~ssq'luTIerii~s --.orrlf----abnormal dysfluency distinction. concern for other reasons. For example, the

Considering stutterers' perceptually fluent question of determining whether the perceptuallyspeech next, there is considerably more data "fluent" speech of severe stutterers is similar toregarding the contrast between stutterers' the fluent speech of control subjects is importantperceptually fluent speech and normal fluent because the answer to the question will helpspeech than in the contrast discussed above. determine whether the stutterers' speech motorHowever, some experimenters conclude that system exhibits generalized spatio-temporalstutterers' perceptually fluent speech is not abnormalities regardless of the perceived fluency,different than control subjects' fluent speech, or whether, alternatively, it behaves normallywhile the majority of experimenters seem to think except during moments of dysfluency. It isthat it is (see, for example, Van Riper, 1982, generally the case that perceptual and acousticChapter 16; Bloodstein, 1987, Chapter 1). Rather data in the absence of simultaneously gatheredthan review a relatively large literature here, it physiological data are not sufficient to answermay be more useful in regards to the development questions about speech motor control.of subject definition and selection criteria to A second reason for the conflict in the results ofdiscuss a few of the reasons underlying the reported literature on the distinction betweenconflicts in the results of these types of stutterers' perceptually fluent speech andexperiments. It should be noted first, however, nonstutterers fluent speech is that the routinelythat the conflicting results are no doubt posed form of the question, "Is stutterers' fluentconfounded by the lack of adequate definitions and speech similar to the fluent speech of adults whocriteria for distinguishing among certain essential do not stutter," is too broad. It is possible that

238 Alfonso

certain aspects of speech do not differentiate thegroups while other aspects do differentiate thegroups. For example, Baken et a!. (1983) found nodifferences between stutterers and controlsubjects in chest wall preposturing maneuversimmediately preceding fluent speech. Yet,significant differences in subglottic pressure (e.g.,Lewis, 1975), flow rates (e.g., Hutchinson, 1975),and lung volume charge and deflation for speech(e.g., Story, 1990; Watson & Alfonso, 1987) havebeen observed between the groups. Thus, it is lessappropriate to ask whether or not respiratorycontrol, per se, (and certainly not speechproduction, in genera!), can be differentiatedbetween the groups. Rather, it might be thatstutterers and normal speakers perform certainspeech respiratory gestures in a similar fashion,namely respiratory preposturing, but that otheraspects of speech respiration, namely themagnitude of the inspiratory charge, areperformed differentially.

A third and perhaps most important reason forthe conflict in the results of these types of experi­ments is that group comparisons are frequentlybased on physiological data that reflect relativelyvariant phonetic level speech gestures. Group databased on phonetic level contrasts are inherentlyunstable since spatial and temporal control of thespeech structures to mark phonetic distinctionsvaries as a function of phonetic context, stress andrate, dialect, and individual speaker preferences.Rather, group comparisons should be based onrelatively stable spatial and temporal character­istics of normal speech dynamics, for examplethose that best reflect organization principles ofspeech motor control. Dynamic parameters thatare relatively invariant across multiple produc­tions of an utterance are thought to be good repre­sentatives of speech motor control parameters(e.g., Gracco & Abbs, 1986). Thus, an appropriatecomparison would be one based on the extent towhich the magnitude of the relatively stable spa­tial and temporal characteristics of normal speechdynamics approximates corresponding character­istics of stutterers' perceptually fluent speech(Alfonso et aI., 1986a, 1987b,c; Caruso et aI.,1988). For example, motor equivalence covariationfor speech is based on the observation that normalsubjects control the relative displacement of indi­vidual articulators (e.g., the tongue and jaw) en­listed in a vocal tract gesture (e.g., alveolar clo­sure) in such a way that the variability of thegesture is less than the variability associated withthe individual articulators comprising the gesture.Accordingly, it is less appropriate to base group

