









Study with the several resources on Docsity
Earn points by helping other students or get them with a premium plan
Prepare for your exams
Study with the several resources on Docsity
Earn points to download
Earn points by helping other students or get them with a premium plan
Community
Ask the community for help and clear up your study doubts
Discover the best universities in your country according to Docsity users
Free resources
Download our free guides on studying techniques, anxiety management strategies, and thesis advice from Docsity tutors
University of Wisconsin-Madison. Preparation of this report was ... The present report describes acoustic correlates for two PVSP codes used to identify.
Typology: Exams
1 / 16
This page cannot be seen from the preview
Don't miss anything!
Phonology Project Technical Report #
Marios Fourakis Heather Karlsson Christie Tilkens Lawrence Shriberg
February, 2010
Phonology Project, Waisman Center,University of Wisconsin-Madison
Preparation of this report was supported by research grant DC00496 from the National Institute on Deafness and Other Communication Disorders, National Institutes of Health (Lawrence D. Shriberg, Principal Investigator)
Background Phonology Project Technical Reports provide technical and substantive information on methods developed for a program of research in speech sound disorders of known and unknown origins. Primary goals of the Phonology Project are to identify etiologic origins, risk and protective factors, and diagnostic markers for eight subtypes of speech sound disorders of currently unknown origin (Shriberg, 2010). The diagnostic instrument used in all Phonology Project studies is termed the Speech Disorders Classification System (SDCS: Shriberg et al., 2010). The SDCS includes a typologic nosology for research and practice in speech sound disorders, and an etiologic nosology for the eight putative subtypes of speech sound disorders of currently unknown origin. Data reduction methods include both perceptual and acoustic methods. Perceptual methods for narrow phonetic transcription of speech are based on extensions to the system described in Clinical Phonetics (Shriberg & Kent, 2003). Perceptual methods to code speakers’ prosody and voice are based on extensions to the system described in The Prosody-Voice Screening Profiles (PVSP: Shriberg, Kwiatkowski, & Rasmussen, 1990) and Phonology Project Technical Report No. 1 (Shriberg, Kwiatkowski, Rasmussen, Lof, & Miller, 1992). Recent methodological focus of the Phonology Project has been on identifying acoustic correlates for all segmental and suprasegmental variables currently transcribed or prosody-voice coded using perceptual methods. The References section includes citations for these published and unpublished papers. The present report describes acoustic correlates for two PVSP codes used to identify inappropriate resonance: PVSP Code 30: Nasal and PVSP Code 32: Nasopharyngeal. The PVSP manual includes extended discussions of each code and provides perceptual guidelines and audio exemplars to train listeners to identify each code. An acoustic correlate for the third resonance
2010). The MSAP includes a conversational speech sample (CSS) obtained in the conventional manner (McSweeny, 1998) and the Challenging Words Task (CWT: Shriberg et al., 2010). The speakers were a mother and daughter, ages 49 and 18, each of whom had a chromosome 7; translocation causing haploinsufficiency in the gene products of FOXP2 (Shriberg et al., 2006). Following PVSP data reduction procedures, the conversational speech sample of each speaker had been processed to yield 24 utterances eligible for PVSP coding (i.e., these utterances were retained for analyses because they did not meet criteria for one or more of 32 exclusion codes; PVSP manual, p. 11-21). All 24 CSS utterances of both speakers were coded PV30: Nasal by a
research transcriber. Additionally, the nasalized diacritic [~] had been used in the transcripts of many imitated vowel responses to CWT stimuli. Acoustic analyses of each of the four corner vowels occurring in the CSS and occurrences of perceptually nasalized /e/ vowels in CWT words were completed. For each vowel token, the frequencies of F1 and F2 were measured from a 20 ms spectral section of the vowel centered at 50% of the vowel’s duration. The spectral peaks were determined using information from a Linear Prediction Coding (LPC) spectrum, a Fast Fourier Transform (FFT) spectrum, and a formant track superimposed on the spectrogram of the utterance. The LPC used a number of coefficients appropriate to the sampling rate (SR +/- 4) for each speaker’s recording. The F1 and F2 values reported by Hillenbrand, Getty, Clark, & Wheeler (1995) for 48 adult women were used as the reference values for non-nasalized vowels. The F1 and F2 measurements in Hillenbrand et al. (1995) were taken from the midpoint of the four corner vowels produced in the context /h_d/. The speakers had read three randomized lists of the words in isolation.
