Professional Documents
Culture Documents
Lee Silverman en Niños
Lee Silverman en Niños
Article
peech disorders have been reported to occur in children with spastic cerebral palsy (CP; Workinger &
Kent, 1991); however, reliable prevalence figures
have not been documented (Pennington, Miller, & Robson,
2009). The most common perceptual characteristics
include consistent hypernasality, breathy voice quality,
monotonous speech, reduced loudness, and uncontrolled
rate and rhythm of voice (Workinger & Kent, 1991). Disordered respiration characterized by short vowel durations, shallow inspirations, and forced expirations also
has been reported (Clement & Twitchell, 1959; Solomon
& Charron, 1998). In addition, disordered articulation has
a
930
Journal of Speech, Language, and Hearing Research Vol. 55 930945 June 2012 D American Speech-Language-Hearing Association
931
dysarthria as evidenced by changes in acoustic measures of voice; (c) do parents perceive changes in perceptual voice and speech characteristics after intensive
voice treatment; and (d) do treatment effects, if any,
last over time?
Participants
A yoked group of typically developing, sex- and agematched children was recruited to participate in this
study. This allowed for a direct comparison between performance of participants with CP and their matched peers,
because data were collected using the same tasks and
methodology. The typically developing peers had no
Method
P1
P2
P3
P4
P5
Age (years;months)
7;10
5;10
6;1
7;7
6;7
Gender
Male
Female
Male
Male
Female
Diagnosis
Spastic quadriparesis
due to CP; secondary
seizure disorder
Spastic quadriparesis
due to CP
Parent-reported
speech and
voice concerns
Improving breath
Weak breath control,
control, improving
soft speech, choppy
clarity of speech,
speech (i.e., took a
maintaining loudness
lot of breaths to finish
and clarity at end
a sentence)
of sentences
Observed speech
and voice
signs (2 SLPs)
Variable loudness,
Reduced loudness,
inconsistent breathymonotone, intermittent
harsh voice quality,
hypernasality, intermittent
inconsistent slow
harshness, imprecise
rate, whispered at
consonants, variable
end of sentences,
slow rate (task dependent),
mild articulatory
prosodic abnormalities
imprecision
Dysarthria type
Overall dysarthria
severity
Study condition
Age (years;months),
gender-matched
control
Spastic quadriparesis
secondary to CP
Spastic quadriparesis
secondary to CP
Spastic quadriparesis
secondary to CP
Variable loudness;
Minimal voicing,
Variable loudness,
variable pitch;
reduced loudness to
irregular articulatory
inconsistent strained
aphonic, monotone,
breakdowns, variable
voice quality,
imprecise articulation,
rate, strained voice
especially at end
spoke only 12 word
quality with occasional
of sentences; mild,
phrases
voice stoppages,
imprecise articulation
intermittent glottal fry
Spastic
Spastic
Mixed spastic/ataxic
Spastic
Spastic
Moderate
Mildmoderate
Moderate
Marked
Moderate
Condition A
4 BASE/2 weeks
Condition C
5 BASE/5 weeks
Condition B
5 BASE/3 weeks
Condition A
4 BASE/3 weeks
Condition D
4 BASE/3 weeks
7;6 Male
6;2 Female
6;1 Male
7;1 Male
7;0 Female
Note.
P1 (P2, etc.) = Participant 1 (2, etc.); CP = cerebral palsy; SLP = speech-language pathologist; BASE = baseline.
932
Journal of Speech, Language, and Hearing Research Vol. 55 930945 June 2012
Design
This study used a nonconcurrent multiple baseline
design with replication across subjects (Barlow & Hersen,
1984; Watson & Workman, 1981). In this design, the
treatment variable (intensive treatment) was applied
sequentially to the same behavior (vocal output) across
different but matched subjects sharing the same environmental conditions. In a nonconcurrent design, the
length of the baseline phase is varied (some short and
some longer) and predetermined (Watson & Workman,
1981). When participants became available, they were
randomly assigned to one of the predetermined baseline
conditions. This provided flexibility for conducting the
research within real-world constraints while maintaining the internal validity achieved with a concurrent design (i.e., controlling for history or maturation variables;
Watson & Workman, 1981).
Procedures
Participant Selection
A telephone screening questionnaire was completed
with parents of potential participants followed by a faceto-face screening session with the child. All the parents
signed a consent form, and children signed an assent
document agreeing to participate. This study received
approval from the University of Arizona Institutional
Review Board. The screening session included (a) a brief
voice and speech screening, (b) an assessment of abilities to follow directions related to the study tasks, and
(c) a hearing screening (500 Hz, 1000 Hz, 2000 Hz, and
4000 Hz at 25 dB HL). In addition, a laryngeal examination was conducted by an otolaryngologist to ensure that
no laryngeal pathology (e.g., vocal nodules or paralysis)
existed. All five participants completed the entire study.
