JSLHR

Article

The Effects of Modified Melodic Intonation

Therapy on Nonfluent Aphasia: A Pilot Study
Dwyer Conklyn,a Eric Novak,b Adrienne Boissy,b Francois Bethoux,b and Kamal Chemalib

Objective: Positive results have been reported with melodic ( p = .02), and a significant difference between groups was found
intonation therapy (MIT) in nonfluent aphasia patients with damage for adjusted total score ( p = .02) favoring the treatment group.
to their left-brain speech processes, using the patient’s intact ability The treatment group also showed a significant change in their
to sing to promote functional language. This pilot study sought to responsive subsection scores ( p = .01) when their pre-tests from
determine the immediate effects of introducing modified melodic Visit 1 to Visit 2 were compared, whereas the control group
intonation therapy (MMIT), a modification of MIT, as an early showed no change, suggesting a possible carry-over effect of
intervention in stroke patients presenting with Broca’s aphasia. MIT treatment.
Method: After a randomized controlled single-blind design, Conclusion: This study provides preliminary data supporting
30 acute stroke survivors with nonfluent aphasia were randomly the possible benefits of utilizing MMIT treatment early in the
assigned to receive MIT treatment or no treatment. A pre/post test, recovery of nonfluent aphasia patients.
based on the responsive and repetition subsections of the Western
Aphasia Battery, was developed for this study.
Results: After 1 session, a significant within-subject change Key Words: aphasia, melodic intonation therapy, modified
was observed for the treatment group’s adjusted total score melodic intonation therapy, stroke, speech rehabilitation

A
ccording to estimates by the National Stroke As- of sentence melody, reduced syntactic capabilities, and im-
sociation, approximately 800,000 people have a paired reading and writing. Treatment for Broca’s aphasia
stroke each year, and of these, 25%–35% develop varies, with results often dependent upon the severity of
aphasia (Dickey et al., 2010; National Stroke Association the aphasia and how quickly the patient’s spontaneous
Website, n.d.). One type of aphasia, referred to as either recall returns (Hillis, 2010; Kelly, Brady, & Enderby,
Broca’s or nonfluent aphasia, often occurs as a result of 2010). Treatment can be intense, 5 or more hours per
damage in the left middle cerebral artery territory, directly week, covering several months with results being incon-
or indirectly affecting the speech area commonly referred sistent (Greener, Enderby, Whurr, & Grant, 1998; Kelly
to as Broca’s area. Patients with Broca’s aphasia are gen- et al., 2010).
erally cognitively intact, with mostly preserved com- For almost 200 years, there have been reports of
prehension, but may present with some or all of the patients with Broca’s aphasia who have retained the abil-
following: word-finding difficulty, impaired fluency, loss ity to sing (Marina, Pasqualetti, & Carlomagno, 2007;
Yamadori, Osumi, Masuhara, & Okubo, 1977), which ulti-
a
The Music Settlement, Cleveland, OH mately led to the development of the speech therapy tech-
b
The Cleveland Clinic Foundation nique called melodic intonation therapy (MIT; Albert,
Correspondence to Dwyer Conklyn, who is now at DBC3 Music Sparks, & Helm, 1973; Sparks & Holland, 1976). MIT is
Therapy, LLC, Independence, OH: [email protected] a stepwise treatment that combines melody with words
Eric Novak is now at Washington University School and phrases in an attempt to take advantage of the
of Medicine, St. Louis, MO. patient’s ability to sing to facilitate speech output. MIT
Editor: Anne Smith appears to work in one of two ways. The first is when
Associate Editor: Wolfram Ziegler the damage to the left-brain speech areas is limited and
Received May 2, 2011 the recruitment of right-brain structures assists in the fa-
Revision received October 21, 2011 cilitation of speech processes as left-brain structures re-
Accepted February 24, 2012 gain their function. The second is when the damage to
DOI: 10.1044/1092-4388(2012/11-0105) the left-brain speech areas is severe, or total, and the

Journal of Speech, Language, and Hearing Research • Vol. 55 • 1463–1471 • October 2012 • D American Speech-Language-Hearing Association 1463

Downloaded from: https://pubs.asha.org University of Zagreb on 02/11/2020, Terms of Use: https://pubs.asha.org/pubs/rights_and_permissions

