Poster Presentation for the ASHA
National Convention in New Orleans, LA (November, 2001)
The Effects of
Articulatory Accuracy on Prosody in Children with Developmental Apraxia of
Speech
Amy Skinder-Meredith,
Ph.D., Carol Stoel-Gammon, Ph.D., Richard Wright, Ph.D., and Edythe Strand,
Ph.D.
Abstract
The following study examines the interaction between segmental accuracy and prosody in 10 children, ages 4-9 years, with Developmental Apraxia of Speech. Stimuli include nonsense words that vary in phonetic complexity. Duration, Fo, and amplitude measures are implemented to examine if achieving articulatory accuracy negatively impacts prosody.
BACKGROUND
v Many researchers have listed
prosodic disturbances as a characteristic of Developmental Apraxia of Speech
(DAS) (Crary, 1993; Robin, Hall, & Jordan 1987; Rosenbek & Wertz, 1972;
Velleman & K. Strand, 1994).
v Researchers have debated that
prosodic errors in apraxic speech may be due to an attempt to compensate for
severe speech production problems (Duffy, 1995; Marquardt & Sussman, 1991;
Velleman & K. Strand, 1994).
v Rationale for a
segmental-suprasegmental relationship
o
A
relationship between prosody and articulation has been noted clinically for
years (Rosenbek & Wertz, 1972; Velleman & Shriberg, 1999; Yoss & Darley,
1974).
o
Neuromotor
control studies (Nelson, 1983), which have examined the ‘trade-off’
between speed and accuracy of movement, also support this potential
relationship.
o
Children
with DAS experience difficulty with accurate speech production at normal rate
(Rosenthal, 1994).
o
Children
with DAS have been observed to incorporate a target phoneme at slower than
normal speech rate but not able to produce it at normal speech rate (Yoss &
Darley, 1974).
v The present study further
investigates the perceptual results from a previous study, which examined the
effects of phonetic complexity and stress patterns on segmental and lexical
stress errors (Skinder, 2000).
v Results of the prior study
indicated a relationship between accuracy of segments and rate of speech, where
decreased rate correlated with increased accuracy.
PURPOSE
v The goal of the present study
is to further examine this potential segmental-suprasegmental relationship by
comparing acoustic measures related to prosody [duration, fundamental frequency
(Fo) and amplitude (dB SPL)] of words produced correctly (i.e.,
words produced with all the correct segments) to words produced incorrectly
(i.e., incorrect segments, including omitted segments).
v A subset of data from
Skinder’s (2000) previous study was used for this analysis
HYPOTHESES
v
It
is predicted that words produced correctly will have required increased
articulatory effort, which will thus have a negative impact on prosody.
v
Peak
Fo and mean Fo differences between stressed and unstressed
syllables will be less for words produced correctly than words produced
incorrectly.
v
Amplitude
differences between stressed and unstressed syllables will be less for words
produced correctly than words produced incorrectly.
v
Total
word durations will be longer in words produced correctly than words produced
incorrectly
METHODS
Subjects:
v Participants include 10 children referred with DAS ages 4;10 to 9;11 years of age.
v Local SLPs were sent a letter asking for children who present with the following characteristics (Davis, Jakielski & Marquardt, 1998):
- limited consonant and vowel repertoire,
- frequent omission errors,
- high incidence of vowel errors,
- inconsistent articulation errors,
- altered suprasegmental characteristics,
- increased errors on longer units of speech output,
- significant difficulty imitating words and phrases,
- predominant use of simple syllable shapes,
- impaired volitional oral movements,
- expressive skills less than receptive skills, and
- reduced diadochokinetic rates.
v Inclusionary criteria: Participants exhibited at least eight of the eleven characteristics previously listed. At least one of these characteristics included one of the three speech characteristics suggested by Davis and colleagues (1998) as being differential characteristics for DAS:
- altered suprasegmental characteristics
- inconsistent articulation errors
- vowel errors
v Additional criteria included (a) normal hearing (b) Standard American dialect; (c) normal receptive language skills and (d) a normal structural functional examination.
v Exclusionary criteria (a) previous acquired neuropathology, disease, or injury; (b) use of neuroleptic drugs; and (c ) cognitive delays.
Children
were split into two levels of impairment groups: severe and mild-moderate. Placement was based on
speech intelligibility, standardized articulation measures, and percent
consonants correct (PCC) scores
Stimuli : The following words were elicited in the carrier phrase, “I can say ______________” six times. Stimuli included nonsense words with three levels of phonetic complexity (easy, moderate, difficult) and 2 levels of lexical stress pattern [trochaic (S-W) and iambic (W-S)].
