Content uploaded by Roger Ingham
Author content
All content in this area was uploaded by Roger Ingham on Aug 18, 2015
Content may be subject to copyright.
AJSLP
Research Article
Efficacy of the Modifying Phonation Intervals
(MPI) Stuttering Treatment Program
With Adults Who Stutter
Roger J. Ingham,
a
Janis C. Ingham,
a
Anne K. Bothe,
b
Yuedong Wang,
a
and Martin Kilgo
a
Purpose: This study compared a new adult stuttering
treatment program (Modifying Phonation Intervals, or MPI)
with the standard of care for reducing stuttered speech in
adults (prolonged speech).
Method: Twenty-seven adults who stutter were assigned to
either MPI or prolonged speech treatment, both of which
used similar infrastructures. Speech and related variables
were assessed in 3 within-clinic and 3 beyond-clinic speaking
situations for participants who successfully completed all
treatment phases.
Results: At transfer, maintenance, and follow-up, the
speech of 14 participants who successfully completed
treatment was similar to that of normally fluent adults.
Successful participants also showed increased self-
identification as a “normal speaker,”decreased self-
identification as a “stutterer,”reduced short intervals of
phonation, and some increased use of longer duration
phonation intervals. Eleven successful participants received
the MPI treatment, and 3 received the prolonged speech
treatment.
Conclusions: Outcomes for successful participants were
very similar for the 2 treatments. The much larger proportion
of successful participants in the MPI group, however,
combined with the predictive value of specific changes in
PI durations suggest that MPI treatment was relatively more
effective at assisting clients to identify and change the
specific speech behaviors that are associated with
successful treatment of stuttered speech in adults.
Literature reviews and meta-analyses consistently
identify prolonged speech (Goldiamond, 1965) as the
most effective option for improving the speech of
adults who stutter (Andrews, Guitar, & Howie, 1980; Bothe,
Davidow, Bramlett, & Ingham, 2006; Ingham, 1984). Boberg
and Kully (1994), for example, reporting on the ISTAR
Comprehensive Stuttering Program (Boberg & Kullyi, 1985),
found that mean percent syllables stuttered (%SS) during a
beyond-clinic telephone conversation changed from 19.59 %
SS pretreatment to 1.29 %SS posttreatment for 17 adults, and
from 14.32 %SS to 1.75 %SS for 25 adolescents, with 14/17
adults and 11/25 adolescents displaying less than 1.0 %SS at
posttreatment. Subsequent evaluations of Boberg and Kully’s
program have continued to show positive results, including
studies using measures of self-identification as “a normal
speaker”versus as “a stutterer”(Boberg & Kully, 1994,
p. 1056) and speech naturalness (Teshima, Langevin, Hagler,
& Kully, 2010), use of the treatment with Dutch speakers
(Langevin et al., 2006), and evaluation of treatment outcomes
in beyond-clinic settings at up to 5 years posttreatment
(Langevin, Kully, Teshima, Hagler, & Prasad, 2010).
Long-term posttreatment results for prolonged speech
studies tend to be complicated, however, by the related
problems of relapse and attrition, or both. In Boberg and
Kully’s (1994) study, mean %SS had risen to 6.03 %SS for
adults and 3.89 %SS for adolescents 12 months posttreat-
ment. The intensive computerized Kassel Program (Euler &
Wolff von Gudenberg, 2000; Euler, Wolff von Gudenberg,
Jung, & Neumann, 2009), similarly, has provided 1–3-year
follow-up data showing approximately 70%–75% sus-
tained and generalized reductions in %SS scores for at least
69/400 enrolled participants. The studies that report better
maintenance of positive treatment effects often provide data
from only a small proportion of participants, suggesting
that if the entire group had remained in the treatment pro-
gram, the mean performance would be poorer. Onslow,
Costa, Andrews, Harrison, and Packman (1996), for exam-
ple, reported zero or near-zero stuttering and naturalness
ratings below 3 (i.e., within the normal range) on the 9-point
a
University of California at Santa Barbara
b
University of Georgia, Athens
Correspondence to Roger J. Ingham: rjingham@speech.ucsb.edu
Editor: Krista Wilkinson
Associate Editor: Nancy Hall
Received June 5, 2014
Revision received October 25, 2014
Accepted January 12, 2015
DOI: 10.1044/2015_AJSLP-14-0076
Disclosure: The authors have declared that no competing interests existed at the time
of publication.
American Journal of Speech-Language Pathology •Vol. 24 •256–271 •May 2015 •Copyright © 2015 American Speech-Language-Hearing Association256
Downloaded From: http://ajslp.pubs.asha.org/ by University of California, Santa Barbara, Roger Ingham on 05/15/2015
Terms of Use: http://pubs.asha.org/ss/Rights_and_Permissions.aspx
speech naturalness scale (Martin, Haroldson, & Triden,
1984) for beyond-clinic recordings of speech at 3 years post-
treatment for an intensive prolonged speech program devised
in the 1980s (Ingham, 1987), but only 12 of 32 original
participants were reflected in these data. Similar problems
confounded the findings of a more recent study by O’Brian,
Onslow, Cream, and Packman (2003).
Other difficulties with some reports of what appear to
be positive prolonged speech treatment outcomes include
the poor methodological quality of many treatment studies
(Bothe et al., 2006; Nye et al., 2013), the only occasional
use of ecologically valid assessments such as speech during
telephone conversations (Bothe et al., 2006) or speaker-
selected difficult speaking tasks (Ingham, Ingham, & Bothe,
2012), and some evidence that the efficacy of prolonged
speech treatment may be influenced by the pretreatment se-
verity of stuttering. Hancock and Craig (1998), for example,
investigated behavioral and personality variables that might
interact with follow-up findings in young adolescents treated
by prolonged speech. They found that only “pre-treatment
stuttering frequency ... and trait anxiety post-treatment sig-
nificantly predicted stuttering frequency 1-year post-treatment”
(p. 31). Boberg and Kully (1994), similarly, reported a .62
correlation between pretreatment %SS scores in telephone
conversations and scores 12 months after treatment. It is
important to note, however, that Boberg and Kully’s data
also show correlations of .63 between %SS immediately
posttreatment and %SS at 12 months posttreatment, and of
.84 between 4 months posttreatment and 12 months post-
treatment. Hence, the trend of speech performance immedi-
ately after treatment may actually account for much more
of the long-term or follow-up variance (71%) than pretreat-
ment performance (38%), a possibility that has not been
systematically investigated in previous studies.
Overall, prolonged speech remains the standard of
care for adults who stutter. It is the best supported of cur-
rent options, and there is no evidence to support such posi-
tive outcomes for other approaches (Bothe et al., 2006).
Nevertheless, the need remains for treatments that can
more predictably result in stutter-free speech that is natural
sounding, acceptable to the speaker or consistent with de-
sirable affective or cognitive changes, and durable across
time and across clinically relevant speaking situations, with-
out the problems of relapse or attrition.
There are also other, more fundamental problems
with prolonged speech—a speech pattern so labelled by
Goldiamond (1965) as resulting from speaking during de-
layed auditory feedback (DAF). These problems were earlier
summarized by Ingham et al. (2001, p. 1230) as follows:
Goldiamond (1967) actually first highlighted one of
the continuing problems with prolonged speech: It
is not a completely predictable response to DAF and
largely depends on instructional training. Indeed, DAF
ultimately became an almost irrelevant adjuvant of
prolonged speech (see Ingham, 1984, Chapter 10).
Furthermore, the features of this pattern that needed
to be trained were only vaguely defined and, probably
for that reason, have never been clearly demonstrated
to be functional within a treatment context (see
Ingham, 1993, for a review). These features are usually
described as “continuous vocalization,”often accom-
panied by “soft contacts,”“gentle onsets,”and “easy
onsets”(see Webster, 1974), but few attempts have
been made to specify their parameters. One legacy of
this situation is Onslow and O’Brian’s (1998) demon-
stration of poor inter- and intraclinician agreement
among clinicians trying to identify these features dur-
ing treatment. Borden, Baer, and Kenney (1985) and
Peters, Boves, and van Dielen (1986) used acoustic
rise time, or the intensity-level gradient, of an utterance
onset to specify the occurrence of “easy onsets”in the
speech of persons who stutter. This is also consistent with
the operation of the Voice Monitor device (Webster,
1977) that is used within the Precision Fluency Shaping
treatment program (Webster, 1980). However, it has
yet to be demonstrated that manipulations in acoustic
rise time have functional control over stuttering.
The endemic problems with prolonged speech described in
this article remain as true today as they were in 2001.
