Content uploaded by Vicki Hammen
Author content
All content in this area was uploaded by Vicki Hammen on May 22, 2017
Content may be subject to copyright.
Journal
of
Speech
and
Hearing
Disorders,
Volume
55, 550-560, August
1990
Improved
intelligibility
is
arguably
the
primary
goal
of
speech
intervention
for
individuals
with
severe
dysar-
thria.
Slowing
the
speaking
rate
of
some
dysarthric indi-
viduals
has
a
beneficial
effect
on
overall
understandabil-
ity
of
their
speech
(Rosenbek &
LaPointe,
1985;
Yorkston
&
Beukelman,
1981;
Yorkston,
Beukelman,
&
Bell,
1988).
Many
techniques
have
been
suggested.
Some
rigid
rate
control
techniques
impose
a
one-word-at-a-time
style
upon
the
speaker and
are
usually
reserved
for
the
most
severely involved
speakers.
These
rigid
techniques
often
involve
some
form
of external pacing
such
as
pacing
boards (Helm,
1979)
or
alphabet
supplementation
(Beu-
kelman
&
Yorkston,
1978;
Crow &
Enderby,
1989).
Other
techniques
are
not
as
disruptive
to
speech naturalness.
These
include
devices
that
provide
delayed
auditory
feedback
to
slow
speaking
rates
(Hanson & Metter,
1980,
1983)
or
speaker
training
with
specific
feedback
such
as
acoustic displays
on
oscilloscopic
screens
(Berry
&
Gos-
horn,
1983).
Two
groups
of
dysarthric
individuals
appear
to
benefit
more
than
others
from
a
reduction
in
speaking
rate.
The
first
are
those
with Parkinson's disease
with
hypokinetic
dysarthria whose
habitual
speaking
rate
is
excessively
rapid.
This
is
the
only
type
of
dysarthria
in
which habitual
speaking
rates
are more
rapid
than those
of
normal speak-
ers.
The speech
patterns
of
these
individuals
have
been
likened
to
the
festinating
gait
pattern
also
exhibited
by
some
individuals
with Parkinsonism. In
a
recent
study
involving kinematic,
acoustic,
and
perceptual
analyses,
the
speech
of
individuals
with
Parkinsonism
is
character-
ized
by
a
general
lack
of
articulatory
movement,
reduced
duration
of
vocalic
segments, and
reduced
formant
tran-
sitions
(Forrest, Weismer,
&
Turner,
1989).
The
second
group
of
individuals
who
may
benefit
from
rate
control
are
individuals
with
ataxic
dysarthria. Unlike
individuals
with Parkinson's disease, persons
with
severe
ataxic
dys-
arthria
speak
more slowly
than
normal
individuals.
Their
failure
to
achieve articulatory
targets
is
the result
of
uncoordinated
movements
rather
than
lack
of
movement.
The
purpose
of
this
study
was to
document the
effect
of
controlling
speaking
rate
on
several
perceptual
aspects of
speech,
specifically
sentence
and
phoneme
intelligibility
and naturalness
of
speech.
A
number of
questions
will
be
addressed.
Do
slowed speaking
rates
result
in
improved
sentence
intelligibility
for
hypokinetic
and
ataxic
dysar-
thric speakers?
If
so,
are
some
strategies of
rate
control
more effective
than
others
in
improving
sentence
intelli-
gibility?
Are
there
parallel
changes
in
phoneme
intelligi-
bility
when
speaking
rates
are
slowed? In
other
words,
does
sentence
intelligibility
change
because
speakers
produce
speech
sounds
more
adequately?
Finally,
be-
cause
maintenance
of
speech naturalness
is
also
an
im-
portant
objective of
intervention,
how does slowing
a
speaker's
rate
affect
perceptual
ratings
of naturalness?
METHOD
Subjects
Dysarthric
subjects.
Eight
dysarthric
individuals
whose
habitual
sentence
intelligibility
was
less
than
90%
O
1990,
American
Speech-Language-Hearing
Association
THE
EFFECT
OF
RATE
CONTROL
ON
THE
INTELLIGIBILITY
AND
NATURALNESS
OF
DYSARTHRIC
SPEECH
KATHRYN
M.
YORKSTON
VICKI
L.
HAMMEN
University
of
Washington,
Seattle
DAVID
R.
BEUKELMAN
University
of
Nebraska, Lincoln
CHARLIE
D.
TRAYNOR
University
of
Washington,
Seattle
Speaking
rates
of
individuals
with
severe
ataxic
dysarthria
(n
=
4)
and severe
hypokinetic
dysarthria
(n
=
4)
were
reduced
to
60%
and
80%
of
habitual
rates
using
four
different
pacing strategies
(Additive
Metered,
Additive
Rhythmic,
Cued
Metered,
and
Cued
Rhythmic).
Effects
of
rate control
on
sentence
and
phoneme intelligibility
and
speech naturalness
were
examined.
Sentence
intelligibility
improved
for
both
groups,
with
metered
pacing conditions associated
with
the
largest
improvement
in
scores.
Similar
improvements
as
speaking
rates
were
reduced
were not
seen
for
the
phoneme intelligibility
task;
however,
one must
recognize
that sentence
and
phoneme
intelligibility
tasks
are
different. Slowing
the
rate
of
dysarthric
speakers
did not
have
as
marked
an
impact
on
speech
naturalness
as
it
did
for
normal
speakers
whose
naturalness
decreased
at
slowed
rates.
Metered
rate
control
strategies were associated with
the
lowest
ratings
of naturalness
for
all
subject
groups.
A
potential
explanation
for
the
discrepancies
between
the
findings
for
sentence
and
phoneme intelligibility
is
offered.
KEY
WORDS:
communication,
dysarthria,
speaking
rate,
intelligibility,
naturalness
L~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
I
0022-4677/90/5503-0550$01.00/0
550
YORKSTON
ET
AL.:
Rate
Control
551
(Yorkston,
Beukelman,
&
Traynor,
1984)
served
as
sub-
jects.
All
had
sufficient
language, visual,
and
reading
abilities
to
perform
the
experimental
tasks.
All
were
native
speakers
of English.
Relevant
clinical
information
for
the
dysarthric
subjects
is
presented
in
Table
1.
Note
that
the
subjects
included
6
men
and
2
women with
ages
ranging
from
30 to
70
years.
Judgments regarding
type
of
dysarthria
were
based
on
perceptual
patterns
of
deviant
dimensions
(Darley,
Aronson,
&
Brown,
1975).
Yorkston
and
Beukelman
judged
4
subjects
to
have
hypokinetic
dysarthria,
2
to
have
ataxic
dysarthria,
and
2
to
have
mixed
dysarthria with
a
component
of
ataxic
dysarthria.