Study Conditions
Participants were randomly assigned to one of four
study conditions with varying durations of baseline
phases according to the nonconcurrent study design. A
minimum of four baseline recordings were planned
across Conditions A (2-week baseline), B (3-week baseline),
and C (4-week baseline). Participant assignment to study
conditions and the actual number of recording sessions
are summarized in Table 1. These conditions all included
16 treatment sessions (4 sessions a week for 4 consecutive weeks), two recording sessions 1 week immediately
following treatment (POST), and two recording sessions
6 weeks after the conclusion of treatment (FUP). Condition D matched Condition A except no treatment was de-
Recording Sessions
Equipment and setup. Data collection procedures
were identical for all baseline (BASE), posttreatment
(POST), and follow-up (FUP) recording sessions. Duration of recording sessions ranged from 30 min to 1 hr.
Data were collected in an Industrial Acoustics Company
sound-treated booth, and all sessions were videotaped.
Four of the participants were comfortably seated in
their own wheelchairs. One participant sat in an adaptive
chair. Audio recordings were made with the participant
wearing a small omni-directional condenser microphone
(Audio-Technica, Model AT 803b) taped to his or her forehead and secured with a soft, elastic cloth headband. The
distance from the corner of the childs mouth to his or her
forehead was measured for placement of the microphone
at each recording session. This allowed for a constant
mouth-to-microphone distance of 4 in. within and across
recording sessions. Microphone signals were recorded
onto a digital audiotape (DAT) recorder (Panasonic Digital Audio Tape Deck, Model SV-3500). Calibration signals
were recorded at the beginning and end of each session by
securing the omni-directional condenser microphone and
sound level meter (SLM) microphone to a Styrofoam head
at a distance of 4 in. from the mouth of the head, mimicking
the setup on the child. A tone generator (KORG Auto Chromatic Tuner, AT-12) was placed in the same plane as the
Styrofoam heads mouth. The generated tone (960.5 Hz)
was recorded to the DAT, and the exact sound pressure
level (dB SPL) reading from the SLM was recorded.
Data were collected by four graduate students and
the first author, who were well trained in the experimental protocol. The investigator who delivered treatment
(the first author) did not collect POST or FUP data. All
attempts to keep the student data collectors blinded to
the purpose of the study were made; however, there
were occasions when the participants revealed aspects
of the treatment they received. All data collectors were
trained to keep the same demeanor and instructions regardless of the participants comments.
Voice and speech tasks. The protocol included two
maximum performance tasks, including a sustained
vowel and maximal frequency range for vowels, as well
as a sentence repetition task (Kent & Kent, 2000; Kent,
Kent, & Rosenbek, 1987; Rvachew, Hodge, & Ohberg,
2005). For the maximal performance tasks, instructions
were Take a deep breath and say ah for as long as you
can, and Take a deep breath and say ah as high / low as
933
934
went from structured to less structured tasks (e.g., naming pictures to playing games requiring spontaneous
speech output). All exercises included a minimum of
15 repetitions of each training task while incorporating
sensory augmentation, such as cueing increased vocal
effort and loudness, and sensory awareness by asking
the children, Did you feel your voice? Did you hear
how you sounded? Children practiced one homework
session on treatment days (lasting 510 min) and two
homework sessions on nontreatment days (lasting 10
15 min). Daily carryover exercises were specific assignments to use the new target voice with someone in
the childs daily living environmentfor example, Say
good morning to the bus driver using your loud voice.
The impact of the carryover assignment was discussed
in the next treatment session (e.g., Did the bus driver understand you yesterday?). Family members assisted
the child in completing homework and carryover assignments. All participants and family members were
encouraged to continue homework routines at the conclusion of treatment.
Participation of Typically
Developing Peers
The typically developing peers participated in two
recording sessions in 1 weeks time, completing the
same tasks as the participants with CP. These children
did not participate in the laryngeal examination or the
treatment phase of this study.
Analysis
AuditoryPerceptual Data (Listener Task)
A group of seven certified speech-language pathologists who had extensive experience in the areas of motor
speech disorders and voice served as judges in a paired
comparison listening task. Three repetitions of the sentences Buy Bobby a puppy, The blue spot is on the
key, or The potato stew is in the pot from the final
BASE, POST, and FUP recording sessions were included
for speech samples. These final sessions were chosen to
ensure that each child would have the greatest familiarity with the task across the study phases. Speech samples
containing three repetitions of each of the sentences were
constructed for BASE, POST, and FUP sessions. The
order of presentation of speech samples was randomized
across study phases (e.g., BASE vs. POST or FUP vs.
BASE).
Listening task. Paired speech samples were presented to listeners via a computer and external speakers.