right-brain structures attempt to take control of the facili- shown positive outcomes might be better utilized earlier
tation of speech processes. in the rehabilitation process.
Positron emission tomography (PET) and functional MIT in its current form has potential limitations. The
magnetic resonance imaging (fMRI) studies have high- first is in the use of a small range of pitches, often just two,
lighted specific right-brain structures for singing and usually separated by a minor or major third. Imaging has
speaking, which include those involved in melody genera- shown prosody of speech to be predominantly processed in
tion (Brown, Martinez, & Parsons, 2006; Jeffries, Fritz, & the right hemisphere (Ethofer et al., 2006; Glasser &
Braun, 2003; Ozdemir, Norton, & Schlaug, 2006; Saito, Rilling, 2008; Meyer, Alter, Friederici, Lohmann, & von
Ishii, Yagi, Tatsumi, & Mizusawa, 2006). Imaging has Cramon, 2002; Mitchell, Elliott, Barry, Cruttenden, &
also shown that although there are bi-hemispherical Woodruff, 2003; Riecker, Wildgruber, Dogil, Grodd, &
areas for both speech and melody processing, the speech Ackermann, 2002). By limiting the pitches and rhythm
areas of the right hemisphere appear underdeveloped in a melodic phrase, a therapist fails to fully incorporate
when compared with their left-sided homologues (Schlaug, the intact right-brain structures responsible for prosody
Marchina, & Norton, 2009), and melody is processed pre- and melody. A second potential limitation is in beginning
dominantly in the right hemisphere (Jeffries et al., 2003). treatment with one or two small words or phrases of two
This helps account for stroke patients’ ability to sing de- to three syllables. If a patient demonstrates the ability to
spite speech impairments. sing “Happy Birthday to You” clearly, then he or she has
Aphasia studies using fMRI have shown homologous shown an ability to vocalize and verbalize six syllables in
speech areas in the right hemisphere that become active the form of a phrase. This puts the onus on the therapist
during the subacute phase of rehabilitation, while the left to design their treatment around accessing this ability in
hemisphere structures affected by stroke are relatively a functional manner. These limitations may be determin-
inactive (Heiss, Kessler, Thiel, Ghaemi, & Karbe, 1999; ing factors in leading the authors of MIT to ascertain that
Heiss & Thiel, 2006; Rosen et al., 2000; Saur et al., it takes 75–90 hour-long sessions to see the full benefits of
2006). This right-brain activation coincides with im- this treatment (Norton et al., 2009; Sparks & Holland,
proved language functioning; the greater the intensity 1976); in today’s current health care climate, such extended
of the activation, the greater the improvement. In some therapy may not be feasible.
subjects, after a return to left-sided activation of language A technique that expands upon the original MIT
functioning, the right-sided homologues continued to protocol, modified melodic intonation therapy (MMIT),
demonstrate an increased activation level when com- shifts the onus to the therapist to use his or her judg-
pared with levels in control subjects (Saur et al., 2006). ment in developing individual treatment plans (Thaut,
This begins to shape the picture of how the right hemi- 2005). Although they represent only minor changes, the
sphere may be recruited to assist the left hemisphere in modifications to the original protocol have the potential
language processing after damage to the left-hemisphere to make this treatment more efficient. The first change
speech areas. is that the treating therapists compose and use novel
PET studies have shown increased activity in Broca’s melodic phrases that closely match the prosody of the spo-
area during MIT training that is not present when stroke ken phrases in both pitch and rhythm to take full advan-
patients are trained to speak, rather than sing, the same tage of those processes that occur in the patient’s intact
group of words (Belin et al., 1996). This increased activity right hemisphere. The next modification is the use of
correlates with improved verbal output and lends further full phrases during initial treatment to allow for access
evidence to how recruiting right-brain structures might to those intact areas that enable patients to sing full
aid in left-brain functioning. Diffusion tensor imaging lines of song. Working within the framework that early
studies have shown increased neural fiber output of the intervention is most effective and that the brain may
right-hemisphere arcuate fasciculus that correlates attempt to recruit right brain language structures within
with improved speech output after MIT training (Schlaug the first few weeks after a stroke (Rosen et al., 2000; Saur
et al., 2009). This increased neural output helps to show et al., 2006), introducing MMIT within this time frame
how it may be possible to adapt right-hemisphere areas may help to facilitate this recruitment, leading to faster
for speech and language processing. and more meaningful recovery. This pilot study sought to
To date, MIT has been predominantly utilized and determine the immediate effects of introducing MMIT
recommended later in a patient’s rehabilitation, often as an early treatment in patients with Broca’s aphasia,
6 months or more after the stroke (Norton, Zipse, Marchina, using a randomized controlled design.
& Schlaug, 2009). However, with the current belief that It was hypothesized that (a) Broca’s aphasia patients
most aphasic patients attain 70% of their potential re- on an acute care unit who received MMIT would demon-
covery within the first 90 days after their stroke (Lazar strate a greater positive change in post-test scores after
et al., 2010), one must ask whether a treatment that has one treatment session when compared with a control

1464 Journal of Speech, Language, and Hearing Research • Vol. 55 • 1463–1471 • October 2012

Downloaded from: https://pubs.asha.org University of Zagreb on 02/11/2020, Terms of Use: https://pubs.asha.org/pubs/rights_and_permissions

