Perceptual and Motor Skills, 1975, 40,63-68.

@ Perceptual and Motor Skills 1975

RELATIONSHIP OF ACOUSTIC PARAMETERS AND PERCEPTUAL RATINGS OF ESOPHAGEAL SPEECH1 MAYNARD D. FILTER3

MELVIN HYMAN

Central Michigan University

Bowling Green State University

Summary.-For 10 male and 10 female esophageal speakers selected acoustic and perceptual dimensions of esophageal speech were investigated to determine the intercorrelations of the following measures: intelligibility, articulation, rate, fundamental frequency, mean relative intensity, and effectiveness. Results indicate that only articulation and effectiveness were significantly correlated with intelligibility, whereas measures of intelligibility, articulation, rate, fundamental frequency, and mean relative intensity were all significantly correlated with effectiveness ratings.

There have been numerous quantitative studies of the speech of the laryngectomized including studies of the development of skills in learning esophageal speech (Wepman, e$ al., 1953), factors related to speech proficiency such as objective measures of articulation and intelligibility (Shames, Font, & Matthews, 1963), clinically measurable skills that are related to adequate esophageal speech (Berlin, 1963), and acoustic parameters of fundamental frequency, intensity, rate, and resonance (Curry & Snidecor, 1959, 1960, 1961; Kytta, 1964; Shipp, 1967 ) . Qualitative studies of the speech of the laryngectomized are scarce in the literature because qualitative or perceptual studies by their very nature employ more subjective-type measurements than do acoustic studies. Hyman (1960) proposed a rating scale for esophageal speech which included rate, intensity, intelligibility, and "social acceptability." Each measure was rated on a five-point scale. Berlin (1965) employed a seven-point scale from "highly acceptable esophageal speech to "unacceptable esophageal speech." Shipp (1967), in one of the first systematic studies relating acoustic parameters with perceptual ratings, investigated the relationship of selected acoustical parameters and perceptual measures to "acceptability" ratings of esophageal speech. He employed a five-point rating scale of acceptability. He found significant correlations between acceptability ratings and the following variables: mean fundamental frequency, duration, respiratory noise racings, the percentage of periodic phonation, and the percentage of silence measures. Shipp suggested that a combination of acceptability ratings and intelligibility scores would provide a strong perceptual specification of lPortions of this study were taken from a Master's thesis by the first author, under the direction of the second at Bowling Green State University, 1966. A paper based on this study was presented at the 44th annual convention of the American Speech and Hearing Association. Portions of the research were supported by the Central Michigan University Faculty Research and Creative Endeavors Committee. 'Address requests for reprints to Maynard D. Filter, Communicarion Disorders, Central Michigan University, Mt. Pleasant, Michigan 48859.

64

M. D. FILTER & M. HYMAN

esophageal speech. Hoops and No11 (1969) studied the relationship of several acoustic parameters to effectiveness and found only rate was significantly correlated with effectiveness. More objective or standardized terms are needed to perceptually describe esophageal speech than the one which are found in the literature, A wide variety of ambiguous subjective terms are found in the literature including the following: "Good, poor, superior, adequate, average, effective, acceptable, efficient, proficient, socially acceptable" plus many other modifications of the above. Before any type of intelligibility-acceptability rating scale can be devised, it would be advantageous to know the relationship of the perceptual measure "intelligibility" with the perceptual measure "acceptability." A review of the literature shows that intelligibility has been related to articulation, loudness, quality, and "acceptability." "Acceptability" has been related to fundamental frequency, duration, percentage of noise, percentage of periodicity, and percentage of silence (Shipp, 1967). Intelligibility, articulation, lack of noise, voice quality, and fluency have also been related to acceptability (Berlin, 1967). It was the purpose of the present study to investigate the relationship of an intelligibility measurement and ratings of acceptability of esophageal speech and to investigate the relationship of articulation scores, rate of speech, fundamental frequency, and mean relative intensity measures to intelligibility scores and effectiveness ratings. In the present study the term "effectiveness" is substituted for the term "acceptability."

METHOD Selection of Subjects Twenty esophageal speakers were chosen, 10 male and 10 female. Each speaker was judged by his clinician and by E as being a "good" esophageal speaker. Each speaker had been using esophageal speech for at least two years and could speak fluently without any apparent effort or strain. Mean age of the speakers was 59 yr., ranging from 30 yr. to 78 yr. Mean number of years in school was 12.1 ranging from G yr. of school to 17.0. Mean number of years of using esophageal speech was 7.8, ranging from 2% yr. to 16.0 yr. Procedure Each S read a standard prose selection, a list of words from an intelligibility test, and a list of nonsense syllables. The recordings of these speech samples were subjected to perceptual and acoustic analyses. All recordings were made in relatively quiet rooms. Auditing of tapes for perceptual analyses was carried out in standard university classrooms with the signal at a comfortable listening level for all auditors. The average signal to noise ratio was 17.5 dB. PerceptuJ Mea~u~ernentr Intelligibi1iiy.-Forms C and D of Black's Multiple-choice Intelligibility Test were employed (Black, 1957). Each speaker read nine groups of three words for a total of 27 words. Standard multiple-choice check sheets were used by 29 speech pathology majors when auditing the tape recordings of the lists of words. The intelligibility score in the