comparison on displacement amplitude and peakvelocity of a single articulator, the tongue or thejaw for example, than it would be to compare therelative organization of the combined tongue-jawdisplacement, because the latter comparisonreflects a relatively stable component of speechdynamics while the former reflects idiosyncraticspeaker preference. To summarize, the question ofwhether stutterers' perceptually fluent speech issimilar to the fluent speech of non-stutteringadults can be appropriately addressed if criteriaare developed that: (a) objectively distinguishamong perceptually fluent, disfluent, anddysfluent utterances, (b) define the question moreprecisely, that is, provide more detail aboutspecific aspects of speech production, (c) requireappropriate types of data, and (d) require groupcomparisons based on appropriate speechbehaviors.

I1.B.4. How can a definition take into accountvariability in stuttering behavior? At least threetypes of variability should be taken into account.The first is associated with the well-knownvariability in the frequency and severity ofdysfluent behaviors as a function of variousexternal conditions. For example, dysfluentbehaviors decrease and perceptually fluentbehaviors increase with manipulation of auditoryfeedback; when the intensity of the feedbacksignal is increased, decreased, or masked byvarious techniques, and when the time of arrivalof the auditory feedback signal to the talkers' earsis delayed. Fluency is enhanced by imposing anexternal rhythmical marker on the speech task,for example, a metronome signal, choral reading,and singing. Finally, fluency is enhanced by theadaptation effect. Most of these conditions willenhance fluency in non-stuttering adults as wellas stutterers. The exception seems to be thedelayed auditory feedback (OAF) paradigm, wherea delay in the signal could result in increaseddysfluency in some non-stuttering adults,although normal speakers vary widely in theirsusceptibility to OAF (Alfonso, 1974). Thus,alterations of the feedback signal, imposition of arhythmical marker, and multiple repetitions ofthe same speech task increase fluency in mosttalkers, stutterers and nonstutterers alike. Itappears possible that all of these conditionsenhance fluency through a similar mechanism,that is, they act to highlight the control of prosodyand perhaps other temporal parameters of thespeech motor plan.

Likewise, there are a number of externalconditions that increase dysfluent behaviors and

Subject Definition and Selection Criteria for Stuttering Research in Adult Subjects 239

decrease perceptually fluent behaviors. It is wellknown that the contextual surround, for example,stressful situations and tasks, and specificlisteners, will increase the frequency and severityof dysfluency. Certain linguistic conditions, suchas long words, content versus function words, andwords carrying high information loading, areknown to promote dysfluency. Once again, theseconditions are known to similarly affect adult non­stutterers, though not to the degree seen instutterers. On the other hand, there are thoseadult nonstutterers (and apparently stutterers)who become more fluent in seemingly stressfulsituations. A likely causal agent for the increasein dysfluency in these conditions for bothstutterers and adults who do not stutter is theeffect of stress on the speech motor system (e.g.,Zimmerman, 1980c).

The second type of variability is associated withchanges in the type, frequency, and/or severity ofdysfluent behaviors that occur across varyingunits of time. As an example of the variation indysfluent behaviors that occur across the longterm, it is generally observed that very youngchildren who stutter do so primarily by repetitionand prolongation, while over the course of 2 to 4years develop a variety of other overt behaviorsnot present at the onset. Thus, stuttering inadults is comparatively idiosyncratic and as suchvaries considerably across the adult stutteringpopulation. There are other examples of long-termchanges in dysfluent behaviors, although the basisfor the change is not sufficiently understood. Withrespect to the example given here, the source andextent of the variability of "core" behaviors, therepetitions and prolongations in very youngchildren, compared to the variability of"secondary" behaviors in older children and adultsapparently remains unresolved (Stromsta, 1986;Van Riper, 1982). Of course, variations indysfluent behaviors occur across much smallerunits of time. In fact, the perception of stutteringfrequency, severity, and type within the samesubject can change significantly as a function ofmany of the external conditions mentioned above(e.g., stressful situations and tasks) over thecourse of hours and days. Thus, as was pointedout in Section II.B.l., the same subject can beclassified as a mild-moderate stutterer in amorning session and using the same severityrating instrument be classified as a severe­moderate stutterer in an afterno'on session. Thisposes a particular problem in the development ofsubject definition and selection criteria andillustrates the importance of accounting for the

variety of conditions that are known to induceshort- and long-term fluency variations.

