Professional Documents
Culture Documents
A Critical Review of Interventions Targeting Prosody
A Critical Review of Interventions Targeting Prosody
COMMENTARY
Abstract
This is a critical review of the research literature pertaining to the treatment of prosody. The studies were located using
electronic databases and were analysed with respect to participants, design, treatment methods, outcome measures, and
findings. Although only 14 studies met the inclusion and exclusion criteria, there was considerable diversity among the
studies with respect to age of participants, type of communication disorder, treatment procedures, and outcomes. Each of
the 14 studies reported at least partial success for outcomes such as increased pitch differential, appropriate production of
affect in sentences, increased sound pressure level or loudness, and appropriate use of stress. The results were interpreted as
supporting the contention that prosody of people with communication impairment can be changed as the result of treatment.
However, several concerns about the rigor of the research were offered. For example, there was limited blinding of data
analysers, inconsistent presentation of reliability data, assignment to treatment groups involved more matching/
Int J Speech Lang Pathol 2009.11:298-304.
Introduction
Method
As Peppé (2009) noted, little empirical evidence
Search
exists detailing the effects of interventions seeking to
improve prosody of speakers with communication Electronic databases (CDIS Dome, CINAHL,
impairments. This dearth exists despite, or perhaps Communication and Mass Media Complete) and
because, problems with prosody extend across the electronic journals of the American Speech-
various types of communication disorders For Language-Hearing Association were searched with
example, descriptions of prosodic disorders often permutations of the search terms: prosody, prosod*,
are listed as a characteristic of a variety of disorders suprasegmental, intonation, rate, pitch, loudness,
such as autism spectrum disorder (Shriberg therapy, intervention, treatment. In addition, the
et al., 2001), childhood apraxia of speech (Shriberg authors checked references of the studies selected for
et al., 2003), and dysarthria (Roth & Worthington, additional potential sources. Only peer-reviewed
2005). intervention studies meeting these four criteria were
It is possible that there is more research pertain- selected for review:
ing to prosody intervention than first meets the
eye. That is, research on prosody intervention could . published in English between 1988 and 2008,
be widely distributed among the literature on . presented evidence of a control group (for
disorders of human communication and not, as group designs) or control condition (for single
usual, located within a single, more accessible, dis- subject designs),
order type. The purpose of this paper is to identify . contained outcomes measuring one or more
and analyse existing research on prosodic interven- prosodic elements, and
tion. It is hoped that this critical review will des- . included participants who were diagnosed as
cribe the extant literature and provide guidance to communicatively impaired. (Interventions for
clinicians and to those wishing to expand the individuals with hearing loss were excluded
research base. from the review.)
Correspondence: Patricia Hargrove, AH 103, Department of Speech, Hearing, and Rehabilitation Services, Minnesota State University, Mankato, Mankato,
MN 56001, USA. Tel: þ1 507 389-1415. Fax: þ1 507 389-2821. E-mail: [email protected]
ISSN 1754-9507 print/ISSN 1754-9515 online ª The Speech Pathology Association of Australia Limited
Published by Informa UK Ltd.
DOI: 10.1080/17549500902969477
Intervention for prosody 299
Overall, over 1050 potential sources were identified. cerebral vascular accident, myotonic dystrophy). The
Review of the abstracts using the criteria reduced the other participants were diagnosed with delays in
number of studies selected for review to fourteen. cognition or unknown origin (one of these partici-
pants was also a second language learner). Descrip-
tions of communication problems also varied with
Review
91% of participants described as dysarthric. Other
The first author developed prototypes of two forms communication problems were labelled as expres-
to analyse the studies selected for review. The forms sive/motor aprosodia, apraxia, specific language
(one for group research, the other for single subject impairment, and selective mutism.
research) then were modified by the reviewers (the Four studies were group designs and the remain-
three authors). The characteristics of the studies that der were single subject studies. Researchers regularly
were reviewed included the type of research, used randomization or counterbalancing to assign or
participant descriptors, internal validity issues (e.g., order treatments. For group studies, there was
similarity of participants before intervention, con- evidence of researchers’ attention to insuring groups
Downloaded from informahealthcare.com by University of California San Francisco on 12/30/14. For personal use only.
effect size correlation, number needed to treat). considerable variation in treatment procedures. The
(Clinical Significance was a categorization by the four most common treatment procedures were visual
reviewer as to whether she would represent the feedback (e.g., SpeechViewer, Visipitch, or hand
treatment as successful if her client were to display cues), imitation, metalinguistics/explanations (de-
identical results following treatment.) The quality scribing how to or when to produce certain prosodic
indicators for single subject studies included (1) elements), and verbal feedback. Only three proce-
Clinical Significance and (2) effect size metric(s). dures (precise articulation, generalization activities,
The effect size metric consisted of the reviewers’ and the encouragement of ‘‘high’’ effort on the part
calculation of the percentage of nonoverlapping data of the speakers/clients) had not been listed in
(PND) using procedures described in Schlosser and Hargrove and McGarr’s (1994) descriptions of
Wendt (2008) and/or a z score which had been prosodic intervention suggesting that the treatment
calculated by the authors of the articles. procedures used in the research were not unique.