Modifying Phonation Intervals
As an attempt to solve some of these problems, the
first author and colleagues have investigated the effects of
what might be described as an opposite approach. Instead
of focusing on prolonging or increasing phonation dura-
tions, persons who stutter (PWS) can reduce the frequency
with which they produce short intervals of phonation while
speaking (Gow & Ingham, 1992; Ingham, Montgomery, &
Ulliana, 1983). The premise underlying these studies was
that part of the effectiveness of prolonged speech might
depend on reducing the frequency of short intervals of pho-
nation, rather than increasing the frequency of longer inter-
vals. In these studies, phonation was operationally defined
as an accelerometer- or electroglottograph-detected signal
recorded from the throat surface at the thyroid prominence
and interpreted as reflecting vocal fold vibration. The dura-
tion of the signal (excluding signal-off periods of 10 ms or
less) constituted a phonation interval (PI). In other words,
an accelerometer-detected vibration for 200 ms, preceded
and followed by more than 10 ms during which no vibration
is detected, would be described as one 200-ms PI. Initial
investigations demonstrated that controlled decreases in
short PIs produced a reduction in stuttering frequency and
that the resulting speech was natural sounding (Ingham
et al., 1983), potentially resolving this fundamental con-
cern about some prolonged speech treatments. Specifically,
Ingham et al. (1983) demonstrated that stuttering could be
increased and decreased in two adult male PWS by teaching
them to increase or decrease, by at least 50% relative to base
rate, the frequency of PIs within the range of 10–150 ms
or 10–200 ms. In subsequent studies, the lower limit of the
duration range used in treatment and research (i.e., target
range PIs, TRPIs) was increased to at least 30 ms because
some head movement and swallowing actions were found
Ingham et al.: Efficacy of Modifying Phonation 257
Downloaded From: http://ajslp.pubs.asha.org/ by University of California, Santa Barbara, Roger Ingham on 05/15/2015
Terms of Use: http://pubs.asha.org/ss/Rights_and_Permissions.aspx
to produce false positive PIs in the 10- to 30-ms range. Fig-
ure 1 shows graphically the measurement of PIs and the de-
termination of one speaker’s TRPIs.
The requirement to reduce by 50% those PIs that
fall into the speaker’s own baseline range of shortest PIs, or
to reduce by 50% those PIs that are between 30 and 150 ms
in duration, has proven to be both functional and predic-
tive. It is important to recognize, however, that there is
no evidence that the distribution of PIs in adult PWS dif-
fers significantly from the distribution that typifies nor-
mally fluent speakers, including PIs within the 30–150-ms
range (Godinho, Ingham, Davidow, & Cotton, 2006).
That is, prior to treatment, PWS do not necessarily produce
more short PIs in their connected speech than normally
fluent speakers. Nevertheless, investigations by Davidow,
Bothe, Andreatta, and Ye (2009), for instance, have shown
that some well-known fluency inducing conditions (chorus
reading, prolonged speech, singing, and rhythmic stimu-
lation) are associated with approximately 50% reductions
in PIs within the range of 51–150 ms. Most important for
the development of improved treatment options, the results
of these PI investigations were used to develop a treatment
known as Modifying Phonation Intervals (MPI). In an eval-
uation of the MPI treatment with five adults who stuttered,
Ingham et al. (2001) demonstrated that all five achieved
stutter-free and natural-sounding speech within and beyond
the clinic environment, and all five maintained those gains
for at least 12 months (these five participants are included
in the present study; see Methods). These successful outcomes
suggested that the MPI treatment could represent an im-
proved variation on prolonged speech treatments.
The present study was designed, therefore, to com-
pare MPI treatment to prolonged speech treatment for
larger groups of participants. This study used a treatment
comparison design that equated all other aspects of treat-
ment infrastructure and delivery for both treatments and
that also allowed the simultaneous investigation of some of
the stuttering treatment outcome issues discussed in previ-
ous paragraphs. Specific research questions are as follows:
1. Are there significant differences in speech performance
and self-identification (as stutterer or normal speaker)
outcomes for participants who successfully complete the
MPI treatment as compared with those who successfully
complete the current best treatment, prolonged speech?
2. Are there significant differences in PI distributions
during treatment that can differentiate between par-
ticipants who will succeed in completing the long-
term maintenance phase of either program and those
who do not succeed?
3. Do participants who succeed in completing either
program achieve levels of speech performance, self-
identification, and PI distributions that are not signif-
icantly different from those found in normally fluent
speakers?
Method
Participants
Twenty-seven PWS (22 men, five women; age range
18–64 years; M= 35.9 years; median = 35 years) and eight
Figure 1. This figure illustrates (a) how the MPI program collects phonated intervals; (b) how the entire corpus of PIs is then ranked
from the longest to the shortest duration PI; (c) how the target range is identified by locating the bottom 20th percentile or quintile
range of the ranked distribution; and (d) how PIs that occur within the TRPI range (now shown in red) are identified in real time
(i.e., during speech) with an audible and visual stimulus being immediately transmitted to the speaker. MPI = Modified Phonation
Intervals treatment; PI = phonation interval.
258 American Journal of Speech-Language Pathology •Vol. 24 •256–271 •May 2015
Downloaded From: http://ajslp.pubs.asha.org/ by University of California, Santa Barbara, Roger Ingham on 05/15/2015
Terms of Use: http://pubs.asha.org/ss/Rights_and_Permissions.aspx
adults who do not stutter (controls, CONT; six men, two
women; age range 20–64 years; M= 37.8 years; median =
32 years) participated in this study, which was conducted in
the Department of Speech and Hearing Sciences, University
of California, Santa Barbara, California, and the Research
Imaging Institute, University of Texas Health Science Center,
San Antonio, Texas. All were volunteers identified from treat-
ment waiting lists at either site or via advertisements in San
Antonio, Austin, and Houston. The first five PWS partici-
pants were assigned to MPI treatment, and most of their data
were reported in Ingham et al. (2001); the next 22 were ran-
domly assigned to either MPI (n= 12) or prolonged speech
(PS) treatment (n= 10).
1
CONT participants were matched,
by gender and age (within ± 5 years), with eight PWS partici-
pants; they completed two data-collection sessions but no
treatment. The CONT group were selected for three pur-
poses: (a) to be used within the brain-imaging section of this
study, the results of which were reported earlier (see Ingham,
Wang, Ingham, Bothe, & Grafton, 2013); (b) to provide
control group PI data for comparison with the PWS PI data
and help establish the consistency of normally fluent speakers’
PI data over time; and (c) to provide speech and self-report
data necessary to be able to interpret the PWS data as similar
to or different from those of speakers who do not stutter.
2
All PWS participants self-reported stuttering since
early childhood and displayed chronic stuttering as con-
firmed by the principal investigator and a certified speech-
language pathologist using standard clinical assessments
(Ingham & Costello, 1985). All participants were right-
handed (>+80 on the Edinburgh Handedness Inventory;
Oldfield, 1971]; displayed no signs of any neurologic disor-
der other than stuttering; reported no other current speech,
language, cognitive, or behavioral disorder; and passed a
hearing screening. All PWS participants had experienced
various therapies, but no participant reported receiving
treatment for stuttering during the preceding 3 years. Ten
of the participants’previous therapies had included a Flu-
ency Shaping procedure that utilized a version of prolonged
speech; they were randomly assigned to either the MPI or
PS group. All PWS participants produced at least 3 %SS
during at least one of three 3-min within-clinic pretreatment
speaking tasks (oral reading, monologue, and a telephone
conversation). All CONT participants met the same selection
criteria, except that they were required to produce 0 %SS
during each of the three speaking tasks and not report either
a history of or the presence of stuttering.
Treatment Format and Programs
Commonalities across MPI and PS treatment programs.
Both the MPI and the PS treatments used in this study have
been reported elsewhere (for MPI: Ingham, 1999; Ingham
et al., 2001; Ingham et al., 2013; for PS: Ingham, 1987;
Onslow et al., 1996). Each is organized as five consecutive
phases: pretreatment, establishment, transfer, maintenance,
and follow-up. A trained clinician directs the pretreatment
phase (and all subsequent phases), which includes at least
three data-collection sessions (see below), scheduled at 2-week
intervals. Unless the client’s stuttering frequency is decreas-
ing (using the average %SS across three within- and three
beyond-clinic speaking tasks on three occasions), treatment
is introduced after the third pretreatment data-collection
session. If stuttering frequency is decreasing, additional
data-collection sessions are scheduled at 2-week intervals
until speech performance stabilizes.
At the conclusion of the pretreatment phase, par-
ticipants begin the establishment phase for their assigned
treatment program. Regardless of the particular treatment
administered, all establishment-phase activities use the
same basic structure, encompassing the infrastructural vari-
ables known to be associated with the most successful treat-
ment of stuttered speech (Bothe et al., 2006). These include
intensive scheduling (3 hours/day, 6 days/week), a sequence
of different speech performance tasks of increasing dura-
tion, self-management of treatment steps by the client,
performance-contingent progress through treatment steps,
and stringent speech performance criteria including 0 %SS
as the pass criterion at all steps, and the implementation
of a requirement for natural-sounding speech indicated
by speech naturalness ratings of 1–3 on the 9-point scale
introduced by Martin et al. (1984). (Speech naturalness
requirements are implemented following the oral reading
segment of the program.)