group without treatment, and (b) that those patients who speech-language pathologists, and two music therapists.
received MMIT would demonstrate greater positive Concerns addressed in this process were complexity,
change from their pretest scores over multiple sessions fatigue, and availability. MMIT appears to address very
when compared with the control group. specific functional speech goals, specifically repetition
and responsiveness, and emphasis was placed on devel-
oping an assessment that would test those areas while
Method also being feasible for use in a wide range of stroke patients
within a 30-minute time frame. The Western Aphasia
Participants Battery has two subtests that were deemed appropriate,
The study was approved by our institution’s institu- one for repetition and one for responsiveness; however,
tional review board. Potential candidates were identified both sections are designed to elicit short answers. Be-
by their acute care team from among patients followed cause of the length of the phrases utilized in MMIT it
on the neurology inpatient units. They were then ap- was decided not to use the exact subtests from the Western
proached by the music therapist, who provided the pa- Aphasia Battery, but instead to design two similar tasks
tient and family an institutional review board–approved that would elicit longer responses.
informational flyer describing the study. Informed con- The responsive section contained the following
sent, both verbal and written, was then obtained from three questions: (a) “When you are thirsty and need a
the patient (when possible) and the primary contact. drink of water, what do you say to the nurse when she
All potential participants were assessed by their treat- comes in?” ( b) “If I come in and introduce myself:
ing neurologist using the National Institutes of Health
‘Hello, my name is _____,’ how do you introduce your-
Stroke Scale (NIHSS), and most potential participants
self ?” and (c) “If you’re sitting here and you realize you
were assessed by a speech-language pathologist prior to
need to urinate or have a bowel movement, what do you
being approached for this study. Patients were considered
say when you press the call button?” Scoring for this sec-
for enrollment if they met the following inclusion criteria:
tion was 0–3 per question, similar to the Western Apha-
18 years of age or older; diagnosis of mild to severe apha-
sia Battery, with the addition of a score of 3 as the
sia (score of 1 or 2 out of 3 on Item 9 on the NIHSS) docu-
questions were designed to elicit longer responses than
mented in the medical records; damage to the left middle
those in the Western Aphasia Battery, making the score
cerebral artery territory documented in the medical re-
range for the responsive section 0–9. The three state-
cords; no previous documented infarcts; any dysarthria
ments in this section were always given first in an at-
noted to be less than their aphasia, as shown on NIHSS
tempt to lessen the likelihood that the participants
Item 10 (i.e., if Item 9 was a 1, then Item 10 had to be a 0);
might remember one of the phrases used in the repeti-
ability to follow commands (e.g., “touch your nose with
tion section.
your left hand,” “hold up two fingers”); ability to sing
at least 25% of the words of “Happy Birthday”; and dem- The repetition section contained the following three
onstrated self-awareness of speech deficits. Exclusion cri- statements: (a) “I need a drink of water,” (b) “Hello, my
teria were as follows: receptive aphasia greater than name is _____________,” and (c) “I have to go to the bath-
expressive aphasia, as noted by the treating neurologist room,” each of which could correspond with the questions
and/or speech language pathologist’s assessment; apha- in the responsive section. Participants were instructed to
sia other than expressive aphasia, Broca’s type; use of a repeat the exact phrase as it was spoken to them. Scoring
tracheotomy collar or ventilator, severe comorbidity that was identical to that used in the Western Aphasia Battery
precluded participation in the study per principal inves- (2 per correct word), making the range for the repetition
tigator’s judgment (e.g., severe cardiac or respiratory section 0–36, for a total possible score of 0–45.
failure), and severe cognitive deficits that precluded in- Each section’s scores were weighted so that they had
formed consent or prevented the participant from follow- equal standing in an adjusted total score that also
ing study procedures. ranged from 0 to 45 (see below under Statistical Analy-
Patients with acquired speech deficits that were sis). Because of their availability and experience in
defined as other than aphasia, for example, dysarthria working with patients with a wide range of neurological
and/or apraxia of speech, were not considered for inclu- deficits, including aphasia, two nursing managers from
sion to this study. All participants identified English as the neurological units where the participants were lo-
their primary language. cated were recruited to administer the pre/post test. The
evaluators, blinded to treatment assignment, gave the
pretest prior to each session and the post-test im-
Speech Assessment (Pre/Post Test) mediately after the session. The evaluators were not
The pre /post test was designed specifically for this present in the room when the treatment or control ses-
study with consultation from three neurologists, two sion was given, and the music therapist, being blinded to

Conklyn et al.: Effects of MMIT on Nonfluent Aphasia 1465

Downloaded from: https://pubs.asha.org University of Zagreb on 02/11/2020, Terms of Use: https://pubs.asha.org/pubs/rights_and_permissions

the test scores until after the post-test was completed for assignment was performed by the music therapist after
each session, was not in the room when the test was enrollment by the nursing manager, who had no prior
administered. knowledge of the ordering of participants.

