- Email: [email protected]
Correlates of sophisticated listener judgments of esophageal air intake noise
J. COMMUN. DISORD. 20 (1987), 25-39
CORRELATES OF SOPHISTICATED LISTENER JUDGMENTS OF ESOPHAGEAL AIR INTAKE NOISE ALBERT
W. KNOX
Kansas City Veterans Administration Medical Center
VINCENT GORDON
ECCLESTON,
JAMES
F. MAURER,
and MARY C.
Portland State University, Portland, Oregon
Twenty-four laryngectomees, ranging widely in speaking ability, read a standard passage for audio recording. Four experienced voice clinicians rated the acceptability of the speakers’ air intake noise. Independently, overall speech proficiency ratings were obtained for 18 of the subjects. Five objective measures of the subjects’ esophageal speech were obtained using a real-time intensity display on a storage oscilloscope. Judges’ reliability was determined by Pearson Product Moment Correlations. Ratings were submitted to multiple regression analysis. The means of air intake noise acceptability were the criterion variables; the objective measures and speech proficiency scores were the predictor variables. Three predictors were positively correlated (<.Ol) with airintakeacceptability: the number of syllables per intake, the sound intensity of the intake, and the rate of speech. Syllables per intake provided the largest share of the variance.
INTRODUCTION It is estimated that 50-70 percent of laryngectomized
individuals utilize esophageal speech as their primary means of communication (Gardner, 1978; Horn, 1962). Studies of listener evaluation of esophageal speech indicate considerable variability in levels of voical proficiency (Shames, Font, and Mathews, 1963; Shipp, 1967; Snidecor, 1968). When compared to laryngeal speech, esophageal speech has inherent limitations in rate, volume, and pitch. Despite these limitations, a skilled esophageal speaker communicates effectively with most listeners. Snidecor (1968) stated that extraneous noises produced by some esophageal speakers may interfere with communication. Many esophageal speakers produce unwanted noise when taking air into the esophagus for use in speech. The injection method of air intake has been empirically Address correspondence
to Dr. Albert W. Knox, 6300 Granada, Prairie Village, KS 66208.
8 1987by Elsevier Science Publishing Co., Inc. 52 Vanderbilt Ave., New York, NY 10017
25 002t-9924lS7iS3SO
26
A. W. KNOX et al.
associated with a low-frequency noise often referred to as a “klunk” (Diedrich and Youngstrom, 1966). This noise has also been described as sounding like “thump,” “punk,” or “clump” (Snidecor, 1968) or as a loud gulp (Greene, 1957; Hyman, 1971). Klunk seems to be the term most widely accepted, and will be used here. The klunk may range from a rarely occurring and barely audible sound to a very distracting noise which may equal or exceed the intensity of the adjacent phonation, and may be produced one or more times per air charge. Gardner (1978) and Shanks (1977) stated that once klunking behavior is established it is very difficult to extinguish. Many authors have suggested clinical approaches to minimize or eliminate klunking, but very little has been published regarding its incidence or etiology. Many questions remain unanswered; for example, what acoustic parameters contribute to the listener’s perception of air intake noise as a klunk? When, if ever, is a klunk severe enough’to interfere with communication? At what stage in the treatment should a clinician overlook the production of air intake noises in deference to other aspects of esophageal voice‘? In the absence of published research information on these questions, these decisions are usually made on the basis of clinical experience. While there is no question of the importance of valid clinical judgment, its validity can only be based upon subjective evaluations of actions resulting in success. Where there is no history of success there is no foundation for clinical judgment. The intention of this study is to provide new objective data in order to enhance the decision process and enable the use of criteria that have been based upon factors of acceptability in the judgment process. The primary research question was: are selected objective measurements of esophageal speech significantly correlated with sophisticated listener judgments of air intake noise acceptability? The specific measurements used in this study are (1) The mean intensity of air intake noise; (2) the mean intensity of speech; (3) the number of syllables produced per audible intake of air; and (4) the rate of speech in syllables per second. A secondary question was: are sophisticated listener judgments of overall esophageal speech proficiency correlated with sophisticated listener judgments of air intake noise acceptability? These questions were formulated to address basic clinical problems; these are: just how important is it to try to manage klunking? Where does klunking stand in the hierarchy of clinical objectives in the acquisition of acceptable esophageal speech? Review of the Literature Numerous authors have identified air intake noise as a problem for many esophageal speakers. Some authors recommend instructional treatment procedures (Diedrich, 1968; Gardner, 1978; Greene, 1957; Hyman, 1971; Kallen, 1934; Luchsinger and Arnold, 1965; Salmon, 1971; Shanks, 1979;
JUDGMENTS
OF ESOPHAGEAL
AIR INTAKE
NOISE
27
Snidecor, 1968; Waldrop, 1956). Diedrich and Youngstrom (1966), Martin (1963), and Shanks (1977, 1979) speculated on the etiology of klunking. The etiology of klunking noise is not clear. A frequent explanation is “too much air taken in too fast and in too tense a manner” Diedrich, (1971). Similar explanations are found in Salmon ( 197 I), Duguay ( 1977)) and Martin (1979). Shanks (1977) stated that the noise is produced by a partial vacuum, created when the pharyngeal-esophageal segment (PE) is rapidely lowered as the air bolus enters the esophagus. Shanks associated klunking with the injection method of air intake. Diedrich and Youngstrom (1966) found support for that association with their cinefluorographic studies. The injection technique relies on the increase of oral air pressure created by a pistonlike movement of the tongue. When intraoral air pressure is sufficient to push the air bolus past the PE segment, the esophagus is partially inflated. Martin (1963) stated that the probable cause of klunking is the speaker’s inability to relax the PE segment, thus causing it to act like a sphincter. No data were found on the incidence of klunking. It was inferred from descriptions of the injection method of air intake that some noise is evident in many beginning speakers (Duguay, 1977; Lauder, 1972). Luchsinger and Arnold (1965) stated that all accomplished esophageal speakers produce a clearly audible “trapping noise” during intake of air. Several investigators reviewed the considerable difficulty encountered in eliminating klunking once it becomes established (Diedrich, 1971; Gardner, 1978; Martin, 1963; Salmon, 1971; Shanks, 1977). Martin stated that intake noise production is probably related to anatomic and physiologic differences among those speakers who produce klunking sounds. Martin stated that even a qualified speech instructor is unlikely to correct the problem. The majority of the authors cited above did not address the question of anatomic and physiologic differences as factors in klunking. Gardner (1978), Greene (1957), and Salmon (1971) refer to the habitual nature of klunking, implying that they believe klunking to be a learned behavior. The identification of klunking as an acquired behavior, if true, would not make it easier to extinguish, but would offer more hope for clinical control of the phenomenon as a consequence of clinical intervention than would the assumption of a physiological or anatomical anomaly. It would also set the direction therapeutic intervention would probably take. One physiological condition that could influence an esophageal speaker’s perception of his klunking is his hearing acuity. Studies by Martin, Hoops, and Shanks (1974) demonstrated a positive correlation between poor audition and poor esophageal speech. Because klunking is generally reported to be low pitched, and most acquired hearing loss is high frequency, one may assume that many esophageal speakers have sufficient low-frequency acuity to perceive the sound. It is possible, however, that many who produce the klunk do in fact perceive it, but have adapted to