PARAMETERS, RATINGS OF ESOPHAGEAL SPEECH

65

present study was the mean number of words heard correctly by auditors as spoken by each subject. Articukion.-A list of 44 nonsense-syllables including all the consonant sounds combined with the vowel [a] according to the usual place in which they are found in the English language, comprised the articulation test. This follows Hyman's method of evaluating the articulation of esophageal speakers (Hyman, 1955). Ten speech pathology majors phonetically transcribed recorded lists of these syllables. The articulation score in the present study was the mean number of nonsense-syllables correctly transcribed by auditors as spoken by each S. Ejfectiueness ratings.-Each speaker read a standard prose selection entitled "My Grandfather" from Vaa Riper (1963). The first two sentences of the prose selection were edited from the tape recordings. Twelve speech pathology students listened to these two sentences and rated each speaker on "effectiveness;" a five-point rating scale was employed, 5 being excellent, &good, 3-average, 2-fair and 1-poor. A specific definition of effectiveness was not given to the auditors. The effectiveness score was the mean rating given each S. Acoustic Measr~rements Rate.-The rate of speech was measured in syllables and words per minute. The number of words and syllables spoken by each S while reading the prose selection was counted. The selection was timed by a stop watch for each reading and the number of words and syllables per minute was calculated. Fundamenial frequency.-An estimate of the fundamental frequency of each speaker's voice was calculated by counting the vertical striations on sonograms of the vowel [a] taken randomly from the recordings of spoken nonsense-syllables. Mean +ela;iue intensity.-The volume of the recorder was set at the same level for each recording of the prose selection. Each speaker was seated with his lips 6 in. from the microphone. Mean relative intensity was measured by averaging the intensity of the syllables of the two sentences edited from the tape recordings. A graphic level recorder was employed to obtain a graphic representation of relative intensity level measured in decibels. Ampex tape recorder and playback system, Model 601, were emEquipment.-An ployed for all recordings. Tape speed was seven and one-half inches per second. An Electra-Voice cardiod microphone, Model 664, was used with the Ampex recorder. A General Radio Company Graphic Level Recorder. Type 1521-B, was employed to measure relative intensity levels. Sonograms for fundamental frequency analyses were made on the Key Electric Company Sono-Graph. Type GOGlA. Analyjis of Data.-Pearson product-moment correlation coefficients were calculated to study the intercorrelations of the measures of intelligibility, articulation, rate, fundamental frequency, mean relative intensity and effectiveness.

RESULTSAND DISCUSSION Table 1 summarizes the mean scores for the 20 esophageal speakers for each of the seven measures. The mean rate of 100 words per minute of the 20 speakers in the present study falls below the mean rate of 113 words per minute of Snidecor and Curry's six superior esophageal speakers. Snidecor and Isshiki (1965) stated that 80 to 128 words per minute was a realistic rate of speech for esophageal speakers. Ninety per cent of the speakers in the present study had a rate of speech which fell within this range.

M. D. FILTER

66

&

M. HYMAN

TABLE 1 MEANS,STANDARD DEVIATIONS, AND RANGES OF SEVEN Acousnc AND PERCEPTUAL OBTA~NELI ON 20 ESOPHAGEAL SPEAKERS MEASUREMENTS Measures

M

SD

Range

Intelligibility Articulation Syllables/minute Words/minute Fundamental frequency Mean relative intensity Effectiveness

21.1 26.1 135.0 100.1 71.8 39.8 3.2

2.9 6.9 31.0 23.4 21.6 4.1 0.7

11.8- 25.2 1 2 . 6 38.6 50.0-175.4 35.9-129.4 45.0-125.0 32.9- 48.3 1.24.5

The mean fundamental frequency of 71.08 of the speaker's voices from the present study seems realistic when compared with the mean fundamental frequency of 62.8 of Snidecor and Curry's six superior esophageal speakers and the fundamental frequency of 74.57 of Shipp's 33 speakers. These differences in mean fundamental frequency may reflect real differences among the three groups of speakers or they may be a result of different methods for obtaining fundamental frequency measurements. The mean effectiveness rating of 3.26 indicates that the speakers of the present study were as a whole rated above average. This is not surprising because these speakers were all judged before the experiment as being "good" esophageal speakers. TABLE 2

1. Intelligence 2. Articulation 3. Syllables/minute

4. Words/minute 5. Fundamental frequency 6. Mean relative intensity 7. Effectiveness

.75*

.16 .42

.13 .39 .99*

.20 .32 .5Ot .5Ot

.17 .36 .50t .50t .31

.67* .76* .6BX .66* .54t .55t

*P.m. tP.06.