A third type ofvariability that could be includedin subject definition and selection criteria is thetype observed at the speech motor level. Greatervariability for stutterers compared to control sub­jects at every accessible level of speech production,namely, acoustic, movement, and neuromuscular,is frequently reported and must be considered arobust observation. Relatively high levels of vari­ability can, of course, be of serious consequence inmeeting the demands of rapid, fluent speech. Highvariability implies that stutterers' control of themovements ofthe speech structures is less precisethan adults who do not stutterer. The lack of pre­cision implies that stutterers lack the flexibilityobserved in normal speech, the flexibility thatforms the basis for co- production of speechsegments and in other human multiarticulatesystems. Thus, the increased variability observedin the kinematics of many structures ofthe speechmechanism, and more importantly, the increasedvariability associated with various organizationalcomponents of fluent speech (e.g., the tongue-jawsynergies referred to in Section II.B.3.) suggestthat stutterers' control of the speech motor systemlacks the flexibility and efficiency to meet the de­mands of rapid, fluent speech. (e.g., Alfonso et al.,1985, 1987c; Stromsta, 1986).

In summary, subject definition and selectioncriteria should account for the variability that isfrequently observed at all levels of stutteringbehaviors. In regards to treatment of the data,descriptions of data dispersion in addition tocentral tendency should be reported. Reportingthe complete data base, perhaps as an appendix,would often be appropriate. Statements regardingthe suspected source or the conditions promotingthe variability, and the effect of the source orconditions on the variability, should be included;for example, fluency enhancement by rhythmicalstimulation, dysfluency increase by psychologicaland physiological stress, accompanied by astatement regarding the magnitude of the fluencychange. Finally, we need to understand better therelationship between the magnitude of thevariability and stuttering severity.

11.B.5. What are the commonalities amongstutterers? There are certain characteristics thatappear to be common among stutterers as a groupand should be considered in developing subjectdefinition and selection criteria. For example, allstutterers seem to. experience the notion ofstutterer-identification discussed above in SectionLB.2. That is, regardless of severity, or the degree

240 Alfonso

to which overt or covert manifestations ofstuttering are realized, stutterers seem to knowthat they are stutterers. If this is indeed true,then the notion of self-identification should be acentral component of the definition of stutteringand stutterers. The stutterers' identification ofthemselves as stutterers and their identification oftheir moments of stuttering should focus thedevelopment of perceptual and physiologicalcriteria.

Based on family history data, many researchershave concluded that the predisposition to stutteris genetically transmitted. For example, VanRiper (1982) concludes: "This incidence is so muchgreater than that found generally in thepopulation (about 5 per cent) that it is difficult tobelieve that environmental factors alone couldaccount for the results" (p. 330). Family historystudies show that more males than femalesstutter, that there is a high incidence of stutteringwithin families, and that the risk of stuttering ishigher if the mother, rather than the father,stutters. Certain of these studies allow thedetermination of risk to relatives; stutteringoccurred in about 20% of male relatives and about5% of female relatives of male stutterers, whereasstuttering occurred in about 25% of male relativesand 12% of female relatives of female stutterers(Kidd et a1., 1978, 1981). However, as Pauls (1990)points out in a paper entitled "A review of theevidence for genetic factors in sl1

tuttering," there are significant limitations tothe family study method. Coupled with the factthat these limitations make it impossible tounderstand the exact nature of the transmission,and that a disorder as common as stuttering isalmost certain to be etiologically and geneticallyheterogeneous, the inclusion offamily history datain subject definition and selection criteria shouldbe cautiously considered.