Acoustic Analysis
Tokens were digitized and analyzed with
Multi-Speech on a CSL 4400
For each stressed and unstressed syllable the
following was measured:
Peak fundamental frequency (Hz): Highest point on the fo
contour in the voiced segment of the syllable.
Mean fundamental frequency (Hz): Mean fo contour
of the voiced segment of the syllable.
Peak amplitude (dB): Highest point in the voiced segment of
the syllable on the amplitude trace.
Mean amplitude (dB): Mean dB of the voiced
segment of the syllable.
Total word duration (ms): Offset of the word –
onset of the word
Data
Reduction:
Tokens had been perceptually judged in the prior study as segmentally correct and incorrect by graduate student clinicians
For total word durations, 104
tokens were used to compare durations of correct to incorrect tokens. Comparisons were made for each
child. I.e. Subject #1’s
incorrect production of ‘stroiful’ to his correct production of
‘stroiful’.
For fundamental frequency and amplitude comparisons, correct and incorrect productions of each token were compared, regardless of the subject. However, most easy word comparisons were from children in the severe group and most moderate and difficult word comparisons were from children in the mild-moderate group.
RESULTS
Hypotheses were supported only on some measures depending on the phonetic complexity and stress pattern.
Figure 1a. Average differences of peak and mean fundamental
frequency between stressed and unstressed syllables for words with easy
phonetic complexity. Words
produced correctly are compared to words produced incorrectly for both trochaic
and iambic stress patterns.
Figure 1b. Average
differences of peak and mean dB between stressed and unstressed syllables for
words with easy phonetic complexity.
Words produced correctly are compared to words produced incorrectly for
both trochaic and iambic stress patterns.
v Easy trochees: Both pitch
(Hz) and amplitude (dB) measures showed greater differences for incorrect than
correct productions.
v Easy iambs: Peak Hz difference was the only measure that showed a higher difference for incorrect production.
Figure 2a. Average differences of peak and mean fundamental frequency between stressed and unstressed syllables for words with moderate phonetic complexity. Words produced correctly are compared to words produced incorrectly for both trochaic and iambic stress patterns.
Figure 2b.
Average differences of peak and mean dB between stressed and unstressed
syllables for words with moderate phonetic complexity. Words produced correctly are compared
to words produced incorrectly for both trochaic and iambic stress patterns
v Moderate trochees: Peak Hz
and peak dB differences were higher
v Moderate iambs: only peak dB
difference was higher
Figure 3a. Average differences of peak and mean fundamental frequency between stressed and unstressed syllables for words with difficult phonetic complexity. Words produced correctly are compared to words produced incorrectly for both trochaic and iambic stress patterns.
Figure 3b. Average differences of peak and mean dB between
stressed and unstressed syllables for words with difficult phonetic
complexity. Words produced
correctly are compared to words produced incorrectly for both trochaic and
iambic stress patterns.
v Difficult trochees: only
maximum Hz difference was higher
v Difficult iambs: Peak and
mean Hz differences were both higher for incorrect productions
v In general, standard
deviations were very high for all of these averages (see appendix)
Total word duration results were influenced by
stress pattern and level of impairment.
Average Total Word Duration Differences
between Correct and Incorrect Productions ms
|
Severe |
trochees |
iambs |
|
avg |
-3ms |
-148.4ms |
|
stdev |
50ms |
141.4 ms |
|
Mild-Moderate |
|
|
|
avg |
130.2ms |
-9ms |
|
stdev |
177.7ms |
363.2ms |
Number
and type of incidences where one category of words was produced with longer
word durations than the other
Severe
Iambs 3/4 correct productions longer
than incorrect productions
Trochees 5/5 incorrect productions longer
than correct productions
Mild to Moderate
Iambs 6/6 correct
productions longer than incorrect productions
Trochees 6/9 incorrect productions longer
than correct productions
v Trochees: Correct productions
were longer in more instances than incorrect production.
v Iambs: Incorrect productions
were longer in more instances than correct productions.
v These patterns appear
stronger in the children in the severe group than the mild-moderate group.