They rated which sample they preferred for the following variables: (a) overall loudness, (b) loudness variability,
(c) overall pitch, (d) pitch variability, (e) overall voice quality, and (f ) articulatory precision. Listeners marked on a
Journal of Speech, Language, and Hearing Research Vol. 55 930945 June 2012
Acoustic Data
Data preparation and digitizing procedures. All
acoustic data were reviewed, and problematic samples,
such as clipped signals (approximately 1015 samples
across all participants and sessions) and faulty microphone signals (one recording session for two participants), were excluded. Data from the DAT tapes were
digitized at 22.05 kHz with the commercial software program Praat (Boersma & Weenick, 2002). Each sample of
a sustained vowel, high vowel, low vowel, or sentence
repetition was saved in an individual .wav file. The calibration tone for each recording session was analyzed for
dB SPL. The value of the calibration tone from Praat was
subtracted from the recorded SLM value, creating a correction factor. The correction factor was applied to all
dB SPL values produced by Praat, providing calibrated
dB SPL at 4 in. across all participants and recording
sessions.
Measurements. Vowel samples (longest, highest,
lowest) from each session were displayed using Praat.
Both the acoustic waveform and spectrogram, including
intensity (dB SPL) and pitch (F0 in hertz) contours, were
displayed for analysis. All acoustic measures were derived from standard Praat algorithms. The following
measures were made for sustained vowel phonation:
935
treatment effect, defined as a visual trend with no overlapping data points between study phases, statistics
were used to confirm the effect; and (c) if there was unclear
evidence of a treatment effect, defined as a visual trend
with overlapping data points, or no visual trend with
variable baseline data, statistics were applied to ascertain whether there was a possible treatment effect not
evident through visual inspection of the graphed data.
We began our statistical analysis by checking for serial
dependency in the data by running a lag-1 auto-correlation
(Pearson) on neighboring data points in the baseline.
If there was no significant auto-correlation, then we
averaged data in each phase and conducted a one-way,
repeated-measures analysis of variance (ANOVA). Least
significant differences were used for post hoc comparisons. If baseline data were significantly correlated,
then a split-middle binomial probabilities test was applied (Siegel, 1956). A lenient post hoc analysis and
p value of <.05 for significance was chosen, allowing liberal tolerances for Type I errors, consistent with Phase 1
treatment research goals (Robey & Schultz, 1998). Effect
size calculations were derived using a modified Cohens
d statistic (Busk & Serlin, 1992), as suggested by Beeson
and Robey (2006) for use with single-subject research
designs. For comparison of participants with CP with
their typically developing peers, a difference in mean beyond 2 SDs was considered significant.
Reliability
Auditory Perceptual Analysis
We determined intrarater reliability for the seven
listeners by repeating 30% of the paired speech samples
per participant in the listening task. Intrarater reliability across listeners ranged from 74% to 89%.
Visual Analysis
We determined intrarater reliability for the three
judges who visually examined data by randomly selecting 10 of the 30 graphs for repeated analysis. The
intrarater reliability ranged from 98% to 100%. We determined interrater reliability by comparing agreements among the three raters. Interrater reliability
was 94.7%.
Parent Ratings
We determined intrarater reliability of parent perceptual ratings by repeating a visual analog scale at
one additional BASE, POST, or FUP recording. Intrarater reliability was available for all participants parents except Participant 1 (P1). Mean difference scores
(MDs) and Pearson productmoment correlations (r)
were calculated. Intrarater reliability fell in the following ranges: MD = 7.00%0.47% (0%100% scale),
r = .68.90.
Results
Auditoryperceptual data and chi-square analyses
are displayed in Table 2. The number of preference
choices per BASE, POST, FUP, or NP (3 Sentence
Pairs 7 Listeners = 21 total choices per perceptual variable) and the accompanying statistic are listed. All
acoustic data are displayed in Table 3. Effect size (ES)
was commensurate with observed direction of change
and statistical significance in all but one case (P1, dB
SPL for sustained phonation). Judgments regarding
magnitude of the ES were not made due to the Phase 1
nature of the study (Beeson & Robey, 2006). Difference
scores across study phases of parent perceptual ratings
(expressed in percentages) are displayed in Table 4. Positive and negative values reflect perceptions of improvement or deterioration, respectively.
Data for P1
Listener Ratings
Acoustic Analysis
We calculated measurement reliability by having
20% of the data reanalyzed by a second measurer.
