Second Opinion

about the necessity of treating every child who

stutters as soon after onset as possible and
about the risks involved if their treatment is
Treatment Decisions for postponed for a while” (p. 8). Some of this
disagreement, according to Curlee and Yairi,
Young Children Who stems from the differences in the prevalence
and the lifetime incidence of stuttering and
from our knowledge about the spontaneous
Stutter: Further Concerns remission of stuttering. They wrote that the
“lifetime incidence of stuttering in the United
and Complexities States and western Europe is generally believed
to approximate 4 to 5%…whereas its preva-
lence [ranges] from 0.5 to 1%” (1997, p. 8). If
these numbers are correct, wrote Curlee and
Yairi, then it is “mathematically transparent
Roger J. Ingham that…75% to 90% of all who begin” (p. 9) to
University of California, Santa Barbara stutter will not continue to do so. In addition,
between 32% and “over 80%” (p. 9) of these
Anne K. Cordes recoveries are said to be “spontaneous or un-
The University of Georgia, Athens aided” (p. 9).
Curlee and Yairi are not the first to summa-
rize the prevalence and incidence literature as
showing a 75–90% recovery rate, but the litera-
ture is not entirely consistent with such an
interpretation. Table 1 summarizes the studies

D
rs. Curlee and Yairi’s (1997) recent listed by Bloodstein (1995, his Tables 3 and 6)
paper provided some interesting and that have been conducted since 1950.1 Esti-
provocative comments regarding mates of the lifetime incidence of stuttering
treatment for children who have been stuttering range from 0.70% to 15.40% across these stud-
for less than approximately 2 years. Their paper ies, whereas estimates of prevalence range from
raised several issues that were important, 0.30% to 2.12%. These differences clearly
complex, and fully deserving of critical assess- reflect substantial and significant variability
ment. It was accompanied by two “Second across studies, and they also argue against
Opinion” commentaries (Bernstein Ratner, 1997; Curlee and Yairi’s (1997, p. 9) claim that it is
Zebrowski, 1997), each of which addressed some “mathematically transparent” that “75% to 90%
Ingham additional pieces of this complex area, and each of all who begin” to stutter will recover: At the
of which essentially agreed with Curlee and Yairi extreme, these numbers actually suggest that
on two general points: that some children who the lifetime incidence of stuttering is only 0.7%
stutter may not need clinical services, and that we (Culton, 1986) at the same time that its preva-
have reason to be dissatisfied with the available lence is somehow as high as 2.12% in junior-
evidence about treatment effectiveness for high and high-school students (Gillespie &
children who stutter. We do not entirely disagree, Cooper, 1973). It is certainly not mathematically
but the purpose of this response is to present an simple to translate such data into a 75–90% re-
opinion about Curlee and Yairi’s paper that covery rate.
challenges their premises and their conclusions in
several areas. These areas include the incidence
and prevalence of stuttering, and whether those Recommending Treatments
data provide evidence of high recovery rates; for Stuttering
several issues related to determining whether The next section of Curlee and Yairi’s paper
Cordes treatments are effective and whether they should argued that it may be appropriate to withhold
be recommended; and several issues related to treatment for young children not only because
the experimental evaluation of treatment of the supposedly high rate of natural recovery
effectiveness. but also because “the efficacy of early interven-
tions with children soon after stuttering onset is
Incidence, Prevalence, and
Mathematically Necessary 1
An additional nine studies in Bloodstein’s Table 3, and
Recovery three studies in his Table 6, were conducted between 1893
and 1942. Inclusion of these studies would only lengthen
Curlee and Yairi (1997) began with the our Table 1 without changing the conclusions drawn, as
premise that there is “considerable disagreement readers may verify.

10 American American
Journal ofJournal
Speech-Language Pathology
of Speech-Language Pathology • Vol.
• Vol. 7 •7 No. 3
• 1058-0360/98/0703-0010 August 1998
© American Speech-Language-Hearing Association
Downloaded From: http://ajslp.pubs.asha.org/ by a Universite Laval User on 09/10/2016
Terms of Use: http://pubs.asha.org/ss/rights_and_permissions.aspx
 TABLE 1. Summary of incidence and prevalence studies of stuttering reported since 1950, selected
from those summarized by Bloodstein (1995, Tables 3 and 6).

Study N Population %

Prevalence studies
Schindler (1955) 22,976 Grades 1–12 0.55
Hull (1969) 6,287 Grades 1–12 0.30
Gillespie & Cooper (1973) 5,054 Grades 7–12 2.12
Leavitt (1974) 10,445 Grades 1–6 0.84
Leavitt (1974) 10,449 Grades 1–6 1.50
Brady & Hall (1976) 187,420 Grades K–12 0.35
Hull et al. (1976) 38,802 Grades 1–12 0.80
Leske (1981) 7,119 Grades 1–6 2.00

Incidence studies
Glasner & Rosenthal (1957) 996 First grade 15.4
Andrews & Harris (1964) 1,000 Birth–16 4.9
Andrews & Harris (1964) 206 Adults 4.8
Sheehan & Martyn (1970) 5,138 Freshmen/graduates 2.9
Dickson (1971) 3,923 K–12 9.4
Cooper (1972) 5,054 High school 3.7
Porfert & Rosenfield (1978) 2,107 University students 5.5
Seider, Gladstien, & Kidd (1983) 1,857 Stutterers’ relatives 13.9
Culton (1986) 30,586 University freshmen 0.7