The onset of stuttering is fairly common acrossstutterers. Most stuttering develops duringchildhood and is usually marked initially bysyllable repetitions and prolongations. Adultstutterers who demonstrate a characteristicchildhood onset should be distinguished from therelatively few adult stutterers who report suddenonset later in life.

Finally, certain physiological characteristics ofthe speech motor system are shared by moststutterers. For example, the stutterers' speechmotor system is inappropriately susceptible topsychological and physiological stress. Fluency,measured at the output of the system primarily byperceptual criteria, is enhanced by rhythmical

stimulation. The reaction-time of the stutterers'speech motor system is slower than normalspeakers. And the stutterers' system may becharacterized by a disability in spatial, temporal,or spatio-temporal coordination, particularly ofthe type discussed in Section II.B.3.

11.B.6. What would be the benefits to research inhaving a consensus definition? While it is likelythat reaching the ultimate consensus definition isnot possible to achieve, the alternative, that is tocontinue without suggestive guidelines would leadto a continuation of the variability in subjectdefinition and selection criteria illustrated inSection II.A One of the aims of this paper is todemonstrate the different ways in which currentdefinitions and criteria are inadequate. Withrespect to definitions of stuttering, there are manyinstances where data have been pooled acrosssuch factors as severity and type so that a clearinterpretation of the results is not possible. Thelack of clearly stated subject selection criteria mayhave even more serious consequence on data in­terpretation because of the heterogeneous natureof the disorder. For example, data are frequentlypooled across such variables as subject severity,varying clinical treatment influences on perceptu­ally fluent speech patterns, and idiosyncraticdysfluent verbal and nonverbal behaviors.

There are, however, obvious and significantlimitations to the consensus approach. The first isrelated to the wide varieties of stutteringbehaviors and subjects that are often examined.Overly precise definitions may exclude criticalmembers of a category while overly generaldefinitions would not make the necessarydistinctions between members of a category.Secondly, standardized subject definition andselection criteria would be in a constant state ofreassessment as our knowledge of the disordercontinues to increase.

Many of the benefits of a consensus definitioncan be achieved by the routine practice ofreporting sufficient details of the experiment sothat, at the very least, the experiment canadequately be replicated. The precise details willvary as a function of the nature of the experiment,yet it would be difficult to imagine many caseswhere basic descriptors, for example, severityratings of the dysfluencies and subjects underexamination, could be omitted. The paperpresented here discusses a number of factors thatbear on subject definition and selection criteria.Undoubtedly, other factors not listed here couldalso be included. Thus, the consensus opinionshould be that subject definition and selection

Subject Definition and Selection Criteria for Stuttering Research in Adult Subjects 241

criteria adequately deal with the heterogeneity ofstuttering, that they be specifically generated as afunction of the experiment at hand, and that untilthe significance of the heterogeneity is betterunderstood, definitions and criteria should err onthe side of over- rather than under-definingcritical details of the experiment. Included in thenotion of over-defining in the context of aheterogeneous behavior is the treatment of thedata. Because the reported data are often used byfuture researchers in a way not related to the aimofthe original experiment, it would be appropriateto publish certain forms of individual subject dataas an appendix, in addition to the typicalreporting of averaged data, especially when thedata represent physiological estimates ofstuttering that are technically difficult to collectand/or potentially hazardous to the subject. Theconsensus, then, becomes one of intent ratherthan the specification of consensus criteria.

REFERENCESAlfonso, P. J. (1974). Differentiating phonemic and spectrographic

speech characteristics of DAF susceptibility. Unpublished M.A.thesis, Western Michigan University.