Establishment phases for both programs are orga-
nized in five progressively demanding steps (oral reading,
monologue, conversation, and two steps of telephone
conversation) that each call for three consecutive trials of
stutter-free and natural-sounding speech at speaking dura-
tions of 1, 2, and 3 min. Self-managed aspects of the treat-
ment require clients to score each trial as including or not
including stuttering and to rate their own speech naturalness.
When three consecutive trials meet all specified behavioral
criteria, the client progresses to the next step. In both pro-
grams, an occurrence of stuttering and a speech natural-
ness rating beyond 3 indicate a failed trial, and the client
returns to the first step of that particular treatment task or,
after multiple failures, returns to a previous speaking task.
Notwithstanding the self-management procedure, the last
task at each step (e.g., a final 3-min conversation) must
also be judged by the clinician as meeting all applicable
speech performance criteria; that is, both programs use self-
management of treatment in combination with clinician
validation. The duration of the establishment phase is per-
formance dependent and ranged from 3 to 12 weeks for
these participants.
1
1Data through the maintenance phase for the first five participants
were presented by Ingham et al. (2001).
2
2We actually recruited sufficient controls so as to have matched numbers,
but many dropped out because they were not willing to complete a
second positron emission tomography (PET) scanning session. Each
PET session took about 3 hours and required the participant to be
injected with a radioactive tracer—something many preferred to avoid
doing a second time. Nonetheless, the eight controls did provide
comparisons for PI counts and speech naturalness ratings for the present
study. Their repeated PET scanning data were crucial for the data
evaluation reported earlier in the Ingham et al. (2013) article.
Ingham et al.: Efficacy of Modifying Phonation 259
Downloaded From: http://ajslp.pubs.asha.org/ by University of California, Santa Barbara, Roger Ingham on 05/15/2015
Terms of Use: http://pubs.asha.org/ss/Rights_and_Permissions.aspx
During establishment, the treatment targets of the
MPI and PS programs are somewhat different (see below);
however, the components of treatment that follow are ex-
actly alike for both programs. Participants who successfully
complete their assigned establishment phase (see Results)
continue to the transfer phase, which requires participants
to complete a sequence of three stutter-free and natural-
sounding recordings in each of six beyond-clinic settings.
These speaking tasks are selected to be relevant to the par-
ticipant’s daily life and are arranged hierarchically. The
duration of the transfer phase is again performance depen-
dent and ranged from 10 to 37 weeks for these participants.
As in the establishment phase, the first two of the three re-
cordings in a beyond-clinic setting are scored by the client,
and the third is scored by the clinician. Participants who
complete the transfer phase enter the maintenance phase
(Ingham, 1980, 1982, 1999), within which they continue the
activities of the last three tasks of transfer. The maintenance
phase is accomplished via implementation of performance-
contingent increases in the time between assessments, begin-
ning 1 week apart and ending with a 16-week period be-
tween assessments. Once again, the client scores two of the
three beyond-clinic recordings, and the clinician scores the
third. Success with all three recordings enables the client
to double the number of weeks before the next maintenance
assessment, but failure on any task means that the client
returns to the first maintenance phase step. The minimum
time to completion is 38 weeks. Actual durations of the
maintenance phase for this study were 38 to 78 weeks.
Follow-up data are collected from successful participants
1 year after the end of their maintenance phase. Partici-
pants who were not successful with the MPI treatment were
offered PS treatment and vice versa. Those participants
were considered unsuccessful with their first treatment and,
therefore, removed from further data analyses and were not
followed for the purposes of this study because of the con-
founding multiple treatments effect (see Results).
The MPI program. Details of the MPI program in
terms of its phases, individual treatment steps, and features
incorporated in the system software were described by Ingham
(1999), Ingham et al. (2001), and Ingham et al. (2013) and
are available to interested readers.
3
A brief overview follows.
Treatment targets for the MPI program are (a) client-produced
connected speech that contains at least a 50% reduction of
short PIs (on the basis of the client’s shortest quintile range;
see Figure 1); (b) 0% SS; and (c) natural-sounding speech.
During the establishment phase, the MPI software provides
auditory and visual biofeedback of all PI durations. While
the client is speaking, the occurrence of a short PI is sig-
naled by an auditory “beep.”Concurrently, the computer
screen displays two boxes, and a mark appears in the left
box for each short PI that occurs; in the right box, marks
appear for all other PI durations. The number of short PIs
allowed is indicated on the screen as well. If that number is
exceeded, the computer stops the progress of the program,
and the step is repeated. At essentially alternating steps in
the program, PI feedback is not provided. Occurrences of
stuttering are judged by the client (a self-managed aspect of
the program), who stops the progress of the program im-
mediately when a stutter occurs. Speech naturalness is also
scored/self-managed by the client at the end of each speak-
ing trial. If self-judged speechnaturalnessisnotwithin
the 1–3 range (Martin et al., 1984), the step is failed and
repeated. As stated above, at crucial steps of the program,
occurrences of stuttering and speech naturalness are also
judged by the clinician. When the requisite number of speak-
ing trials meeting all three criteria has occurred (see above),
the program automatically initiates the next step (i.e.,
performance-contingent progress through treatment steps).
The PS program. The PS program used in this study
is identical to a previously published and evaluated pro-
gram that can be found in Ingham (1987) and Onslow et al.
(1996). As previously stated, treatment targets for the PS
program are client-produced connected speech that con-
tains (a) 0% SS and (b) natural-sounding speech. During
the oral reading part of the establishment phase, tape-
recorded models of prolonged speech (see Footnote 3) are
presented in a progressive sequence that begins with models
of speech at 70 syllables per minute (SPM) and progresses
by 30-SPM increments to 130 SPM. At each step of oral
reading (i.e., 1, 2, and 3 min), the client first reads in chorus
with the tape-recorded model and then reads at the same
speech/prolongation rate without the model. After comple-
tion of oral reading, only stuttering and speech naturalness
are monitored, and clients progress through the treatment
steps as described above.
Measures of Speech Performance,
Self-Identification, and PIs
Speech performance, self-identification, and PI data
were collected for PWS participants at least three times in
pretreatment and twice each in establishment, transfer, and
maintenance, including at a predetermined phase midpoint
and at the end of each of these phases; and at one follow-up
point 12 months after completion of the maintenance phase.
Members of the control group each completed two data-
collection sessions, once at a time referred to as pretreat-
ment and again at a time referred to as the end of transfer,
with the timing of those assessments determined by yoking
CONT participants individually to their paired PWS (see
Participants, above).
Speech performance measures. At each data-collection
session, Treatment Evaluation Speaking Tasks (TEST)
assessed speech performance and PI data through the use of
three within-clinic and three beyond-clinic speaking tasks,
each 3 min in duration. Within-clinic speaking tasks in-
cluded oral reading from a novel (READ), a monologue on
a self-chosen topic (MONO), and a telephone conversation
3
3The MPI program is now available as an app for iPad or iPod touch
use at http://mpi-stuttering-treatment.com. The PS program manual
(Ingham, Moglia, Kilgo, & Felino, 2007), including model recordings,
is available as electronic supplemental files and on request from the
first author. Readers interested in the MPI software, peripherals, and
program details should also contact the first author.
260 American Journal of Speech-Language Pathology •Vol. 24 •256–271 •May 2015
Downloaded From: http://ajslp.pubs.asha.org/ by University of California, Santa Barbara, Roger Ingham on 05/15/2015
Terms of Use: http://pubs.asha.org/ss/Rights_and_Permissions.aspx
with a research assistant (PHONE). All within-clinic tasks
were audiovisually recorded. Beyond-clinic speaking tasks
were speaking with a family member or acquaintance
(FAMILY), a telephone conversation with a business
(PHONE), and a self-selected difficult speaking situation
(SELF SEL). For the SELF SEL task, the participant was
instructed to “choose a very hard and, for you, a very im-
portant speaking task that is able to be recorded and would
indicate that the treatment was or was not working.”Typical
SELF SEL tasks were speaking before an audience and an-
swering their questions, ordering in a restaurant, and making
phone calls to strangers. Beyond-clinic tasks were audio re-
corded with a miniature cassette recorder that was voice acti-
vated, but participants were instructed to inform others that
they were being recorded. All TEST recordings were com-
pleted without performance-contingent feedback, without
performance requirements or contingencies, and outside
of the treatment location used for the establishment phase.
All TEST recordings were evaluated at the University
of Georgia by trained observers who had no direct contact
with the participants and were kept masked as to the par-
ticipant’s group, assigned treatment program (MPI or PS),
and treatment phase. TEST recordings for the first five MPI
participants were evaluated independently by two trained
research assistants at the University of California, Santa
Barbara, who were also masked to the participant’streat-
ment phase (see Ingham et al., 2001). All recordings were
assessed for percent syllables stuttered (%SS), stutter-free
syllables spoken per minute (SFSPM), and speech natural-
ness (NAT; Martin et al., 1984). All measures were made
independently by two judges for each recording and ac-
cording to the definitions and methods available in a stan-
dard and freely available audiovisual stuttering judgment
training program known as the SMS (Ingham, Bakker,
Ingham, Kilgo, & Moglia, 1999). All judges had been
trained on the SMS program.