936
Journal of Speech, Language, and Hearing Research Vol. 55 930945 June 2012
BASEFUP
BASE
POST
NP
c2
BASE
FUP
NP
c2
Overall loudness
Loudness variability
Overall pitch
Pitch variability
Overall voice quality
Articulatory precision
2
2
0
0
2
1
14*
11
13*
13*
14*
19*
5
8
8
8
5
1
.004
.050
.002
.002
.004
.000
11.14
6.0
12.28
12.28
11.14
30.86
9
9
8
11*
11
8
2
2
1
1
2
4
10
10
12*
9
8
9
.066
.066
.012
.018
.050
.368
5.43
5.43
8.86
8.0
6.0
2.0
Overall loudness
Loudness variability
Overall pitch
Pitch variability
Overall voice quality
Articulatory precision
1
2
3
2
1
0
12*
11*
6
10*
9*
4
1
1
5
2
4
10*
.000
.002
.607
.010
.030
.004
15.39
13.0
1.0
9.14
7.0
10.86
4
8
3
4
3
0*
14*
13*
10
13*
14*
11
3
0
8
4
4
10
.005
.002
.156
.021
.005
.005
10.57
12.27
3.71
7.71
10.57
10.57
Overall loudness
Loudness variability
Overall pitch
Pitch variability
Overall voice quality
Articulatory precision
2
2
0
1
1
0
14*
14*
8
11*
12*
14*
5
5
13*
9
8
7
.004
.004
.002
.018
.012
.001
11.14
11.14
12.29
8.0
8.86
14.0
10
3
1
0*
1
5
8
13*
5
11
13*
7
3
5
15*
10
7
9
.156
.018
.001
.005
.006
.565
3.71
8.0
14.86
10.57
10.29
1.14
Overall loudness
Loudness variability
Overall pitch
Pitch variability
Overall voice quality
Articulatory precision
1
0
0
0
0
0
19*
18*
13*
11*
20*
14*
1
3
8
10
1
7
.000
.0001
.002
.005
.0001
.0009
30.86
26.57
12.29
10.57
36.28
14.0
0
0
2
2
2
1*
16*
17*
6
7
17*
11
5
4
13*
12*
2
9
.000
.000
.012
.028
.000
.018
19.14
22.57
8.86
7.14
21.42
8.0
Overall loudness
Loudness variability
Overall pitch
Pitch variability
Overall voice quality
Articulatory precision
1
1
1
1
3
1
3
4
3
6
5
3
17*
16*
17*
14*
13*
17*
.000
.000
.000
.002
.018
.000
21.74
18.0
21.71
12.29
8.00
21.74
4
5
3
4
3
6
4
5
6
6
8
3
13*
11
12*
11
10
12*
.021
.180
.050
.156
.156
.050
7.71
3.43
6.0
3.71
3.71
6.0
Variable
P1
P2
P3
P4
P5
Note. For each perceptual variable, there were a total of 21 choices (3 Sentence Pairs 7 Listeners) for each comparison BASE to POST
and BASE to FUP (with the exception of P2 for POST). The values represent the number of listeners who preferred either a BASE, POST, FUP,
or NP speech sample. Boldfaced and asterisked values indicate the preference (BASE, POST, FUP, or NP) that was statistically significant.
P4 repeated only single words versus the entire sentence in this task. POST = posttreatment; FUP = follow-up; NP = no preference.
preferences were not maintained at FUP. Listener comments for preference of POST samples included louder,
more variable, pitch more stable, natural, less
strain, decreased nasal emissions, more crisp, and
less effort. Comments for preference of BASE samples
included more consistent loudness, louder, and less
gurgle.
Parent Ratings
Acoustic Analysis
The measures for which a visual treatment effect
was observable and for which a statistically significant
937
BASEPOST
BASE
POST
FUP
TP
Visual
Max performance ah
Duration (s)
Max performance
F0 range (Hz)
Sustained phonation
dB SPL of 3 longest ahs
Sustained phonation
HNR of 3 longest ahs
Sentence repetition
dB SPL
Sentence repetition
F0 range (Hz)
2.4
(1.0)
200
(42)
87.4
(4.9)
22.1
(2.3)
75.0
(1.6)
62
(7)
3.5
(1.0)
436
(139)
87.4
(2.9)
22.9
(0.7)
76.4
(1.3)
76
(3)
4.4
(0.5)
301
(45)
87.4
(7.2)
24.1
(0.8)
76.0
(0.74)
66
(16)
16.1
(2.8)
884
(234)
96.5
(1.3)
26.3
(0.1)
74.9
(3.6)
96.1
(49)
Max performance ah
Duration (s)
Max performance
F0 range (Hz)
Sustained phonation
dB SPL of 3 longest ahs
Sustained phonation
HNR of 3 longest ahs
Sentence repetition
dB SPL
Sentence repetition
F0 range (Hz)
1.8
(0.6)
289
(215)
86.3
(1.5)
22.0
(2.3)
73.0