28
A. W. KNOX er al.
it and ignore the behavior. The degree to which air intake noise may influence communication is largely unreported in the literature. Because klunking occurs immediately prior to an utterance, some authors describe klunking as a distraction rather than a direct interference to speech (Hyman, 1979; Shanks, 1979). No references identify the specific characteristics of klunking behavior that could cause it to be distracting. Hyman (1979) stated that a klunking noise may not affect intelligibility, but may adversely affect oral communication. Shanks (1979) included klunking with other distracting behaviors that are “not phonemic and not truly speech elements,” but that influence intelligibility by distracting the listener’s attention from the message. In summary, the literature has identified klunking as a problem for many esophageal speakers, but has neither determined its incidence nor quantified its severity. Some research workers have suggested etiology but have not confirmed their conjectures with experimental data. Treatment procedures have been described but appear to be based upon anecdotal evidence. An increased understanding of priorities appropriate to the clinical management of klunks might be facilitated by the identification of measurable correlates to listener acceptability of these noises. Of the activities ordinary and necessary to the clinical management of esophageal voice, how time or cost effective are the attempts at management of undesirable air intake noise? The apparent lack of objective criteria for listener judgments was a primary motivation for this study. METHOD Subjects Twenty-four volunteer laryngectomees attending the XIV Annual Institute in Laryngectomee Rehabilitation at Eastern Washington University, Cheney Washington, 1981 were the subjects. The criterion for participation was the ability to read the sample paragraph aloud. Recording Audiotape recording was accomplished in the audiometric suite at the University. Equipment consisted of a tape recorder (Sony model 366), a dynamic microphone (Electrovoice model 631B), and a high fidelity recording tape (Maxell UDXL). The input level from the microphone was adjusted to cause the VU meter to peak at zero for each subject. The tape speed was set at 7.5 ips. The stand-mounted microphone was placed approximately 1 foot in front and 1 foot to the side of the subject’s mouth. This placement minimized the effect of microphone sensitivity to extraneous body movement and stoma noise, while it provided high-quality recording of speech and the air intake noises.
JUDGMENTS
OF ESOPHAGEAL
AIR INTAKE
NOISE
29
Recordings were made between the hours of 9:00 a.m. and 4:30 p.m. and at least 45 minutes after eating to minimize any influence of digestion on esophageal voice function. After explaining the purpose and securing written consent, each subject was instructed to read aloud the first paragraph of the “Rainbow Passage” (Fairbanks, 1960) twice for practice. This practice reading enabled the subjects to adjust to the acoustic chamber and to “warm-up” their esophageal voices. Each subject notified the investigators when he was ready to begin. Judges and Ratings Four judges were obtained from the guest faculty of the Esophageal Voice Institute. Each judge had professional experience with hundreds of laryngectomees over a minimum of 20 years as a speech clinician. Hearing sensitivity of the judges was within normal limits for speech. Air Intake Acceptability Ratings Each judge had an individual listening session conducted in the control booth of the acoustic chamber. They were achieved by playing the recorded passages on the same tape recorder with the signal output to TDH 39 headphones. The recordings had been randomized onto several different tapes and were played back at a mean output level of 70 dB SPL. The tape was stopped for 2 minutes after each 10 minutes of play to minimize listener fatigue. Esophageal air intake noise was rated on a five-point equal appearing interval scale with 1 representing the least acceptable air intake noise and 5 indicating the most acceptable. Procedure for Overall Proficiency Ratings Overall proficiency ratings were available for 18 of the 24 subjects. These assessments were made by the faculty of the Laryngectomee Institute as part of the normal proceedings. Although the proficiency ratings and the air intake acceptability ratings took place during the same week and shared some of the same judges, they were conducted independently. The proficiency was judged during direct observation of subjects as each spoke briefly to an audience in a lecture hall. Those using esophageal voice were rated on a five-point rating scale; the rest were not considered for the research project. Instrumentation An acoustic analysis of the recorded speech samples was made with the aid of a Visi-Pitch model 6087, frequency and amplitude extractor mounted in and connected to a Tektronic storage oscilloscope. This in-
30
A. W. KNOX et al.
strument produced a graphic display of relative intensity over time. A 4set display time was used for this study. All segments of each subject’s performance were photographed for a permanent record. Each photograph was phonetically annotated to ensure correspondence with the segment of speech displayed on the oscilloscope. Fundamental frequency analysis established that the lowest klunk frequency of the subject group was 66 Hz, well above the 50-Hz low cutoff frequency of the Visi-Pitch. Procedures Air intake noise was readily identified on the oscilloscope as it typically preceded and followed signals of greater amplitude and duration. Examples of these klunk images are observable in Figure 1. The relatively smooth, straight sections of the trace represent the noise floor of the recorded image. Syllables are seen as the higher peaks, while the intake noises are displayed as the much lower peaks of brief duration. Intensity peaks were easily read on the grid scale. On the vertical axis of the display each large grid square represented 10 dB SL. Each horizontal grid square represented 0.4 sec. The number of syllables uttered in each sample was divided by the number of quantified audible air intakes in the sample. The resultant values were expressed as syllables per intake.
IO dB,
SPL
t .4
SECONDS
*
!
INTO
*
1 1 MANY*
*
BEAUTIFUL
Figure 1. Overlay traced from a Polaroid photograph of the Vi&Pitch, model 6087, display of an esophageal speech sample. Asterisks represent esophageal air
intakes.
JUDGMENTS
OF ESOPHAGEAL
AIR INTAKE
NOISE
31
The rate of speech was calculated by timing each sample with a quartz crystal chronograph. Each sample’s time measurement was started automatically by the zero-cross output pulse generated by a level zone detector (Grason Stradler 1270), upon detecting the onset of the subject’s voice. The timing was stopped manually when the utterance was concluded. The number of syllables in each sample was divided by their timed length. The rate of speech shown is in syllables per second. The number of syllables uttered per esophageal air intake has been identified as one of the skills associated with the acquisition of esophageal speech (Berlin, 1963; Damste, 1979; Gardner, 1978; Snidecor and Curry, 1959). This measure is usually obtained by counting a subject’s repetitions of a designated monosyllable between air intakes. The repetition of a monosyllable obviously differs from the connected speech task used in this study. The purpose of using syllables of connected speech between intakes was to provide a simple and sensitive indicator of the frequency of occurrence of audible air intakes.
STATISTICAL ANALYSIS Interjudge Reliability The Pearson product-moment correlation coefficient (r) compared the ratings from each judge with the mean rating from all judges. These ranged from .78 to .86 (Table I).
Intrajudge Reliability The internal consistency of the judges was measured by having each of them rejudge four of the subjects. The r values and corresponding significance levels for the test-retest scores ranged from 88 to .98 (Table 2).