essentially unknown” (p. 10). In expanding on the chances that children will recover.2
this idea, Curlee and Yairi drew on a confer- Curlee and Yairi (1997), on the other hand,
ence presentation by one of us (Ingham, 1996) concluded that treatment may not need to be
that has since become a book chapter by both offered to young children and that delaying
of us (Ingham & Cordes, in press). In this chap- treatment will have no deleterious effects:
ter, we criticized current stuttering treatment “Our contention is that active monitoring of
research, asserting that our discipline and our young preschoolers during the first 2 years of
clients are poorly served when researchers stuttering…rather than intervening actively,
promote treatments that are supported by inad- permits the unaided remission of stuttering for
equate treatment outcome data. We provided most of them and does not adversely affect later
specific examples of articles that recommended treatment of stuttering for those who do not
treatment procedures that were undocumented, stop” (Curlee & Yairi, 1997, p. 12). Clearly,
at best, and shown to be ineffective, at worst. Ingham and Cordes (in press) and Curlee and
We also pointed out that there is, in fact, an Yairi (1997) have drawn diametrically opposed
established and growing body of treatment conclusions from the same research and clinical
research studies that do satisfy a relatively literature, making opposite recommendations
standard evaluation framework, providing for young children who stutter. There would
repeated speech measures from before, during, appear to be several factors underlying this dis-
and after treatment and from both within- and agreement, several of which we explore below.
beyond-clinic conditions (e.g., Craig et al.,
1996; Kully & Boberg, 1991; Lincoln, Onslow,
Lewis, & Wilson, 1996; Martin, Kuhl, & Distinguishing Between
Haroldson, 1972). These studies consistently Disfluency and Stuttering
report clinically significant benefits relative to One of the first issues that deserves further
untreated control conditions, some for very consideration is as simple as the basic fact that
young children, and these benefits may also be 2
evaluated through comparisons with other The more important implication of these findings, actually,
is that parents should be given information about the
reports of children who have not received pro- benefits that may occur if they do try to intervene in their
fessional treatment. Finally, we presented a re- child’s stuttering. Studies reported by Onslow and
analysis of available data that we believed colleagues, and others, strongly suggest that children’s
strongly suggested two conclusions: that treat- stuttering can be reduced if parents employ relatively mild
verbal contingencies for occasions of stuttering and for
ment should be offered to young children who periods of fluent speech production. There is, on the
stutter, and that delaying treatment, even for as contrary, absolutely no evidence that such procedures will
little as 15 months after onset, may diminish cause stuttering to increase or become chronic.

Ingham • Cordes 11
Downloaded From: http://ajslp.pubs.asha.org/ by a Universite Laval User on 09/10/2016
Terms of Use: http://pubs.asha.org/ss/rights_and_permissions.aspx
 a disfluency is not necessarily a stuttering. Starkweather et al. (1990) discussed very little
Measuring disfluencies cannot be equated with data at all (as Curlee and Yairi pointed out).
measuring stuttering, because measuring all
disfluencies would include measuring both
normal and stuttered disfluencies. Systems that Distinguishing Among General
distinguish between stuttered disfluencies and Approaches to Treatment
nonstuttered disfluencies (e.g., Conture, 1990; The determination of what constitutes treat-
Ryan, 1974; Wingate, 1964) also fail, because ment is equally critical to a meaningful discus-
it is consistently possible to find examples of sion about whether to provide treatment for
speech behaviors that cannot be appropriately children who stutter. Three related issues are
classified under such systems (the attempt to important here: whether treatment must be
define all between-word disfluencies as normal administered by a professional, whether treat-
provides one clear example; Cordes & Ingham, ment includes direct or indirect procedures, and
1995). This problem has led some researchers whether a procedure should be referred to as a
to measure “stutter-type” (Meyers, 1986) or treatment if there is little or no evidence that it
“stutter-like” (Yairi & Ambrose, 1992) dis- will be effective.
fluencies, an equivocating approach that does The question of whether treatment must be
not solve the problem of determining whether administered by a professional was raised by
the reported data are meant to represent stutter- Curlee and Yairi in the context of spontaneous
ing or not. Much other current research about recovery, and it is related to previous sugges-
stuttering is conducted simply in terms of dis- tions that parent- or self-initiated procedures
fluencies, with no attempt to provide data about might be a relevant factor in recoveries that
stutterings at all. appear to occur without professional interven-
The implications of such basic definitional tion (Finn, 1996; Ingham, 1983). Curlee and
problems are enormous, especially for very Yairi argued that “most advocates of this view
young children. They lead to diagnostic criteria do not address the apparent failure of care-
that label children as stuttering if their speech givers’ interventions” (p. 11) when stuttering
includes “10% or more total disfluency” (Gre- persists. In fact, Ingham (1983) addressed this
gory & Hill, 1993, p. 28), whether or not that very issue: “there is also no evidence that these
speech includes any atypical, abnormal, or procedures are necessarily effective in reducing
stuttered disfluencies. They also lead to important stuttering in all children. All that we have to
difficulties in evaluating the effects of treatment: look to is some evidence that they may be ef-
A report that disfluencies were reduced is not fective with some children who stutter” (p.
necessarily a report that stuttering was reduced 123). The importance of this point is high-
and should not be interpreted as such. lighted by the fact that Curlee and Yairi used it
Given these complexities, any meaningful to support another suggestion that early inter-
discussion about whether children are stutter- vention for stuttering may be unnecessary:
ing, or about whether children need treatment “If…intervention or treatment is restricted to
for stuttering, should begin by affirming that activities that are undertaken by or under the
normal disfluency will not be confused with supervision of certified clinicians trying to
stuttering and by documenting that the children remediate stuttering, current evidence indicates
in question were, in fact, stuttering. Curlee and that 60% to 70% of preschool age children who
Yairi (1997) did not address this seemingly begin to stutter stop within the first 2 years of
basic issue of defining their terms, a problem onset without having received any kind of such
that may have led them to some questionable professionally directed intervention” (p. 11).
conclusions. They wrote, for example, that “a The premises underlying this statement de-
variety of different treatment procedures may serve some thoughtful scrutiny. First, we see no
be capable of eliminating or significantly reduc- reason to reserve the label “treatment” for cases
ing almost all young children’s stuttering” (p. where a professional was involved. Adults often
10), basing this statement on the evidence that “a provide treatments for themselves or for their
number of different intervention procedures…are children, for everything from headaches to
reported to have high rates (e.g., >85%) of behavioral, emotional, or learning problems.
success or ‘recovery’ (Fosnot, 1993; Martin, Second, we know of no data to suggest that
Kuhl, & Haroldson, 1972;…Starkweather, “60% to 70%” of all children who begin to
Gottwald, & Halfond, 1990)” (pp. 9–10). The stutter receive no professional intervention at
definitional problem here is that some of these all. Third, there is evidence that some children
papers did not report data in terms of children’s who recover without professional intervention
stuttering: Fosnot (1993) discussed disfluencies are exposed to procedures that are known to
and stutter-type disfluencies, so her data may or reduce stuttering (Finn, 1996; Ingham, 1976,
may not be related to changes in stuttering, and 1983; Martin & Lindamood, 1986). In addition,