(1.9)
109
(16)
5.7
(0.4)
724
(147)
94.6
(1.1)
22.0
(2.8)
82.0
(0.4)
131
(23)
7.0
(1.4)
1097
11.4
(0.6)
2242
(61)
95.3
(2.3)
27.9
(1.6)
78.6
(0.4)
180
(29)
Max performance ah
Duration (s)
Max performance
F0 range (Hz)
Sustained phonation
dB SPL of 3 longest ahs
Sustained phonation
HNR of 3 longest ahs
Sentence repetition
dB SPL
Sentence repetition
F0 Range (Hz)
5.3
(0.8)
796
(127)
82.4
(2.7)
19.2
(4.9)
80.4
(2.9)
118
(18)
6.8
(0.6)
633
(112)
84.2
(2.8)
21.7
(0.5)
79.3
(2.9)
128
(32)
13.4
Max performance ah
Duration (s)
Max performance
F0 range (Hz)
Sustained phonation
dB SPL of 3 longest ahs
Sustained phonation
HNR of 3 longest ahs
Sentence repetition
dB SPL
Sentence repetition
F0 range (Hz)
1.0
(0.7)
137
(34)
83.1
(10.1)
15.6
(4.4)
70.4
(4.4)
34
(18)
1.2
(0.2)
211
(21)
87.5
(2.5)
19.3
(2.0)
75.0
(0.7)
42
(9)
STAT
BASEFUP
ES
Visual
STAT
ES
p < .12a
1.1
p < .035a
2.0
p < .01a
5.6
p < .17a
2.4
p < .0001b
0.0
p < .0001b
0.0
N/A
0.3
N/A
0.8
N/A
0.9
N/A
0.6
N/A
2.0
N/A
0.6
p < .001a
6.5
p < .0001a
5.2
p < .0001b
2.0
p < .0001b
3.8
p < .0001a
5.5
p < .0001a
8.7
p < .87a
0.0
p < .451a
0.9
p < .005a
4.7
p < .002a
5.5
p < .23a
1.4
p < .003a
7.3
p < .05a
1.9
p < .001a
10.1
p < .14a
1.3
p < .07a
1.6
N/A
0.7
N/A
1.8
N/A
0.5
N/A
0.8
p < .861a
0.4
p < .626a
0.4
N/A
0.6
N/A
0.9
N/A
0.3
N/A
0.7
p < .0001b
2.2
p < .0001b
1.4
N/A
0.4
N/A
0.6
p < .17a
0.8
p < .03a
2.0
p < .001b
1.1
p < .001b
1.5
N/A
0.4
N/A
1.0
P1
P2
99.4
(1.0)
24.0
(1.3)
83.5
(4.4)
225
(26)
P3
1005
87.2
23.3
82.9
134
15.7
(1.0)
2853
(48)
93.9
(2.6)
26.4
(4.0)
74.4
(3.7)
107
(38)
P4
1.5
(0.4)
185
(18)
89.4
(1.4)
24.5
(0.7)
76.8
52
16.9
(0.1)
1638
(1099)
94.3
(0.5)
26.8
(0.5)
75.7
(0.6)
118
(6)
938
Journal of Speech, Language, and Hearing Research Vol. 55 930945 June 2012
BASEPOST
BASEFUP
BASE
POST
FUP
TP
Visual
STAT
ES
Visual
STAT
ES
6.7
(1.4)
860
(146)
95.1
(1.4)
22.6
(4.7)
78.9
(2.4)
132
(38)
7.4
(1.3)
519
(81)
91.2
(5.04)
22.9
(0.5)
78.5
(1.9)
106
(8)
7.3
(0.8)
719
(69)
95.3
(4.2)
21.7
(2.47)
78.4
(2.5)
96
(141)
15.8
(5.3)
1066
(110)
99.7
(3.6)
28.9
(0.2)
86.9
(0.1)
165
(0.7)
N/A
0.5
N/A
0.4
p < .02a
2.3
p < .24a
0.1
p < .20a
2.8
p < .95a
0.1
N/A
0.1
N/A
0.2
N/A
0.2
N/A
0.2
N/A
0.7
N/A
0.9
P5
Max performance ah
Duration (s)
Max performance
F0 range (Hz)
Sustained phonation
dB SPL of 3 longest ahs
Sustained phonation
HNR of 3 longest ahs
Sentence repetition
dB SPL
Sentence repetition
F0 range (Hz)
Note. Mean data across BASE, POST, FUP, and TP are presented with SDs in parentheses. Effect sizes were calculated using a Cohens (1988) d statistic
modified by Busk and Serlin (1992). Direction of arrows and triangles reflects direction of change (up indicating improvements, down indicating
worsening). TP = typically developing peer; Visual = summary of the visual trend analysis; STAT = results of the statistical analysis, if applied; r or 4 = visual
trends with nonoverlapping data points; j or , = visual trends with overlapping data points; = no visual trend in the data, including instances of variable
baseline data; HNR = harmonics-to-noise ratio; N/A = statistics not applied.
a
p value derived through analysis of variance. bp value derived through split-middle binomial probability test.
comments for preference of POST or FUP samples included louder, more variable, not strained, less
monotone, more precise, and less hoarse. Comments when a BASE sample was preferred over FUP
included not yelling, which referred to her FUP repetition of Buy Bobby a puppy.
percentage increases from BASE to FUP were not maintained at 5% and 2%, respectively.