Approximately eight pairs of judges were used to per-
form ratings at different times throughout the study. Reli-
ability (replicability) of experimental data was assured
by using as the data for all analyses the means of the two
independent judges’ratings. In addition, all recordings
on which the two judges’data differed by more than 10%
were identified and re-rated before the experimental
data were finalized. In the case of the NAT scores, re-
ratings were made if the judges differed by more than
one unit on the 9-point scale.
Self-identification measures.Perkins’s (1981; see also
Boberg & Kully, 1994) Speech Performance Questionnaire
(SPQ) was adapted for this study to create a 3-item mea-
sure that addressed the speaker’s self-perceptions and self-
identification with respect to speech and stuttering. The
three items (“I am now able to speak normally without think-
ing about controlling my speech,”“I now feel like a normal
speaker,”and “I now consider myself a stutterer”) were each
rated on a 4-point scale with each point defined (never,some-
times,almost always,always). The SPQ was presented to the
participant as a paper-and-pencil instrument at each TEST
evaluation and required less than a minute to complete.
PI measures. PI data were gathered during all within-
clinic TEST recordings for all participants using MPI
software and an accelerometer housed within a Velcro
neckband and wired to a customized preamplifier unit. The
software includes preliminary steps for establishing that
the accelerometer signal during phonation meets intensity
level criteria and that the noise level within the computer
system does not exceed prescribed limits. Routine procedures
were used following each resetting of the accelerometer to
ensure that a predictable speech sequence with a known
number of PIs (counting from 1 to 10) registered the correct
number of PIs. To further ensure the validity of PI data,
six PWS participants and six CONT participants each
provided 20 recorded tokens that were analyzed using the
PRAAT (version 4.2.07) acoustic analysis program (Boersma
& Weenink, 2003) and by the MPI system. In total, 240 PIs
were recorded, ranging from 36 to 778 ms according to the
PRAAT analysis. The PI duration of 197 intervals (82%)
differed between the PRAAT system and MPI software by
not more than 20 ms. The remaining 43 intervals showed
the following differences: 23 differed by 22–26 ms, and
20 differed by 31–40 ms. The overall mean duration differ-
ence was 17.6 ms. The MPI system ignores PIs of less than
10-ms duration, so the accuracy of the PI measures used
for this study was considered satisfactory.
Results
Speech Performance and Self-Identification
Outcomes of MPI and PS Treatment Programs
Summarized speech performance data for partici-
pants who had successfully completed each phase are pro-
vided in Table 1 and Figure 2, with additional detail in
Figure 3. Table 1 also shows the number of participants
who completed the performance-contingent parts of each
treatment phase and were therefore eligible to progress to
the next phase of the program, and their mean %SS scores
for within- and beyond-clinic tasks (TEST assessments) at
the conclusion of each phase.
4
There was no significant difference between the MPI
group and PS group within-clinic ( p> .05) or beyond-clinic
(p> .05) mean %SS scores at pretreatment. The MPI group
displayed significantly higher within-clinic ( p= .008) and
beyond-clinic ( p= .021) mean %SS scores than the PS group
4
4Recall that TEST data collection was not directly linked to treatment;
participants who had passed the final step of a treatment phase would
have done so by demonstrating no stuttering and natural-sounding
speech in treatment conditions, but TEST data were generalization
measures and, therefore, free to vary. One possible confounding effect
on the TEST data is that the within-clinic data were derived from
audio-visual recordings, whereas the beyond-clinic data were obtained
from audio-only recordings. Conceivably, the audio recordings could
have caused the SMS raters to “miss”occurrences of stuttering only
distinguished by visual features (e.g., Williams, Wark, & Minifie, 1963).
However, if that were the case, then it might have been expected that
there would have been differences between the within- and beyond-clinic
%SS trends in Figure 3. That was obviously not the case.
Ingham et al.: Efficacy of Modifying Phonation 261
Downloaded From: http://ajslp.pubs.asha.org/ by University of California, Santa Barbara, Roger Ingham on 05/15/2015
Terms of Use: http://pubs.asha.org/ss/Rights_and_Permissions.aspx
at the end of establishment, but no differences existed be-
tween groups following completion of transfer, mainte-
nance, or follow-up. In addition, it is clear that participants
in both groups who completed their respective programs
(MPI = 11; PS = 3) achieved a reduction of stuttering to
less than 1 %SS. This reduction did not involve reduced
speed of speech or unnatural sounding speech. It also oc-
curred in multiple beyond-clinic speaking situations and
was an improvement that generally was maintained across
time (Figures 2 and 3). By follow-up, however, there was
some deterioration in speech naturalness within the PS
group (see Figure 3). At follow-up, 12 months after the
active maintenance phase had ceased, two participants
(both MPI-treated) had zero stuttering on all six within and
beyond-clinic speaking tasks (Figure 3). On the self-selected
task, 6/11 MPI participants had zero stuttering with
all NAT scores below 3. (Most of these tasks involved
public speaking before an audience of strangers who also
asked questions, a task deemed highly difficult by the
participants.)
Participants who completed either treatment also re-
ported changes in self-identification, from modal responses
of never or sometimes feeling “like a normal speaker”at
pretreatment to sometimes or almost always feeling “like a
normal speaker”at the end of maintenance and at follow-up,
and from “consider[ing]”themselves to be “a stutterer”
always at pretreatment to only sometimes or almost always
at maintenance (see Table 2). These changes occurred over
the complete treatment time of a mean of 159 weeks (ex-
cluding data from four participants who were forced to
make a temporary break from the program for 4–9 weeks
in the transfer phase because of illness), with the mean and
the modal self-identification responses both continuing
to change after speech performance had stabilized at low
levels (compare Table 1 and Figures 2 and 3 with Table 2).
There was no significant difference between mean total
number of weeks that successful participants in the MPI
(M= 155 weeks) and PS (M= 168) treatment programs re-
quired to complete their programs.
PIs as Predictors of Treatment Success
Despite the success of the 14 participants described
in the previous section (see above and Table 1), almost half
(13/27) of the participants in this study failed to complete
allfivetreatmentphases.Nineofthesemaybecategorizedas
treatment failures at various stages: Four of 17 participants
originally assigned to the MPI program were unable to meet
that program’s requirements (two during transfer and two
during maintenance), and five of 10 assigned to the PS pro-
gram failed to meet its requirements (two could not com-
plete its establishment phase, and three others could not
complete transfer). The other four participants who did not
complete their treatment programs (two MPI and two PS
participants) dropped out for reasons unrelated to the treat-
ment programs (domestic or legal situations, or concerns
about the brain-imaging procedure associated with a related
study).
To examine whether changes in PIs predicted success
in treatment, PI data from the 14 participants who suc-
cessfully completed the maintenance phase were compared
with those from unsuccessful participants. The 11 success-
ful participants from the MPI program and three from the
PS program were combined for the purposes of these anal-
yses, because their speech performance was essentially
identical through transfer and maintenance and because
so few PS participants met the criteria to be included in
these analyses.
Figure 4 shows the number of PIs produced per
minute in each of the five duration ranges that defined the
quintiles (derived separately for each speaker), for each
speaking condition and from all three pretreatment TEST
occasions. That is, Figure 4 allows visual comparisons of PI
durations from later occasions with PI durations from pre-
treatment by using each speaker’s pretreatment quintiles
as the referent. Figure 5 shows the results of statistical tests
of differences between mean number of PIs per minute
within each quintile range and the variance in PIs per min-
ute within each quintile range. As shown in Figures 4 and 5,
during pretreatment speech, the mean number of PIs per
minute in the first quintile (Q1, which includes the shortest
PI durations and is also the TRPI range for MPI treatment)
was higher for participants who were ultimately successful
(shown in black in Figure 4) than for those who were ulti-
mately unsuccessful (shown in red). At the end of establish-
ment and continuing into transfer, however, compared to
their pretreatment frequencies, successful participants typi-
cally showed significantly fewer PIs in this range, signifi-
cantly less within-group variability in this range, or both, as
compared with unsuccessful participants. Significant changes
in the number of PIs, or in within-group variability, also
occurred for some longer duration PIs, especially in the range
of durations that had defined the speakers’fourth quin-
tiles (Q4) during pretreatment. Specifically, comparison of
Table 1. Number (and percentage) of participants who completed each phase of the Modifying Phonation Intervals (MPI) and the Prolonged
Speech (PS) treatments, with mean percent syllables stuttered (%SS) from within- and beyond-clinic Treatment Evaluation Speaking Tasks
(TEST; Ingham et al., 2001) assessments after each phase.