12 American Journal of Speech-Language Pathology • Vol. 7 • No. 3 August 1998

Downloaded From: http://ajslp.pubs.asha.org/ by a Universite Laval User on 09/10/2016
Terms of Use: http://pubs.asha.org/ss/rights_and_permissions.aspx
 Ingham (1983) and Martin and Lindamood of some types of modifications to the home com-
(1986) showed that direct parent intervention munication environment” (p. 30). The problem
was generally reported more frequently among with this suggestion is that these studies simply
recovered than among nonrecovered stutterers. do not provide experimentally compelling evi-
The conclusion that a group of children stopped dence that the procedures mentioned can improve
stuttering without professional intervention, in children’s stuttering. The Stephenson-Opsal
other words, should not be equated with the and Bernstein Ratner study, for example, em-
conclusion that they all stopped stuttering in the ployed an AB treatment design, with no experi-
absence of actively administered procedures mental control. The subjects were 2 boys who
that are known to reduce stuttering in some stuttered: one aged 3;3 who may have con-
children—that is, depending on how the word founded the results by being concurrently
is defined, without treatment. enrolled in another “behavior management
This discussion is also related to the differ- program” (1988, p. 50) and one aged 6;2. They
ence between direct and indirect treatments for were recorded for unspecified amounts of time
children’s stuttering. Direct treatments, in this in mother-child dyads, and the “palliative ef-
context, are usually defined as those that target fects” were a mean 28% and 45% decrease
the child’s speech behaviors (whether through (respectively) in the children’s “mean percent
the parent or through a clinician); indirect treat- time disfluent.” No evidence was presented to
ments are usually defined as those that target show that this decrease was stable, that it ex-
only environmental or parental behaviors. ceeded base-rate variability, that it reflected
Curlee and Yairi did not distinguish between beyond-clinic performance, or that it reflected
the two, but Zebrowski (1997) made several changes in the children’s stuttering (as op-
recommendations specifically for indirect treat- posed to normal disfluency; see above). In
ment. Bernstein Ratner (1997) also noted that short, this study provided no evidence that the
indirect treatments are “undoubtedly preferred “intervention techniques…can show relatively
to direct treatment by clinicians” (p. 30) for immediate reductions in stutter frequency that
early stuttering. She suggested several possible are time-linked to the intervention manipula-
reasons for this preference, including the com- tion” (Bernstein Ratner, 1997, p. 32).
ment that “it is unclear to me whether most of Winslow and Guitar (1994) obtained similar
the major tasks of direct stuttering treatment results. One boy who stuttered, aged 5;9, com-
can be performed by the average child as young pleted an ABAB experimental design, where
as 2;6 years…for the simple reason that they the A condition was routine interaction and the
are not metalinguistically capable of many of B condition was “structured turn taking” during
the behavioral features of these programs” (p. conversations at meal time in the child’s home.
30). There are at least two large classes of di- Results showed, at best, a 50% reduction in
rect treatments, however, that are ideally suited “stuttering-type disfluencies” during the struc-
to very young children, that require no meta- tured turn-taking condition in the home (from a
linguistic awareness, and that have been shown mean of 7–20 stuttering-type disfluencies per
to be effective at reducing young children’s 300 words to a mean of 5–10 stuttering-type
stuttering. These include punishment of unde- disfluencies per 300 words), with no data from
sired responses combined with reinforcement other settings, no follow-up data, and no evi-
of desired responses, which was discussed dence that the child’s stuttering (not stuttering-
above (Martin et al., 1972; Onslow, Andrews, type disfluencies) was decreasing. This change
& Lincoln, 1994), and procedures that depend might be “palliative,” but it cannot be justified
on controlling the length or complexity of the as an experimentally verified and effective
child’s utterances (by controlling the speaking treatment for children’s stuttering, as we would
situation; no metalinguistic awareness is re- define those terms.
quired) (Costello, 1983; Ryan, 1974). Zebrowski’s (1997) response to Curlee and
The distinctions between professionally Yairi was similar to Bernstein Ratner’s in this
directed and parent-directed treatment, and respect and raises similar issues. Zebrowski
between direct and indirect treatment strategies, provided a series of decision “streams,” including
are both related to the question of whether a suggesting that clinicians could introduce indirect
treatment should be referred to as a treatment if treatments for certain children. Suggested proce-
it has not been demonstrated to be effective. dures included “modeling a reduced speaking
Bernstein Ratner (1997), for example, recom- rate and increased turn-switching pause duration,
mended some indirect home management strat- allowing the child to finish his conversational
egies, because “some controlled experimental turn without interruption, and so forth” (1997, p.
data (Stephenson-Opsal & Bernstein Ratner, 25), methods that Zebrowski justified by refer-
1988) as well as observational data (Winslow ence to publications by Kelly (1993, 1995) and
& Guitar, 1994), validate the palliative effects Zebrowski, Weiss, Savelkoul, and Hammer