Etiologies
included
Parkinson's disease
(n
=
3),
traumatic
brain
injury
(n =
2),
cerebral
palsy
(n =
1),
tumor
resection
(n
=
1),
and
cerebellar
degeneration
(n
=
1).
All
subjects
were
at
least
2
years
postonset.
Normal
subjects.
Four
individuals
with
no
history of
neurologic
disorder
served
as
control subjects
for
the
ratings
of
speech
naturalness.
Control
subjects
were
matched
for age
and
gender
to
dysarthric subjects
H1,
H4,
A2,
and
A4.
Speech
Samples
Sentence
intelligibility.
Sentence
intelligibility
was
measured
using
samples randomly
generated
from
the
Computerized
Intelligibility
of
Dysarthric Speech
(York-
ston
et
al.,
1984).
Because
of
the
length of the
recording
protocol,
an
attempt
was
made
to
reduce
the number
of
stimulus
items
by
recording
1
sentence
from
each
sen-
tence
length
(5
words-15
words).
Thus,
the
total
sentence
sample
for
each
task
was
11
sentences
and
110
words.
A
different random sample
was
generated
and audio
re-
corded
for
each
speaking
condition.
See
Appendix
A
for
an
example
of
a
sentence
intelligibility
sample.
Three
judges
orthographically
transcribed
each audio
recorded
sample.
Scores
are
reported
in
percentage
of
words
cor-
rectly
transcribed,
averaged
across
judges.
Judges
for
the
sentence
and
phoneme intelligibility
samples
were
grad-
uate
students
in
the
Department
of
Speech and Hearing
Sciences who
had
judged
a
series of
training
tapes
of
other
dysarthric
speakers
for
both
the
sentence
and
pho-
neme intelligibility
tasks
prior
to
judging
the
experimen-
tal
sample.
Phoneme
intelligibility.
Phoneme
intelligibility
was
measured
using
a
Phoneme
Identification
Task
(Yorkston,
Beukelman,
Honsinger,
& Mitsuda,
1989;
Yorkston,
Beu-
kelman,
& Traynor,
1988;
Yorkston,
Honsinger,
Beukel-
man, &
Taylor,
1989).
In
this
task,
subjects
read
19
sentences
in
which
words
containing
a
total
of
57
vowels
and
singleton
consonants
were
embedded.
A
different
random sample
was
generated
and
recorded
for
each
speaking condition.
See
Appendix
B
for
an
example.
The
score
sheets
contained
the
sentence
and
word
frames
with
the
target
phoneme
deleted.
Judges
were
asked
to
identify
the
missing
sound.
Scores,
averaged
across
three
judges,
were
reported
as
the
percentage
of consonants
and
vowels
correctly
identified.
Speech
naturalness.
In
addition
to
the
8
dysarthric
speakers,
4
nonimpaired
speakers
served
as
normal
con-
trols
for
this
phase of the
project.
Each
subject
read
a
paragraph
(see
Appendix
C)
at
habitual and
slowed
rates.
The
paragraph
was
constructed
for
this
investigation
and
included
sentences
of
varying
lengths
and
stress
or
into-
national
patterns.
A
three-sentence
segment
from
the
midportion
of the
paragraph
served
as
the
speech
sample
and
was
dubbed
onto
a
judging
tape.
One
judging
tape
was
created
for
each
subject
at
each
rate.
The
judging
tape contained
six
samples: one
segment
was
dubbed
from
the
habitual production,
and
one
from
each rate
control
strategy
for
each
targeted
rate
(80%
or
60%
of
habitual).
Habitual productions,
which
were
included
on
both
tapes,
served
as
reliability
items.
The
presentation
order
of
the individual
speech
samples
was
randomized
within
each tape.
The
order
in
which
the
tapes
were
judged
was
also
randomized. Judges were naive
to
the
recording
conditions
of the
samples
they
were
judging.
Natural
speech
was
defined
for
this
study
as
conven-
tional
in
terms of
intonation,
voice
quality,
rate,
rhythm,
or
intensity
adjustments. Unnatural
or
bizarre
speech,
on
the
other
hand,
was
defined
as
markedly
deviating
from
the
expected
or
conventional
pattern.
Naturalness,
or
lack
of
it,
is
a
multidimensional
phenomenon
whose
features
TABLE
1.
Characteristics of
the
dysarthric
subjects.
Gender
Dysarthria
Years
Subject
Age Type
Etiology
Postonset
H1
M/56
Hypokinetic Parkinson's disease
6
H2
M/65
Hypokinetic Parkinson's disease
6
H3
M/70
Hypokinetic Parkinson's disease
8
H4
F/30
Hypokinetic Cerebral
palsy
with
dystonic
29
posturinga
Alb
F/56
Ataxic
Cerebellar
degeneration
4
A2
M/38
Ataxic/Spastic
Traumatic
brain
injury
(TBI)
3
A3
M/40
Ataxic/Flaccid
Tumor resection
4
A4
M/35
Ataxic
TBI
17
aAlthough
it
is
somewhat
unusual
for
individuals
with cerebral
palsy to
exhibit hypokinetic
dysarthria,
this
subject's
speech
was
characterized
by
rapid
rushes,
inappropriate
silences,
monopitch, monoloudness, and
little
articulatory
excursion.
All
of these
features
are
consistent
with
hypokinetic
dysarthria.
bOnly
sentence
intelligibility and
paragraph
samples
were
obtained
from
this
subject because
a
change
in
medical
status
prevented
completion of
the
entire
protocol.
552
Journal
of
Speech
and
Hearing
Disorders
vary
from
one
dysarthric speaker
to
another.
Ratings
of
naturalness
were
made
using
a
7-point,
equal-appearing
interval
scale
as
described
by
Darley
et
al. (1975).
Be-
cause
the
focus
of
this
project
was
a
comparison
of
a
number of
speaking conditions
within
subjects,
a
single
naturalness
judgment
tape
was
constructed
for
each
sub-
ject
and
rate.
Thus,
judges
were
comparing the
natural-
ness
of
one
subject
across
a
number of
speaking
condi-
tions.
Nine
judges,
who
were generally
familiar
with
dysarthric
speakers
but
not with
the
specific
subjects
in
this investigation,
served
as
raters.
All
judges
were
either
certified
speech-language
pathologists
or
graduate
stu-
dents
in
the
Department
of Speech
and
Hearing
Sci-
ences.
Judges
were
instructed
to
listen
to
the
entire
judging
tape
once
prior
to
rating
each
sample
in
order
to
familiarize
themselves
with
the
range
of samples
on
each
tape.
Scores
reported
are
mean
scale
scores
averaged
across
the
nine
judges.
Rate
Control
Conditions
Speaking
rates
were controlled
via software
developed
for
an
Apple
IIe
microcomputer.