Variable n% MPI within %SS MPI beyond %SS n% PS within %SS PS beyond %SS N
Pretreatment 17 100 8.6 6.7 10 100 10.5 7.3 27
Establishment 17 100 2.1 2.7 8 80 0.8 1.2 25
Transfer 15 88.2 0.8 0.7 4 40 0.8 0.3 19
Maintenance 11 64.7 0.4 0.5 3 30 1.0 0.1 14
Follow-up 11 64.7 0.3 0.3 3 30 0.6 0.6 14
262 American Journal of Speech-Language Pathology •Vol. 24 •256–271 •May 2015
Downloaded From: http://ajslp.pubs.asha.org/ by University of California, Santa Barbara, Roger Ingham on 05/15/2015
Terms of Use: http://pubs.asha.org/ss/Rights_and_Permissions.aspx
Figure 2. Mean and range of %SS, SFSPM, and NAT scores from each TEST assessment for successful participants who completed MPI (left)
or prolonged speech (PS, right). %SS = percent syllables stuttered; SFSPM = syllables per minute in stutter-free speech; NAT = speech naturalness;
TEST = Treatment Evaluation Speaking Tasks (Ingham et al., 2001); MPI = Modifying Phonation Intervals treatment.
Ingham et al.: Efficacy of Modifying Phonation 263
Downloaded From: http://ajslp.pubs.asha.org/ by University of California, Santa Barbara, Roger Ingham on 05/15/2015
Terms of Use: http://pubs.asha.org/ss/Rights_and_Permissions.aspx
Figure 3. Mean %SS, SFSPM, and NAT scores from TEST assessments (cf. Figure 2), shown separately for three within-clinic speaking tasks
(orange = READ;blue=MONO;black=PHONE) and three beyond-clinic speaking tasks (green = PHONE;blue=CONV (family);red=SELF-SEL).
%SS = percent syllables stuttered; SFSPM = syllables per minute in stutter-free speech; NAT = speech naturalness; TEST = Treatment Evaluation
Speaking Tasks (Ingham et al., 2001); MPI = Modifying Phonation Intervals treatment.
264 American Journal of Speech-Language Pathology •Vol. 24 •256–271 •May 2015
Downloaded From: http://ajslp.pubs.asha.org/ by University of California, Santa Barbara, Roger Ingham on 05/15/2015
Terms of Use: http://pubs.asha.org/ss/Rights_and_Permissions.aspx
Figures 4 and 5 shows that, for several combinations of
speaking condition and TEST occasion from the midpoint
of the establishment phase (marked as E1 in the figures)
through to the end of the transfer phase, participants who
were ultimately successful at completing their treatment
program produced more PIs in their pretreatment fourth
quintile, produced fewer PIs in what had been their longest
(fifth) quintile at pretreatment, or showed significantly re-
duced within-group variability with a trend toward these
differences between the means, as compared with participants
who were not ultimately successful.
Comparisons to Control Group
All eight CONT participants produced stutter-free
and highly natural sounding speech (NAT < 3) on all within-
clinic and beyond-clinic recordings. In addition, there were
no significant changes in the CONT group’s PI data from
their first TEST to their second, which occurred on average
35 weeks later (see Figure 4). As a group, successful treatment
participants showed similar speech performance (i.e., means
of < 1 %SS and naturalness < 3; see Table 1 and Figures 2
and 3). Two of the 14 successful treatment participants pro-
duced stutter-free speech on all within- and beyond-clinic
TEST speaking tasks, and eight of the 14 self-rated their
speech as natural sounding (naturalness ratings of 3 or less
on all six TEST speaking tasks). Clinicians also rated those
same recordings within the same range. The small amounts
of residual stuttering are consistent with the speakers’self-
reports of feeling sometimes to almost always “like a normal
speaker”and also sometimes to almost always considering
themselves to be persons who stuttered (see Table 2).
Discussion
Each of the three questions that were the focus of this
study was addressed. In summary, the comparison between
MPI treatment and PS treatment showed that partici-
pants who completed either program achieved similar and
substantial reductions in the frequency of stuttering and
Table 2. Most common response (and range of responses, as text, using N = never,S=sometimes,AA=almost always,A=always, unless the
mode and range are identical) and mean response (converted to a 4-point ordinal scale where never = 0 and always = 3) to three items on the
Speech Performance Questionnaire at the end of each phase, for each group.
Treatment Phase MPI PS CONT
“I can now speak normally without thinking about controlling my speech”
Pretreatment Never/sometimes Sometimes (N-S) Always (AA-A)
0.50 0.90 2.88
Establishment Never/sometimes (N-AA) Sometimes (N-AA)
0.50 1.63
Transfer Sometimes (N-AA) Sometimes (N-A) Always (AA-A)
1.00 1.00 2.88
Maintenance Sometimes (S-AA) Almost always (S-AA)
1.25 1.67
Follow-up Sometimes (S-AA) Sometimes/almost always
1.67 1.50
“I now feel like a normal speaker”
Pretreatment Never (N-S) Sometimes (N-AA) Always (AA-A)
0.42 0.70 2.88
Establishment Never/sometimes Sometimes (N-A)
0.50 1.38
Transfer Sometimes (N-AA) Sometimes (N-A) Always (AA-A)
0.83 1.25 2.88
Maintenance Sometimes (S-AA) Sometimes (S-AA)
1.25 1.33
Follow-up Almost always (S-AA) Sometimes/almost always
1.67 1.50
“I now consider myself a stutterer”
Pretreatment Always (AA-A) Always (S-A) Never
2.92 2.60 0.00
Establishment Always (AA-A) Sometimes (S-A)
2.67 1.88
Transfer Always (S-A) Sometimes/almost always (N-A) Never
2.33 1.50 0.00
Maintenance Almost always (AA-A) Sometimes (N-S)
2.25 0.67
Follow-up Almost always (S-A) Sometimes
2.00 1.00
Note. MPI = Modifying Phonation Intervals treatment; PS = Prolonged Speech treatment; CONT = control.
Ingham et al.: Efficacy of Modifying Phonation 265
Downloaded From: http://ajslp.pubs.asha.org/ by University of California, Santa Barbara, Roger Ingham on 05/15/2015
Terms of Use: http://pubs.asha.org/ss/Rights_and_Permissions.aspx
required essentially the same amount of time to complete
their programs. All successful participants, from either pro-
gram, maintained zero or very low stuttering, with speech
rates and speech naturalness levels that were similar to
those produced by control speakers, throughout extended
transfer and maintenance phases that totaled an average of
more than 3 years. Successful participants also shifted from
viewing themselves consistently as a person who stutters,
and as not feeling like normal speakers, to sometimes or
almost always feeling “like a normal speaker.”Most speech
performance gains were maintained throughout an addi-
tional 12-month follow-up period, during which self-
identification either stabilized or continued to move further
away from “stutterer”and toward “normal speaker.”
In short, the speech outcomes of both programs were out-
standing and well maintained for successful participants,
as has previously been reported for MPI in one preliminary
study (Ingham et al., 2001) and for prolonged speech in
Figure 4. Mean (and 95% confidence limits) number of PIs per minute within the duration ranges that defined the quintiles individually derived
for each speaker from pretreatment speech, shown for participants who successfully completed their treatment program (black), participants
who did not (red), and controls (green), for within-clinic TEST assessments in READ, MONO, and PHONE conditions at the end of pretreatment
(PT), the middle and end of establishment (E1, E2), and the middle and end of transfer (T1, T2). PI = phonation interval; TEST = Treatment
Evaluation Speaking Tasks (Ingham et al., 2001); MPI = Modifying Phonation Intervals treatment; READ = reading from a novel; MONO =
monologue on a self-chosen topic; PHONE = telephone conversation.
266 American Journal of Speech-Language Pathology •Vol. 24 •256–271 •May 2015
Downloaded From: http://ajslp.pubs.asha.org/ by University of California, Santa Barbara, Roger Ingham on 05/15/2015
Terms of Use: http://pubs.asha.org/ss/Rights_and_Permissions.aspx
many studies and reviews (see Introduction), and these
changes were associated with changes in self-image and
self-identification.
One major difference between the two treatments,
however, was that more than twice the proportion of partic-
ipants succeeded via the MPI program. This result is also
consistent with previous research; the current finding of
success for 3/10 adults assigned to the PS treatment approx-
imates, for example, Onslow et al.’s (1996) report that
12/32 of their adult PWS completed the program. The re-
sults of this study also provide one possible explanation for
this finding. Successful maintenance of stutter-free, natural-
sounding speech appears to be associated with both a re-
duction in the number of very short PIs (those of durations
that defined the speaker’s shortest 20% at baseline) and also
an increase in the use of mid- to long-range PIs, but not
the longest (Q4 but not Q5; Figures 4 and 5). MPI treat-
ment requires the first of these changes explicitly, whereas
the initial focus of prolonged speech treatments might be
described as emphasizing the use of only very long intervals
of exaggerated, extended phonation. Thus, participants in
MPI treatment may be receiving one of two critical ele-
ments directly and efficiently, but participants in PS treat-
ment may be receiving only indirect approximations of the
necessary changes that are associated with long-term main-
tenance of positive treatment outcomes. Those indirect
approximations appear to be sufficient for about one third
of participants, but the finding from this study—that two
PS participants, but no MPI participants, failed to meet
performance criteria during the establishment phase—
suggests that MPI treatment may be more directly, effectively,
and efficiently providing participants with the information
they need at initial stages of learning than prolonged speech
approaches can do.