The reviewers independently read each study and Moreover, with two exceptions (‘‘high’’ effort and
completed the appropriate form. After the indepen- contrastive stress drill), the prosodic treatment
dent reviews, the reviewers met to discuss their procedures are commonly used in the treatment of
interpretations. Disagreements were resolved by a wide variety of communication disorders such as
consensus. fluency, speech sound disorders, motor speech, and
voice. What made the treatments unique was their
focus on prosodic outcomes and they way the
Results and discussion
researchers combined or packaged the procedures
Table I summaries participant and design character- to create their own treatment program.
istics of the studies. The studies are organized based The researchers selected outcomes associated with
on the aspect of prosody that treatment outcome(s) expressive prosody. Moreover, most (71%) of the
targeted. Some of the studies are listed in multiple outcomes involved speakers’ production of isolated
sections because their outcomes focused on more behaviours (usually a word or sentence) in controlled
than one aspect of prosody (e.g., pitch and loud- contexts such as imitating a sentence or answering a
ness). Overall the studies involved 155 participants question and only 29% of the outcomes could be
(139 with communication disorders and 16 with classified as focusing on prosodic function. Eight
‘‘normal’’ communication skills). The majority (57%) of the studies used instruments to assess
(95%) of the participants were adults. Ten of the outcomes, yielding measurement such as fundamen-
studies involved treatment of adults; four studies tal frequency (F0) and sound pressure level (SPL).
focused on children or adolescents. All researchers claimed at least partial success for
There was considerable variability in the aetiology treatment approaches. It is difficult to make a
type/co-occurring problems of the participants. Most definitive statement about the relative effectiveness
(97%) of the participants experienced neurogenic about the treatment procedures because quality
problems (Parkinson disease, traumatic brain injury, indicators were not consistent in judging direction
Int J Speech Lang Pathol 2009.11:298-304.
Downloaded from informahealthcare.com by University of California San Francisco on 12/30/14. For personal use only.
Preinterv.
Source Featuresa Disorder typeb Evidence typec Assignment similarity Blindingd Reliabilitye
Affective Prosody
Leon et al. (2005) N ¼ 3 CA ¼ 49–57 Gen ¼ 2f, 1m expressive aprosodia (CVA) SS-ABAC Counter-balanced P, T – No A – Yes Intra. - .75 Inter. - .75 Tr. Fid. - No
Rosenbek et al. (2004) N ¼ 3 CA ¼ 19–85 Gen ¼ 2m, 1f expressive aprosodia (CVA) SS-ABAC Randomized P, T – No A – Yes Intra. - .75 Inter. - .79 Tr. Fid. - No
Stringer (1996) N ¼ 1 CA ¼ 36 Gen ¼ F motor aprosodia (TBI) SS-ABC P, T – No A – Yes (test) No
Pitch
P. Hargrove et al.
Bouglé, Ryalls, & Le Dorze (1995) N ¼ 2 CA ¼ 23–28 Gen ¼ 1f, 1m ataxic dysarthria (TBI) SS-AT Counter-balanced No No
Johnson & Pring (1990) N ¼ 16 CA ¼ 59–72 Grps ¼ 3 dysarthria (PD) Groups: treatment; Matched (limited) Unclear P, T – No A – Unclear No
(2 PD; 1 normal) Gen ¼ 5m, no treatment;
1f in PD grps normal
Le Dorze, Dionne, Ryalls, Julien, & N ¼ 1 CA ¼ 74 Gen ¼ f dysarthria (PD) SS-MB No No
Ouellet (1992)
Ramig, Countryman, Thompson, & N ¼ 45 CA ¼ 32–83 Grps ¼ 2 dysarthria (PD) Groups: treatment; Randomized Yes P , T– No A – Unclear Intra. - .78- .94 Inter. - .87- 1.00
Horii (1995) Gen ¼ 12f, 33m comparison Tr. Fid. - No
Stringer (1996) N ¼ 1 CA ¼ 36 Gen ¼ F motor aprosodia (TBI) SS-ABC P, T – No A – Yes (test) No