Correlation of Objective Measurements with Listener Judgments The mean-of-judges ratings for acceptability for each of the 24 subjects was entered into a multiple regression analysis (Kim and Kohout, 1975), which compared the ratings with objective measures of esophageal speech Table 1. Correlation
between Each Judge’s Rating and Mean of Judges’ Ratings Judge
r
Significance level
A B C D
0.86 0.78 0.80 0.80
.Ol .Ol .Ol .Ol
A. W. KNOX et al.
32 Table 2. Correlation between Judge’s Test and Retest Ratings Judge
r
Significance level
A B C D
0.96 0.88 0.95 0.98
.OS .I0 .05 .02
used in the study. This analysis yielded the Pearson product moment correlation coefficient (r). The regression analysis provided the coefficient of determination (R2). In the complex relationship of the multiple regression the degree to which any of the objective measurements of speech can be predicted from the judges’ ratings is indicated by the correlation coefficient. Additionally, the coefficient of determination indicates the percentage of the total variance that can be accounted for by the measurements included in the regression application. F ratios were computed for each measurement variable and compared with a table of critical values for F at the .Ol level. Association of Listener Judgments of Speech Proficiency and Noise Acceptability Independently conducted speech proficiency ratings for 18 of the subjects were compared with the intake noise ratings obtained from the study. The overall speech proficiency scores were included as predictor variables in the multiple regression analysis described above. RESULTS AND DISCUSSION Are Selected Objective Measures of Esophageal Speech Signilkantly Correlated with Sophisticated Listener Judgments of Ah- Intake Noise Acceptability? Intake noise acceptability ratings were found to be significantly correlated with the four measurements selected. These measurements were: the number of syllables uttered per air intake, the rate of speech, the mean intensity of air intake noise, and the mean speech intensity. The multiple regression analysis indicated that approximately 83 percent of the total variance in the listener ratings of air intake noise acceptability could be accounted for by the measured variables. The measurement that exhibited the most predictive strength was the number of syllables uttered per audible intake. Computed F values for each variable exceed the critical
JUDGMENTS
OF ESOPHAGEAL
AIR INTAKE
NOISE
33
values of F beyond the .Ol level. The results of the statistical analysis for each parameter are presented in Tables 3 and 4.
The Mean Intensity of Air Intake Noise Intake noise intensity was found to be inversely associated to the judges’ ratings of intake noise with a correlation coefficient (r) of - .71. The greater the intensity of the intake noise, the lower was the level of listener acceptability. This was significant at the .Ol level with 20 degrees of freedom. The coefficient of determination (R2), or the change caused by including this variable in the regression analysis, was .13, indicating that this variable accounted for 13 percent of the variance of the ratings.
The Mean Intensity of Speech Mean speech intensity was negatively correlated with intake acceptability ratings with an r value of - .14. The R2 change for this measure accounts for less than 1 percent of the total variance of the listener ratings.
The Number of Syllables per Audible Intake of Air This was the strongest prediction variable in the study. The positive r of .81 was significant at the .Ol level with 20 degrees of freedom. This variable accounted for 65 percent of the variance in listener ratings of intake noise.
Table 3. Correlation
Matrix for Esophageal
Intake noise acceptability Intake noise acceptability Intake intensity Speech intensity Syllables per intake Rate in syllabledsec Esophageal speech proficiency
l.C!0OOO
Air Intake Noise Acceptability
Intake intensity
Speech intensity
- 0.70765
- 0.13747
Syllables Per intake 0.80794
Rate in Syllables/ set
Esophageal speech proficiency”
0.63313
0.16642 0.07299
- 0.70765
l.OWOO
0.10942
-0.48397
- 0.37300
-0.13747
0.10942
l.OOOWJ
-0.32334
0.01262
0.80794
-0.48397
-0.32334
-0.09612
l.OOOKl
0.68037
0.50961
0.63313
0.37300
0.01262
0.68037
l.OOOOo
0.47564
0.16642
0.07299
0.09612
0.50961
0.47564
1.om
E Sample size (N) for esophageal speech proficiency was 18; for all objective measures N = 24.
.88667
.89961
.90827
.91084
2. Intake
3. Skill
4. Rate
5. Speech
.82963
.82495
.80930
.78618
.65276
R2
.004
,016
.023
.133
.652
variance
Percent total regression residual regression residual regression residual regression residual regression residual
1 22 2 21 3 20 4 19 5 18
Degrees of freedom
12.64558 6.72682 15.23016 4.14224 15.67802 3.69437 15.93180 3.39109 16.07194 3.30045
Sums of squares
12.64558 0.30576 7.61058 0.19725 5.22601 0.18472 3.99533 0.17848 2.67866 0.19414
Mean square
13.79725
22.38546
28.29171
38.00631
41.35725
F
.Ol
.Ol
.Ol
.Ol
.Ol
Significance level
.-
- ._ -
-
.,.
_
-
-
-
0 Dependent variable: acceptability. b (1) “Syll.” is syllables per audible intake of air (klunk). (2) “Intake” is the intensity of the intake sound. (3) “Skill” is the mean judges’ ratings of esphageal speech. (4) “Rate” is syllables per second. (5) “Speech” is intensity of speech in dBSPL.
.80794
R
variableb
1. Syll.
Multiple
Independent
Table 4. Multiple Regression and Analysis of Variance”
^.
JUDGMENTS OF ESOPHAGEAL AIR INTAKE NOISE
35
The Rate of Speech Rate was positively correlated with acceptability ratings at the .Ol level with an I value of .63. The multiple regression coefficient of determination, R*,showed only 1.5 percent of the total variance was accounted for by rate.
The Correlation of Listener Judgments of Esophageal Speech Proficiency and Air Intake Noise Acceptability The statistical analysis revealed that ratings of overall esophageal speech proficiency were only weakly correlated with ratings of air intake noise acceptability. The r value of. 17 was indicative of weak correlation. The multiple regression analysis indicated that overall speech proficiency ratings accounted for about 2 percent of the total variance of the intake noise acceptability ratings.
Discussion The number of syllables uttered per audible air intake was by far the strongest predictor of sophisticated listener acceptance of air intake noise. Mean intensity of intake noise was also predictive, but to a lesser extent. The rate of speech was significantly correlated with listener acceptance of intake noise, but was a weak predictor of such acceptance. Mean speech intensity was not significantly correlated with listener acceptance of intake noise. The predictive power of independent variables in a multiple regression analysis is affected by the variables selected for computation. In a perceptual task, such as judging voice, it is unlikely that all appropriate measurable variables will be included. It is also unlikely that all the variance of intake noise acceptability can be accounted for by physical correlates. Within the limits imposed by these probabilities the results presented here reveal clear differences in the predictive values of the measures chosen for this study. The strong correlation between syllables per intake and intake noise acceptability is explained if the syllables-per-intake parameter is viewed as the rate of occurrence of audible air intakes. From this perspective the listeners tended to judge intake noise less critically as the number of intakes decreased. While it is not surprising that listener acceptability of intake noise improved as the quantity decreased, it is interesting that syllables per audible intake was a much stronger predictor of acceptability than the intensity of the intake noise. Apparently the listeners could accept the higher intake noise intensity if the audible intakes did not occur too frequently.
36
A. W. KNOX et al.
Many clinicians are familiar with esophageal speakers who klunk but otherwise have superior esophageal voices. It was hypothesized that if esophageal speech intensity were sufficiently high it would compensate for above average intake noise intensity. The assumption was that higher speech intensity would be predictive of greater acceptability. The results failed to support this hypothesis. The mean speech intensity measurements were not highly correlated with either intake noise acceptability or overall speech proficiency ratings. The low correlation of speech intensity with overall speech proficiency is consistent with the findings of Hoops and No11 (1979). Rate of speech was moderately correlated with intake noise acceptability, but accounted for less than 2 percent of the total variance, providing support to Snidecor and Curry (1959) and Hoops and No11 (1969). This association of rate and proficiency does not ensure that a faster rate will be predictive of greater intake noise acceptability, as intake noise acceptability is just one of many factors that influence overall speech proficiency ratings. The Correlation of Listener Judgments of Esophageal Speech Proficiency and Listener Judgments of Intake Noise Acceptability The findings of this study support the concept that the sophisticated listeners’ ratings of overall esophageal speech proficiency had little association with their ratings of air intake noise acceptability. While klunking has frequently been described as a problem requiring clinical intervention, this study found that klunking has little impact on sophisticated listener judgment of overall proficiency. A rationale for this mild influence may be found in the prephonatory characteristic of esophageal inflation. Klunking may distract the listener, but apparently it is not sufftciently distracting to impair communications seriously. Implications Two main implications emerged from these results, and the findings of a previous investigation were supported. The first finding was that the number of syllables uttered per intake was a stronger predictor of intake noise acceptability than was mean intake intensity. This implies that clinical and practice time might be better spent improving the duration of speech per audible intake rather than working directly on reducing the intensity of the intake noise. Even if the intake noise intensity remains constant, the increase in the production of syllables per intake may improve both intake noise acceptability and overall speech proficiency. The second finding, that intake noise acceptability is not a significant correlate of overall esophageal speech proficiency, also implies that prac-
JUDGMENTS
OF ESOPHAGEAL
AIR INTAKE
NOISE
37
tice should be concentrated on other aspects of speech, such as duration, quality, and inflection. The finding that intensity of esophageal speech is not highly correlated to proficiency, which ‘supported Hoops and No11 (1969), should be of interest to those who arduously struggle to increase volume at the expense of other more contributory factors to success of the esophageal speech process.