Ingham • Cordes 13
Downloaded From: http://ajslp.pubs.asha.org/ by a Universite Laval User on 09/10/2016
Terms of Use: http://pubs.asha.org/ss/rights_and_permissions.aspx
 (1996). Kelly offered less critical reviews of study have varied from publication to publica-
this literature than Nippold and Rudzinski tion (Onslow & Packman, in press). Yairi and
(1995), who could find little supporting evi- Ambrose (1992) required, among other criteria,
dence for these procedures. Zebrowski et al. that a child display not more than 2.99 SLDs
(1996) did offer some data on the effects of per 100 words to be classified as “recovered.”
slowing the mother’s speech rate, from a study Later studies, however, have allowed as many
of 5 children who stuttered (aged 2;10–7;5) as 4 SLDs (see Paden & Yairi, 1996, p. 983;
and their mothers, but those data showed only Watkins & Yairi, 1997, pp. 387–388; Ambrose,
variable and nonsignificant effects on the Cox, & Yairi, 1997, p. 569). At the same time,
children’s disfluencies in a clinic setting. In the opposite problem is also present: In Yairi
other words, as Nippold and Rudzinski (1995) and Ambrose’s (1992, 1996) data, 4 of the 8
concluded previously, not one of the indirect continuing stutterers displayed less than 2.99
treatment strategies recommended by either SLDs at Visit 5 in the clinic but were still re-
Bernstein Ratner (1997) or Zebrowski (1997) garded as nonrecovered (presumably because
can be justified on the grounds of data that of parental reports that these children still stut-
should be interpreted as showing clinically tered outside the clinic). These and other com-
significant reductions in children’s stuttering. plexities in the data from Yairi and colleagues
would be unremarkable were it not for the fact
that these are the numbers used by Curlee and
The Desirable Outcomes for Yairi (1997, p. 10) to claim that the rate of recov-
Children Who Stutter ery in untreated stutterers within the longitudinal
Arguments about whether treatment is desir- study is as high as 89%. In a more general sense,
able or necessary for young children who stutter these conflicting reports simply demonstrate the
are further complicated by arguments about the importance of establishing and following clear
desired treatment outcome. Whether adminis- criteria for determining when a child will be said
tered by a professional or not, whether direct or to have recovered from stuttering, as well as the
indirect, whether administered immediately or difficulties inherent in doing so.3
delayed, we assume that the goals of treatment
for a child who stutters are twofold: to eliminate
the stuttering and to allow the child to grow up The Effects of Delaying Treatment
without any of the social or emotional conse- As Curlee and Yairi and others have ob-
quences of living with a communication disor- served, there is unquestionably some number of
der. We also believe that accomplishing the children who stutter for several months and
former is among the most straightforward ways then stop stuttering without formal treatment.
to achieve the latter (Lincoln & Onslow, 1997). For example, of the 43 children who ever stut-
One of the many complications in this area tered in Andrew and Harris’s 1964 longitudinal
involves determining whether stuttering has study, approximately 18 children stopped stut-
been eliminated. Clearly, this determination tering within 6 months.4 They labeled most of
cannot be made from brief speech samples these children “Transient Nonfluent,” but such
recorded within a single treatment environ-
ment; there are sufficient reports of stuttering
3
varying across time and place, and of “clinic- The issue is complicated by the fact that Yairi and
Ambrose (1996) published a correction to the Yairi and
bound fluency,” that this point should be self- Ambrose (1992) data that, itself, may require correction. In
evident (Andrews & Ingham, 1972; J. Ingham, a personal communication to one of us (R. I.) regarding the
1989; Ingham & Packman, 1977; Lincoln & original correction, Ambrose (personal communication,
Onslow, 1997; Lincoln et al., 1996; Martin et April 22, 1996) reported that by Visit 5 there were 13
Recovered and 8 Continuing stutterers. The mean SLD
al., 1972; Onslow, Costa, & Rue, 1990; Onslow scores for both groups of subjects, as shown in the 1996
et al., 1994; Reed & Godden, 1977; Ryan, Erratum, correspond precisely to that number of subjects
1971, 1974; Ryan & Van Kirk Ryan, 1983). within each category. Yairi and Ambrose (1996) also
Equally, this determination cannot be made if reported, however, that at Visit 5 there were 14 Recovered
data are reported only in terms of disfluencies, and 7 Continuing stutterers.
4
rather than in terms that make it clear whether Readers should be aware that the Andrews and Harris study
the child is stuttering or not, as discussed above. was not without its problems (Ingham, 1976). The data from
this study were first displayed in a figure (Andrews & Harris,
The longitudinal study of Yairi and col- 1964, p. 31) that has been reproduced in various editions of
leagues, referred to by Curlee and Yairi (1997), Bloodstein’s A Handbook on Stuttering. In 1984, however,
exemplifies some of the difficulties in this area Andrews (1984, p. 4) made unexplained changes to the years
and is directly relevant to Curlee and Yairi’s of stuttering onset and recovery for 10 of the 43 children. A
comparison between the original table (Bloodstein, 1981, p.
recommendations about delaying treatment. 84) and the revised table (Bloodstein, 1987, p. 94) shows
First, the performance criteria for defining several changes in the longitudinal data for which no
children as “recovered” in the longitudinal explanation has ever been provided.