Each speech
sample
(sentence
intelligibility, phoneme
intelligibility,
and
speech naturalness
paragraph)
was
recorded
under
nine
experimental
conditions
(one
habitual
production
and
four
rate
control
strategies
at
two
speaking
rates).
Subjects
were audio-taped
in
a
sound-treated
room
using
a
TEAC
A-2300SX
reel-to-reel
recorder
and
a
Sony
ECM-50PS
electret
condenser
microphone.
The
microphone
was
mounted
in
a
headgear
device
to
keep
the
microphone
at
a
constant
lip-to-microphone
distance
of
10
cm
despite
head
movements
by
some
subjects.
Each recording
ses-
sion
lasted
no more
than
2
hr.
Because
of the
large
number
of
samples
required,
several
recording
sessions
were
usually
needed
to
complete
the
entire
experimental
protocol.
The
majority
of
the
subjects
completed
the
protocol
in
three
2-hr sessions.
Speaking
rates.
Subjects
were
recorded
at
three
speak-
ing
rates:
habitual,
80%,
and
60%
of
habitual. The
habit-
ual
rate
was
defined
in
this
project
as
the
rate
at
which
subjects
read
stimulus
passages
when
given
no
instruc-
tions
regarding speaking
rate.
The
80%
of
habitual
speak-
ing rate
was
chosen
because it
represented
a
rate
that
was
not
excessively
slow
even
for
those
individuals
with
ataxic
dysarthria
whose
habitual
rate
was
less
than
half
of
a
normal
speaking
rate.
Sixty
percent
of
habitual
was
selected
to
represent
a
noticeable
change
in
speaking
rate
even
for
individuals
with
hypokinetic
dysarthria
whose
habitual
rates
exceeded
that of
normal
speakers.
Record-
ings
at
habitual
speaking
rates
were
obtained
first,
fol-
lowed by
paced
samples
at
80%
and
60%
of
habitual
rate.
A
randomly
ordered
recording
sequence
for
the
rate-
controlled
conditions
was
selected
for
each subject.
Rate
control
strategies.
Because
the
effect
of
rate
con-
trol
may
differ
depending
on
the
strategy
used
to
slow
the
subject's
speech, both
presentation
style
and
timing
rela-
tionships were
varied.
Rate
control
strategies
included
both additive
and
cued
presentation
styles,
which
are
described
below.
These
two
styles
represented
different
types of
prompting.
It
was
felt
that the additive
style
would
be
most
useful
for
those
individuals
who
needed
a
rigid,
powerful
prompt
to
control
their
speaking
rate
and
that the
cued
style
would
be
more
appropriate
for
those
for
whom
a
less
rigid strategy
was
adequate.
Rate
control
strategies
also
included
both
metered
and
rhythmic
pac-
ing. Both
were
included
because
many
pacing
tech-
niques,
such
as
metronomes
or
pacing
boards,
encourage
metered
pacing
strategy,
yet
many
clinicians
believe
that
maintaining
the
rhythmic characteristics of speech
is
important
for
naturalness.
Varying
presentation
style
and
timing
relationships
resulted
in
the
following
four
strate-
gies.
1.
Additive
Metered
(AM):
A
speech
sample
was
dis-
played
in
its
entirety
on
a
computer
monitor
so
that
subjects
could
familiarize
themselves
with
the
passage.
Activation
of
a
control
switch by
the
examiner
cleared
the
screen,
and
the
passage
was
presented
by
adding
one
word
at
a
time
at
the
rate
selected
by
the
examiner.
During
this
condition,
each word
was
presented
on
the
screen
with
equal
duration.
2.
Additive
Rhythmic
(AR):
This strategy
was
similar
to
the
Additive
Metered, except
that the
passage
was
pre-
sented
on
the
screen
with
timing
patterns
simulating
normal
speech rather
than
equal
durations
for
each
word.
Rhythmic
characteristics
were
assigned
to
all
experimen-
tal
samples
via
a
program
in
which
a
trained
technician,
following
the
natural rhythm
of
an
audio
recording of
a
nonimpaired
speaker,
tapped
out
the
relative durational
relationships
of
the
sentence. The
program
then
assigned
a
relative
durational
value
to
each
word of
the sentence.
These
values
were
used
to
slow
speaking
rates
while
preserving
a
normal
relative durational pattern.
More
recently,
computer
software
has
been
developed
that
automates
specifying
rhythmic patterns
by
having
a
pro-
gram
assign
relative
durational relationships
by
estimat-
ing
the
number
of syllables in
a
word
(Beukelman,
Yorkston,
&
Tice,
1988).
3.
Cued
Metered
(CM):
In
this
strategy,
the
entire
stimulus
passage
was
presented
on
the
screen
for
subject
familiarization.
Activation
of
a
switch
initiated
the
under-
lining
of
the
passage,
cuing
the production
of
each
word.
Each
word
was
cued
with
an
equal
duration
at
a
rate
selected
by
the
examiner.
4.
Cued
Rhythmic
(CR):
This
strategy
was
similar
to
Cued
Metered, except
that
the
words
of
the
passage were
cued
with
timing patterns that
simulate
normal
speech.
Verification
of
Aspects
of
the
Experiment
Before
proceeding
with
the presentation
and
discus-
sion
of
results,
the
adequacy
of
a
number
of
aspects of
this
experiment
need
to
be
verified.
Of
particular
concern
is
how
well
the
pacing
program
actually
controlled
the
speaking
rates
of
both
the
neurologically impaired
and
nonimpaired
subjects.
Also
of
concern
is
the
reliability
of
the
various
perceptual
judgments.
55
550-560
August
990
YORKSTON
ET
AL.:
Rate
Control
553
Achievement
of
target
speaking rates.
All
of
the
non-
impaired and
6
of the
8
dysarthric
subjects
were
able
to
perform
the
pacing
task
successfully
at
both
80%
and
60%
of
their
habitual speaking
rate.
One
of
the
dysarthric
subjects who
could
be
paced
at
only
one
of
the
reduced
rates
was
a
hypokinetic
dysarthric
speaker
(H3)
whose
habitual
speaking
rate
was
excessively
rapid
(240
words
per
minute).
For
him,
the
80%
condition
was
ineffective
in
changing
his
habitual
rate.
Note
that
80%
of
240
is
192
words
per
minute
(wpm).
Thus,
even
this
"slowed"
rate
was
close
to
the
190
wpm
mean speaking
rate
of
normal
subjects
on
this
task.
The
other
dysarthric subject
who
could
be paced
at
only
a
single
slowed
rate
was
A2,
a
traumatically
brain-injured
individual
with
ataxic
dysar-
thria. He
was
unable
to
tolerate
the
extremely
slow
rate
of
49
wpm,
which
was
the
target
rate
when
his
speech
was
paced
at
60%
of
habitual.