Loudness
Facon, Sahiri, & Rivière (2008) N ¼ 1 CA ¼ 12 Gen ¼ m selected mutism (developmental SS-CC No No
delay; 2nd language learner)
Johnson & Pring (1990) N ¼ 16 CA ¼ 59–72 Grps ¼ 3 dysarthria (PD) Groups: treatment; Matched (limited) Unclear P, T– No A – Unclear No
(2 PD; 1 normal) Gen ¼ 5m, no treatment;
1f in PD grps normal
Ramig et al. (1995) N ¼ 45 CA ¼ 32–83 Grps ¼ 2 dysarthria (PD) Groups: treatment; Randomized Yes P, T– No A – Unclear Intra. ¼ .78- .94 Inter. ¼ .87- 1.00
Gen ¼ 12f, 33m comparison Tr. Fid. ¼ No
Ramig, Sapir, Fox, & Countryman N ¼ 43 CA ¼ 67–71 Grps – 3 dysarthria (PD) Groups: treatment; Randomized Yes P, T – No A – Unclear No
(2001) (2 PD, 1 normal) no treatment;
Gen ¼ 22f; 21m normal
Sapir, Ramig, Hoyt, N ¼ 35 CA ¼ (mean) 63 & 65 dysarthria (PD) Groups: treatment; Counter-balanced Yes P, T – No A – Yes No
Countryman, O’Brien & Hoehn Grps – 2 Gen ¼ Unk comparison
(2002)
Rate
Le Dorze et al. (1992) N ¼ 1 CA ¼ 74 Gen ¼ f dysarthria (PD) SS-MB P, T, A – No
Thomas-Stonell, McClean, & Hunt N ¼ 3 CA ¼ 5–18 Gen ¼ 1f, 2m dysarthria (2 TBI; SS-MB No
(1991) 1 myotonic dystrophy)
Stress
Dworkin, Abkarian, & Johns (1988) N ¼ 1 CA ¼ 57 Gen ¼ f apraxia (CVA) SS-MB No Intra. ¼ No Inter. ¼ not 590%
Tr. Fid. ¼ No
Hargrove, Roetzel, & Hoodin N ¼ 1 CA ¼ 6 Gen ¼ m SLI (unknown) SS-MB No Intra. ¼ 87% Inter. ¼ 76%
(1989) Tr. Fid. ¼ 98%- 100%
Shea & Tyler (2001) N ¼ 2 CA ¼ 3 Gen ¼ Unknown SLI (unknown) SS-MB No Intra. ¼ No Inter. ¼ .89 Tr. Fid. ¼ No
a
Notes: N ¼ number of participants; CA ¼ chronological age; GEN ¼ gender; f ¼ female; m ¼ male.
b
CVA ¼ cerebrovascular accident; PD ¼ Parkinson disease; SLI ¼ speech language impairment; TBI ¼ traumatic brain injury; ( ) within parentheses indicates etiology.
c
SS ¼ single subject; AT ¼ alternating treatments; CC ¼ changing criterion; MB ¼ multiple baseline.
d
A ¼ analyzer; P ¼ participant; T ¼ therapist.
e
Intra. ¼ intraobserver reliability; inter. ¼ interobserver reliability; Tr. Fid. ¼ treatment fidelity.
Intervention for prosody 301
and/or degree of clinical success. That is, 50% of the prehension outcomes. However, as Peppé (2009)
time the measures of treatment success (i.e., quality had suggested, interventions may not be able to be
indicators) were in agreement but 50% they contra- addressed thoroughly until we have a clearer picture
dicted one another with one quality indicator of how prosodic productions map onto meanings
suggesting the treatment was successful and the and how speakers with impaired communication
other indicating limited or no improvement. We compensate for their prosodic problems.
could not detect a pattern to these disagreements. At
times clinical significance was the more liberal
assessor of improvement and at other times, statis- References
tical significance or effect size was more liberal. Bouglé, F., Ryalls, J., & Le Dorze, G. (1995). Improving
Two sets of procedures were investigated in fundamental frequency modulation in head trauma patients:
multiple studies thus allowing more confidence in A preliminary comparison of speech-language therapy con-
ducted with and without IBM’s SpeechViewer. Folia Phonia-
their findings. Clearly, the strongest evidence of trica et Logopaedica, 47, 24–32.
treatment success is the work of Ramig and her Dworkin, J. P., Abkarian, G. G., & Johns, D. F. (1988). Apraxia of
Downloaded from informahealthcare.com by University of California San Francisco on 12/30/14. For personal use only.
colleagues (Ramig, Countryman, Thompson, & speech: The effectiveness of a treatment regimen. Journal of
Horii, 1995; Ramig, Sapir, Fox, & Countryman, Speech and Hearing Disorders, 53, 280–294.