14 American Journal of Speech-Language Pathology • Vol. 7 • No. 3 August 1998

Downloaded From: http://ajslp.pubs.asha.org/ by a Universite Laval User on 09/10/2016
Terms of Use: http://pubs.asha.org/ss/rights_and_permissions.aspx
 children may display both speech behaviors and withholding treatment from children who are
emotional and social reactions that are indistin- clearly not recovering and then concluding,
guishable from those exhibited by older chil- when they are older, that untreated and unre-
dren or adults who stutter. Given that these data covered children must have had a predisposi-
do suggest that some children may stutter for tion to display chronic stuttering.
something less than 6 months and then stop, we Both Zebrowski (1997) and Bernstein
would agree with the principle of waiting to Ratner (1997) also made recommendations that
initiate treatment to this extent: It certainly does are relevant to decisions about when to intro-
make sense for clinicians to consider waiting duce treatment. Zebrowski presented five po-
perhaps 3 to 6 months from the time of stutter- tential treatment plans, each based on a combi-
ing onset before initiating intervention, if wait- nation of suggested risk factors. It is only the
ing does not introduce social or emotional last of these, Plan E, that includes any mention
pressures for the child or for the parents and if of direct treatment, designed to decrease the
there is some indication that the individual child’s stuttering, and the child must be “24–36
child in question is improving without treat- months postonset” (p. 26) to qualify. Children
ment. One issue that Curlee and Yairi did not in Plan C are allowed to stutter for up to 24
mention, however, is that 3 to 6 months or months without direct treatment; parents here
more has often passed before parents bring a “can be informed about changes in their speech
child to a speech-language pathologist in the that might have an indirect fluency-inducing
first place, making the question of a 6-month effect on the child” (p. 26). Children in Plan D
waiting period essentially moot in many cases. have also been stuttering for up to 24 months,
We emphasize here that the most effective and their parents are “often…beginning to
use of a waiting period would not be simply to suspect that the child may not ‘grow out of it’”
wait in the hope or expectation that the child (p. 26). We simply see no justification for sug-
will recover spontaneously. Instead, the parents gesting that a child who is clearly not “growing
and the clinician could collect recorded base-rate out of it” should be required to wait for 2 full
speech samples, at least monthly and in mul- years before being given access to treatments
tiple speaking situations, to provide a rational that are known to be more effective if used
and data-based source for deciding whether within 15 months of stuttering onset.
stuttering is increasing or decreasing for a par- Again, as we interpret the literature, there is
ticular child. Again, Curlee and Yairi (1997) compelling evidence that the probability of
reported data from Yairi and colleagues’ longi- successful treatment is lower for older children,
tudinal study that are relevant to this point and and the decision to delay stuttering treatment
that we would interpret as potentially confirm- should be made with full awareness of this
ing the need for treatment within 12 months evidence. The comparisons that we presented in
after onset. Repeated measurements of children the Ingham and Cordes (in press) paper showed
in the longitudinal study showed that percent- that 81.8% of preschoolers who received treat-
ages of SLDs declined over the first 12 months ment met our criteria for satisfactory outcome,
in those children who stopped stuttering within but only 54.2% of school-age children who
12 months, whereas the percentage of SLDs received treatment met these criteria, for treat-
“either increased or remained at relatively ment studies that provided data from treatment
stable levels throughout this period among evaluation procedures. Similar comparisons
those whose stuttering persisted” (Curlee & showed that 85.7% of children who received
Yairi, 1997, p. 12). We would argue that these treatment within 15 months of stuttering onset
data provided ample justification for introduc- met our recovery criteria (7/9 boys and 5/5
ing treatment for the children whose stuttering girls), whereas only 59.4% of children who
was persisting, especially if the failure to pro- received treatment starting more than 15 months
vide treatment was related to these children’s after stuttering onset did so (16/28 boys and 3/4
“increased risk of stuttering for another 2 years girls). Other evidence about the importance of
or longer” (Curlee & Yairi, 1997, p. 12). Curlee early treatment can be found in the reports of
and Yairi noted that older children “most likely Onslow and colleagues, for example, who have
constitute a subgroup of stuttering children who reported comparatively better results at long-
have passed through the primary remission term follow-up for children aged 2–4 years
period of the stuttering population” (1997, p. (Onslow et al., 1994) than for children aged
10). They continued, however, with the state- 7–12 years (Lincoln et al., 1996) who received
ment that these older children “are, therefore, essentially the same treatment.
much more likely to possess characteristics or Curlee and Yairi argued, in commenting on
predispositions that result in chronic stuttering” these comparisons, that there are a “number of
(1997, p. 10). We would argue that providing risks involved in combining data from different
treatment for younger children is preferable to studies…[including] such factors as subjects’