Results Table 1 includes F1 and F2 means and standard deviations for the total number of eligible vowel tokens obtained from the older (Panel A) and younger (Panel B) speakers with the FOXP disruption. Descriptive statistics from the 48 reference speakers’ productions (Hillenbrand et al.,
Table 1. Acoustic findings for vowel tokens from two speakers with perceived nasal resonance and nasalized vowels (Shriberg et al., 2006) and from a reference group (Hillenbrand et al., 1995). Panel A. Older speaker with FOXP2 disruption. F1 F Vowel No. Tokens M SD M SD //iq/ (^) / 610 402777 6667 26591842 107128 /e/ 10 722 73 1540 233 /u/ 6 453 42 1341 154
Panel B. Younger speaker with FOXP2 disruption. F1 F Vowel No. Tokens M SD M SD /i/ 2 430 0 2823 13 /q/ 3 867 43 1780 286 /e/ 6 725 50 1255 99 /u/ 4 513 49 1422 349
Panel C. Hillenbrand et al. (1995) vowel data for adult females. F1 F Vowel No. Speakers M SD M SD /i/ 48 437 71 2761 147 /q/ 48 676 69 2334 159 //eu// (^4848 921459 9739 15251105 )
Figure 1. F1 and F2 measurements of the corner vowels in a Conversational Speech Sample (CSS) and in the Challenging Words Task (CWT) in the Madison Speech Assessment Protocol for the older of the two speakers. The reference vowel space for the four corner vowels is definedby the mean F1 and F2 frequencies reported in Hillenbrand et al. (1995) for adult women.
z
z
z
z zz
{ {
{
{
{
S S
S
S
S S
1000
800
600
400
(^3000 2500 2000 1500 ) F2 frequency in Hz
CSS - CWT - z CSS - { CWT - S CSS - CSS -
/ e /
/ q /
Figure 2. F1 and F2 measurements of the corner vowels in a Conversational Speech Sample (CSS) and in the Challenging Words Task (CWT) in the Madison Speech Assessment Protocol for the younger of the two speakers. The reference vowel space for the four corner vowels isdefined by the mean F1 and F2 frequencies reported in Hillenbrand et al. (1995) for adult women.
zz
z
S S
1000
800
600
400
(^3000 2500 2000 1500 ) F2 frequency in Hz
CSS - CWT - z CSS - S CSS - CSS -
/ e /
/ q /
Conclusion Findings indicate that F1 lowering on /e/ tokens was a robust acoustic correlate of vowel tokens perceived as nasalized from both speakers in both sampling contexts on the Madison Speech Assessment Protocol. The percentage of acoustic change from the reference data for non- nasalized vowels was acceptably close to the theoretical prediction for nasalization in Stevens (1999). These findings are interpreted as support for the use of F1 lowering on /e/ vowels (i.e.,
F1 and F2 frequencies for all vowels in utterances coded PV32 were measured using the same procedures as described in the first section of this paper reporting an acoustic correlate for the percept of nasal resonance. The LPC used a number of coefficients appropriate to the sampling rate (SR+/- 4) for each speaker’s recording, which ranged from 20 to 44.1 kHz, depending on the speaker group and the recording date. Results Table 2 includes the mean and standard deviation F1 and F2 values for each corner vowel for each of the tokens obtained from speakers in each of the three groups.