DISCUSSION
v Although the compared average
values are interesting, it is important to note that the standard deviations
were very high.
v High standard deviations
indicate high inter-subject variability.
v Further analysis of the
individual children may lead to more interesting findings.
v The segmental-suprasegmental
relationship is revealed in different ways that the current studies methods
were not always sensitive to.
v Different case scenarios
v Children who struggled while
saying the word incorrectly, thus increasing total word duration and decreasing
Hz and dB differences
v Children who say words
correctly with minimal effort, keeping suprasegmental aspects in tact.
v Children who struggle to say
the word correctly and do so at the expense of suprasegmental aspects
v Children who say the word
incorrectly and don’t struggle and keep suprasegmental aspects in tact.
v The present study worked off
of a hypothesis based on the latter two scenarios, but not the first two.
v To better investigate the
relationship between articulation and prosody, one needs to take into account
ways to better measure articulatory effort.
REFERENCES:
Crary, M. (1993). Developmental Motor Speech Disorders, Singular Publishing
Group, Inc., San Diego, CA.
Davis, B., Jakielski, K., & Marquardt, T. (1998). Developmental apraxia of speech: determiners of differential diagnosis. Clinical Linguistics & Phonetics, 12, 25-45.
Duffy, J. R., (1995). Motor Speech Disorders: Substrates, Differential Diagnosis, and Management, Mosby, St. Louis, MO. pp. 259-281.
Marquardt, T. & Sussman, H. (1991). Developmental apraxia of speech: theory and
practice. In D. Vogel and M. Cannito (Eds. ) Treating Disordered Speech Motor Control.
Texas, Pro-ed. 341-390.
Nelson, W. L. (1983). Physical principles for economies of skilled movement. Biological Cybernetics, 46(2), 135-137.
Robin, D. A., Hall, P.K., & Jordan, L. S. (1987). Prosodic impairment in developmental verbal apraxia. Presentation to the American Speech-Language-Hearing Association, New Orleans.
Rosenbek, J., & Wetz, R. T. (1972). A
review of 50 cases of developmental apraxia of
speech. Language, Speech and Hearing Services in Schools, 3, 23-33.
Rosenthal, J. B. (1994). Rate control therapy for
developmental apraxia of speech. Clinics
in Communication Disorders, 3, 190-200.
Skinder, A. (2000). The relationship of prosodic and articulatory errors produced by children with developmental apraxia of speech. Unpublished dissertation, University of Washington, Seattle.
Rosenbek, J. & Wertz, R. (1972). A review of 50 cases of developmental apraxic speech. Journal of Speech and Hearing Research, 26, 231-249.
Velleman, S. & Shriberg, L. (1999). Metrical Analysis of the Speech of Children with suspected developmental apraxia of speech. Journal of Speech, Language, and Hearing Research, 42, 1444-1460.
Velleman, S. & Strand, K. (1994). Developmental verbal dyspraxia. In J. Bernthal (Ed.),
Phonological Characteristics of Special Populations. New York: Thieme Medical Publishers.
Yoss, K. A. & Darley, F. L. (1974). Developmental apraxia of speech in children with defective articulation. Journal of Speech and Hearing Research, 17, 399-416.
Appendix
|
correct easy trochees |
|
incorrect easy trochees |
||||
|
|
AVER |
SD |
|
|
avg |
stdev |
|
max hz |
16.1175 |
37.25838 |
|
max hz |