Ingham • Cordes 15
Downloaded From: http://ajslp.pubs.asha.org/ by a Universite Laval User on 09/10/2016
Terms of Use: http://pubs.asha.org/ss/rights_and_permissions.aspx
 ages, time since onset, sex, and family history” reasons (both of which Curlee and Yairi also
(p. 10).5 We agree, but the comparisons pre- recognized): an untreated control group may be
sented by Ingham and Cordes (in press) were, scientifically unnecessary, and an untreated con-
in fact, completed separately for boys and girls trol group may be unethical (Lewin, 1995). These
of different ages (preschoolers and school-age) issues are complex, in part because the overriding
and of different times since onset [more or less principle of treatment research should be to pro-
than 15 months, a dividing line chosen based vide the best possible treatment to all participants
on Yairi and Ambrose’s suggestion that (World Medical Assembly, 1989). If it truly is
“chances for chronicity increase approximately not known whether the treatment in question will
15 months after onset” (1996, p. 73)]. The be any better than no treatment at all, then a no-
claim that a particular treatment was solely and treatment control group can provide a scientifi-
directly responsible for the recovery of the cally relevant and entirely ethical comparison.
children who received it is different from the This is not the only option, however; investiga-
claim that the children who received a treat- tors can compare different components or levels
ment recovered, and this is an important dis- of a treatment, or compare newer and older treat-
tinction. Nevertheless, and even given the many ments, both procedures that have been used ef-
terminological difficulties that we have raised fectively in evaluating treatments for stuttering in
here, our comparisons consistently showed that adults (e.g., Ingham, Andrews, & Winkler, 1972;
children who received their treatment relatively Perkins, Rudas, Johnson, Michael, & Curlee,
early displayed more substantial treatment 1974).
benefits than children who received their treat- Large-scale treatment trials may also be
ment relatively late, even when “late” treatment appropriately abandoned when treatment ef-
was defined as only 15 months postonset. Noth- fects become so obvious that it is no longer
ing in Curlee and Yairi’s comments on this defensible to deny those benefits to the control
issue alters this conclusion. group. Such decisions are often made using the
same logic that Sidman (1960) employed to
determine the number of subjects that are
Treatment Research needed before researchers will accept that a
Methodology powerful treatment has been identified. If a
Curlee and Yairi’s final section on “Re- treatment is introduced, for example, to a rela-
search Needs and Ethical Issues” begins by tively small number of similarly afflicted pa-
acknowledging that treatment for very young tients, all of whom had shown stable (or dete-
children who stutter can be effective, and that riorating) base-rate levels, if all immediately
some interventions can be “causally related to reduce their disorder when treatment is intro-
the changes reported in some children’s stutter- duced, and if this effect is replicated across
ing” (p. 14). Nevertheless, they still hesitate to clinics or laboratories, then it is not necessary
recommend treatment for young children, be- to repeat the procedure with too many other
cause of the “absence of scientifically credible patients before clinical scientists should accept
treatment outcome data” (p. 15). We do agree, that the treatment will most probably have the
as discussed above, that there is an absence of same positive effects on the next patient. The
credible data to support many of the procedures possibility certainly does exist that the next
that are recommended as treatments for chil- subject might respond differently, but similar
dren who stutter, but these issues also deserve problems of external validity are evident in the
to be more carefully addressed. assumption that averaged findings from a large
Curlee and Yairi suggested that the use of group are applicable to any one particular sub-
“randomly assigned, untreated control groups has ject. More importantly, sequential analysis
long been viewed as essential for evaluating (Wald, 1947) can establish the probability that
treatment effectiveness” (p. 15) and that it is a sequence of consistently positive (or nega-
“critically important…that such important scien- tive) treatment responses could not have oc-
tific standards not be abandoned until there is curred by chance (Bross, 1952), without the
sufficient evidence available” (p. 15). Untreated need for the sampling-theory assumptions of
control groups are by no means universally ac- large-group research.
cepted, however, for two distinct but overlapping Curlee and Yairi did suggest that the studies
needed to provide credible treatment outcome
5
They also rightly observed that one of the studies included data could begin with “single-subject experi-
in our comparisons (Ramig’s 1993 historical study of mental designs…to assess the effects of differ-
untreated children) was a study of older children, not of ent treatment procedures in an unequivocal
very young preschoolers. But Ramig’s study also included
8 children, including 2 girls, who were assessed within 15
manner” (p. 16). At least eight studies over the
months of reported onset—and of those 8, only 1 boy past two decades have done precisely that, and
recovered. these were cited by Curlee and Yairi. They then

16 American Journal of Speech-Language Pathology • Vol. 7 • No. 3 August 1998

Downloaded From: http://ajslp.pubs.asha.org/ by a Universite Laval User on 09/10/2016
Terms of Use: http://pubs.asha.org/ss/rights_and_permissions.aspx
 proceeded to suggest, however, not only that reasonably be expected to insist on treatments
this information is not sufficient without the that have been demonstrated to be effective in
support of group studies, but also that satisfac- well-designed experimental analyses. One of
tory treatment designs would require control- the greatest strengths of Curlee and Yairi’s paper,
group subjects to receive no treatment for 2 in this light, may be that they have delivered a
years. It is critically important to recognize, in much-needed “reality check” about the status of
addition to recognizing that a no-treatment stuttering treatment research, and we thank them
control group might not be necessary at all, that for their thought-provoking contributions.
the standard pretest-posttest control-group
design does not require the control group to Author Note
remain in no-treatment conditions for the entire
duration of any possible untreated recovery Authorship is equal. Preparation of this paper was
period. In fact, the design is meant precisely to supported in part by research grant number 5 R01
DC 00060 from the National Institute of Deafness
allow for untreated recovery, and improvement
and Other Communication Disorders, National
in the no-treatment group does not, by any Institutes of Health, awarded to the first author.
means, prevent a meaningful comparison from
being made: The question is whether, during
the time period covered by the study, the treat- References
ment group shows more improvement than the Ambrose, N. G., Cox, N. J., & Yairi, E. (1997).
no-treatment group (Schiavetti & Metz, 1997). The genetic basis of persistence and recovery in
There is absolutely no reason to accept that stuttering. Journal of Speech, Language, and
experimentally sound stuttering treatment re- Hearing Research, 40, 567–580.
search would require large groups of children to Andrews, G. (1984). The epidemiology of stutter-
remain as untreated controls for 24 full months ing. In R. Curlee & W. H. Perkins (Eds.), Nature
and treatment of stuttering: New directions (pp.
postonset, as Curlee and Yairi (1997, p. 16)
1–12). San Diego, CA: College-Hill.
would have us believe. Andrews, G., & Harris, M. (1964). The syndrome
of stuttering. London: Heinemann.
Conclusions Andrews, G., & Ingham, R. J. (1972). An approach
to the evaluation of stuttering therapy. Journal of
Curlee and Yairi’s (1997) article was pub- Speech and Hearing Research, 15, 296–302.
lished in the “Second Opinion” section of this Bernstein Ratner, N. (1997). Leaving Las Vegas:
journal. Zebrowski (1997) and Bernstein Ratner Clinical odds and individual outcomes. American
(1997) presented second and third opinions, Journal of Speech-Language Pathology, 6(2),
both of which concurred with some of Curlee 29–33.
Bloodstein, O. (1981). A handbook on stuttering
and Yairi’s major premises. The purpose of our
(3rd ed.). Chicago: The National Easter Seal
response has been to present another opinion Society.
that has diverged sharply in several places from Bloodstein, O. (1987). A handbook on stuttering
those presented in the original three papers. We (4th ed.). Chicago: The National Easter Seal
also suggested that many of the issues raised by Society.
Curlee and Yairi are more complex and more Bloodstein, O. (1995). A handbook on stuttering
important than the original papers appeared to (5th ed.). San Diego, CA: Singular.
recognize. Most importantly, where Curlee and Bross, I. (1952). Sequential medical plans. Biomet-
Yairi argued that our discipline’s data base and rics, 8, 188–205.
our discipline’s professional and research ethics Conture, E. (1990). Stuttering (2nd ed.). Englewood
Cliffs, NJ: Prentice-Hall.
support the idea of waiting to provide treatment
Cordes, A. K., & Ingham, R. J. (1995). Stuttering
to some children who stutter, we have argued includes both within-word and between-word
essentially the opposite. disfluencies. Journal of Speech and Hearing
If our discipline chooses to advocate with- Research, 38, 382–386.
holding treatment from children who may not Costello, J. M. (1983). Current behavioral treat-
recover without it or chooses to advocate the ments for children. In D. Prins & R. J. Ingham
use of indirect treatments or treatments that (Eds.), Treatment of stuttering in early child-
cannot be shown to be clinically effective, then hood: Methods and issues (pp. 69–112). San
we should not be surprised to find ourselves in Diego, CA: College-Hill.
the position of needing to “affirm our role as Craig, A., Hancock, K., Chang, E., McCready, C.,
Shepley, A., McCaul, A., Costello, D., Harding,
the ‘experts’ in the identification and manage-
S., Kehren, R., Masel, C., & Reilly, K. (1996).
ment of stuttering” (Zebrowski, 1997, p. 21). A controlled clinical trial for stuttering in persons
We suggest only that the “experts” on the man- aged 9 to 14 years. Journal of Speech and Hear-
agement of any disorder might reasonably be ing Research, 39, 808–826.
expected to base their treatment decisions on Culton, G. L. (1986). Speech disorders among
critical analyses of all available data and might college freshmen: A 13-year survey. Journal of