The
durations
of
all
habitual
and
paced
samples
were
measured,
and
actual
speaking
rates
were
computed. The
actual
speaking
rates
for
the sentence
intelligibility
tasks
were
determined
using the
timing
function
within the
software
program
(Yorkston
et
al.,
1984).
The
phoneme
intelligibility
and
paragraph
tasks
were
timed
using
a
stopwatch.
In
this
analysis,
intersentence
pause
times
were
eliminated
for
the sentence
and
phoneme intelligi-
bility
task,
but
not
for
the
paragraph
task.
The three-
sentence
segment
used
for
the
naturalness
judgments
was
timed
to
determine
actual
speaking.
rates
for
that
task.
Speaking
rates
for
each
task
were computed
in
wpm.
Because
speakers with
hypokinetic
dysarthria
tend
to
have
speaking
rates
so
different
from
those with
ataxic
dysarthria,
subjects
were
grouped
by type of
dysarthria.
Figures
1A,
B,
and
1C
illustrate
mean speaking
rates
averaged
across
subjects
for
the
sentence
and
phoneme
intelligibility
and
naturalness
judgment
tasks.
Data
are
presented
for
habitual,
80%,
and
60%
conditions.
For
the
slowed conditions,
data
are
averaged
across
the
four
pacing
strategies
for
each
rate.
Also
illustrated
are the
target
rates
for
each
of the
slowed
rate
conditions.
A
review
of
Figure
1A
suggests
that
the
average
habit-
ual
speaking
rate
for
the
sentence intelligibility
task
produced
by
the
ataxic
group
was
70
wpm
and
by the
hypokinetic
group
was
201
wpm,
as
compared
with
a
normal
speaking
rate
for
this
task
of
190
wpm.
An
exam-
ination
of individual
subject
data
under
rate
control
suggested
that
in
all cases,
the
actual
speaking
rate
achieved
was
within
10%
of
the
target
rate.
The
greatest
discrepancy
between
actual
and target
rate
occurred
in
the hypokinetic
group
at
80%
of
habitual,
where
the
average
target
rate
was
161
wpm
and
the
average
actual
rate
was
143
wpm.
Thus,
when
the
target
rate
was
80%
of
habitual,
the
actual
rate
was
71%
of
habitual.
A
similar
pattern
is
seen
when
reviewing
the
data
from
the
Phoneme
Intelligibility
Task
(Figure
1B)
with
aver-
age
habitual
rates
of
66
wpm
and
167
wpm
for
the
ataxic
and
hypokinetic
group,
respectively.
The
largest
discrep-
ancy
between
actual
and
target
rates
was
found
in
the
ataxic
group
at
60%
of habitual.
Although
the
actual
rate
was
71%
of habitual rather
than
the
targeted
60%,
this
represents
a
difference of
only
7
wpm.
Figure
1C
illus-
FIGURE
1.
Speaking
rates in
words
per
minute
(wpm)
for
the
sentence
intelligibility
(A),
phoneme
intelligibility
(B),
and
naturalness
judgment
(C)
tasks
for ataxic
and
hypokinetic
dysar-
thria
groups at
habitual
speaking
rate
and
when
paced
at
80%
and
60%
of habitual.
Also
included
in
C
are
data
from
the
normal
control
group.
Columns
indicate
actual
speaking
rates.
Target
rates
for
the
slowed conditions
are
shown.
trates
the
actual
and
target
speaking
rates
for
the
natural-
ness
judgment
task.
Note
that
for
the
normal subjects,
actual
speaking
rates
were
slightly
higher
than
target
rate.
For
hypokinetic
subjects, actual
speaking
rates
were
slightly
slower
than
targets;
and
for
ataxic
subjects,
target
and
actual
rates
were
nearly
identical. Thus,
for
all
subject
groups,
speaking
rate
was
effectively
paced
by
the
computer
software.
Reliability.
Because
of
the length
of
the
recording
protocol,
an
effort
was
made
to
reduce
the
impact
of
fatigue
by recording
a
shortened
version
of
Computerized
Assessment of
Intelligibility
of Dysarthric Speech. In this
shortened
version,
one
rather
than
two
sentences
of
each
word
length
was
recorded.
In
order
to
verify
that
the
scores
of
the
shortened
version were similar
to
those
of
554
Journal
of
Speech
and
Hearing Disorders
the
standard
version,
transcripts
of
30
previously
judged
samples
were reviewed.
The
differences
between
scores
generated
from
the shortened
sample
(11
sentences)
and
scores
from
the
total
sample
(22
sentences)
were
com-
puted.
Results
of this
analysis
indicated that the
differ-
ence
between
the
shortened
and
total
sample averaged
3.2%
with
a
standard deviation
of
2.8%.
A
simple t-test
comparison
indicated that the shortened
samples
were
not
different
from
the
total
sample,
with
p
>
.01.
Reliability
of
the
Phoneme
Intelligibility
Task
(York-
ston
et
al.,
1988)
and
Computerized
Assessment of
Intel-
ligibility
of
Dysarthric
Speech
(Yorkston,
Beukelman, &
Traynor,
1984)
is
reported elsewhere.
Because
the
Com-
puterized
Assessment
of
Intelligibility
of
Dysarthric
Speech
was
shortened,
a
measure of the
dispersion
of
judges'
scores
was
obtained.
Average
range
from
the
highest
score
to
the
lowest
across
speakers,
tasks,
and
rates
was
calculated.
Results
indicate
an
average range
of
8.9%
with
a
standard deviation
of
6.4%.
A
similar measure
of
dispersion
of
judges'
scores
was
obtained
for
consonant
and
vowel
intelligibility
for
the phoneme intelligibility
task.
The
average
range
of
judges'
scores
for
consonant
and
vowel
intelligibility
was
9.6%
(SD
=
9.3)
and
17.9%
(SD
=
14.9),
respectively.
In
order
to
examine
the
intrajudge
reliability
of
the
naturalness
judgments,
10%
of the
samples
were
judged
a
second
time.
A
comparison of
first
versus
second
judg-
ments
indicated
that
responses were
the
same
or
one
scaled
score
different
88%,
91%,
and
89%
of
the
time
for
the
ataxic,
hypokinetic,
and
normal groups,
respectively.
A
measure
of
interjudge
reliability
was
obtained
by
com-
puting the
average
standard deviation
across
the
nine
judges
for
each
of
the
samples rated.
Results
of
this
computation
indicate
that the
mean
standard deviation
was 0.97
scale
points.
RESULTS
AND
DISCUSSION
The
Effect
of
Rate
Control
on
Sentence
Intelligibility
Our
first
question
was
"Do
slowed speaking
rates
result
in
improved
sentence
intelligibility
for
the
two groups
of
dysarthric speakers
studied
in
this
project?"
In
order
to
answer
this
question,
sentence
intelligibility
scores
were
averaged
across
judges
and
tasks
for
each
speaking
rate
(habitual
and
80%
and
60%
of
habitual).