Procedure
Statistical Analysis
MMIT was administered by a board-certified music
The statistical analysis plan was designed by a biosta-
therapist trained in the technique. Those participants en-
tistician familiar with the study. Descriptive statistics
rolled in the treatment group received a 10- to 15-min
were generated (mean, standard deviation, frequency
music therapy session directly after their pretest, consist-
counts). An adjusted total score was created that equally
ing of the music therapist teaching the participant a
balanced both components (responsive and repetitive).
melodic phrase. The first session consisted only of the
The adjusted total score was calculated as follows: Ad-
first phrase listed above (“I need a drink of water”). The
justed total = (2.5 × responsive) + (0.625 × repetitive). The
music therapist spoke the phrase one time when intro-
adjusted total score takes on values 0–45, just as the orig-
ducing the procedure to the participant, after which the
inal raw total score.
participant only heard the phrase sung. The music ther-
apist modeled the phrase multiple times and then Each primary measure was evaluated at Visits 1, 2,
instructed the participant to sing the phrase. While sing- and 3, separately. For each visit, the change score (post
ing the phrase, the music therapist assisted the partici- minus pre) within group was tested to be different from
pant in tapping the rhythm of the phrase with his or zero by a paired t test. A two-sample t test was used to
her left hand to provide an added cue that is also believed compare the change between groups.
to assist in motor planning for the vocalizing of the phrase All significance tests used a type I error of a = .05.
(Thaut, 2005). The music therapist determined in subse- The expectation was that MMIT would produce imme-
quent sessions whether it was appropriate to introduce diate, positive effects; therefore, tests to evaluate the
an additional phrase. The second session never consisted change were all one-sided, as were tests to compare
of more than the first two phrases, whereas in the third group differences. Higher scores indicated improve-
session the participant had the possibility of learning the ment. Because this is considered a pilot study looking
third phrase. for data trends, no correction for multiple comparison
To control for possible placebo effects in the treat- tests were applied. All analyses were conducted in SAS
ment group, those participants enrolled in the control version 9.2.
group received 10–15 min with the music therapist,
who discussed the participant’s impairment, different
forms of treatment, different outcomes, and various
issues that can result from aphasia, such as depression
Results
and withdrawal. Thirty-nine patients were approached, and 32 (82%)
agreed to participate. Two patients did not meet the in-
Other Descriptive Variables clusion criteria. All patients enrolled participated in at
least one session. There were 30 participants available
and Outcome Measures for analysis, 14 in the control group (“controls”) and 16
The following information was collected: demo- in the treatment group (“treatments”). Descriptive sta-
graphics (age and gender), disease characteristics (type tistics are provided in Table 1. There were no statistically
and location of stroke, date of symptom onset, date of ad- significant differences in baseline measures between
mission, date of initial session), active comorbidities, and treatment and control groups, although there was a
number of speech therapy sessions. trend for younger age in the treatment group. Of the
Along with the pre/post test scores, treatment group 30 participants, 25 received speech assessments from a
data included the number of times a participant was speech-language pathologist, and of those, six received
prompted to sing a melodic phrase, how many times the follow-up treatment sessions.
phrase was completed, and how many partial phrases Some participants had incomplete or missing data
were sung. for responsive, repetitive, and /or total scores. As this
was a small sample pilot study, no attempt to extrapolate
missing values was made. Out of the 14 controls, 10 had
Study Design both pre and post scores at Visit 1, and eight had pre and
The study followed a randomized, controlled single- post scores at Visit 2. For the treatment group, 14 out of
blind design. The randomization table was generated by the 16 had both pre and post scores at Visit 1, and nine
a biostatistician prior to the start of the study. Random had pre and post scores at Visit 2. Only patients who

1466 Journal of Speech, Language, and Hearing Research • Vol. 55 • 1463–1471 • October 2012

Downloaded from: https://pubs.asha.org University of Zagreb on 02/11/2020, Terms of Use: https://pubs.asha.org/pubs/rights_and_permissions

 Table 1. Baseline comparisons of participants.

Control Treatment

Variable n M (SD ) n M (SD ) p

Age (years) 14 66.9 (11.77) 16 56.8 (17.11) .07

10 65.3 (11.01) 14 58.29 (16.68) .22
HB word count 14 10.6 (4.41) 16 11.9 (4.46) .43
10 10.5 (4.50) 14 11.9 (4.73) .46
Days since onseta 14 28.4 (67.84) 16 32.2 (93.42) .90
12 5.75 (3.65) 15 4.07 (2.31) .09
10 11.1 (19.11) 14 10 (23.10) .89
9 5.10 (2.67) 13 3.85 (1.95) .24
Visit 1 pre scores
Total (0–45) 11 17.8 (17.44) 15^ 20.8 (17.29) .66
10 19.3 (17.64) 14 21.4 (17.97) .78
Adjusted total (0–45)b 11 16.3 (17.12) 15^ 18.0 (14.81) .78
10 17.5 (17.51) 14 18.9 (14.91) .83
Responsive (0–9) 11 2.7 (3.47) 16 2.5 (2.80) .85
10 2.9 (3.60) 14 2.8 (2.88) .93
Repetitive (0–36) 11 15.1 (14.27) 15^ 18.1 (15.59) .61
10 16.4 (14.32) 14 19.1 (15.66) .66

Control Treatment

n % n % p

Male gender 9 64 7 43 .29

8 80 6 42
Attempted to respond 11 78 16 100 .08
10 100 14 100

Note. The top rows indicate all participants, the bottom rows are only those included in the Visit 1 analysis. Differences in age, Happy
Birthday (HB) word count, days since onset, and the pretest scores were evaluated with t tests for independent groups (two-sided); for gender
and attempted status, with Fisher’s exact test.
a
Two control participants (257, 65) and one treatment participant (90) were over 60 days from onset; all others (in italics) were within
13 days. bAdjusted total = (responsive score × 2.5) + (repetitive score × 0.625). c One treatment participant was not asked to complete the
repetitive section during the pretest.