Ingham • Cordes 17
Downloaded From: http://ajslp.pubs.asha.org/ by a Universite Laval User on 09/10/2016
Terms of Use: http://pubs.asha.org/ss/rights_and_permissions.aspx
 Speech and Hearing Disorders, 51, 3–7. Kully, D., & Boberg, E. (1991). Therapy for school-
Curlee, R. F., & Yairi, E. (1997). Early intervention age children. Seminars in Speech and Language,
with early childhood stuttering: A critical exami- 12, 291–300.
nation of the data. American Journal of Speech- Lewin, D. I. (1995, December). Are placebo-based
Language Pathology, 6(2), 8–18. trials unethical? The Journal of NIH Research, 7,
Finn, P. (1996). Establishing the validity of recover- 30–33.
ing from stuttering without formal treatment. Lincoln, M., & Onslow, M. (1997). Long-term
Journal of Speech and Hearing Research, 39, outcome of early intervention for stuttering.
1171–1181. American Journal of Speech-Language Pathol-
Fosnot, S. M. (1993). Research design for examin- ogy, 6(1), 51–58.
ing treatment efficacy in fluency disorders. Jour- Lincoln, M., Onslow, M., Lewis, C., & Wilson, L.
nal of Fluency Disorders, 18, 221–252. (1996). A clinical trial of an operant treatment for
Gillespie, S. K., & Cooper, E. B. (1973). Preva- school-age children who stutter. American Jour-
lence of speech problems in junior and senior nal of Speech-Language Pathology, 5(2), 73–85.
high schools. Journal of Speech and Hearing Martin, R. R., Kuhl, P., & Haroldson, S. K.
Research, 16, 739–743. (1972). An experimental treatment with two
Gregory, H. H., & Hill, D. (1993). Differential preschool stuttering children. Journal of Speech
evaluation—Differential therapy for stuttering and Hearing Research, 15, 743–752.
children. In R. F. Curlee (Ed.), Stuttering and Martin, R. R., & Lindamood, L. R. (1986). Stutter-
related disorders of fluency (pp. 23–44). New ing and spontaneous recovery: Implications for the
York: Thieme. speech-language pathologist. Language, Speech,
Ingham, J. C. (1989). Generalization in the treat- and Hearing Services in Schools, 17, 207–218.
ment of stuttering. In L. V. McReynolds & J. E. Meyers, S. C. (1986). Qualitative and quantitative
Spradlin (Eds.), Generalization strategies in the differences and patterns of variability in dis-
treatment of communication disorders (pp. 63– fluencies emitted by preschool stutterers during
81). Toronto: B. C. Decker. dyadic conversations. Journal of Fluency Disor-
Ingham, R. J. (1976). “Onset, prevalence and recov- ders, 11, 293–306.
ery from stuttering”: A reassessment of findings Nippold, M. A., & Rudzinski, M. (1995). Parents’
from the Andrews and Harris Study. Journal of speech and children’s stuttering: A critique of the
Speech and Hearing Disorders, 41, 280–281. literature. Journal of Speech and Hearing Re-
Ingham, R. J. (1983). Spontaneous remission of search, 38, 978–989.
stuttering: When will the emperor realize he has Onslow, M., Andrews, C., & Lincoln, M. (1994).
no clothes on? In D. Prins & R. J. Ingham (Eds.), A control/experimental trial of operant treatment
Treatment of stuttering in early childhood: Meth- for early stuttering. Journal of Speech and Hear-
ods and issues (pp. 113–140). San Diego, CA: ing Research, 37, 1244–1259.
College-Hill. Onslow, M., Costa, L., & Rue, S. (1990). Direct
Ingham, R. J. (1996, May). On watching a disci- early intervention with stuttering: Some prelimi-
pline shoot itself in the foot: Some observations nary data. Journal of Speech and Hearing Disor-
on current trends in stuttering treatment re- ders, 55, 405–416.
search. Paper presented at the meeting of the Onslow, M., & Packman, A. (in press). Recovery
American Speech-Language-Hearing Associa- from early stuttering with and without treatment:
tion Special Interest Division 4, Monterey, CA. The need for consistent methods in collecting and
Ingham, R. J., Andrews, G., & Winkler, R. interpreting data. Journal of Speech, Language,
(1972). Stuttering: A comparative evaluation of and Hearing Research.
the short-term effectiveness of four treatment Paden, E., & Yairi, E. (1996). Phonological charac-
techniques. Journal of Communication Disor- teristics of children whose stuttering persisted or
ders, 5, 91–117. recovered. Journal of Speech and Hearing Re-
Ingham, R. J., & Cordes, A. K. (in press). On search, 39, 981–990.
watching a discipline shoot itself in the foot: Perkins, W. H., Rudas, J., Johnson, L., Michael,
Some observations on current trends in stuttering W. B., & Curlee, R. F. (1974). Replacement of
treatment research. In C. Healey & N. Bernstein stuttering with normal speech: III. Clinical effec-
Ratner (Eds.), Stuttering treatment efficacy. New tiveness. Journal of Speech and Hearing Disor-
York: Lawrence Erlbaum. ders, 39, 416–428.
Ingham, R. J., & Packman, A. (1977). Treatment Ramig, P. (1993). High reported spontaneous recov-
and generalization effects in an experimental ery rates: Fact or fiction? Language, Speech, and
treatment for a stutterer using contingency man- Hearing Services in Schools, 24, 156–160.
agement and speech rate control. Journal of Reed, C. G., & Godden, A. L. (1977). An experi-
Speech and Hearing Disorders, 42, 394–407. mental treatment using verbal punishment with
Kelly, E. M. (1993). Speech-rate and turn-taking two preschool stutterers. Journal of Fluency
behaviors of children who stutter and their par- Disorders, 2, 225–233.
ents. Seminars in Speech and Language, 14(3), Ryan, B. P. (1971). Operant procedures applied to
203–214. stuttering therapy for children. Journal of Speech
Kelly, E. M. (1995). Parents as partners: Including and Hearing Disorders, 36, 264–280.
mothers and fathers in the treatment of children Ryan, B. P. (1974). Programmed therapy for stutter-
who stutter. Journal of Communication Disor- ing in children and adults. Springfield, IL: Charles
ders, 28, 93–106. C. Thomas.