Table 2. nasopharyngeal resonance and from a reference group of typically-developing speakers. Acoustic findings for vowel tokens from two speaker groups with perceived
Panel A. Conversational speech produced by adolescent speakers with Fragile X syndrome. F1 F Vowel No. Speakers M SD M SD /i/ 7 386 47 2233 227 /q/ 6 648 118 1786 198 /e/ 8 692 134 1501 114 /u/ 6 396 42 1418 146
Panel B. Conversational speech produced by adolescent speakers with Down syndrome. F1 F Vowel No. Speakers M SD M SD /i/ 8 379 43 2234 204 /q/ 8 635 73 1803 286 /e/ 8 703 93 1404 120 /u/ 8 405 46 1366 177
Panel C: Conversational speech produced by typically-developing adolescent speakers. F1 F Vowel No. Speakers M SD M SD //iq/ (^) / 1010 396674 3879 24841772 204122 /e/ 10 691 51 1305 66 /u/ 10 412 43 1592 178
Figure 3 is a display of the mean F1 and F2 frequencies of the vowels produced by the speakers with Fragile X syndrome (open circles) and the typically-developing speakers (open triangles), each connected by lines to circumscribe the vowel spaces. In comparison to the values from the reference group, Fragile X speakers’ lower F2 values for the high vowels /i/ and /u/ are consistent with backing of the tongue. One-tailed effect sizes for these between-group differences (i.e., to assess the directional prediction) are 1.18 (90% CI = 0.25-1.99) for /i/ and 1.05 (90% CI = 0.10-1.89) for /u/. Mean F1-F2 values for the low vowel /q/ are generally similar for both groups. F1 and F2 values for the low vowel /e/ are higher for Fragile X speakers (one-tailed effect size for F2 = 2.17; 90% CI = 1.10-3.03).
F2 values, and the two speaker groups have similar values for the low vowel /q/. The one-tailed between-group F2 effect size for /i/ is 1.23 (90% CI =0.32-2.01) and for /u/ is 1.15 (90% CI =0.26-1.93). The one-tailed between-group F2 effect size for /e/ is 1.06 (90% CI =0.18-1.84).
Figure 4. Mean F1 and F2 frequencies of corner vowels produced by adolescent speakers with Down syndrome (squares) and by typically-developing adolescent speakers (triangles).
Conclusion The acoustic findings in Table 2 and Figures 3 and 4 are interpreted as support for lowered F2 values on high vowels as a reliable acoustic correlate of the percept of nasopharyngeal resonance. In both speaker groups with complex neurodevelopmental disorders, the percept of nasopharyngeal resonance was associated with F2 values further back in the vowel space than the corresponding vowels of typically-developing adolescent speakers. Findings also supported fronting of the low-back vowel, but effects were stronger for the speakers with Fragile X syndrome than for speakers with Down syndrome. As with acoustic findings emerging for all perceptual constructs in this research series, studies in progress will attempt to cross validate these acoustic marker findings for nasal and nasopharyngeal resonance.
Shriberg, L. D., Campbell, T. F., Karlsson, H. B., Brown, R. L., McSweeny, J. L., & Nadler, C. J. (2003). A diagnostic marker for childhood apraxia of speech: The lexical stress ratio. Clinical Linguistics and Phonetics , 17 , 549-574. Shriberg, L. D., Fourakis, M., Hall, S., Karlsson, H. K., Lohmeier, H. L, McSweeny, J., et al. (2010). Extensions to the Speech Disorders Classification System (SDCS). Manuscript submitted for publication. Shriberg, L. D., Flipsen, P., Jr., Karlsson, H. B., & McSweeny, J. L. (2001). Acoustic phenotypes for speech-genetics studies: An acoustic marker for residual / 6 / distortions. Clinical Linguistics and Phonetics , 15 , 631-650. Shriberg, L. D., Green, J. R., Campbell, T. F., McSweeny, J. L., & Scheer, A. (2003). Adiagnostic marker for childhood apraxia of speech: The coefficient of variation ratio. Clinical Linguistics and Phonetics , 17 , 575-595. Shriberg, L. D., & Kent, R. D. (2003). Clinical phonetics (3rd ed.). Boston: Allyn & Bacon. Shriberg, L. D., Kent R. D., Karlsson, H. B., McSweeny, J. L., Nadler, C. J., & Brown, R. L. (2003). A diagnostic marker for speech delay associated with otitis media with effusion: Backing of obstruents. Clinical Linguistics and Phonetics , 17 , 529-547. Shriberg, L. D., Kwiatkowski, J., & Rasmussen, C. (1990). Profile. Tucson, AZ: Communication Skill Builders. The Prosody-Voice Screening
Shriberg, L. D., Kwiatkowski, J., Rasmussen, C., Lof, G. L., & Miller, J. F. (1992). The Prosody- Voice Screening Profile (PVSP): Psychometric data and reference information for children Wisconsin-Madison. (Tech. Rep. No. 1). Phonology Project, Waisman Center, University of
Stevens, K. (1999). Acoustic phonetics. Cambridge, MA: The MIT Press.