completed both components (responsive and repetitive) in increased in the treatment group and decreased slightly
both pre and post assessments were considered in the in the control group.
following analysis. Table 1 provides baseline data for The overall mean (SD) number of days between Visits 1
those participants included in the analysis of Visit 1. and 2 was 1.3 (0.84), 1.2 (.083) for the control group and
Data are not given for Visit 3 due to the small number 1.4 (.88) for the treatment group. Table 3 shows the
of participants (one control, three treatments). change in pretest results from Visit 1 to Visit 2. Only
A significant change from pre- to post-test was found patients with pretest scores for both Visit 1 and Visit 2
in the treatment group at Visit 1 for adjusted total score were considered for this analysis. Significant differences
(change = 5.3, p = .02). No other pre- to post-test changes were found in both the control group and the treatment
were found to be significant within the treatment or con- group. The control group improved in adjusted total score
trol groups (see Table 2). (change = 4.1, p = .03) and repetitive score (change = 6.0,
The difference in change in adjusted total score be- p = .04). The treatment group improved in adjusted total
tween the treatment and control groups was also signifi- score (change = 8.1, p < .01) and responsive score (change =
cant (difference = 6.4, p = .02) at Visit 1, confirming the 1.9, p = .01). Except for repetitive score, the treatment
first hypothesis: that Broca’s aphasia patients receiving group change was always greater than that of the control
MMIT will demonstrate a greater improvement in post- group, but only the responsive score change was found to
test scores after one treatment session. Average scores be significantly greater (difference = 1.7, p = .02), partially

Conklyn et al.: Effects of MMIT on Nonfluent Aphasia 1467

Downloaded from: https://pubs.asha.org University of Zagreb on 02/11/2020, Terms of Use: https://pubs.asha.org/pubs/rights_and_permissions

Table 2. Analysis for Visit 1.

Control Treatment

M (SD ) M (SD )
Difference Effect
Score n Pre Post Changea n Pre Post Changea in changeb sizec

Adjusted total
(0–45)d 10 17.5 (17.5) 16.4 (19.6) –1.1 (5.4) 14 18.9 (14.9) 24.2 (16.1) 5.3 (8.9) 6.4 0.83
Items 1–3 p = .73 p = .02 p = .02
(0–30)d 10 11.1 (12.6) 10.6 (13.6) –0.5 (4.3) 14 12.1 (9.9) 15.6 (10.9) 3.5 (5.1) 4.0
Items 2 and 3 p = .36 p = .01 p = .02
Responsive
(0–9) 10 2.9 (3.6) 2.7 (4.1) –0.2 (0.9) 14 2.8 (2.9) 3.7 (3.4) 0.9 (2.4) 1.1 0.57
Items 1–3 p = .74 p = .08 p = .06
(0–6) 10 1.9 (2.6) 1.8 (2.9) –0.1 (.6) 14 1.6 (1.8) 2.4 (2.2) 0.8 (1.4) 0.9
Items 2 and 3 p = .29 p = .02 p = .02
Repetitive
(0–36) 10 16.4 (14.3) 15.4 (15.4) –1.0 (5.8) 14 19.1 (15.7) 23.9 (14.4) 4.7 (11.1) 5.7 0.62
Items 1–3 p = .70 p = .06 p = .07
(0–24) 10 10.2 (9.8) 9.8 (10.6) –0.4 (5.2) 14 12.7 (10.6) 15.3 (9.9) 2.6 (7.4) 3.0
Items 2–3 p = .40 p = .10 p = .13

a
Change is shown as post minus pre (one-sided t test). bDifference is shown as treatment change minus control change (one-sided t test). cEffect size was
calculated as the absolute value of the mean difference divided by the pooled SD. dAdjusted total = (responsive score × 2.5) + (repetitive score × 0.625).

confirming the second hypothesis: that patients who total score (change = 3.5, p = .01) and responsive score
received MMIT would demonstrate greater continued (change = 0.8, p = .02). The treatment group also showed
positive changes over multiple sessions when compared significant differences when compared with the control
to the control group. group for adjusted total score (difference = 4.0, p = .02)
Post hoc analyses were done to look for possible train- and responsive score (difference = 0.9, p = .02).
ing effects. Table 2 shows analysis of Visit 1 for Items 2 The adjusted means (responsive mean score × 2.5,
and 3. Because Item 1 of the repetition section was the repetitive mean score × 0.625) were examined for the
phrase used during the MMIT training, an argument six individual items at pre- and post-test for Visit 1 and
can be made that positive results on this test item could for pretest for Visit 2. Table 4 demonstrates that the
indicate a training effect, thus biasing the overall results. treatment group improved across all items both
Removal of Item 1 from both sections, repetitive and re- within Visit 1 and from Visit 1 to Visit 2, whereas the
sponsive, and from both groups reveals significant and control group showed some improvements to their rep-
more robust changes in the treatment group for adjusted etition but little to no change in their responsive scores,

Table 3. Change in pre scores from Visit 1 to Visit 2.