Future Research It is unlikely that the parameters included in this study exhaust the possibilities for variance in the air intake phenomena. Certainly the impact of visually perceived behaviors accompanying air intake could be observed and rated to determine whether the klunking variance extends beyond the auditory impression, a limitation of the audiotape recording method of study. Additional comparison and correlation could be made of the studied phenomenon using naive listeners to determine whether naive and sophisticated listeners apply the same standards. Acoustically oriented research workers may be interested in exploring the effects of spectral components and temporal relationships. Intake latency, pause time, position of intake-within-pause, and rate of multiple intakes are all addressable parameters that have not been systematically investigated. Finally, spectral analysis of esophageal air intake noise may yield a more precise description of the klunk and a clearer insight into its etiology than the conjectural descriptions of klunking available to clinicians today. The authors unite in expressing their gratitude to the faculty and staff of Eastern Washington University, particularly Dr. Herbert E. Gunderson, to the laryngectomized individuals who freely gave their time and efforts to this research, and to the judges whose volunteer efforts, willingly given, are testimony to the honors that have accrued to them individually. This research was supported in part with funds from the Research and Development Committee of the Kansas City Veterans Administration Medical Center Project No. 589-103-473-361661-006 and the Communications Research Laboratory.
REFERENCES American Cancer Society (1978). Facts on Cancer of the Larynx. New York: American Cancer Society. Berlin, C. I. (1963). Clinical measurement of esophageal speech: I. Methodology and curves of skill acquisition. J. Speech Hear. Disord. 28:42-51. Berlin, C. I. (1965). Hearing loss, palatal function, and other factors in post leayngectomy rehabilitation. J. Chronic Disord. 30: 174-183. Brown, F. L., Amos, J. R. and Mink, 0. G. (1975). Statistical Concepts, A Basic Program. New York: Harper and Row.
38
A. W. KNOX et al.
Damste, P. H. (1979) Rehabilitation score-list for the laryngectomized. In R. L. Keith and F. L. Darley (eds.), Laryngectomee Rehabilitntion. Houston: College Hill Press. Diedrich, W. M. (1968). The mechanism Sci. 15.5:303-317.
of esophageal speech. Ann. N. Y. Acad.
Diedrich, W. M. (1971). Primary stages of teaching. In S. Rigrodsky, J. Lerman and E. Morrison (eds.), Therapy for the Laryngectomized Patient. New York: Teachers College Press. Diedrich, W. M., and Youngstrom, IL: Charles C. Thomas.
K. A. (1966). Alaryngeal Speech. Springfield,
Duguay, M. J. (1977). Esophageal speech. In M. Cooper and M. H. Cooper (eds.), Approaches to Vocal Rehabilitation. Springfield, IL: Charles C. Thomas. Fairbanks, G. (1960. Voice and Articulation Drill Book (2nd edition). Philadelphia: Harper and Row. Filter, M. H. (1971). Rating systems for esophageal toral dissertation, Bowling Green University. Gardner, W. H. (1978). Laryngectomee IL: Charles C. Thomas. Greene,
speakers.
Unpublished
Speech and Rehabilitation.
M. C. L. (1957). The Voice and Its Disorders. London:
doc-
Springfield,
Pitman.
Hoops, H. R., and Noll, J. D. (1969). Relationship of selected acoustic variables to judgements of esophageal speech. J. Commun. Disord. 2:1-13. Horn, D. (1962). Laryngectomee survey report. Presented at the Eleventh Annual Meeting, International Association of Laryngectomees, Memphis, Tennessee, August 21. Hyman, M. (1979). Factors influencing intelligibility of alaryngeal speech. In R. L. Keith and F. L. Darley (eds.), Laryngectomee Rehabilitation. Houston: College Hill Press. Hyman, M. (1971). Intermediate stage of teaching alaryngeal speech. In S. Rigrodsky, J. Lerman and E. Morrison (eds.), Therapy for the Laryngectomized Patient. New York: Teachers College Press. Kallen, L. A. (1934). Vicarious vocal mechanisms, the anatomy, physiology and development of speech in laryngectomized persons. A.M.A. Arch. Otolaryngol. 20:460-503. Kim, J. and Kohout, F. J. (1975). Multiple regression analysis: Sub-program regression. In W. R. Klecka, N. H. Nie and C. H. Hull (eds), SPSS primer: Statistical Package for the Social Sciences. New York: McGraw-Hill. Lauder, E. (1972). Self-Help for the Laryngectomee. Lauder.
San Antonio:
Edmund
Luchsinger, D. and Arnold, G. (1965). Voice-Speech-Lanaguage, Clinical Communicology: Its Physiology and Pathology. Belmont, CA: Wordsworth. Martin, D. E. (1979). Evaluating esophageal speech development and proficiency. In R. L. Keith and F. L. Dailey (eds.), Laryngectomee Rehabilitation. Houston: College Hill Press.
JUDGMENTS
OF ESOPHAGEAL
AIR INTAKE
NOISE
39
Martin, D. E., Hoops, H. R. and Shanks, J. C. (1974). The relationship between esophageal speech proficiency and selected measures of auditory function. J. Speech Hear. Res. 74:80-85. Martin, H. (1963). Rehabilitation
of the laryngectomee.
Cancer 16:559-574.
Salmon, S. J. (1971). Use of imagery in teaching esophageal Commun. Disord. 2~17-24.
speech.
Cal6
J.
Shames, G. H., Font, J. and Mathews, J. (1963). Factors related to speech proficiency of the laryngectomized. J. Speech Hear. Disord. 28:273-287. Shanks, J. C. (1977). Variations on a vocal theme through interpersonal therapy with alaryngal patients. In M. Cooper and M. H. Cooper (eds.), Approaches to Vocal Rehabilitation. Springfield, IL: Charles C. Thomas. Shanks, J. C. (1979). Developing esophageal communication. In R. L. Keith and F. L. Darley (eds.), Laryngectomee Rehabilitation. Houston: College Hill Press. Shipp, T. (1967). Frequency, duration and perceptual memasures in relation to judgements of alaryngeal speech acceptability. J. Speech Hear. Res. 10:417427. Snidecor, J. C. (1968). Speech rehabilitation IL: Charles C. Thomas.
of the laryngectomized.
Springfield,
Snidecor, J. C. and Curry, E. T. (1959). Temporal and pitch aspects of superior esophageal speech. Ann. Otol. Rhinol. Laryngol. 68:1-14. Waldrop, W. and Gould, M. A. (1956). Your New Voice. Chicago, American Cancer Society.