Data for P2
Listener Ratings
Chi-square analysis revealed statistically significant
preferences for both POST over BASE and FUP over
BASE samples for overall loudness, loudness variability, pitch variability, and overall voice quality. Listener
Acoustic Analysis
The measures that had a visual treatment effect and
also were statistically significant included maximum
Table 4. Parents perceptual ratings of their childs voice, speech, and communication most of the time.
Difference scores in parent perceptual ratings
P1
P2
P3
P4
Perceptual characteristic
BP
BF
BP
BF
BP
BF
BP
44
8
20
5
31
17
18
18
23
23
5
2
22
7
2
19
10
71
72
10
39
11
1
32
39
7
20
3
7
3
40
13
2
38
39
5
24
5
5
5
33
4
6
3
22
28
13
17
14
26
23
1
2
15
12
19
0
2
2
13
32
18
31
6
1
2
9
0
22
25
P5
BF
BP
BF
8
14
26
1
6
2
30
11
10
2
Note. B P is the difference score in percentage rating from BASE to POST; B F is the difference score in percentage rating from BASE to FUP.
Positive values reflect improved perceptions; negative values reflect worsening perceptions.
939
Parent Ratings
Perceptual variables with the greatest percentage
increase from BASE to POST included always loud
enough (39%) and never breathy voice (39%). These
percentage increases from BASE to FUP were maintained at 40% and 39%, respectively.
Data for P3
Listener Ratings
Chi-square analysis revealed statistically significant preferences for the POST over BASE samples for
all variables except overall pitch. Preference for loudness
variability and overall voice quality were maintained at
FUP. Listener comments for preference of POST or FUP
samples included more precise, louder, consistent
loudness, less breathy, less strain and strangle,
and clearer voice quality.
Acoustic Analysis
The measure that had a visual treatment effect,
which also was statistically significant, was maximum
sustained ah from BASE to FUP. P3 was 2 SDs
below the mean value of his matched typically developing peer for all variables at BASE, POST, and FUP,
except for HNR of sustained phonation and dB SPL
and F0 range of sentence repetition at all study phases.
Parent Ratings
Perceptual variables with the greatest percentage
increase from BASE to POST included always loud
enough (33%) and never strained voice (28%). These
percentage increases from BASE to FUP were somewhat
maintained at 23% and 19%, respectively.
Data for P4
P4 was unable to repeat entire sentences; therefore,
he repeated the last word of each sentence (puppy,
pot, key) during the sentence repetition task. There
are no data for sentence repetition at BASE 2 due to P4s
unwillingness to perform the task and at FUP 2 due to
microphone failure.
940
Listener Ratings
Chi-square analysis revealed statistically significant preferences for the POST over BASE single-word
repetition samples across all six perceptual variables
rated. Preference for overall loudness, loudness variability, and overall voice quality were maintained at FUP.
Listener comments for preference of POST or FUP samples included louder, more variable, less breathy,
less pressed, easier for child to produce, more precise, and more natural pitch.
Acoustic Analysis
The measures that had a visual treatment effect and
also were statistically significant included maximum
F0 range from BASE to POST and maximum F0 range,
HNR of sustained phonation, and dB SPL of his singleword repetition from BASE to FUP. P4 was two SDs
below the mean value of his matched typically developing peer for all variables at BASE, POST, and FUP, except for maximum F0 range at all phases and dB SPL of
sentences at POST and FUP.
Data for P5
P5 was an untreated comparison participant in this
study. Therefore, the study phases of POST and FUP
were actually additional BASE recordings following
1 month and an additional 6 weeks of no treatment.
Listener Ratings
Chi-square analysis revealed statistically significant preferences for the NP category for all variables
rated from BASE to POST and for overall loudness, overall pitch, and articulatory precision at the FUP session.
Acoustic Analysis
The measure that had a visual treatment effect and
also was statistically significant was a decreasing performance for maximum F0 range from BASE to POST.
P5 was 2 SDs below the mean value of her matched typically developing peer for all variables at BASE, POST,
and FUP except maximum duration at all phases, maximum F0 range at BASE, and dB SPL of sustained phonation at BASE and FUP.
Journal of Speech, Language, and Hearing Research Vol. 55 930945 June 2012
Parent Ratings
The perceptual variable with the greatest percentage increase from BASE to POST was for never hoarse,
scratchy voice (26%). The next largest percentage
change was a decrease (worsening) for the perception
of always speaks so others can understand (30%).
Data for FUP measures from the same parent who completed the BASE and POST ratings were not available.