Control M (SD ) Treatment M (SD )

Difference Effect
Score n Visit 1 Pre Visit 2 Pre Changea n Visit 1 Pre Visit 2 Pre Changea in changeb sizec

Adjusted total (0–45) 7 10.9 (14.21) 15.0 (16.60) 4.1 (4.83) 8 19.1 (15.41) 27.2 (15.71) 8.1 (6.88) 4.0 0.67
p = .03 p < .01 p = .10
Responsive (0–9) 7 1.4 (2.51) 1.6 (3.36) 0.1 (1.07) 8 2.6 (2.72) 4.5 (3.54) 1.9 (1.96) 1.7 1.08
p = .36 p = .01 p = .02
Repetitive (0–36) 7 11.7 (13.54) 17.7 (16.75) 6.0 (8.08) 8 20.0 (17.50) 25.5 (15.00) 5.5 (12.41) –0.5 0.05
p = .04 p = .12 p = .53

a
Change is shown as Visit 2 pre minus Visit 1 pre (one-sided t test). bDifference is shown as treatment change less control change (one-sided t test). cEffect size
was calculated as the absolute value of the mean difference divided by the pooled SD.

1468 Journal of Speech, Language, and Hearing Research • Vol. 55 • 1463–1471 • October 2012

Downloaded from: https://pubs.asha.org University of Zagreb on 02/11/2020, Terms of Use: https://pubs.asha.org/pubs/rights_and_permissions

providing evidence of a treatment effect versus spon- changes within the treatment group, as well as those be-
taneous speech recall. tween the two groups, were stronger than with the in-
clusion of the first item. Spontaneous recall would be
represented by positive and consistent changes in the con-
trol group from Visit 1 to Visit 2, which was not observed.
Discussion Although far from equivocal, the improvements in the
In comparing MMIT with no treatment in acute stroke treatment group on the items not used during the treat-
patients with nonfluent aphasia, we found significant im- ment period point more toward a treatment effect and
mediate improvements in speech output after one session potential generalization of MMIT treatment rather than
of MMIT training, supporting our first hypothesis. Al- a training effect of the first phrase used during the session.
though both groups showed significant improvements Also of note is the fact that none of the participants,
when their pretest scores from Visit 1 were compared treatment or control, attempted to sing their responses
with those from Visit 2, the control group’s gains were during any pre/post test, lending further argument
entirely from an improvement in their repetition scores, against a training effect. The only time during the treat-
whereas the treatment group showed similar gains in ment session in which any of the phrases were spoken
repetition and significant gains in their responsiveness was at the very beginning of each session when the music
scores, partially supporting our second hypothesis. As therapist said which phrase they were going to start with.
the control group demonstrated a minimal gain in this After that, all repetitions by the music therapist were sung,
area, the above results suggests a carry-over effect of as were all prompted repetitions by the participants.
MMIT in the area most important to patients with non- One would expect at least a small percentage of sung
fluent aphasia, their ability to express themselves. responses to indicate a training effect, for as similar as
In an attempt to differentiate between training ef- the sung phrases may be to actual speech, the process of
fect, treatment effect and /or spontaneous recall, we an- speaking is still different from the process of singing.
alyzed group comparisons after removing the scores Traditional aphasia therapy appears to be most suc-
from the first item of both assessment sections (see cessful when the main brain structures used for speech
Table 2) and compared adjusted means for the individual remain largely intact after the stroke, therefore leading
items of the assessment (see Table 4). A training effect to spontaneous recall, and/or when right brain structures
would mean that with the removal of the item used during are spontaneously recruited (Winhuisen et al., 2005). As
treatment, there should be no differences between the mentioned earlier, when the speech process is impaired,
groups on the items that were not used. However, the improvements in speech output may occur (a) when the
damage to the left-brain speech areas is limited and the
recruitment of right-brain structures assists in the facili-
Table 4. Mean group adjusted scores for each assessment item. tation of speech processes as left-brain structures regain
their function, or (b) when the damage to the left-brain
Treatment Control speech areas is severe, or total, and the right-brain struc-
tures attempt to take control of the facilitation of speech
Visit Visit Visit Visit Visit Visit
processes. MMIT, by its very nature, may assist in this
Item type 1–pre 1–post 2–pre 1–pre 1–post 2–pre
right-brain recruitment process, allowing it to be success-
Responsivea ful in both cases. This is demonstrated by its positive
Item 1 results with patients with varied severities of aphasia.
(0–7.5) 2.500 2.813 3.325 1.785 1.608 1.75 By using training in musical composition, novel, melodic
Item 2 phrases are developed that retain many of the prosodic
(0–7.5) 1.408 2.656 3.600 1.608 1.785 0.750
characteristics of the spoken phrase. This optimizes the
Item 3
right-brain function for prosody while mobilizing the
(0–7.5) 2.344 2.656 4.445 1.965 1.608 0.750
Repetitiveb
brain’s structures for melodic processing and may provide
Item 1 the template necessary for the brain to shift to these alter-
(0–7.5) 3.833 5.000 5.833 2.857 3.214 3.375 nate right-brain language pathways when the more widely
Item 2 used left-brain pathways are damaged. This also allows
(0–6.25) 2.833 3.333 4.167 2.054 2.232 2.250 for a smooth transition from the sung phrase to the spoken
Item 3 phrase later in the treatment process. Introducing this
(0–8.75) 4.583 5.583 6.389 2.500 3.393 2.875 technique early after a person’s stroke may help to make
a the rehabilitation process more efficient and consistent.
Adjusted means for responsive scores = (responsive score × 2.5). bAdjusted
means for repetitive scores = (repetitive score × 0.625). Hospital staff have many demands to meet in serving
the needs of their patients. For example, speech-language