CORRELATES OF SOPHISTICATED LISTENER JUDGMENTS OF ESOPHAGEAL AIR INTAKE NOISE ALBERT
W. KNOX
Kansas City Veterans Administration Medical Center
VINCENT GORDON
ECCLESTON,
JAMES
F. MAURER,
and MARY C.
Portland State University, Portland, Oregon
Twenty-four laryngectomees, ranging widely in speaking ability, read a standard passage for audio recording. Four experienced voice clinicians rated the acceptability of the speakers’ air intake noise. Independently, overall speech proficiency ratings were obtained for 18 of the subjects. Five objective measures of the subjects’ esophageal speech were obtained using a real-time intensity display on a storage oscilloscope. Judges’ reliability was determined by Pearson Product Moment Correlations. Ratings were submitted to multiple regression analysis. The means of air intake noise acceptability were the criterion variables; the objective measures and speech proficiency scores were the predictor variables. Three predictors were positively correlated (<.Ol) with airintakeacceptability: the number of syllables per intake, the sound intensity of the intake, and the rate of speech. Syllables per intake provided the largest share of the variance.
INTRODUCTION It is estimated that 50-70 percent of laryngectomized
individuals utilize esophageal speech as their primary means of communication (Gardner, 1978; Horn, 1962). Studies of listener evaluation of esophageal speech indicate considerable variability in levels of voical proficiency (Shames, Font, and Mathews, 1963; Shipp, 1967; Snidecor, 1968). When compared to laryngeal speech, esophageal speech has inherent limitations in rate, volume, and pitch. Despite these limitations, a skilled esophageal speaker communicates effectively with most listeners. Snidecor (1968) stated that extraneous noises produced by some esophageal speakers may interfere with communication. Many esophageal speakers produce unwanted noise when taking air into the esophagus for use in speech. The injection method of air intake has been empirically Address correspondence
to Dr. Albert W. Knox, 6300 Granada, Prairie Village, KS 66208.
8 1987by Elsevier Science Publishing Co., Inc. 52 Vanderbilt Ave., New York, NY 10017
25 002t-9924lS7iS3SO
26
A. W. KNOX et al.
associated with a low-frequency noise often referred to as a “klunk” (Diedrich and Youngstrom, 1966). This noise has also been described as sounding like “thump,” “punk,” or “clump” (Snidecor, 1968) or as a loud gulp (Greene, 1957; Hyman, 1971). Klunk seems to be the term most widely accepted, and will be used here. The klunk may range from a rarely occurring and barely audible sound to a very distracting noise which may equal or exceed the intensity of the adjacent phonation, and may be produced one or more times per air charge. Gardner (1978) and Shanks (1977) stated that once klunking behavior is established it is very difficult to extinguish. Many authors have suggested clinical approaches to minimize or eliminate klunking, but very little has been published regarding its incidence or etiology. Many questions remain unanswered; for example, what acoustic parameters contribute to the listener’s perception of air intake noise as a klunk? When, if ever, is a klunk severe enough’to interfere with communication? At what stage in the treatment should a clinician overlook the production of air intake noises in deference to other aspects of esophageal voice‘? In the absence of published research information on these questions, these decisions are usually made on the basis of clinical experience. While there is no question of the importance of valid clinical judgment, its validity can only be based upon subjective evaluations of actions resulting in success. Where there is no history of success there is no foundation for clinical judgment. The intention of this study is to provide new objective data in order to enhance the decision process and enable the use of criteria that have been based upon factors of acceptability in the judgment process. The primary research question was: are selected objective measurements of esophageal speech significantly correlated with sophisticated listener judgments of air intake noise acceptability? The specific measurements used in this study are (1) The mean intensity of air intake noise; (2) the mean intensity of speech; (3) the number of syllables produced per audible intake of air; and (4) the rate of speech in syllables per second. A secondary question was: are sophisticated listener judgments of overall esophageal speech proficiency correlated with sophisticated listener judgments of air intake noise acceptability? These questions were formulated to address basic clinical problems; these are: just how important is it to try to manage klunking? Where does klunking stand in the hierarchy of clinical objectives in the acquisition of acceptable esophageal speech? Review of the Literature Numerous authors have identified air intake noise as a problem for many esophageal speakers. Some authors recommend instructional treatment procedures (Diedrich, 1968; Gardner, 1978; Greene, 1957; Hyman, 1971; Kallen, 1934; Luchsinger and Arnold, 1965; Salmon, 1971; Shanks, 1979;
JUDGMENTS
OF ESOPHAGEAL
AIR INTAKE
NOISE
27
Snidecor, 1968; Waldrop, 1956). Diedrich and Youngstrom (1966), Martin (1963), and Shanks (1977, 1979) speculated on the etiology of klunking. The etiology of klunking noise is not clear. A frequent explanation is “too much air taken in too fast and in too tense a manner” Diedrich, (1971). Similar explanations are found in Salmon ( 197 I), Duguay ( 1977)) and Martin (1979). Shanks (1977) stated that the noise is produced by a partial vacuum, created when the pharyngeal-esophageal segment (PE) is rapidely lowered as the air bolus enters the esophagus. Shanks associated klunking with the injection method of air intake. Diedrich and Youngstrom (1966) found support for that association with their cinefluorographic studies. The injection technique relies on the increase of oral air pressure created by a pistonlike movement of the tongue. When intraoral air pressure is sufficient to push the air bolus past the PE segment, the esophagus is partially inflated. Martin (1963) stated that the probable cause of klunking is the speaker’s inability to relax the PE segment, thus causing it to act like a sphincter. No data were found on the incidence of klunking. It was inferred from descriptions of the injection method of air intake that some noise is evident in many beginning speakers (Duguay, 1977; Lauder, 1972). Luchsinger and Arnold (1965) stated that all accomplished esophageal speakers produce a clearly audible “trapping noise” during intake of air. Several investigators reviewed the considerable difficulty encountered in eliminating klunking once it becomes established (Diedrich, 1971; Gardner, 1978; Martin, 1963; Salmon, 1971; Shanks, 1977). Martin stated that intake noise production is probably related to anatomic and physiologic differences among those speakers who produce klunking sounds. Martin stated that even a qualified speech instructor is unlikely to correct the problem. The majority of the authors cited above did not address the question of anatomic and physiologic differences as factors in klunking. Gardner (1978), Greene (1957), and Salmon (1971) refer to the habitual nature of klunking, implying that they believe klunking to be a learned behavior. The identification of klunking as an acquired behavior, if true, would not make it easier to extinguish, but would offer more hope for clinical control of the phenomenon as a consequence of clinical intervention than would the assumption of a physiological or anatomical anomaly. It would also set the direction therapeutic intervention would probably take. One physiological condition that could influence an esophageal speaker’s perception of his klunking is his hearing acuity. Studies by Martin, Hoops, and Shanks (1974) demonstrated a positive correlation between poor audition and poor esophageal speech. Because klunking is generally reported to be low pitched, and most acquired hearing loss is high frequency, one may assume that many esophageal speakers have sufficient low-frequency acuity to perceive the sound. It is possible, however, that many who produce the klunk do in fact perceive it, but have adapted to