Discussion
This Phase 1 treatment study examined the effects
of an intensive voice treatment in children with spastic
CP and dysarthria. A nonconcurrent single-subject multiple baseline design with replication across participants
was used. Listeners preferred POST speech samples for
loudness, pitch variability, and voice quality over BASE
speech samples in all four treated participants. Treated
participants made a significant gain in at least one acoustic
measure during maximum performance tasks; improvements were not as frequently observed for speech. All
parents rated an improved percentage for always loud
enough from BASE to POST. Maintenance of these perceptual and acoustic changes at FUP varied across participants and are detailed below. The untreated participant
did not make improvements over the course of the study.
The implications of these findings and directions for future research are discussed below.
Listener Preferences
Listeners preferred POST over BASE speech samples from the sentence repetition task for most of the perceptual variables rated, including five of six for P1, four
of six for P2, five of six for P3, and six of six for singleword repetitions for P4. These listener ratings support
an immediate therapeutic effect of the intervention.
However, these findings should be interpreted with caution
for a number of reasons. First, a lenient p value, which is
consistent with Phase 1 treatment research goals, increased the potential for Type 1 error (treatment effect
identified when none exists). Second, perceptual variables rated in this study were not specifically defined
for the listeners. Although loudness, pitch, voice quality,
and articulatory precision are standard perceptions, listeners may have interpreted these variables differently.
Third, the intrarater reliability for some of the listeners
was borderline acceptable. Finally, the model for sentence
repetition was provided by verbal output from data collectors. Although these data collectors were well trained on
presentation style of the sentences, we cannot eliminate
the possibility that differences in models may have affected the participants speech. Future research using
prerecorded samples, perhaps with animated characters
Acoustic Measures
Overall there were minimal changes in acoustic
measures across study phases, with the exception of P2.
In addition, most participants with CP were still 2 SDs
below the mean performance of their typically developing peers for maximum performance tasks after treatment. These findings are consistent with those of Wit,
Maassen, Gabreels, and Thoonen (1993), who reported
that children with spastic CP were significantly reduced
in their performance envelope compared with typically
developing peers. These findings also are different
from the pattern of results found in adults with PD who
nearly always have large and significant changes in maximum performance tasks, which are trained targets in
LSVT LOUD (Sapir, Ramig, & Fox, 2011). This was
most likely related to the different etiology of the dysarthria in PD versus CP. In PD, the soft voice is often due to
deficits in scaling amplitude of motor output and sensory
perception of normal loudness (Sapir et al., 2011). Targeting vocal loudness in people with PD may cue or access a relatively intact motor system that is capable of
scaling up output to produce normal vocal loudness. In
CP, the soft voice is related to generalized weakness of
respiratory and laryngeal systems, coupled with a lack
of coordination between the respiratory and laryngeal
subsystems (Ansel & Kent, 1992; Workinger & Kent,
1991). Thus, unlike in PD, there may be inherent neuromuscular limitations in CP that impact gains on tasks
targeting the maximum performance envelope.
The acoustic variables measured for sentence repetition (mean vocal SPL and F0 range) may not have been
adequate to capture what the listeners perceived in
speech, when they preferred POST speech samples.
Selecting different or additional acoustic measures in future studies may better capture potential changes in vocal
motor functioning that underlie the listener-perceived
changes. For example, including the entire speech envelope in vocal SPL analysis for sentences versus a focus on
the voiced segments may capture the whispered elements of some participants speech at the end of sentences.
Although the acoustic measures did not document significant changes for most participants, written comments
from listeners often indicated that children spoke with
less effort, less strain and strangle, or less of a
struggle. These perceptions of decreased effort and
strain are not reflected in measures of vocal SPL or F0
range and may have been the basis of listener preference
of POST samples.
Using multiple strategies to detect change in singlesubject designs may provide the best interpretation of
941
942
Journal of Speech, Language, and Hearing Research Vol. 55 930945 June 2012
Summary
Improving speech in children with CP is challenging. Many of these children have a range of medical problems, multiple speech mechanism disorders, and
cognitive deficits, all of which may limit the magnitude
and long-term effects of treatment outcomes. This Phase 1
Acknowledgments
This work was supported, in part, by National Multipurpose Research and Training Center Grant DC-01409 from
the National Institute of Deafness and Other Communication
Disorders and by the Final Project Fund from the Graduate
College at the University of Arizona. This article is based on a
doctoral dissertation completed by Cynthia Marie Fox at the
University of Arizona under the direction of Carol Ann Boliek
and Jeannette Hoit. We thank Jeannette Hoit, Brad Story,
Leslie Tolbert, Becky Farley, and Lorraine Ramig for their
expert advice. Special thanks go to the children and families
who participated in this study.