Conklyn et al.: Effects of MMIT on Nonfluent Aphasia 1469

Downloaded from: https://pubs.asha.org University of Zagreb on 02/11/2020, Terms of Use: https://pubs.asha.org/pubs/rights_and_permissions

pathologists carry the dual tasks of performing swallow and suggests that this method has short-term beneficial
evaluations in addition to speech evaluations and treat- effects. When utilized as early stroke intervention in
ment. For various reasons, including relatively short hos- patients with nonfluent aphasia, MMIT demonstrates
pital stays and priorities of acute care treatment, the significant positive results in patients’ overall ability to
immediate concern becomes the swallow evaluations, fol- verbally respond following one session and continued
lowed by initial speech assessments. In this study, 25 of improved verbal output after 24 hr. Follow-up research
the 30 participants received a speech assessment; only is needed to identify MMIT’s long-term potential as a
six of these patients received follow-up visits while still precursor and adjunct to other aphasia therapies.
in the hospital. Follow-up sessions for treatment of
language issues seldom occur until patients have been
discharged to a rehabilitation facility. Utilizing MMIT Acknowledgments
while a patient is still in the hospital can provide an
early intervention geared to improving his or her lan- This research project was internally funded; no specific
guage and communication skills, and potentially make grant from any funding agency in the public, commercial, or
not-for-profit sectors was received or used. We would like
for a smoother transition into rehabilitation.
to thank Lisa Gallagher, Lisa Stellmacher, Angie Hamm,
Patients who develop nonfluent aphasia as a result and Angelia Watley for their hard work and dedication to
of a stroke find that within a short period of time, they this project. We also acknowledge The Music Settlement and
have lost their usual means of communication. Persis- both the Arts and Medicine and Neurologic Institutes for
tent, unsuccessful attempts at speaking lead to increased their support.
frustration, and possibly withdrawal and depression
(Pitts & Sheridan, 2009). Of the 16 participants who re-
ceived MMIT training as part of the treatment group, all References
but one were able to sing at least part of the first phrase,
“I need a drink of water.” Of the remaining 15 partic- Albert, M. L., Sparks, R. W., & Helm, N. A. (1973). Melodic
intonation therapy for aphasia. Archives of Neurology, 29,
ipants who were able to sing at least part of the phrase, 130–131.
12 (75%) sang at least one complete and accurate phrase
Belin, P., Van Eeckhout, P., Zilbovicius, M., Remy, P.,
during the first session. Therefore, providing MMIT Francois, C., Guillaume, S., . . . Samson, Y. (1996). Re-
training early may help to lessen frustration and with- covery from nonfluent aphasia after melodic intonation
drawal by showing the patient that he or she can produce therapy: A PET study. Neurology, 47, 1504–1511.
meaningful and accurate words. Brown, S., Martinez, M. J., & Parsons, L. (2006). Music and
Our protocol allowed for up to five sessions; however, language side by side in the brain: A PET study of the gen-
because of scheduling challenges present in acute hospi- eration of melodies and sentences. European Journal of
Neuroscience, 23, 2791–2803.
tal care and the relatively short hospital stays of our
patients, we were only able to achieve more than two ses- Dickey, L., Kagan, A., Lindsay, M. P., Fang, J., Rowland, A.,
& Black, S. (2010). Incidence and profile of inpatient stroke-
sions with four of our participants, limiting the data col-
induced aphasia in Ontario, Canada. Archives of Physical
lected. Other limitations of this study include a relatively Medicine and Rehabilitation, 91, 196–202.
small sample size and absence of long-term follow-up.
Ethofer, T., Anders, S., Erb, M., Herbert, C., Wiethoff, S.,
Also, we mostly used custom-made outcome measures Kissler, J., . . . Wildgruber, D. (2006). Cerebral pathways in
designed to closely fit our intervention. Finally, we did processing of affective prosody: A dynamic causal modeling
not perform a precise assessment of the brain damage study. NeuroImage, 30, 580–587.
caused by the stroke, or study potential changes in brain Glasser, M. F., & Rilling, J. K. (2008). DTI tractography of the
function underlying the improvement of speech out- human brain’s language pathways. Cerebral Cortex, 18,
put. Further studies comparing MMIT with traditional 2471–2482.
speech therapy that encompass early intervention and Greener, J., Enderby, P., Whurr, R., & Grant, A. (1998).
continue over 2–3 months, include more standard apha- Treatment for aphasia following stroke: Evidence for effec-
sia assessments, include patients at different stages of tiveness. International Journal of Language & Communi-
cation Disorders [Supplement], 33, 158–161.
stroke recovery, and look at brain structures and activa-
tion are warranted. Although input was sought from Heiss, W.-D., Kessler, J., Thiel, A., Ghaemi, M., & Karbe, H.
several neurologists, speech-language pathologists, (1999). Differential capacity of left and right hemispheric
areas for compensation of poststroke aphasia. Annals of
and music therapists, the tool developed and used was Neurology, 45, 430–438.
pilot tested for this study. This instrument, which pur-
Heiss, W.-D., & Thiel, A. (2006). A proposed regional hierar-
ports to more accurately measure the specific effects of chy in recovery of post-stroke aphasia. Brain and Language,
MMIT, needs to be further refined and validated. 98, 118–123.
Our study demonstrates the feasibility of MMIT early Hillis, A. E. (2010). The standard for poststroke aphasia re-
after a stroke in an acute neurology inpatient setting, covery. Stroke, 41, 1316–1317.