28
A. W. KNOX er al.
it and ignore the behavior. The degree to which air intake noise may influence communication is largely unreported in the literature. Because klunking occurs immediately prior to an utterance, some authors describe klunking as a distraction rather than a direct interference to speech (Hyman, 1979; Shanks, 1979). No references identify the specific characteristics of klunking behavior that could cause it to be distracting. Hyman (1979) stated that a klunking noise may not affect intelligibility, but may adversely affect oral communication. Shanks (1979) included klunking with other distracting behaviors that are “not phonemic and not truly speech elements,” but that influence intelligibility by distracting the listener’s attention from the message. In summary, the literature has identified klunking as a problem for many esophageal speakers, but has neither determined its incidence nor quantified its severity. Some research workers have suggested etiology but have not confirmed their conjectures with experimental data. Treatment procedures have been described but appear to be based upon anecdotal evidence. An increased understanding of priorities appropriate to the clinical management of klunks might be facilitated by the identification of measurable correlates to listener acceptability of these noises. Of the activities ordinary and necessary to the clinical management of esophageal voice, how time or cost effective are the attempts at management of undesirable air intake noise? The apparent lack of objective criteria for listener judgments was a primary motivation for this study. METHOD Subjects Twenty-four volunteer laryngectomees attending the XIV Annual Institute in Laryngectomee Rehabilitation at Eastern Washington University, Cheney Washington, 1981 were the subjects. The criterion for participation was the ability to read the sample paragraph aloud. Recording Audiotape recording was accomplished in the audiometric suite at the University. Equipment consisted of a tape recorder (Sony model 366), a dynamic microphone (Electrovoice model 631B), and a high fidelity recording tape (Maxell UDXL). The input level from the microphone was adjusted to cause the VU meter to peak at zero for each subject. The tape speed was set at 7.5 ips. The stand-mounted microphone was placed approximately 1 foot in front and 1 foot to the side of the subject’s mouth. This placement minimized the effect of microphone sensitivity to extraneous body movement and stoma noise, while it provided high-quality recording of speech and the air intake noises.
JUDGMENTS
OF ESOPHAGEAL
AIR INTAKE
NOISE
29
Recordings were made between the hours of 9:00 a.m. and 4:30 p.m. and at least 45 minutes after eating to minimize any influence of digestion on esophageal voice function. After explaining the purpose and securing written consent, each subject was instructed to read aloud the first paragraph of the “Rainbow Passage” (Fairbanks, 1960) twice for practice. This practice reading enabled the subjects to adjust to the acoustic chamber and to “warm-up” their esophageal voices. Each subject notified the investigators when he was ready to begin. Judges and Ratings Four judges were obtained from the guest faculty of the Esophageal Voice Institute. Each judge had professional experience with hundreds of laryngectomees over a minimum of 20 years as a speech clinician. Hearing sensitivity of the judges was within normal limits for speech. Air Intake Acceptability Ratings Each judge had an individual listening session conducted in the control booth of the acoustic chamber. They were achieved by playing the recorded passages on the same tape recorder with the signal output to TDH 39 headphones. The recordings had been randomized onto several different tapes and were played back at a mean output level of 70 dB SPL. The tape was stopped for 2 minutes after each 10 minutes of play to minimize listener fatigue. Esophageal air intake noise was rated on a five-point equal appearing interval scale with 1 representing the least acceptable air intake noise and 5 indicating the most acceptable. Procedure for Overall Proficiency Ratings Overall proficiency ratings were available for 18 of the 24 subjects. These assessments were made by the faculty of the Laryngectomee Institute as part of the normal proceedings. Although the proficiency ratings and the air intake acceptability ratings took place during the same week and shared some of the same judges, they were conducted independently. The proficiency was judged during direct observation of subjects as each spoke briefly to an audience in a lecture hall. Those using esophageal voice were rated on a five-point rating scale; the rest were not considered for the research project. Instrumentation An acoustic analysis of the recorded speech samples was made with the aid of a Visi-Pitch model 6087, frequency and amplitude extractor mounted in and connected to a Tektronic storage oscilloscope. This in-
30
A. W. KNOX et al.
strument produced a graphic display of relative intensity over time. A 4set display time was used for this study. All segments of each subject’s performance were photographed for a permanent record. Each photograph was phonetically annotated to ensure correspondence with the segment of speech displayed on the oscilloscope. Fundamental frequency analysis established that the lowest klunk frequency of the subject group was 66 Hz, well above the 50-Hz low cutoff frequency of the Visi-Pitch. Procedures Air intake noise was readily identified on the oscilloscope as it typically preceded and followed signals of greater amplitude and duration. Examples of these klunk images are observable in Figure 1. The relatively smooth, straight sections of the trace represent the noise floor of the recorded image. Syllables are seen as the higher peaks, while the intake noises are displayed as the much lower peaks of brief duration. Intensity peaks were easily read on the grid scale. On the vertical axis of the display each large grid square represented 10 dB SL. Each horizontal grid square represented 0.4 sec. The number of syllables uttered in each sample was divided by the number of quantified audible air intakes in the sample. The resultant values were expressed as syllables per intake.
IO dB,
SPL
t .4
SECONDS
*
!
INTO
*
1 1 MANY*
*
BEAUTIFUL
Figure 1. Overlay traced from a Polaroid photograph of the Vi&Pitch, model 6087, display of an esophageal speech sample. Asterisks represent esophageal air
intakes.
JUDGMENTS
OF ESOPHAGEAL
AIR INTAKE
NOISE
31
The rate of speech was calculated by timing each sample with a quartz crystal chronograph. Each sample’s time measurement was started automatically by the zero-cross output pulse generated by a level zone detector (Grason Stradler 1270), upon detecting the onset of the subject’s voice. The timing was stopped manually when the utterance was concluded. The number of syllables in each sample was divided by their timed length. The rate of speech shown is in syllables per second. The number of syllables uttered per esophageal air intake has been identified as one of the skills associated with the acquisition of esophageal speech (Berlin, 1963; Damste, 1979; Gardner, 1978; Snidecor and Curry, 1959). This measure is usually obtained by counting a subject’s repetitions of a designated monosyllable between air intakes. The repetition of a monosyllable obviously differs from the connected speech task used in this study. The purpose of using syllables of connected speech between intakes was to provide a simple and sensitive indicator of the frequency of occurrence of audible air intakes.
STATISTICAL ANALYSIS Interjudge Reliability The Pearson product-moment correlation coefficient (r) compared the ratings from each judge with the mean rating from all judges. These ranged from .78 to .86 (Table I).
Intrajudge Reliability The internal consistency of the judges was measured by having each of them rejudge four of the subjects. The r values and corresponding significance levels for the test-retest scores ranged from 88 to .98 (Table 2).