5
It remains problematic, however, that
four participants who completed the MPI’s establishment
phase were not successful during their transfer or mainte-
nance phases. Indeed, this failure at a later stage to attain
beyond-clinic success, for clients who have achieved stutter-
free, natural sounding speech in an initial establishment
Figure 5. Results of comparisons between participants who successfully completed their treatment program and those who did not,
for measures of the number of phonation intervals (PIs) per minute within the duration ranges derived by quintile from pretreatment
speech (cf. Figure 4). Significant Student ttests for differences between the groups’means, where p≤.05, are labeled M; significant
Ftests for differences in the variance, where p≤.05, are labeled V. Student ttests were used to compare means between the two
groups, and an Ftest was used to compare variances between two groups. A quantile-quantile plot or normal probability plot was
used to check the normality assumption (Wilk & Gnanadesikan, 1968). Wilcoxon rank sum test was also applied, and the results
remained the same. READ = reading from a novel; MONO = monologue on a self-chosen topic; PHONE = telephone conversation.
5
5Arguably, the use of a biofeedback procedure might have contributed
to the differences between the MPI- and PS-treatment outcomes. First,
it is important to recognize (see Abstract) that this study was designed to
compare a new treatment with the current “standard of care”stuttering
treatment for adults, that is, prolonged speech. Second, no attempt was
made either in the design or the result analysis to “unpack”the MPI
treatment nor to prove that the new treatment was effective because
it contained a biofeedback system. We simply wanted to know if it
produced a result that was superior to the current standard of care. If
we had wanted to test whether biofeedback produced a functional
difference between the two treatment programs, then we would have
designed an entirely different study.
Ingham et al.: Efficacy of Modifying Phonation 267
Downloaded From: http://ajslp.pubs.asha.org/ by University of California, Santa Barbara, Roger Ingham on 05/15/2015
Terms of Use: http://pubs.asha.org/ss/Rights_and_Permissions.aspx
phase, might be viewed as even more problematic than dif-
ficulties that make themselves apparent early in treatment.
This issue is related to questions about relationships
among pretreatment performance, immediately posttreat-
ment performance, and long-term outcomes. As mentioned
in the Introduction, previous authors have suggested that
pretreatment severity and/or immediately posttreatment re-
sults might predict long-term performance. For the present
study, pretreatment stuttering frequency was not highly pre-
dictive of success (the point-biserial correlation between
pretreatment stuttering frequency and completion or non-
completion of the maintenance phase as a dichotomous
variable was −.096; pretreatment mean stuttering was
7.31 %SS, range 1.06–20.2, for participants who ultimately
completed the maintenance phase and mean 9.24 %SS,
range 2.08–30.41, for those who did not). Similarly, corre-
lations between pretreatment stuttering frequency and later
TEST data were all moderate (r= .571–.580) for post-
establishment, posttransfer, and follow-up, and pre-
treatment and postmaintenance stuttering frequency were
unrelated (r= .086) for participants who successfully com-
pleted maintenance.
Several notable differences emerged at the end of
the establishment phase, however, that point to the im-
portance of clients being able to transfer their skills to
nontreatment conditions at that early, or immediately post-
establishment, point. Specifically, those participants who
would go on to complete the maintenance phase averaged
only 0.8 %SS in nontreatment TEST speaking tasks at the
end of the establishment phase, whereas those who would
prove unable to complete a later phase averaged 3.8 %SS in
the postestablishment TEST assessment. In addition, post-
establishment stuttering frequency was highly correlated
with stuttering frequency at transfer (r= .807). (For later
phases, correlations are difficult to interpret because of the
reduced range of stuttering frequency scores.) In short, pre-
treatment stuttering frequency did not predict treatment
outcome in this study, but success in transferring stutter-
free speech to nontreatment speaking tasks early in a treat-
ment program (i.e., speech performances on TEST tasks)
may be critical to clients’ability to continue to manage their
improved speech.
Stuttering frequency at maintenance (on TEST tasks),
which might be considered the final posttreatment point
for these programs, also appears to be highly predictive of
stuttering frequency at follow-up, for the participants who
successfully completed the maintenance phase. Stuttering
frequency for individual participants ranged from 0.0 to
1.6 %SS in maintenance and from 0.0 to 1.3 %SS at follow-
up, with seven participants increasing %SS from mainte-
nance to follow-up, seven participants decreasing, and a
mean change from maintenance to follow-up of +0.026.
This finding is of some interest, in part because it demon-
strates that relapse is not inevitable for adults who stutter;
successful participants in this study continued to stutter less
and continued to view themselves as more typical speakers
for a full year after the maintenance program was completed.
It has been claimed that allowing a 12-month follow-up
evaluation period after the completion of the maintenance
phase would allow uncontrolled variables to intrude and
confound ultimate outcome data. Such claims may have
some face validity (see Ingham, 1981; O’Brian, Packman,
Onslow, & Menzies, 2012), and indeed it cannot be said
with any certainty that nothing occurred during the follow-up
year to influence participants. Nevertheless, the long-term
follow-up data are crucial and here serve the very necessary
purpose of establishing that treatment gains were main-
tained after all treatment contingencies and schedules had
been removed, something that cannot be measured or deter-
mined while treatment is still being administered.
6
The present results therefore suggest several possible
improvements to future treatments for adults who stutter.
First, the need to incorporate generalization (Stokes &
Osnes, 1989) from the beginning of treatment is reaffirmed
by these data; participants who remained successful in the
long term were those who could, after only 3–12 weeks
of treatment, produce the same essentially stutter-free and
natural-sounding speech in multiple beyond-clinic speaking
situations (i.e., TEST tasks) that they had learned to pro-
duce during therapy. Regardless of the specific treatment ap-
proach being used, this emphasis on beyond-clinic work
from the beginning of treatment appears to be crucial.
It also appears from these results that the MPI pro-
gram could be improved by incorporating feedback to
the speaker about both a required reduction in very short
PIs and a required increase in a specific mid-to-long range
of PIs (the fourth pretreatment quintile). Alternatively,
the MPI software could be used to focus only on the fourth
quintile, a change that might be simpler for clients than
the two-quintiles approach and that would essentially
be using biofeedback to teach a less exaggerated form of
prolonged speech than has traditionally been possible
when that speech pattern is taught only by audiotaped or
clinician-produced models. Such an emphasis on “longer
but not too long”phonation would be consistent with Hillis
and McHugh’s (1998) demonstrations that biofeedback
to reduce the frequency of short pauses or “pauses per min-
ute of acoustic energy”resulted in clinically important ben-
efits with single subjects, a possibility that was never fully
explored.
Another set of improvements that might be investi-
gated on the basis of these results arises from the related is-
sues of the time that participants required to complete the
entire treatment program and several other features of the
overall treatment infrastructure used for both treatments
in this study. The mean total time required for successful
6
6We find a logical flaw in the suggestion by O’Brian et al. (2012) that
valid treatment outcome data can be derived from evaluations made during
the conduct of a performance-contingent or other active maintenance
schedule. Maintenance phases are a part of treatment, and there is
an obvious logical contradiction in claiming to assess the long-term
durability of a treatment while the treatment is still being administered.
Long-term outcome data must be gathered at a follow-up point well
after all treatment, including maintenance phase treatment, has ceased.
268 American Journal of Speech-Language Pathology •Vol. 24 •256–271 •May 2015
Downloaded From: http://ajslp.pubs.asha.org/ by University of California, Santa Barbara, Roger Ingham on 05/15/2015
Terms of Use: http://pubs.asha.org/ss/Rights_and_Permissions.aspx
treatment, from pretreatment through the end of a success-
ful maintenance phase, was a little over 3 years (159 weeks).
Most of the variation occurred, however, in the transfer
(10–37 weeks) and especially the maintenance (38–78 weeks)
phases; establishment phases were a relatively efficient 3–
12 weeks for all successful participants. Much of the reason
for this was because participants were permitted consider-
able latitude in implementing their self-managed schedules
during transfer and maintenance. Lacking evidence-based
guidelines for implementing self-management, it was de-
cided to simply inform participants that they should com-
plete a scheduled speaking task in the program when they
were ready to do so. They were encouraged to proceed
through the program at a steady rate while also using self-
managed practice, especially during the establishment phase
(MPI program participants had access to the MPI hard-
ware through the transfer phase, another variable worthy
of investigation). It is not possible to determine from these
data, however, whether the approximately 1–3-year duration
of focus and practice was necessary to long-term success or
whether that time could have been shortened.