18 American Journal of Speech-Language Pathology • Vol. 7 • No. 3 August 1998

Downloaded From: http://ajslp.pubs.asha.org/ by a Universite Laval User on 09/10/2016
Terms of Use: http://pubs.asha.org/ss/rights_and_permissions.aspx
 Ryan, B. P., & Van Kirk Ryan, B. (1983). Pro- Declaration of Helsinki IV, Hong Kong.
grammed stuttering therapy for children: Com- Yairi, E., & Ambrose, N. (1992). A longitudinal
parison of four establishment programs. Journal study of stuttering in children: A preliminary
of Fluency Disorders, 8, 291–321. report. Journal of Speech and Hearing Research,
Schiavetti, N., & Metz, D. E. (1997). Evaluating 35, 755–760.
research in communicative disorders (3rd ed.). Yairi, E., & Ambrose, N. (1996). Erratum. Journal
Boston: Allyn & Bacon. of Speech and Hearing Research, 39, 826.
Sidman, M. (1960). Tactics of scientific research. Zebrowski, P. M. (1997). Assisting young children
New York: Basic Books. who stutter and their families: Defining the role
Starkweather, C. W., Gottwald, S. R., & Halfond, of the speech-language pathologist. American
M. (1990). Stuttering prevention. Englewood Journal of Speech-Language Pathology, 6(2),
Cliffs, NJ: Prentice-Hall. 19–28.
Stephenson-Opsal, D., & Bernstein Ratner, N. Zebrowski, P. M., Weiss, A., Savelkoul, E., &
(1988). Maternal speech rate modification and Hammer, C. (1996). The effect of maternal rate
childhood stuttering. Journal of Fluency Disor- reduction on the stuttering speech rates and lin-
ders, 13, 49–56. guistic productions of children who stutter: Evi-
Wald, A. (1947). Sequential analysis. New York: dence from individual dyads. Clinical Linguistics
John Wiley. and Phonetics, 10, 189–206.
Watkins, R. V., & Yairi, E. (1997). Language
production abilities of children whose stuttering Received October 20, 1997
persisted or recovered. Journal of Speech, Lan- Accepted March 19, 1998
guage, and Hearing Research, 40, 385–399.
Wingate, M. E. (1964). A standard definition of Contact author: Roger J. Ingham, Department of
stuttering. Journal of Speech and Hearing Disor- Speech and Hearing Sciences, University of Cali-
ders, 29, 484–489. fornia, Santa Barbara, Santa Barbara, CA 93106
Winslow, M., & Guitar, B. (1994). The effects of Email: [email protected]
structured turn-taking on disfluencies: A case
study. Language, Speech, and Hearing Services
in Schools, 25, 251–257. Key Words: stuttering, stuttering treatment,
World Medical Assembly. (September, 1989). early intervention

Ingham • Cordes 19
Downloaded From: http://ajslp.pubs.asha.org/ by a Universite Laval User on 09/10/2016
Terms of Use: http://pubs.asha.org/ss/rights_and_permissions.aspx