Correlation of Objective Measurements with Listener Judgments The mean-of-judges ratings for acceptability for each of the 24 subjects was entered into a multiple regression analysis (Kim and Kohout, 1975), which compared the ratings with objective measures of esophageal speech Table 1. Correlation
between Each Judge’s Rating and Mean of Judges’ Ratings Judge
r
Significance level
A B C D
0.86 0.78 0.80 0.80
.Ol .Ol .Ol .Ol
A. W. KNOX et al.
32 Table 2. Correlation between Judge’s Test and Retest Ratings Judge
r
Significance level
A B C D
0.96 0.88 0.95 0.98
.OS .I0 .05 .02
used in the study. This analysis yielded the Pearson product moment correlation coefficient (r). The regression analysis provided the coefficient of determination (R2). In the complex relationship of the multiple regression the degree to which any of the objective measurements of speech can be predicted from the judges’ ratings is indicated by the correlation coefficient. Additionally, the coefficient of determination indicates the percentage of the total variance that can be accounted for by the measurements included in the regression application. F ratios were computed for each measurement variable and compared with a table of critical values for F at the .Ol level. Association of Listener Judgments of Speech Proficiency and Noise Acceptability Independently conducted speech proficiency ratings for 18 of the subjects were compared with the intake noise ratings obtained from the study. The overall speech proficiency scores were included as predictor variables in the multiple regression analysis described above. RESULTS AND DISCUSSION Are Selected Objective Measures of Esophageal Speech Signilkantly Correlated with Sophisticated Listener Judgments of Ah- Intake Noise Acceptability? Intake noise acceptability ratings were found to be significantly correlated with the four measurements selected. These measurements were: the number of syllables uttered per air intake, the rate of speech, the mean intensity of air intake noise, and the mean speech intensity. The multiple regression analysis indicated that approximately 83 percent of the total variance in the listener ratings of air intake noise acceptability could be accounted for by the measured variables. The measurement that exhibited the most predictive strength was the number of syllables uttered per audible intake. Computed F values for each variable exceed the critical
JUDGMENTS
OF ESOPHAGEAL
AIR INTAKE
NOISE
33
values of F beyond the .Ol level. The results of the statistical analysis for each parameter are presented in Tables 3 and 4.
The Mean Intensity of Air Intake Noise Intake noise intensity was found to be inversely associated to the judges’ ratings of intake noise with a correlation coefficient (r) of - .71. The greater the intensity of the intake noise, the lower was the level of listener acceptability. This was significant at the .Ol level with 20 degrees of freedom. The coefficient of determination (R2), or the change caused by including this variable in the regression analysis, was .13, indicating that this variable accounted for 13 percent of the variance of the ratings.
The Mean Intensity of Speech Mean speech intensity was negatively correlated with intake acceptability ratings with an r value of - .14. The R2 change for this measure accounts for less than 1 percent of the total variance of the listener ratings.
The Number of Syllables per Audible Intake of Air This was the strongest prediction variable in the study. The positive r of .81 was significant at the .Ol level with 20 degrees of freedom. This variable accounted for 65 percent of the variance in listener ratings of intake noise.
Table 3. Correlation
Matrix for Esophageal
Intake noise acceptability Intake noise acceptability Intake intensity Speech intensity Syllables per intake Rate in syllabledsec Esophageal speech proficiency
l.C!0OOO
Air Intake Noise Acceptability
Intake intensity
Speech intensity
- 0.70765
- 0.13747
Syllables Per intake 0.80794
Rate in Syllables/ set
Esophageal speech proficiency”
0.63313
0.16642 0.07299
- 0.70765
l.OWOO
0.10942
-0.48397
- 0.37300
-0.13747
0.10942
l.OOOWJ
-0.32334
0.01262
0.80794
-0.48397
-0.32334
-0.09612
l.OOOKl
0.68037
0.50961
0.63313
0.37300
0.01262
0.68037
l.OOOOo
0.47564
0.16642
0.07299
0.09612
0.50961
0.47564
1.om
E Sample size (N) for esophageal speech proficiency was 18; for all objective measures N = 24.
.88667
.89961
.90827
.91084
2. Intake
3. Skill
4. Rate
5. Speech
.82963
.82495
.80930
.78618
.65276
R2
.004
,016
.023
.133
.652
variance
Percent total regression residual regression residual regression residual regression residual regression residual
1 22 2 21 3 20 4 19 5 18
Degrees of freedom
12.64558 6.72682 15.23016 4.14224 15.67802 3.69437 15.93180 3.39109 16.07194 3.30045
Sums of squares
12.64558 0.30576 7.61058 0.19725 5.22601 0.18472 3.99533 0.17848 2.67866 0.19414
Mean square
13.79725
22.38546
28.29171
38.00631
41.35725
F
.Ol
.Ol
.Ol
.Ol
.Ol
Significance level
.-
- ._ -
-
.,.
_
-
-
-
0 Dependent variable: acceptability. b (1) “Syll.” is syllables per audible intake of air (klunk). (2) “Intake” is the intensity of the intake sound. (3) “Skill” is the mean judges’ ratings of esphageal speech. (4) “Rate” is syllables per second. (5) “Speech” is intensity of speech in dBSPL.
.80794
R
variableb
1. Syll.
Multiple
Independent
Table 4. Multiple Regression and Analysis of Variance”
^.
JUDGMENTS OF ESOPHAGEAL AIR INTAKE NOISE
35
The Rate of Speech Rate was positively correlated with acceptability ratings at the .Ol level with an I value of .63. The multiple regression coefficient of determination, R*,showed only 1.5 percent of the total variance was accounted for by rate.
The Correlation of Listener Judgments of Esophageal Speech Proficiency and Air Intake Noise Acceptability The statistical analysis revealed that ratings of overall esophageal speech proficiency were only weakly correlated with ratings of air intake noise acceptability. The r value of. 17 was indicative of weak correlation. The multiple regression analysis indicated that overall speech proficiency ratings accounted for about 2 percent of the total variance of the intake noise acceptability ratings.
Discussion The number of syllables uttered per audible air intake was by far the strongest predictor of sophisticated listener acceptance of air intake noise. Mean intensity of intake noise was also predictive, but to a lesser extent. The rate of speech was significantly correlated with listener acceptance of intake noise, but was a weak predictor of such acceptance. Mean speech intensity was not significantly correlated with listener acceptance of intake noise. The predictive power of independent variables in a multiple regression analysis is affected by the variables selected for computation. In a perceptual task, such as judging voice, it is unlikely that all appropriate measurable variables will be included. It is also unlikely that all the variance of intake noise acceptability can be accounted for by physical correlates. Within the limits imposed by these probabilities the results presented here reveal clear differences in the predictive values of the measures chosen for this study. The strong correlation between syllables per intake and intake noise acceptability is explained if the syllables-per-intake parameter is viewed as the rate of occurrence of audible air intakes. From this perspective the listeners tended to judge intake noise less critically as the number of intakes decreased. While it is not surprising that listener acceptability of intake noise improved as the quantity decreased, it is interesting that syllables per audible intake was a much stronger predictor of acceptability than the intensity of the intake noise. Apparently the listeners could accept the higher intake noise intensity if the audible intakes did not occur too frequently.
36
A. W. KNOX et al.
Many clinicians are familiar with esophageal speakers who klunk but otherwise have superior esophageal voices. It was hypothesized that if esophageal speech intensity were sufficiently high it would compensate for above average intake noise intensity. The assumption was that higher speech intensity would be predictive of greater acceptability. The results failed to support this hypothesis. The mean speech intensity measurements were not highly correlated with either intake noise acceptability or overall speech proficiency ratings. The low correlation of speech intensity with overall speech proficiency is consistent with the findings of Hoops and No11 (1979). Rate of speech was moderately correlated with intake noise acceptability, but accounted for less than 2 percent of the total variance, providing support to Snidecor and Curry (1959) and Hoops and No11 (1969). This association of rate and proficiency does not ensure that a faster rate will be predictive of greater intake noise acceptability, as intake noise acceptability is just one of many factors that influence overall speech proficiency ratings. The Correlation of Listener Judgments of Esophageal Speech Proficiency and Listener Judgments of Intake Noise Acceptability The findings of this study support the concept that the sophisticated listeners’ ratings of overall esophageal speech proficiency had little association with their ratings of air intake noise acceptability. While klunking has frequently been described as a problem requiring clinical intervention, this study found that klunking has little impact on sophisticated listener judgment of overall proficiency. A rationale for this mild influence may be found in the prephonatory characteristic of esophageal inflation. Klunking may distract the listener, but apparently it is not sufftciently distracting to impair communications seriously. Implications Two main implications emerged from these results, and the findings of a previous investigation were supported. The first finding was that the number of syllables uttered per intake was a stronger predictor of intake noise acceptability than was mean intake intensity. This implies that clinical and practice time might be better spent improving the duration of speech per audible intake rather than working directly on reducing the intensity of the intake noise. Even if the intake noise intensity remains constant, the increase in the production of syllables per intake may improve both intake noise acceptability and overall speech proficiency. The second finding, that intake noise acceptability is not a significant correlate of overall esophageal speech proficiency, also implies that prac-
JUDGMENTS
OF ESOPHAGEAL
AIR INTAKE
NOISE
37
tice should be concentrated on other aspects of speech, such as duration, quality, and inflection. The finding that intensity of esophageal speech is not highly correlated to proficiency, which ‘supported Hoops and No11 (1969), should be of interest to those who arduously struggle to increase volume at the expense of other more contributory factors to success of the esophageal speech process.