Alfonso, P. J., Watson, B. c., & Baer, T. (1984). Muscle, movement,and acoustic measurements of stutterers' laryngeal reactiontimes. In M. Edwards (Ed.), Proceedings of the 19th Congress ofthe International Association of Logopedics and Phoniatrics (Vol. II,pp. 580-585). Perth, Scotland: Danscott Print Umited.

Alfonso, P. J., Watson, B. c., Baer, T., Sawashima, M., Hirose, H.,Kiritani, S., Niimi, S., Itoh, K., Sekimoto, S., Honda, K., &Imagawa, H. (1985). The organization of supralaryngealarticulation in stutterers' fluent speech production: Apreliminary report. Annual Bulletin of the Research Institute ofLogopedics and Phoniatrics, 19,191-200.

Alfonso, P. J., Watson, B. c., Seider, R. A. (19800). Tongue and jawarticulation in stutterers' fluent speech. ASHA, 28-10,166.

Alfonso, P. J., & Seider, R. A. (1986b). Laryngeal and respiratoryphysiological characteristics of inaudible and audible dysfluentproduction. In S. Hibi, D. Bless, & M. Hirano (Eds.), Proceedingsof the International Congress on Voice (1, pp. 13-22). Kurume,Japan: Kurume University.

Alfonso, P. J., Watson, B. c., & Baer, T. (1987a). Measuringstutterers' dynamical vocal-tract characteristics by x-raymicrobeam pellet-tracking. In H. F. M. Peters & W. Hulstijn(Eds.), Speech motor dynamics in stuttering (pp. 141-150).Wien/New York: Springer-Verlag Press..

Alfonso, P. J., Story, R. S., & Watson, B. C. (1987b). Theorganization of supralaryngeal articulation in stutterers' fluentspeech production: A second report. Annual Bulletin of theResearch Institute of Logopedics and Phoniatrics, 21, 117-129.

Alfonso, P. J., Story, R. S., & Watson, B. C. (1987c). Spatial­temporal variability of tongue- jaw movements in stutterers'fluent speech. ASHA, 29-10, 159.

Andrews, G., Craig, A., Feyer, A. M., Hoddinott, S., Howie, P., &Neilson, M. (1983). Stuttering: A review of research findingsand theories circa 1982. Journal of Speech and Hellring Disorders,48,226- 246.

Baer, T., & Alfonso, P. J. (1984). On simultaneous neuromuscular,movement, and acoustic measures of speech articulation. In R.

G. Daniloff (Ed.), Articulation assessment and trelltment issues.San Diego: College Hill Press.

Baken, R. J., McManus, D. A., & Cavallo, S. A. (1983).Prephonatory chest wall posturing in stutterers. Journal ofSpeech and Hellring Research, 26, 444-450.

Bloodstein, O. (1987). A handbook on stuttering. Chicago, IL: TheNational Easter Seal Society.

Caruso, A. J., Abbs, J. H., & Gracco, V. L. (1988). Kinematicanalysis of multiple movement coordination during speech instutterers. Brain, 111,439-455.

Conture, E. (1990). Childhood stuttering: What is it and who doesit? ASHA,18, 2-14.

Conture, E., McCall, G., & Brewer, D. (1977). Laryngeal behaviorduring stuttering. Journal of Speech and Hellring Research, 20, 661­668.

Conture, E., Schwartz, H. D., & Brewer, D. (1985). Laryngealbehavior during stuttering: A further study. Journal of Speech andHearing &search, 28, 233- 240.

Fairbanks, G. (1960). Voice and articulation drillbook. New York:Harper & Row.

Freeman, F. J., & Ushijima, T. (1978). Laryngeal muscle activityduring stuttering. Journal of Speech and Hellring Research, 21, 538­562.

Gracco, V. L., & Abbs, J. H. (1986). Variant and invariantcharacteristics of speech movements. Experimental Brain Research,65,156-166.

Guitar, B., Guitar, c., Neilson, P., O'Dwyer, N., & Andrews, G.(1988). Onset sequencing of selected lip muscles in stutterers andnonstutterers. Journal of Speech and Hearing ReSellTch, 31,28-35.