References
Ansel, B. M., & Kent, R. D. (1992). Acoustic-phonetic contrasts and intelligibility in the dysarthria associated with
mixed cerebral palsy. Journal of Speech and Hearing
Research, 35, 296308.
Barlow, D. H., & Hersen, M. (1984). Single case experimental
designs: Strategies for studying behavior change. New York,
NY: Pergamon.
Beeson, P. M., & Robey, R. R. (2006). Evaluating singlesubject treatment research: Lessons learned from the
aphasia literature. Neuropsychology Review, 16, 161169.
Boeckstyns, M. E., & Backer, M. (1989). Reliability and validity of the evaluation of pain in patients with total knee
replacement. Pain, 38, 2933.
Boersma, P., & Weenick, D. (2002). Praat (Version 4.0.7)
[Computer software]. Retrieved from http://www.praat.org/.
Bower, E., McLellan, D. L., Arney, J., & Campbell, M. J.
(1996). A randomized controlled trial of different intensities
of physiotherapy and different goal-setting procedures in
44 children with cerebral palsy. Developmental Medicine
and Child Neurology, 38, 226237.
Busk, P. L., & Serlin, R. C. (1992). Meta-analysis for singlecase research. In T. Kratochwill & J. R. Levin (Eds.), Singlecase research design and analysis (pp. 187212). Hillsdale,
NJ: Erlbaum.
Butler, C., & Darrah, J. (2001). Effects of neurodevelopmental treatment (NDT) for cerebral palsy: An AACPDM evidence report. Developmental Medicine and Child Neurology,
43, 778790.
Clement, M., & Twitchell, T. E. (1959). Dysarthria in cerebral palsy. Journal of Speech and Hearing Disorders, 24,
118122.
Cohen, J. (1988). Statistical power analysis for the behavioral
sciences (2nd ed.). Hillsdale, NJ: Erlbaum.
943
Lenz, F. A., & Byl, N. N. (1999). Reorganization in the cutaneous core of the human thalamic principal somatic sensory
nucleus (ventral caudal) in patients with dystonia. Journal of
Neurophysiology, 82, 32043212.
Damiano, D. L., Kelly, L. E., & Vaughan, C. L. (1995). Effects of quadriceps femoris muscle strengthening on crouch
gait in children with spastic diplegia. Physical Therapy, 75,
658671.
Fox, C. M., & Ramig, L. (1997). Vocal sound pressure level and
self-perception of speech in men and women with idiopathic
Parkinson disease. American Journal of Speech-Language
Pathology, 6(2), 8594.
Fox, C., Ramig, L., Ciucci, M., Sapir, S., McFarland, D., &
Farley, B. (2006). The science and practice of LSVT/ LOUD:
Neural plasticity-principled approach to treating individuals
with Parkinson disease and other neurological disorders.
Seminars in Speech and Language, 27, 283299.
Sapir, S., Pawlas, A., Ramig, L., Seeley, E., Fox, C., &
Corboy, J. (2001). Effects of intensive phonatory-respiratory
treatment (LSVT) on voice in individuals with multiple
sclerosis. Journal of Medical Speech-Language Pathology,
9(2), 3545.
944
Sapir, S., Ramig, L., & Fox, C. (2011). Intensive voice treatment in Parkinsons disease: Lee Silverman Voice Treatment. Expert Review of Neurotherapeutics, 11, 815830.
Sapir, S., Spielman, J., Countryman, S., Ramig, L., Hinds,
S., Fox, C., & Story, B. (2003). Phonatory and articulatory
changes in ataxic dysarthria following intensive voice
therapy with the LSVT: A single subject study. American
Journal of Speech-Language Pathology, 12, 387399.
Sapir, S., Spielman, J., Ramig, L., Story, B., & Fox, C.
(2007). Effects of intensive voice treatment (LSVT) on vowel
articulation in dysarthric individuals with idiopathic Parkinsons disease: Acoustic and perceptual findings. Journal
of Speech, Language, and Hearing Research, 50, 899912.
Schiffman, S., Reynolds, M. L., & Young, F. W. (1981).
Introduction to multidimensional scaling: Theory, methods,
and applications. New York, NY: Academic Press.
Schindl, M. R., Forstner, C., Kern, H., & Hesse, S. (2000).
Treadmill training with partial body weight support in
Journal of Speech, Language, and Hearing Research Vol. 55 930945 June 2012
Schmidt, R. A., & Fitzpatrick, P. (1996). Dynamical perspective on motor learning. In H. N. Zelaznik (Ed.), Advances
in motor learning and control (pp. 195223). Champaign,
IL: Human Kinetics.
945
Copyright of Journal of Speech, Language & Hearing Research is the property of American Speech-LanguageHearing Association and its content may not be copied or emailed to multiple sites or posted to a listserv
without the copyright holder's express written permission. However, users may print, download, or email
articles for individual use.