Future Research It is unlikely that the parameters included in this study exhaust the possibilities for variance in the air intake phenomena. Certainly the impact of visually perceived behaviors accompanying air intake could be observed and rated to determine whether the klunking variance extends beyond the auditory impression, a limitation of the audiotape recording method of study. Additional comparison and correlation could be made of the studied phenomenon using naive listeners to determine whether naive and sophisticated listeners apply the same standards. Acoustically oriented research workers may be interested in exploring the effects of spectral components and temporal relationships. Intake latency, pause time, position of intake-within-pause, and rate of multiple intakes are all addressable parameters that have not been systematically investigated. Finally, spectral analysis of esophageal air intake noise may yield a more precise description of the klunk and a clearer insight into its etiology than the conjectural descriptions of klunking available to clinicians today. The authors unite in expressing their gratitude to the faculty and staff of Eastern Washington University, particularly Dr. Herbert E. Gunderson, to the laryngectomized individuals who freely gave their time and efforts to this research, and to the judges whose volunteer efforts, willingly given, are testimony to the honors that have accrued to them individually. This research was supported in part with funds from the Research and Development Committee of the Kansas City Veterans Administration Medical Center Project No. 589-103-473-361661-006 and the Communications Research Laboratory.
REFERENCES American Cancer Society (1978). Facts on Cancer of the Larynx. New York: American Cancer Society. Berlin, C. I. (1963). Clinical measurement of esophageal speech: I. Methodology and curves of skill acquisition. J. Speech Hear. Disord. 28:42-51. Berlin, C. I. (1965). Hearing loss, palatal function, and other factors in post leayngectomy rehabilitation. J. Chronic Disord. 30: 174-183. Brown, F. L., Amos, J. R. and Mink, 0. G. (1975). Statistical Concepts, A Basic Program. New York: Harper and Row.
38
A. W. KNOX et al.
Damste, P. H. (1979) Rehabilitation score-list for the laryngectomized. In R. L. Keith and F. L. Darley (eds.), Laryngectomee Rehabilitntion. Houston: College Hill Press. Diedrich, W. M. (1968). The mechanism Sci. 15.5:303-317.
of esophageal speech. Ann. N. Y. Acad.
Diedrich, W. M. (1971). Primary stages of teaching. In S. Rigrodsky, J. Lerman and E. Morrison (eds.), Therapy for the Laryngectomized Patient. New York: Teachers College Press. Diedrich, W. M., and Youngstrom, IL: Charles C. Thomas.
K. A. (1966). Alaryngeal Speech. Springfield,
Duguay, M. J. (1977). Esophageal speech. In M. Cooper and M. H. Cooper (eds.), Approaches to Vocal Rehabilitation. Springfield, IL: Charles C. Thomas. Fairbanks, G. (1960. Voice and Articulation Drill Book (2nd edition). Philadelphia: Harper and Row. Filter, M. H. (1971). Rating systems for esophageal toral dissertation, Bowling Green University. Gardner, W. H. (1978). Laryngectomee IL: Charles C. Thomas. Greene,
speakers.
Unpublished
Speech and Rehabilitation.
M. C. L. (1957). The Voice and Its Disorders. London:
doc-
Springfield,
Pitman.
Hoops, H. R., and Noll, J. D. (1969). Relationship of selected acoustic variables to judgements of esophageal speech. J. Commun. Disord. 2:1-13. Horn, D. (1962). Laryngectomee survey report. Presented at the Eleventh Annual Meeting, International Association of Laryngectomees, Memphis, Tennessee, August 21. Hyman, M. (1979). Factors influencing intelligibility of alaryngeal speech. In R. L. Keith and F. L. Darley (eds.), Laryngectomee Rehabilitation. Houston: College Hill Press. Hyman, M. (1971). Intermediate stage of teaching alaryngeal speech. In S. Rigrodsky, J. Lerman and E. Morrison (eds.), Therapy for the Laryngectomized Patient. New York: Teachers College Press. Kallen, L. A. (1934). Vicarious vocal mechanisms, the anatomy, physiology and development of speech in laryngectomized persons. A.M.A. Arch. Otolaryngol. 20:460-503. Kim, J. and Kohout, F. J. (1975). Multiple regression analysis: Sub-program regression. In W. R. Klecka, N. H. Nie and C. H. Hull (eds), SPSS primer: Statistical Package for the Social Sciences. New York: McGraw-Hill. Lauder, E. (1972). Self-Help for the Laryngectomee. Lauder.
San Antonio:
Edmund
Luchsinger, D. and Arnold, G. (1965). Voice-Speech-Lanaguage, Clinical Communicology: Its Physiology and Pathology. Belmont, CA: Wordsworth. Martin, D. E. (1979). Evaluating esophageal speech development and proficiency. In R. L. Keith and F. L. Dailey (eds.), Laryngectomee Rehabilitation. Houston: College Hill Press.
JUDGMENTS
OF ESOPHAGEAL
AIR INTAKE
NOISE
39
Martin, D. E., Hoops, H. R. and Shanks, J. C. (1974). The relationship between esophageal speech proficiency and selected measures of auditory function. J. Speech Hear. Res. 74:80-85. Martin, H. (1963). Rehabilitation
of the laryngectomee.
Cancer 16:559-574.
Salmon, S. J. (1971). Use of imagery in teaching esophageal Commun. Disord. 2~17-24.
speech.
Cal6
J.
Shames, G. H., Font, J. and Mathews, J. (1963). Factors related to speech proficiency of the laryngectomized. J. Speech Hear. Disord. 28:273-287. Shanks, J. C. (1977). Variations on a vocal theme through interpersonal therapy with alaryngal patients. In M. Cooper and M. H. Cooper (eds.), Approaches to Vocal Rehabilitation. Springfield, IL: Charles C. Thomas. Shanks, J. C. (1979). Developing esophageal communication. In R. L. Keith and F. L. Darley (eds.), Laryngectomee Rehabilitation. Houston: College Hill Press. Shipp, T. (1967). Frequency, duration and perceptual memasures in relation to judgements of alaryngeal speech acceptability. J. Speech Hear. Res. 10:417427. Snidecor, J. C. (1968). Speech rehabilitation IL: Charles C. Thomas.
of the laryngectomized.
Springfield,
Snidecor, J. C. and Curry, E. T. (1959). Temporal and pitch aspects of superior esophageal speech. Ann. Otol. Rhinol. Laryngol. 68:1-14. Waldrop, W. and Gould, M. A. (1956). Your New Voice. Chicago, American Cancer Society.