Results
of this
analysis
are
presented
in
Figure
2.
Note
that
for
both
groups
mean
sentence
intelligibility increased
as
speaking
rate
was
reduced.
For
the
ataxic
group,
mean
sentence
intelligibility
improved
from
40.9%
at habitual
rates
to
73.7%
at
60%
of
habitual.
For
the
hypokinetic
group, mean
sentence intelligibility
improved
from
60.7%
at habitual
speaking
rates
to 81.2%
at
60%
of habitual.
Data
presented
in
Figure
2
represent
the
mean
scores
across
subjects.
Because
subject
groups
are
diverse,
an
examination
of
individual
performance
was
also
under-
taken
in
an
effort to
verify
the
consistency of
the
overall
FIGURE
2.
Sentence
intelligibility
at
habitual
and slowed
speak-
ing
rates
for
ataxic
and
hypokinetic
dysarthric
speakers.
Data
for
the slowed speaking
rates
were
averaged
across
rate
control
strategies.
trend.
Figure
3
illustrates
the
relationship
between
actual
speaking
rate (averaged
across
all
tasks
for a
given
speak-
ing
rate)
and
sentence
intelligibility.
A
review
of the
figure
suggests
a
highly
consistent
pattern. Sentence
intelligibility
scores
of
all
subjects
increased
as
speaking
rates
decreased.
Thus
far,
our
data
suggest
a
strong
rate
effect
for
sentence intelligibility.
Our
next
question
related
to
whether
certain
rate
control
strategies
resulted
in more
improvement
in
sentence
intelligibility
than
did
others.
In
order
to
identify possible
differences
among
the
vari-
ous
rate control
strategies,
data
from
the
60%
condition
were averaged
across
subjects
in
the
ataxic
and
hypoki-
netic
dysarthria
groups.
Data
from
the
60%
condition
were
selected
for
this
analysis
because
the
largest
changes
in
intelligibility
scores
were noted
for
this
con-
dition.
Results
of
this
analysis are
displayed
in
Figure
4.
Note
that the
metered
conditions
produced
higher
mean
sentence
intelligibility
scores
for
both
groups
of
subjects
than
did
the
rhythmic conditions.
In
order
to
examine
the
consistency
of
this
trend
across
subjects,
the
sentence
intelligibility
scores
for
each
sub-
ject
and speaking
rate
were
rank
ordered
across
the
four
strategies, Additive
Metered
(AM),
Additive
Rhythmic
(AR),
Cued Metered
(CM),
and
Cued
Rhythmic
(CR).
In
other
words,
the
proportion
of
time
that
each
strategy
produced the
highest sentence
intelligibility
score,
the
next
highest
score,
and
so
on
was
computed.
Results
of
this
rank
ordering
are
presented
in
Table
2.
Of
particular
interest
is
the
comparison
of
the
Cued
Metered
with the
Cued
Rhythmic
Strategy.
Note
that
over
half
the
time
(54%),
the
Cued
Metered
strategy
resulted
in
the highest
sentence
intelligibility
scores.
Compare this
with
the
Cued
Rhythmic strategy
that
never
resulted
in
the
high-
est
or
the
next
to
highest
rank.
The
Effect
of
Rate
Control
on
Phoneme
Intelligibility
One
potential
explanation
for
the
improved
sentence
intelligibility
noted
at
reduced
speaking
rates
is
that
55
550-560 August
1990
YORKSTON
ET
AL.:
Rate Control
555
100
-
80
-
60
-
Sentence
Intelllgibllty
(%)
40
-
20
-
0
A4
H3
H4
*
REDUCED
RATES
*
HABITUAL
A1
100
200
300
Speaking
Rate
(wpm)
FIGURE
3.
A
plot
of
sentence intelligibility
as
a
function
of
speaking
rate
for
4
hypokinetic and
4
ataxic
dysarthric
speakers.
Data
for
the
slowed speaking
rates
were
averaged
across
rate
control
strategies.
Readers
should
note
that
for
2
of
the
subjects
(A2
and
H3)
data
are
presented
for
only
one rate
controlled
condition.
slowing
speech
improves
the
speakers'
ability
to
achieve
precise
articulatory
targets.
If
this
is
the
case,
then
one
would expect
to
see
an
improvement
in
phoneme
intelli-
gibility
at
slowed
rates.
In
order
to
explore
the
relation-
ship
between
speaking
rate
and
phoneme intelligibility,
consonant and
vowel
intelligibility
scores
were
averaged
across
judges
(n
=
3)
and
tasks
(n
=
4)
for
each
speaking
rate.
Results
of
this
analysis
are
presented
in
Figures
5A
and
5B.
Note
that
reduced
speaking
rates
were
not
asso-
ciated with
improved consonant
or
vowel
intelligibility.
In
fact,
mean
phoneme intelligibility
scores
change little
as
a
function
of
slowed speaking
rates.
In
no
case
were
mean
consonant
or
vowel
intelligibility
scores
more
than
5%
different
for
habitual
versus
60%
condition.
An exam-
ination of
individual
data
(Figures
6A
and
6B)
suggests
that individual
responses
to
slowed
rate
are
quite
vari-
able.
For
some
subjects,
phoneme
intelligibility
de-
creased
at
slowed speaking
rates.
This
was
the
case
for
Subject
A4
for
both
consonants
and
vowels.
For
others,
phoneme
intelligibility
increased
at
slowed
rates
as
was
the
case
for
Subject
A3.
It
is
interesting
to
note
that
vowel
intelligibility
decreased
more
than
10%
for
over
half
of
the subjects at
slowed
rates.
For
Subject
A4,
vowel
intelligibility
decreased by
more
than
25%
at
the
slowed
FIGURE
4.
Sentence
intelligibility
scores
averaged
across
slowed
rates
for
the
rate control
strategies: Additive
Metered
(AM),
Additive Rhythmic
(AR),
Cued
Metered
(CM),
and
Cued
Rhythmic
(CR).
Also
indicated
are
the
mean
habitual
speaking
rates
for
subjects
in
the
ataxic
and
hypokinetic dysarthria
groups.
····
1
c
556
Journal
of
Speech
and
Hearing
Disorders
100-
80
60
Consonant
Intelllglbility
(%)
40-
20-
0
55
550-560
August
1990
A
H2
A3
*
Habitual
*
Reduced Rate
100
200
300
Speaking
Rate (wpm)
H3
---
H4
H2
A A4
H1
A4
too
200
FIGURE
5.
Consonant
(A)
and
vowel
(B)
intelligibility
at
habitual
and
slowed
speaking
rates
for
ataxic
and
hypokinetic
dysarthric
speakers.
Data
for
the
slowed speaking
rates
were
averaged
across
rate control
strategies.
rate.