Many other variables might be necessary or sufficient
for maximizing treatment efficacy, particularly for a behav-
ioral treatment such as the MPI. Many have been explored
and investigated by the first author and colleagues (see
Ingham et al., 2012). These include identifying functional
performance measures (e.g., speech naturalness, speech effort)
and endeavoring to integrate them into a performance-
contingent management framework. In addition, as brain-
imaging research on stuttering has proceeded to search
for neural systems that are critical to stuttering, it has be-
come increasingly important to find ways to integrate those
research findings with treatment, and especially to guide
treatment so as to increase the probability that the known
plasticity of the nervous system can be functionally harnessed
during treatment (see Ingham, Finn, & Bothe, 2005). With
increasing indications that PWS exhibit white-matter ab-
normalities (e.g., Cai et al., 2014; Chang, Horwitz, Ostuni,
Reynolds, & Ludlow, 2011; Cykowski, Fox, Ingham,
Ingham, & Robin, 2010), treatments that can be shown to
engage neural regions relevant for the change toward fluent
speech become a point of considerable interest. Thus, in
the course of the present research program, the authors re-
ported (Ingham et al., 2013) positron emission tomography
(PET) scan findings indicating that decreases in cerebral
blood flow (CBF) in left putamen during the establishment
phase are highly predictive of successful progress through
later parts of the program, and that CBF increases in
left putamen predict just the opposite. The link between
these changes and the TRPI frequency is made more plau-
sible for two reasons. The first are the findings reported
by Riecker, Kassubek, Groschel, Grodd, and Ackermann
(2006), which showed that increases in the frequency of
phonated utterances (compatible with improved speech pro-
duction) in normal speakers are correlated with decreases
in CBF activity in left and right putamen. The second,
however, seems from the present study’s results to be less
straightforward. During the period when CBF activity in
putamen was decreasing in the successfully treated cohort,
there was actually a decrease in TRPI frequency. However,
PI frequencies outside of the TRPI range (Q2–Q5) showed
no such decrease and were accompanied by systematic
increases in speech rate. In short, these studies have shown
that it may be a combination of TRPI reductions, fourth-
quintile PI increases, and reductions in putamen activity
that, as a set, could predict within a 12-week establishment
phase whether extended self-managed practice in ensuing
transfer and maintenance phases will result in successful
and durable acquisition of stutter-free, natural-sounding
speech. This possibility, and the modifications that need to
be made for those clients whose performance at 12 weeks
does not meet these criteria, deserves further investigation.
One additional variable that was not systematically
investigated within the present study was the amount of
self-managed practice that occurred among participants,
and especially the amount that differentiated the successful
and unsuccessful participants involved in either the MPI
or PS program. Informal and frequent comments by partic-
ipants made it evident that successful participants incor-
porated substantial and consistent speech practice, usually
oral reading, to ensure they could readily produce self-
judged stutter-free and natural-sounding speech. Some
practice routines appeared to greatly exceed the amount of
formal treatment time. The amount of practice time varied
greatly, but it was often increased markedly after perfor-
mance failure caused a retreat within the treatment phase.
Practice is a much-overlooked factor in stuttering treat-
ment and warrants formal investigation as to its functional
value.
In general, the findings of this study are promising
as to adult treatment in general, provide important guidance
as to the structure of any treatment, and are especially
encouraging of future research on the MPI. There are a
number of ways in which the program can and will be im-
proved. Some have already been developed and imple-
mented, including an application for an iPad or iPhone, an
automatic server-based data exchange of recordings and
treatment status between client and clinician, and incorpo-
ration of a system for automatically prompting a client to
complete transfer and maintenance phase speaking tasks
(Ingham & Student, 2014).
7
The latter system is expected to
resolve the problem of slow passage through the program.
Finally, the recently discovered prediction of successful
passage through the MPI program by using critical neural
region change is especially interesting (see Ingham et al.,
2013) because it offers the prospect of a more fruitful link
between brain imaging and stuttering treatment.
7
7In 2011, a product known as “MPIStutter iPhone App”began to be
advertised on the web and purports to be the MPI stuttering treatment
program. This product is essentially a crude imitation of just one part
of the MPI program and was developed and made available on Apple’s
App Store without the program author’s permission. There is, for
example, no performance-contingent schedule of tasks nor clinician-
management operations incorporated in this product, features that
constitute necessary components of the authentic MPI program.
Ingham et al.: Efficacy of Modifying Phonation 269
Downloaded From: http://ajslp.pubs.asha.org/ by University of California, Santa Barbara, Roger Ingham on 05/15/2015
Terms of Use: http://pubs.asha.org/ss/Rights_and_Permissions.aspx
Acknowledgments
This study was completed with the support of RO1 Grant
DC007893 from the National Institute on Deafness and Other
Communication Disorders, awarded to the first author. Special
thanks are offered to Peter Fox for providing the facilities for parts
of this study and for his continuing support for imaging research
on developmental stuttering. We also wish to acknowledge the de-
velopmental work on the MPI by Jack Montgomery at Cumberland
College in Australia, Robert Bandes at the University of Washington,
and Don Devan and Peter Frank at the University of California,
Santa Barbara. Thanks are also due to Robin Selman Hampton,
Krystal Purkheiser, Annie New, Kate Paolini, Gina Pecile,
Amanda Petros, and multiple research assistants at the University
of Georgia for their assistance in gathering data and background
material.
References
Andrews, G., Guitar, B., & Howie, P. (1980). Meta-analysis of the
effects of stuttering treatment. Journal of Speech and Hearing
Disorders, 45, 287–307.
Boberg, E., & Kully, D. (1985). Comprehensive stuttering program.
San Diego, CA: College-Hill.
Boberg, E., & Kully, D. (1994). Long-term results of an intensive
treatment program for adults and adolescents who stutter.
Journal of Speech and Hearing Research, 37, 1050–1059.
Boersma, P., & Weenink, D. (2003). PRAAT (Version 4.2.07)
[Computer software]. Amsterdam, the Netherlands: Institute
of Phonetics Sciences, University of Amsterdam. Retrieved
from http://www.praat.org
Borden, G. J., Baer, T., & Kenney, M. K. (1985). Onset of voicing
in stuttered and fluent utterances. Journal of Speech and Hear-
ing Research, 28, 363–372.
Bothe, A. K., Davidow, J. H., Bramlett, R. E., & Ingham, R. J.
(2006). Stuttering treatment research, 1970–2005: I. Systematic
review incorporating trial quality assessment of behavioral,
cognitive, and related approaches. American Journal of
Speech-Language Pathology, 15, 321–341.
Cai, S., Tourville, J. A., Beal, D. S., Perkell, J. S., Guenther, F.
H., & Ghosh, S. S. (2014). Diffusion imaging of cerebral white
matter in persons who stutter: Evidence for network level
anomalies. Frontiers in Human Neuroscience, 8, 54. doi:10.3389/
fnhum.2014.00054
Chang, S.-E., Horwitz, B., Ostuni, J., Reynolds, R., & Ludlow,
C. L. (2011). Evidence of left inferior frontal-premotor struc-
tural and functional connectivity deficits in adults who stutter.
Cerebral Cortex, 21, 2507–2518.
Cykowski, M., Fox, P. T., Ingham, R. J., Ingham, J. C., & Robin,
D. A. (2010). A study of the reproducibility and etiology of
diffusion anisotropy differences in developmental stuttering:
A potential role for impaired myelination. NeuroImage, 52,
1495–1504.
Davidow, J. H., Bothe, A. K., Andreatta, R. D., & Ye, J. (2009).
Measurement of phonated intervals during four fluency-inducing
conditions. Journal of Speech, Language, and Hearing Research,
52, 188–205.
Euler, H. A., & Wolff von Gudenberg, A. (2000). Die Kasseler
Stottertherapie (KST): Ergebnisse einer computergestützten
Biofeedbacktherapie für Erwachsene [The Kassel Stuttering
Therapy: Investigation of a computer-assisted biofeedback
therapy for adults]. Sprache Stimme Gehör, 24, 71–79.
Euler, H. A., Wolff von Gudenberg, A., Jung, K., & Neumann, K.
(2009). Computergestützte Therapie bei Redeflussstörungen:
Die langfristige Wirksamkeit der Kasseler Stottertherapie (KST)
[Computer assisted therapy for fluency disorders: The long-term
effectiveness of the Kassel Stuttering Therapy]. Sprache Stimme
Gehör, 33, 193–201.
Godinho, T., Ingham, R. J., Davidow, J., & Cotton, J. (2006). The
distribution of phonated intervals in the speech of stuttering
speakers. Journal of Speech, Language, and Hearing Research,
49, 161–171.
Goldiamond, I. (1965). Stuttering and fluency as manipulatable
operant response classes. In L. Krasner & L. P. Ullman (Eds.),
Research in behavior modification (pp. 106–156). New York,
NY: Holt, Rinehart, & Winston.