Ham, R. E. (1989). What are we measuring? Journal of FluencyDisorders, 14,231-243.

Hutchinson, J. (1975); Aerodynamic patterns of stuttered speech.In L. M. Webster & L. C. Fust (Eds.), Vocal tract dynamics anddysfluency. New York: Speech and Hearing Institute.

Johnson, W., Darley, F., & Spriestersbach, D. (1963). Diagnosticmethods in speech pathology. New York: Harper and Row.

Kalinowski, J. S., & Alfonso, P. J. (1987). Improvement in laryngealreaction time with improvement in fluency. ASHA, 29-10, 124.

Kent, R. D. (1984). Stuttering as a deficiency in temporal pro­gramming. In W. H. Perkins & R. Curlee (Eds.), Nature and trellt­ment of stuttering: New directions. San Diego: College Hill Press.

Kidd, K. K., Kidd, J. R., & Records, M. A. (1978). The possiblecauses of the sex ratio in stuttering and its implications. JournalofFluency Disorders,3, 13-24.

Kidd, K. K., Heimbuch, R. c., & Records, M. A. (1981). Verticaltransmission of susceptibility to stuttering with sex-modifiedthreshold. Proceedings of the National Aauiemy of Sciences, 78,606-610.

Lewis, J. J. (1975). An aerodynamic study of 'artificial' fluency instutterers. Unpublished doctoral dissertation, PurdueUniversity.

Ludlow, C. L. (1990). Research procedures for measuringstuttering severity. ASHA, 18,26-31.

Martin, R. R., & Haroldson, S. K. (1988). An experimental increasein stuttering frequency. Journal of Speech and Hearing Research,31,272-274.

Metz, D. E., Samar, V. J., & Sacco, P. R. (1983). Acoustic analysis ofstutterers' fluent speech before and after therapy. Journal ofSpeech and Hellring Research, 26, 531-536.

McOean, M. D. (1987). Surface EMG recording of the perioralreflexes: Preliminary observations on stutterers andnonstutterers. Journal of Speech and Hearing Research, 3D, 283-287.

Pauls, D. L. (1990). A review of the evidence of genetic factors instuttering. ASHA, 18, 38.

Perkins, W. H. (1983). The problem of definition: Commentary on'stuttering: Journal of Speech and Hearing Disorders, 48,246-249.

242 Alfonso

Perkins, W. H. (1984). Stuttering as a categorical event: Barkingup the wrong tree-reply to Wingate. Joul7Ul1 of Speech lindHeAring Disorders, 49, 431-433.

Peters, H. F. M., &: Boves, L. (1987). Aerodynamic functions influent speech utterances of stutterers in different speechconditions. In H. F. M. Peters &: W. Hulstijn (&is.), Speech motordynIImics in stuttering. Wien: Springer Verlag.

Peters, H. F. M., &: Boves, L. (1988). Coordination of aerodynamicand phonatory processes in fluent speech utterances ofstutterers. Joul7Ul1 ofSpeech lind HeAring Research, 31, 352-361.

Riley, G. (1972). A stuttering severity instrument for children andadults. Joul7Ul1 of Speech lind Ht'lIring Disorders, 37,314-322.

Riley, G. (1980). Stuttering severity instrument. Tigard, Oregon: CCPublications.

Ryan, B. (1974). Progrllmmed therllpy for stuttering in children lindlIdults. Springfield, IL: Charles C. Thomas.

Sacco, P. R., &: Metz, D. E. (1987). Changes in stutterers'fundamental frequency contours following therapy. Joul7Ul1 ofFluency Disorders, 12, 1-8.

Shapiro, A. I. (1980). An electromyographic analysis of the fluentand dysfluent utterances of several types of stutterers. Joul7Ul1 ofFluency Disorders, 5, 2~231.