A
review
of this
ataxic
dysarthric speaker's
audio
recording and
judges'
responses
suggested
the
judges
heard
more
diphthongs
at
the
slowed
as
compared
to
habitual
rates.
Because
no
consistent
rate
effect
for
pho-
neme
intelligibility
was
found,
further
analysis
of
data
for
the
various control
strategies
was
not
undertaken.
TABLE
2.
Each
cell
indicates
the percentage
of
time
that
a
particular
rate control
strategy
resulted
in
highest
to
lowest
rank
ordering
of
sentence
intelligibility
scores.
Rank
order
of
sentence
intelligibility
scores
Highest
-
-
Lowest
Additive
31% 54%
15%
0
Metered
Additive
Additive
15%
23% 39% 23%
Rhythmic
Strategy
Cued
Metered
54%
31%
8% 8%
Metered
Rhythmic
Speaking
Rate (wpm)
FIGURE
6.
A
plot
of
consonant
(A)
and
vowel
(B)
intelligibility
as
a
function
of
speaking
rate
for
4
hypokinetic
and
3
ataxic
dysarthric
speakers.
Data
for
the
slowed speaking
rates
were
averaged
across
rate control
strategies.
The
Effect
of
Rate
Control
on
Speech
Naturalness
Our next
research
question
relates
to
the
issue of
rate
control
and
speech naturalness.
Specifically, how does
slowing
a
speaker's
rate
influence
naturalness?
In
addi-
tion
to
the
two
groups
of
dysarthric
speakers,
a
normal
control
group
was
added
for
this
task.
Figure
7
illustrates
the
mean
naturalness
ratings
across
judges,
subjects,
and
rate control
strategies.
An
examination of
the
figure
sug-
gests
that the
ataxic
group
was
judged
to
be
the
least
natural,
followed
by
the
hypokinetic
dysarthria
group
and
the
normal
control group.
When speaking
rates
were
controlled,
the
greatest
reduction
in
naturalness
was
seen
in
the
normal group
where mean
naturalness
ratings
decreased
from
1.8
(habitual
rate) to
2.7
(60%
condition).
The
hypokinetic dysarthria
group
showed
a
similar
trend,
but
the
difference
between
habitual
and
60%
was
only
0.2
scaled
points
with mean
naturalness
scores
of
4.3
for
habitual
and
4.5
for
the
60%
condition.
The
greatest
difference
between
habitual
naturalness
and
the
slowed
condition
for
the
ataxic
group
was
also
small
(from
5.0
for
the
80%
condition
to
5.3
for
the habitual
production).
100-
80-
60-
Vowel
Intellilglbllty
(%)
40-
20
-
0
B
*
Reduced Rate
*
Habitua
300
I
· · · ·
· ·
__
O
YORKSTON
ET
AL.:
Rate
Control
Speaking Rate
o
Habitual
e]
80
%
*
60%
Standard
Deviation
Ataxic
Hypokinetic
Normal
Group
FIGURE
7.
Mean ratings
of
speech
naturalness
averaged
across
nine judges
for
ataxic,
hypokinetic,
and
normal groups
at
habitual,
80%
and
60%
of habitual
conditions.
Thus, it would
appear
that
both
dysarthric
groups
were
habitually
quite
unnatural
and
that
slowing
their
rates
did
not
cause
substantial
further
deterioration.
Normal speak-
ers,
who
were,
of
course,
judged
to
be
more
natural
at
habitual
rates
than
were
the
dysarthric
speakers,
were
considered
less
natural
as
their
speaking
rates
were
slowed.
Because
data
related
to
sentence
intelligibility
sug-
gested that the metered
pacing
strategy
produced
the
greatest
changes
in
intelligibility,
differences
in
natural-
ness ratings
as
a
function
of
pacing
strategy
were
also
explored.
Figure
8
illustrates
the
mean
naturalness
rat-
ings
across
judges,
subjects, and
rates
for
the
metered
and
rhythmic pacing
strategies.
Note
that
for all
groups
the
metered
pacing
strategy
was
associated
with
the
least
natural
productions.
This
trend
was
the
most
marked
in
the
normal
group.
CONCLUSIONS
Slowing
the
speaking
rate
of
severely
involved
dysar-
thric
individuals
resulted
in
consistent
improvement
in
sentence
intelligibility
scores
for
the
ataxic
and
hypoki-
netic
groups
studied
in
this
project.
Somewhat
surpris-
ingly,
the
strategies
that
employed
metered
pacing were
associated
with
the
largest
improvement
in
sentence
intelligibility
scores.
These
metered
strategies
tended
to
be
judged
as
slightly
less
natural
than habitual speech
or
speech
produced
with rhythmic
pacing. Slowing
the
Least
7
6
5
Naturalness
4
Strategies
o
Habitual
*
Metered
El
Rhythmic
Standard
Deviation
2
Most
1
Ataxic
Hypokinetic Normal
Group
FIGURE
8.
Mean
naturalness
ratings
averaged
across
nine judges
for
the
ataxic,
hypokinetic,
and
normal groups
at
habitual
rate
and
when
paced with
metered
and rhythmic
strategies.
Least
7
6
5
Naturalness
4
3
2
Most
I
557
558
Journal
of
Speech
and
Hearing
Disorders
speaking
rate
did
not
have
as
marked
an
impact
on
naturalness
for
dysarthric speakers
as
it did
for
normal
speakers.
Presumably,
severely
dysarthric speech
is
al-
ready
so
unnatural that
changing
rate
did
not
have
a
further detrimental
effect.
In the
clinical
setting
a
slight
reduction
in
speech
naturalness
may
be
an
acceptable
price
to
pay
for
a
substantial
improvement
in
intelligibil-
ity.
Obviously,
continued
effort
should
focus
on
identify-
ing
pacing
strategies
that
are effective
in
increasing
speech
intelligibility
and
at
the
same
time
do
not
reduce
speech naturalness.
A
number
of
additional
research
projects
are
needed
if
one wishes
to
apply
these techniques
to
the
improvement
of
speech
intelligibility
in
the
clinical
population.
Studies
of
the
ability
of
dysarthric
speakers
to
learn
to
pace
their
spontaneous
speech production
is
an
important
next step.
Investigations
of
the
impact of
various types
of feedback
on
speakers' ability
to
control
speaking
rate
are
only
beginning
to
appear
in
the
literature
(Hyland
&
Weismer,
1988).
Also
of
critical
interest
to
clinicians
is
the
devel-
opment
of
profiles
of
dysarthric
speakers
who will
benefit
from
rate control
and
who,
perhaps
more
importantly,
will
learn
to
maintain
the
slowed
rates
when
pacing
is
re-
moved.
Other
questions
also
arise
from
the
results
of
the
current
project.