2001; Sapir, Ramig, Hoyt, Countryman, O’Brien, & Facon, B., Sahiri, S., & Rivière, N. (2008). A controlled-single
case treatment of severe long-term selective mutism in a child
Hoehn, 2002) who used multiple group studies to with mental retardation. Behavior Therapy, 39, 313–321.
explore the effectiveness of focused effort to increase Hargrove, P. M., & McGarr, N. S. (1994). Prosody management of
loudness. In addition, Rosenbek and his associates communication disorders. San Diego, CA: Singular Publishing.
(Leon et al., 2005; Rosenbek et al., 2004) presented Hargrove, P. M., Roetzel, K., & Hoodin, R. (1989). Modifying the
two articles using single subject designs that prosody of a language-impaired child. Language, Speech, and
Hearing Services in Schools, 20, 245–258.
supported the effectiveness of treating expressive Johnson, J. A., & Pring, T. R. (1990). Speech therapy and
aprosodia. Parkinson’s disease: A review and further data. British Journal
Int J Speech Lang Pathol 2009.11:298-304.
of speech: Coefficient of variation ration for childhood AOS. Stringer, A. Y. (1996). Treatment of motor aprosdia with pitch
Clinical Linguistics and Phonetics, 17, 575–795. biofeedback and expression modelling. Brain Injury, 10, 583–
Shriberg, L. D., Paul, R., McSweeny, J., Klin, A., Cohen, D., & 590.
Volkmar, E. (2001). Speech and prosody characteristics of Thomas-Stonell, N., McClean, M., & Hunt, E. (1991). Evalua-
adolescents and adults with high functioning autism and tion of the SpeechViewer computer-based speech training
Asperger syndrome. Journal of Speech, Language, and Hearing system with neurologically impaired individuals. Journal of
Research, 44, 1097–1115. Speech-Language Pathology and Audiology, 15(4), 47–56.
Appendix A
Characteristicsa
Bouglé et al. Method: In Subject-Verb-Object sentences: Clinical Significance:
(1995) . auditory feedback (A) . increase Fo range (SD) . A - yes/range; variable/SD
. visual feedback using . modulate perceived F0 . V- variable/range and SD
SpeechViewer (V) . judges differentiate pre- and
. both interventions involved post-treatment of 1 participant
imitation and self monitoring Effect Size; PND
A/range ¼ 50–100% (ineffective-highly
effective)
A/SD ¼ 75–100% (fairly- highly effective)
V/range ¼ 75–100% (fairly-highly
effective)
Int J Speech Lang Pathol 2009.11:298-304.
(continued)
Intervention for prosody 303
(Continued)
4 (Continued)
Appendix
high effort reading and . increase Fo in reading and more pre and post intervention
speaking tasks. Visual feedback monologues measures than R.
was provided. . increase maximum duration
. LSVT—Targeted increasing . increase loudness and pitch
vocal fold adduction in high variability
effort speaking and reading
tasks. Visual and auditory cues
were provided.
Ramig et al. Method: . increase SPL in sustained Clinical Significance: LSVT more effective
(2001) . see LVST from Ramig et al. phonation, reading, than no treatment.
(1995) monologue, and picture Statistical Significance:
description tasks. LSVT – significantly better than no
treatment for all post-treatment
outcomes (p .05).
Sapir et al. Method: . increase perceived loudness Clinical Significance: LSVT more effective
(2002) . same as Ramig et al. (1995) level 12 months after treatment than R
cessation (FU12) Statistical Significance:
LSVT, but not R, pre and follow up
12 months later data—significantly
different (p .0001).
Shea & Tyler Method: . production of wS words and Clinical Significance: unclear
(2001) . differentiated big/little, phrases Effect Size: PND ¼ 100% (all outcomes)
loud/soft . production of Sw(S) words and
. practiced SwS phrases phrases
. practiced wS phrases
. auditory bombardment
(modelling)
(continued)
304 P. Hargrove et al.
(Continued)
4 (Continued)
Appendix
Notes: aLSVT ¼ Lee Silverman Voice Treatment; P ¼ participant/patient; SwS ¼ strong-weak-strong stress pattern; T ¼ therapist; wS ¼ weak-
strong stress pattern.
Int J Speech Lang Pathol 2009.11:298-304.
b
Fo ¼ fundamental frequency; SD ¼ standard deviation; SPL ¼ sound pressure level; SwS ¼ strong-weak-strong stress pattern; wS ¼ weak-
strong stress pattern.
c
Clinical Significance ¼ a subjective measure indicating whether the reviewers would have claimed improvement if they had collected the
data in treatment; Effect Size ¼ a measure of the magnitude of the change as the result of treatment; NS ¼ not significant; PND ¼ percentage
of nonoverlapping data (PND) using procedures described in Schlosser and Wendt (2008); Statistical Significance ¼ the p value associated
with inferential statistics is equal to or less than .05.