1470 Journal of Speech, Language, and Hearing Research • Vol. 55 • 1463–1471 • October 2012

Downloaded from: https://pubs.asha.org University of Zagreb on 02/11/2020, Terms of Use: https://pubs.asha.org/pubs/rights_and_permissions

Jeffries, K. J., Fritz, J. B., & Braun, A. R. (2003). Words in Rosen, H. J., Petersen, S. E., Linenweber, M., Snyder, A. Z.,
melody: An H215O PET study of brain activation during White, D. A., Chapman, L., . . . Corbetta, M. (2000). Neural
singing and speaking. NeuroReport, 14, 749–754. correlates of recovery from aphasia after damage to left in-
Kelly, H., Brady, M. C., & Enderby, P. (2010). Speech and ferior frontal cortex. Neurology, 55, 1883–1894.
language therapy for aphasia following stroke [review]. The Saito, Y., Ishii, K., Yagi, K., Tatsumi, I. F., & Mizusawa, H.
Cochrane Library, 7, 1–170. (2006). Cerebral networks for spontaneous and synchronized
Lazar, R. M., Minzer, B., Antoniello, D., Festa, J. R., singing and speaking. NeuroReport, 17, 1893–1897.
Krakauer, J. W., & Marshall, R. S. (2010). Improvement Saur, D., Lange, R., Baumgaertner, A., Schraknepper, V.,
in aphasia scores after stroke is well predicted by initial Willmes, K., Rijntjes, M., & Weiller, C. (2006). Dynamics of
severity. Stroke, 41, 1485–1488. language reorganization after stroke. Brain, 129, 1371–1384.
Marina, A., Pasqualetti, P., & Carlomagno, S. (2007). Pat- Schlaug, G., Marchina, S., & Norton, A. (2009). Evidence for
terns of language improvement in adults with non-fluent plasticity in white matter tracts of patients with chronic
aphasia after specific therapies. Aphasiology, 21, 164–186. Broca’s aphasia undergoing intense intonation-based speech
Meyer, M., Alter, K., Friederici, A. D., Lohmann, G., & therapy. Annals of the New York Academy of Sciences, 1169,
von Cramon, D. Y. (2002). fMRI reveals brain regions me- 385–394.
diating slow prosodic modulations in spoken sentences. Sparks, R. W., & Holland, A. L. (1976). Method: Melodic in-
Human Brain Mapping, 17, 73–88. tonation therapy for aphasia. Journal of Speech and Hearing
Mitchell, R. L., Elliott, R., Barry, M., Cruttenden, A., & Disorders, 41, 287–297.
Woodruff, P. W. (2003). The neural response to emotional Thaut, M. H. (2005). Rhythm, music and the brain: Scientific
prosody, as revealed by functional magnetic resonance foundations and clinical applications. New York, NY:
imaging. Neuropsychologia, 41, 1410–1421. Routledge.
National Stroke Association. (n.d.). Stroke 101 fact sheet. Winhuisen, L., Thiel, A., Schumacher, B., Kessler, J.,
Retrieved from www.stroke.org/site/DocServer/STROKE_ Rudolf, J., Haupt, W. F., & Heiss, W. D. (2005). Role of the
101_Fact_Sheet.pdf ?docID=4541. contralateral inferior frontal gyrus in recovery of language
Norton, A., Zipse, L., Marchina, S., & Schlaug, G. (2009). function in poststroke aphasia: A combined repetitive tran-
Melodic intonation therapy: Shared insights on how it is scranial magnetic stimulation and positron emission to-
done and why it might help. Annals of the New York Academy mography study. Stroke, 36, 1759–1763.
of Sciences, 1169, 431–436. Yamadori, A., Osumi, Y., Masuhara, S., & Okubo, M.
Ozdemir, E., Norton, A., & Schlaug, G. (2006). Shared and (1977). Preservation of singing in Broca’s aphasia. Journal of
distinct neural correlates of singing and speaking. Neuro- Neurology, Neurosurgery, & Psychiatry, 40, 221–224.
Image, 33, 628–635.
Pitts, B., & Sheridan, S. (2009). Coping with aphasia [Self-
published book]. Cleveland, OH: Authors.
Riecker, A., Wildgruber, D., Dogil, G., Grodd, W., &
Ackermann, H. (2002). Hemispheric lateralization effects
of rhythm implementation during syllable repetitions: An
fMRI study. NeuroImage, 16, 169–176.

Conklyn et al.: Effects of MMIT on Nonfluent Aphasia 1471

Downloaded from: https://pubs.asha.org University of Zagreb on 02/11/2020, Terms of Use: https://pubs.asha.org/pubs/rights_and_permissions