It
is
difficult
and
perhaps impossible
without
further
data
to
explain
why
the sentence intelli-
gibility
of
selected
dysarthric speakers
consistently
im-
proved with slowed
rate
and
phoneme intelligibility
does
not.
The
following
suggestion
is
offered
as
a
possible
but
certainly
not
definitive
explanation.
The
first
consider-
ation
is
the
fact
that
at
habitual
speaking
rates,
sentence
intelligibility
is
lower
than
phoneme
intelligibility
(com-
pare
Figure
2
with
Figures
5A
and
5B).
This
may
be
partially
explained
by
the
fact
that
the
judging
tasks
for
sentence
intelligibility
and
phoneme
intelligibility
are
more
different
from
one
another
than
they
may
initially
appear.
When
performing
the
sentence intelligibility
judging
task,
listeners report that
they
search
for
the
complete meaning
of
the
utterance;
if
they
understand
the
general
content,
then
they
can
fill
in
the
details
and
in
effect
guess at
some
words
that
would otherwise
be
unintelligible.
In
the phoneme intelligibility
task,
the
judges
are
already
provided
with
the
context
or
the
frame
of
the sentence
and
they
need
only
attend
to
one
segment
of
the
utterance,
namely
the
target phoneme.
In
addition,
slowing
the
speaker's
rate
may give
the
listener
extra
processing
time
to
extract
the
general
content
of
meaning
of
the
sentence
and
thus
facilitate "filling in
the
missing
pieces."
This explanation
has
also
been
put
forth
by
Parkhurst
and Levitt
(1978),
who
inserted
short
pauses
into
sentences produced
by
deaf
speakers. Using
multi-
ple
regression
analyses,
pause insertion
was
positively
correlated
with
orthographically
transcribed intelligibil-
ity
scores.
Their
interpretation
of
the
results
included
the
suggestion
that
the
pauses
may
have
allowed
listeners
more
time
to process
distorted
speech.
An
increase
in
the
number
of pauses
that
may
occur
as
the
result
of slowing
the
speaking
rate
may
also
help
to
identify
word
boundaries
for
the
listener.
Again,
re-
searchers
interested
in
deaf
speech
have
begun
to
pursue
this
theory. Maasen
(1986)
investigated
the
effect
of
marking word
boundaries
with
short pauses
on
ortho-
graphically
transcribed
speech
intelligibility
measures.
A
small,
significant
increase
in
intelligibility
was
obtained
for
the pause-inserted
sentences.
Word
boundary
infor-
mation
is
more
likely
to
influence
sentence intelligibility
than
phoneme
intelligibility
when
the listener
is
already
given
the sentence
and
word
frame.
A
number
of
questions that
await
future research
are
raised
by
the
finding
that
phoneme
intelligibility
does
not
consistently
change
as
a
function
of
speaking
rate.
Are
speakers
inserting
more
pauses
into
their
speech
at
the
slowed
rates?
If
so,
are
they
marking
word
boundaries
more
accurately?
Finally,
does
artificially
inserting
pause
time
into
the
unintelligible
utterances
of
dysarthric
speakers improve
intelligibility
scores?
Evidence
that
it
does
would
suggest
that
listener-processing
variables
are
an
important
contributor
to
improved
scores.
The
current
project
focuses
exclusively
on
the perceptual
impact
of
rate
control.
Future
research
is
also
needed
to
examine
the
consequences
of
slowing
the
speaking
rate
on
acous-
tic
segment
characteristics
(e.g.,
voice
onset
time,
stop
closure
duration),
as
well
as
on
the
physiologic
processes
underlying
speech
production.
ACKNOWLEDGMENTS
This project
was
supported
in
part
by
Grant
#5-RO1-NS-
19417-03
from
NINCDS and Grant
#H133B80081
from
the
National
Institute of
Disability
and Rehabilitation
Research,
Department of
Education,
Washington,
DC.
The
authors
wish to
thank
Patricia
Dowden,
Sandy
Hirsch,
Inge
Anema,
Julie
Bar-
ber-Bene,
Marcie
Maloney,
Amy
Smith,
Tamara Coulson,
Kathy
Smith,
and
Marsha
Adams
for
their
assistance.
REFERENCES
BERRY,
W.
R.,
&
GOSHORN,
E.
L. (1983).
Immediate
visual
feedback
in
the
treatment of
ataxic
dysarthria:
A
case
study.
In
W.
R.
Berry
(Ed.),
Clinical
dysarthria
(pp.
253-266).
Boston:
College-Hill
Press.
BEUKELMAN,
D.,
YORKSTON,
K.,
&
TICE,
R.
(1988).
Pacer/Tally.
Tucson,
AZ:
Communication
Skill
Builders.
CROW,
E.,
&
ENDERBY,
P.
(1989).
The
effects
of
an
alphabet
chart
on
the
speaking
rate
and
intelligibility
of speakers with
dysarthria.
In
K. M.
Yorkston
&
D.
R.
Beukelman
(Eds.)
Re-
cent
advances
in
clinical
dysarthria
(pp.
99-108).
Boston:
College-Hill
Press.
DARLEY,
F.,
ARONSON,
A.,
&
BROWN,
J.
(1975).
Motor
speech
disorders.
Philadelphia:
W.
B.
Saunders Company.
FORREST, K.,
WEISMER,
G.,
&
TURNER,
G.
S.
(1989).
Kinematic,
acoustic,
and
perceptual
analyses
of
connected
speech
pro-
duced
by
Parkinsonian
and
normal geriatric
adults.
Journal
of
the Acoustic
Society
of
America,
85,
2608-2622.
HANSON,
W.,
&
METTER,
E.
(1980).
DAF
as
instrumental
treat-
ment
of
dysarthria in
progressive
supranuclear
palsy:
A
case
report.
Journal
of
Speech
and Hearing
Disorders,
45,
268-276.
HANSON,
W.,
&
METTER,
E.
(1983).
DAF
speech
rate
modifica-
tion
in
Parkinson's
disease:
A
report
of
two cases.
In
W.
Berry
(Ed.),
Clinical
dysarthria
(pp.
231-252).
Boston:
College-Hill
Press.
HELM,
N.
E.
(1979).
Management
of
palilalia
with
a
pacing
board.
Journal
of
Speech
and
Hearing
Disorders,
44, 350-353.
55
550-560
August
990
YORKSTON
ET
AL.:
Rate
Control
559
HYLAND,
J.
D.,
&
WEISMER,
G.
(1988).
The
effects
of
three
feedback modes
on
the ability
of
normal
geriatric
individuals
to
match
speaking
rate.
Journal
of
Speech
and
Hearing
Disor-
ders,
53, 271-279.
MAASEN,
B.
(1986).
Marking
word
boundaries
to
improve
the
intelligibility
of
the
speech
of
the
deaf.
Journal
of
Speech
and
Hearing
Research,
29,
227-230.
PARKHURST,
B.