Table 2 summarizes the intercorrelations among the seven measures obtained from analysis of the recordings of the 20 esophageal speakers. The COefficient of .67 berween intelligibility and effectiveness indicates a significant positive relation between these two perceptual measures. This result supports-~erlin's (1965) findings; his auditors listed intelligibility as a qualiry on which they based their ratings of acceptability of samples of esophageal speech.

PARAMETERS, RATINGS OF ESOPHAGEAL SPEECH

67

Coefficients of correlation from Table 2 indicate that, besides effectiveness ratings, articulation scores are the only measures significantly correlated with intelligibility scores. These dara agree with the observations of Snidecor (1969) where he stated, that given reasonable loudness and quality, the intelligibility of esophageal speech seems mainly a function of articulation. Moderate to low intercorrelations of Table 2 indicate that intelligibility scores, articulation scores, measures of rate, mean fundamental frequency, and mean relative intensity are somewhat though significantly correlated in a positive direction with ratings of effectiveness. Probably ratings of effectiveness reflect a wide variety of acoustic and perceptual aspects of esophageal speech. When evaluating esophageal speech, one must take into account the criteria by which to measure it. If one's criterion is merely whether or not listeners recognize words spoken by an esophageal speaker, i.e., whether or not the person's speech is intelligible (including good articulation), then the instrument to employ would be an intelligibility test, such as the one used in the present study. If one is interested in additional information other than just intelligibility or articulation, then an effectiveness rating scale which takes into account a wide variety of perceptual aspects of esophageal speech could be employed. It appears that there are more perceptual and acoustic measures significantly correlated with effectiveness ratings than there were with intelligibility scores. Even though effectiveness ratings and intelligibility scores are significantly correlated, they share only 45% common variance, therefore each measures other dimensions. It is suggested that an evaluation of esophageal speech include effectiveness ratings and a measure of intelligibility. REFERENCES BERLIN,C. I. Clinical measurement of esophageal speech: I. Methodology and curves of skill acquisition. I . speech hear. Dis., 1963, 28, 42-51. BERLIN,C. I. Clinical measurement of esophageal speech: Ill. Performance of nonbiased groups. 1. speech hear. Dis., 1965, 30, 174-182. BLACK,J. W. Multiple-choice intelligibility tests. 1. speech hear. Dis.,1957, 22, 213235. CURRY,E. T., & SNWECOR,J. C. Physical measurement and pitch perception in esophageal speech. Laryngoscope. 1961, 7 1, 41 5-424. HOOPS,H. R., & NOLL,J. D. Relationship of selected acoustical variables to judgments of esophageal speech. J. comm. Dis., 1969, 2, 1-13. H W N , M. An experimental study of artificial larynx and esophageal speech. J . speech heas. Dis., 1955, 20, 291-299. HYMAN, M. A rating scale for esophageal speech. Presented at the 6th post-graduate course in disorders of voice and speech and esophageal speech, Miami, Fla.. June 20-24. 1960. KYTTA. J. Finnish esophageal speech after laryngectomy: sound spectrographic and cineradiographic studies. Acta Oiolasyng., 1964, Supplement 195. SHAMES.G. H., FONT. J., & MATIHEWS, J. Factors related to speech proficiency of the laryngectomized. 1. speech hear. Dis., 1963, 28, 273-387.

68

M. D. FILTER

&

M. HYMAN

SHIPP, T. Frequency, duration, and perceptual measures in relation to judgments of alaryngeal speech acceptability. I. speech hear. Res., 1967, 10, 417-427. SNIDECOR,J. C. (Ed.) Speech rehabilitation of the laryngectomized. (2nd ed.) Springfield, Ill.: Thomas, 1969. SNIDECOR, J. C.,& CURRY,E. T. Temporal and pitch aspects of superior esophageal speech. Ann. Otol. Rhinol. Laryngol., 1959, 69, 623-636. SNIDECOR,J. C., & CURRY,E. T. How effectively can the laryngectomee expect to speak? Norms for effective speech. Lmyngoscope, 1960, 70, 62-67. SNIDECOR, J. C., & ISSHIKI, N. Air volume and air flow relationships of six male esophageal speakers. 1. speech hear. Dis.,1965, 30, 205-215. VAN RIPER, C. Speech correction. (4th ed.) Englewood Cliffs: Prentice-Hall, 1963. WEPMAN,J. M., MCGAHAN,J. A,, RICHARD,J. C, & SKELTON,N . W. The objective measurement of progressive esophageal speech development. 1. speech hear. Dis., 1953, 18, 247-251. Accepted October 17, 1974.

Relationship of acoustic parameters and perceptual ratings of esophageal speech.

For 10 male and 10 female esophageal speakers selected acoustic and perceptual dimensions of esophageal speech were investigated to determine the inte...
223KB Sizes 0 Downloads 0 Views