27.4649 |
61.21707 |
|
Hzexc |
-0.1336 |
9.583335 |
|
Hzexc |
-3.0541 |
62.35906 |
|
mean hz |
16.1088 |
34.42051 |
|
mean hz |
39.3265 |
57.03438 |
|
meanfo |
16.3319 |
34.69792 |
|
meanfo |
41.6664 |
60.25113 |
|
|
|
|
|
|
|
|
|
max dB |
1.9449 |
4.100872 |
|
max dB |
4.8791 |
9.46579 |
|
dBexcur |
-3.8687 |
6.408364 |
|
dBexcur |
1.1296 |
8.395381 |
|
mean dB |
2.6874 |
4.04583 |
|
mean dB |
4.2296 |
6.501341 |
|
|
|
|
|
|
|
|
|
correct easy iambs |
|
|
incorrect easy iambs |
|||
|
|
avg. |
stdev |
|
|
avg |
stdev |
|
max hz |
5.410583 |
14.55373 |
|
max hz |
10.1905 |
21.27774 |
|
Hzexc |
5.784833 |
18.67807 |
|
Hzexc |
25.51708 |
24.81689 |
|
mean hz |
4.738417 |
15.87201 |
|
mean hz |
0.829083 |
14.6304 |
|
meanfo |
4.554167 |
15.47975 |
|
meanfo |
-0.10867 |
14.59625 |
|
|
|
|
|
|
|
|
|
max dB |
0.617083 |
4.424099 |
|
max dB |
0.616333 |
5.170878 |
|
dBexcur |
2.294917 |
4.696744 |
|
dBexcur |
3.187667 |
5.35308 |
|
mean dB |
1.041917 |
3.717542 |
|
mean dB |
-0.42942 |
4.747411 |
|
|
|
|
|
|
|
|
|
correct moderate
trochees |
|
incorrect moderate
trochees |
||||
|
|
avg. |
stdev |
|
|
avg |
stdev |
|
max hz |
46.4761 |
53.83985 |
|
max hz |
54.75742 |
77.61593 |
|
Hzexc |
15.7014 |
29.85462 |
|
Hzexc |
23.94142 |
16.28918 |
|
mean hz |
39.6376 |
49.86828 |
|
mean hz |
25.35975 |
48.08008 |
|
meanfo |
39.9582 |
50.18175 |
|
meanfo |
25.0785 |
48.6575 |
|
|
|
|
|
|
|
|
|
max dB |
4.0058 |
2.9656 |
|
max dB |
9.749167 |
19.33399 |
|
dBexcur |
3.6562 |
9.309486 |
|
dBexcur |
1.583 |
7.093766 |
|
mean dB |
10.929 |
20.85805 |
|
mean dB |
5.2635 |
3.236549 |
|
|
|
|
|
|
|
|
|
correct moderate iambs |
|
incorrect moderate
iambs |
||||
|
|
avg |
stdev |
|
|
avg. |
stdev |
|
max hz |
26.18073 |
40.46635 |
|
max hz |
14.22067 |
32.86981 |
|
Hzexc |
23.21318 |
20.28083 |
|
Hzexc |
9.205833 |
15.67648 |
|
mean hz |
38.93573 |
72.32214 |
|
mean hz |
5.71325 |
35.1999 |
|
meanfo |
40.00345 |
76.90316 |
|
meanfo |
5.698583 |
35.34041 |
|
|
|
|
|
|
|
|
|
max dB |
4.722545 |
4.195045 |
|
max dB |
6.185167 |
21.15819 |
|
dBexcur |
3.017727 |
12.99188 |
|
dBexcur |
3.576417 |
6.844719 |
|
mean dB |
4.722545 |
26.59304 |
|
mean dB |
-1.516 |
2.837961 |
|
|
|
|
|
|
|
|
|
correct difficult
trochees |
|
incorrect difficult
trochees |
||||
|
|
avg |
stdev |
|
|
avg |
stdev |
|
max hz |
76.1374 |
71.22074 |
|
max hz |
93.27409 |
77.2951 |
|
Hzexc |
-0.1599 |
23.41936 |
|
Hzexc |
-11.8812 |
17.64695 |
|
mean hz |
69.8973 |
65.03138 |
|
mean hz |
69.57027 |
59.7241 |
|
meanfo |
71.2037 |
65.82518 |
|
meanfo |
70.906 |
60.65311 |
|
|
|
|
|
|
|
|
|
max dB |
5.5586 |
2.836316 |
|
max dB |
4.654818 |
4.91614 |
|
dBexcur |
2.5052 |
5.854321 |
|
dBexcur |
2.354818 |
3.404689 |
|
mean dB |
4.6356 |
4.268778 |
|
mean dB |
4.521 |
4.405127 |
|
|
|
|
|
|
|
|
|
correct difficult iambs |
|
incorrect difficult
iambs |
||||
|
|
avg. |
stdv |
|
|
avg |
stdev |
|
max hz |
-22.7177 |
68.1075 |
|
max hz |
31.08507 |
48.42557 |
|
Hzexc |
16.6446 |
42.36394 |
|
Hzexc |
32.70047 |
45.85923 |
|
mean hz |
-24.1876 |
47.93952 |
|
mean hz |
23.42373 |
41.31184 |
|
meanfo |
-25.8302 |
48.31514 |
|
meanfo |
21.9182 |
40.20569 |
|
|
|
|
|
|
|
|
|
max dB |
-1.0123 |
6.351219 |
|
max dB |
4.2866 |
5.522375 |
|
dBexcur |
0.8327 |
5.577814 |
|
dBexcur |
1.557733 |
7.637645 |
|
mean dB |
0.4261 |
4.462514 |
|
mean dB |
3.544933 |
4.990934 |