G.,
&
LEVITT,
H.
(1978).
The
effect
of
selected
prosodic
errors
on
the intelligibility of
deaf
speech.
Journal
of
Communication
Disorders,
11,
249-256.
ROSENBEK,
J.
C.,
&
LAPOINTE,
L.
L.
(1985).
The
dysarthrias:
Description,
diagnosis,
and
treatment.
In
D.
F.
Johns
(Ed.),
Clinical
management
of
neurogenic
communication
disorders
(pp.
97-152).
Boston:
Little,
Brown
&
Company.
YORKSTON,
K.,
&
BEUKELMAN,
D.
(1981).
Ataxic
dysarthria:
Treatment
sequences based
on
intelligibility
and
prosodic
considerations.
Journal
of
Speech
and
Hearing
Disorders,
46,
398-404.
YORKSTON,
K.,
BEUKELMAN,
D.,
&
BELL,
K.
(1988).
Clinical
management
of
dysarthric
speakers.
Boston:
College-Hill
Press.
YORKSTON,
K.,
BEUKELMAN,
D.,
HONSINGER,
M.
J.,
&
MITSUDA,
P.
A.
(1989).
Perceived
articulatory adequacy
in
velopharyn-
geal
function
in
dysarthric
speakers.
Archives
of
Physical
and
Rehabilitation
Medicine,
70,
313-317.
YORKSTON,
K.,
BEUKELMAN,
D.,
&
TRAYNOR,
C.
(1984).
Com-
puterized
assessment
of
intelligibility
of
dysarthric
speech.
Austin,
TX:
Pro-ed.
YORKSTON,
K.,
BEUKELMAN,
D.,
&
TRAYNOR,
C. (1988).
Articu-
latory
adequacy
in
dysarthric
speakers:
A
comparison
of judg-
ing
formats.
Journal
of
Communication Disorders,
21,
351-
361.
YORKSTON,
K.,
HONSINGER,
M.,
BEUKELMAN,
D.,
&
TAYLOR,
T.
(1989).
The
effects
of
palatal
lift fitting
on
perceived
articula-
tory
adequacy
of
dysarthric
speakers.
In
K.
Yorkston
&
D.
Beukelman
(Eds.),
Recent
advances
in
dysarthria
(pp.
85-98).
Boston:
College-Hill
Press.
Received
September
15,
1989
Accepted
December
6,
1989
Requests
for
reprints
should
be
sent
to
Kathryn
M.
Yorkston,
Ph.D.,
Department of
Rehabilitation
Medicine, University
of
Washington,
RJ-30,
Seattle,
WA
98195.
560
Journal
of
Speech
and
Hearing
Disorders
APPENDIX
A
Sentence
intelligibility
sample randomly
generated
from
Yorkston,
Beukelman,
and
Traynor
(1984)
Word
length
Sentence
5
They
will
make
many friends.
6
We
all
sat
down
and
relaxed.
7
They had
no
natural
interest
in
sports.
8
Night
after
night,
they
received
annoying
phone
calls.
9 I
never
worried about
being
in
someone
else's
way.
10
I
typed the letter
and
put
it
on his
desk.
11
As
the
day flicked
by,
he
asked
a
thousand
crucial
questions.
About
that
time,
two
young
men
were
painting
my
neighbor's
house
12
white.
13
Her
position
is
basically
the
same
as
it
was
a
few
weeks
ago.
14
No
one
will
ever
play
what
you
could
consider
a
perfect
game
of golf.
15
He
was
so
good
at
it
that
I
urged
him
to
have his
recipes
printed.
APPENDIX
B
Phoneme Identification
Task
Sample
(Yorkston,
Beukelman,
&
TRAYNOR,
1988)
1. "MADGE"
is
between
"LOAN"
and
"THOUGHT."
2.
"MOTH"
and
"RUNG"
are
first
and
"MAN"
is
last.
3.
He
said "MUM";
she
said
"LOVE"
or
"WHO'LL."
4.
"MASH"
and
"HAM"
come
before
"BAKE."
5.
I
thought
it
was
"NEAT"
or
"BATHE";
you
thought
"RAIN."
6.
Spell
the
words
"SHOWED," "TOM,"
and
"VAT."
7.
Spell
"PAN,"
then
say
"FUEL"
and
"SHOULD."
8.
"HEAL"
is
before "TAUGHT"
or
"LOAF."
9.
"THERE"
and
"HOWL"
were
correct;
"FAN"
was
spelled
wrong.
10.
Remember the
words
"GAIN,"
"JAN,"
and
"MAIN."
11.
First
come
"MOSS"
and
"LUG"; later
comes
"LOB."
12.
I
said
"YEAR"
and
"MEN" not
"MAP."
13.
Write
the
words
"ZONE,"
"SEAT,"
and
"LOT."
14.
"LOWELL,"
"WIN,"
and "DANE"
are
the
words
he spelled.
15.
Say
"BOY'
and
"MATCH,"
then
say
"MALE."
16.
"BAN"
is
after
"HAL"
and
"WILL."
17.
You
said,
"MALL";
I
said,
"LEARN"
and
"MOD."
18.
Read
the
words
"SHAM"
and "MAZE";
say
the
word
"MOM."
19.
He
spelled
"CHIN";
I
spelled
"CANE" and
"BY."
Note.
Target
phonemes
are
underlined.
APPENDIX
C
Mount
Rainier
Passage
Believe
me,
my
goal
is
not
to
perfect
your
knowledge
of
nature
or
any
of
its
attributes,
but
let's
record
the
story
of
Sam's
first
day
on
the
mountain.
Unbelievably,
this
was
the
first
time
that
he
had
seen
Mt.
Rainier.
It
was a
perfect
day.
I
didn't
know
what
to
show
him
first.
"Show
Sam
some
snow," someone
said.
I
pointed
to
the
mountain
covered with
snow,
but
someone
said,
"Don't
try
to
show
him
all
that
snow, show
Sam
some
snow."
I
picked
up
a
handful
of
snow
and
began
to
show
everyone,
but
they
interrupted
again
and
said,
"Show
Sam
some
snow."
Pretending
to
be
angry
I
questioned,
"Show
Sam
some
snow?"
What
a
ridiculous idea.
55
550-560
August
990
1990;55;550-560 J Speech Hear Disord
Traynor
Kathryn M. Yorkston, Vicki L. Hammen, David R. Beukelman, and Charlie D.
Speech
The Effect of Rate Control on the Intelligibility and Naturalness of Dysarthric
http://jshd.asha.org/cgi/content/abstract/55/3/550#otherarticles
free at:
This article has been cited by 3 HighWire-hosted article(s) which you can access for
This information is current as of October 18, 2011
http://jshd.asha.org/cgi/content/abstract/55/3/550
located on the World Wide Web at:
This article, along with updated information and services, is