Smith, A., &: Weber, C. (1988). The need for an integratedperspective on stuttering. ASHA, 30-32.

Starkweather, C. W. (1987). Fluency lind stuttering. EnglewoodGifts, NJ: Prentice-Hall.

Story, R. S. (1990). Pre- lind post-therllpy comptlrison of respirlltory,lIIryngtll1, lind sUprlllllryngeid kinemlltics of stutterers' fluent speech.Unpublished doctoral dissertation, University of Connecticut,Storrs. .

Story, R. S., &: Alfonso, P. J. (1988). Changes in respiratorykinematics following an intensive stuttering therapy program.ASHA, 30-32.

Stromsta, C. (1986). Elements of stuttering. Oshtemo, Michigan:Atsmorts Publishing.

Van Riper, C. (1982). The 7lIIture of stuttering (2nd ed.). EnglewoodOiffs, NJ: Prentice-Hall.

Watson, B. c., &: Alfonso, P. J. (1982). A comparison of LRT andVOT values between stutterers and nonstutterers. Jour7llll ofFluency Disorders, 7,219-241.

Watson, B. c., &: Alfonso, P. J. (1983). Foreperiod and stutteringseverity effects on acoustic laryngeal reaction time. Joul7Ul1 ofFluency Disorders, 8, 183-205.

Watson, B. c., &: Alfonso, P. J. (1987). Physiological bases ofacoustic LRT in nonstutterers, mild stutterers, and severestutterers. Joul7Ul1 of Speech lind Ht'lIring Re5eQrch, 30, 434-447.

Wingate, M. E. (1964). A standard definition of stuttering. Jou1'7lll1of Speech II7ld HeAring Disorders, 29, 484-489.

Wingate, M. E. (1984a). Definition is the problem. Joul7Ul1 of SpeechII7ld Ht'lIring Disorders, 49,429-431.

Wingate, M. E. (1984b). Fluency, disfluency, dysfluency andstuttering. Joul7Ul1 ofFluency Disorders, 9,163-168.

Wingate, M. E. (1988). The structure of stuttering: A psycholinguisticlI7lII1ysis. New York: Springer- Verlag.

World Health Organization (1977). MRnUlll of the intnnlltio7llllstlltistiClll dllSsification ofdisases, injuries, II7ld CIIU5eS ofdeath (Vol.1). Geneva: World Health Organization.

Zimmerman, G. (1980a). Articulatory dynamics of fluentutterances of stutterers and nonstutterers. Joul7Ul1 of Speech lindHeiaring Re5eQrch, 23, 95-1<J7.

Zimmerman, G. (198Ob). Articulatory behaviors associated withstuttering: Onefluorographic analysis. Jour7llll of Speech lindHeiaring Research, 23, 108-121.

Zimmerman, G. (198Oc). Stuttering: A disorder of movement.Joul7Ul1 of Speech lind HeAring Re5eQrch, 23, 122- 136.

Zimmerman, G., &: Hanley, J.M. (1983). A cinefluorographicinvestigation of repeated fluent productions of stutterers in anadaptation procedure. Joul7Ul1 of Speech lind Heiaring ReseRrch, 26,35-42.

FOOTNOTES-Appears in J. Cooper (Ed.), ASHA REPORTS monograph

"Research Needs in Stuttering: Roadblocks and FutureDirections" (18, pp. 15-24). Rockville, MD: American Speech­Language-Hearing Association (1990).

t Also University of Connecticut, Storrs. .IHowever, a paper entitled "Research procedures for measuringstuttering severity" by Ludlow (1990) takes a different approachto the core versus secondary behavior distinction.

2For a discussion of the problems associated with gatheringfamilial history data, see a paper entitled "A review of theevidence of genetic factors in stuttering" by Pauls (1990).

3For a detailed discussion of the problems associated withseverity estimates see a paper entitled "Procedures for assessingthe severity of stuttering for research" by Ludlow (1990).