Goldiamond, I. (1967). Supplementary statement to operant
analysis control of fluent and non-fluent verbal behavior (Public
Health Service Application No. MH-8876-03). Washington,
DC: Department of Health, Education, and Welfare.
Gow, M. L., & Ingham, R. J. (1992). The effect of modifying elec-
troglottograph identified intervals of phonation on stuttering.
Journal of Speech and Hearing Research, 35, 495–511.
Hancock, K., & Craig, A. (1998). Predictors of stuttering relapse
one year following treatment for children aged 9 to 14 years.
Journal of Fluency Disorders, 23, 31–48.
Hillis, J. W., & McHugh, J. (1998). Theoretical and pragmatic
considerations for extraclinical generalization. In A. K. Cordes
and R. J. Ingham (Eds.), Treatment efficacy for stuttering:
A search for empirical bases (pp. 243–292). San Diego, CA:
Singular Publishing Group.
Ingham, R. J. (1980). Modification and maintenance of generaliza-
tion during stuttering treatment. Journal of Speech and Hearing
Research, 23, 732–745.
Ingham, R. J. (1981). Evaluation and maintenance in stutter-
ing therapy: A search for ecstasy with nothing but agony.
In E. Boberg (Ed.). Maintenance of fluency (pp. 179–218).
New York, NY: Elsevier.
Ingham, R. J. (1982). The effects of self-evaluation training on
maintenance and generalization during stuttering treatment.
Journal of Speech and Hearing Disorders, 47, 271–280.
Ingham, R. J. (1984). Stuttering and behavior therapy: Current
status and empirical foundations. San Diego, CA: College-Hill.
Ingham, R. J. (1987). Residential prolonged speech stuttering ther-
apy manual. Santa Barbara, CA: Department of Speech and
Hearing Sciences, University of California.
Ingham, R. J. (1993). Current status of stuttering and behavior
modification II: Principal issues and practices in stuttering
therapy. Journal of Fluency Disorders, 18, 57–79.
Ingham, R. J. (1999). Performance-contingent management
of stuttering in adolescents and adults. In R. Curlee (Ed.),
Stuttering and related disorders of fluency (pp. 200–221).
New York, NY: Thieme.
Ingham, R. J., Bakker, K., Ingham, J. C., Kilgo, M., & Moglia, R.
(1999). Stuttering measurement system (SMS) [Software, man-
ual, and training materials]. Santa Barbara, CA: University
of California, Santa Barbara. Available at http://sms.id.ucsb.
edu/index.html
Ingham, R. J., & Costello, J. M. (1985). Stuttering treatment
outcome evaluation. In J. M. Costello (Ed.), Speech disorders
in adults: Recent advances (pp. 189–223). San Diego, CA:
College-Hill.
Ingham, R. J., Finn, P., & Bothe, A. K. (2005). Roadblocks
revisited: Neural change, stuttering treatment, and recovery
from stuttering. Journal of Fluency Disorders, 30, 91–107.
Ingham, R. J., Ingham, J. C., & Bothe, A. K. (2012). Integrating
functional measures with treatment: A tactic for enhancing
personally significant change in the treatment of adults and
270 American Journal of Speech-Language Pathology •Vol. 24 •256–271 •May 2015
Downloaded From: http://ajslp.pubs.asha.org/ by University of California, Santa Barbara, Roger Ingham on 05/15/2015
Terms of Use: http://pubs.asha.org/ss/Rights_and_Permissions.aspx
adolescents who stutter. American Journal of Speech-Language
Pathology, 21, 264–277.
Ingham, R. J., Kilgo, M., Ingham, J. C., Moglia, R., Belknap, H.,
& Sanchez, T. (2001). Evaluation of a stuttering treatment
based on reduction of short phonation intervals. Journal of
Speech, Language, and Hearing Research, 44, 1229–1244.
Ingham, R. J., Moglia, R., Kilgo, M., & Felino, A. (2007). Pro-
longed speech treatment program (accompanied by phonation in-
terval recording). Santa Barbara, CA: Department of Speech
and Hearing Sciences, University of California.
Ingham, R. J., Montgomery, J., & Ulliana, L. (1983). The effect
of manipulating phonation duration on stuttering. Journal of
Speech and Hearing Research, 26, 579–587.
Ingham, R. J., & Student, F. (2014). Modifying Phonation Inter-
vals (MPI) Stuttering Treatment iOS application. Stuttgart,
Germany: Logera Solutions GmbH. Available at http://mpi-
stuttering-treatment.com
Ingham, R. J., Wang, Y., Ingham, J. C., Bothe, A. K., & Grafton,
S. T. (2013). Regional brain activity change predicts responsive-
ness to treatment for stuttering in adults. Brain and Language,
127, 510–519.
Langevin, M., Huinck, W. J., Kully, D., Peters, H. F. M.,
Lomheim, H., & Tellers, H. (2006). A cross-cultural, long-term
outcome evaluation of the ISTAR Comprehensive Stuttering
Program across Dutch and Canadian adults who stutter.
Journal of Fluency Disorders, 31, 229–256.
Langevin, M., Kully, D., Teshima, S., Hagler, P., & Prasad, N. G. N.
(2010). Five-year longitudinal treatment outcomes of the ISTAR
Comprehensive Stuttering Program. Journal of Fluency Disorders,
35, 123–140.
Martin, R. R., Haroldson, S. K., & Triden, K. A. (1984). Stuttering
and speech naturalness. Journal of Speech and Hearing Disorders,
49, 53–58.
Nye, C., Vanryckeghem, M., Schwartz, J. B., Herder, C., Turner,
H. M., & Howard, C. (2013). Behavioral stuttering interventions
for children and adolescents: A systematic review and meta-
analysis. Journal of Speech, Language, and Hearing Research,
56, 921–932.
O’Brian, S., Onslow, M., Cream, A., & Packman, A. (2003). The
Camperdown Program: Outcomes of a new prolonged-speech
treatment model. Journal of Speech, Language, and Hearing
Research, 46, 933–946.
O’Brian, S., Packman, A., Onslow, M., & Menzies, R. (2012).
Measuring outcomes following the Camperdown Program
for stuttering: A response to Dr. Ingham. Journal of Speech,
Language, and Hearing Research, 55, 310–312.
Oldfield, R. C. (1971). The assessment and analysis of handed-
ness: The Edinburgh Inventory. Neuropsychologia, 25,
965–969.
Onslow, M., Costa, L., Andrews, C., Harrison, E., & Packman, A.
(1996). Speech outcomes of a prolonged-speech treatment
for stuttering. Journal of Speech and Hearing Research, 39,
734–749.
Onslow, M., & O’Brian, S. (1998). Reliability of clinicians’judg-
ments about prolonged-speech targets. Journal of Speech,
Language, and Hearing Research, 41, 969–975.
Perkins, W. (1981). Measurement and maintenance of fluency.
In E. Boberg (Ed.), Maintenance of fluency (pp. 147–178).
New York, NY: Elsevier.
Peters, H. F. M., Boves, L., & van Dielen, I. C. (1986). Perceptual
judgment of abruptness of voice onset in vowels as a function
of the amplitude envelope. Journal of Speech and Hearing
Disorders, 51, 299–308.
Riecker, A., Kassubek, J., Groschel,K.,Grodd,W.,&Ackermann,H.
(2006). The cerebral control of speech tempo: Opposite relation-
ship between speaking rate and BOLD signal changes at striatal
and cerebellar structures. NeuroImage, 29, 46–53.
Stokes, T. F., & Osnes, P. G. (1989). An operant approach to gen-
eralization. Behavior Therapy, 20, 337–355.
Teshima, S., Langevin, M., Hagler, P., & Kully, D. (2010). Post-
treatment speech naturalness of Comprehensive Stuttering Pro-
gram clients and differences in ratings among listener groups.
Journal of Fluency Disorders, 35, 44–58.
Webster, R. L. (1974). A behavioral analysis of stuttering treat-
ment and theory. In K. S. Calhoun, H. E. Adams, & K. M.
Mitchell (Eds.), Innovative treatment methods in psychopathology
(pp. 17–61). New York, NY: Wiley.
Webster, R. L. (1977). A few observations on the manipulation
of speech response characteristics in stutterers. Journal of Com-
munication Disorders, 10, 73–76.
Webster, R. L. (1980). The precision fluency shaping program:
Speech reconstruction for stutterers [Clinician’s program guide].
Roanoke, VA: Communication Development Corporation.
Wilk, M. B., & Gnanadesikan, R. (1968). Probability plotting
methods for the analysis of data. Biometrika, 55, 1–17.
Williams, D. E., Wark, M., & Minifie, F. D. (1963). Ratings of
stutterings by audio, visual, and audiovisual cues. Journal of
Speech and Hearing Research, 6, 91–100.
Ingham et al.: Efficacy of Modifying Phonation 271
Downloaded From: http://ajslp.pubs.asha.org/ by University of California, Santa Barbara, Roger Ingham on 05/15/2015
Terms of Use: http://pubs.asha.org/ss/Rights_and_Permissions.aspx
