- Email: [email protected]
The perceptual identity and characteristics of aging in normal male adult speakers
JOURNAL
OF COMMUNICATION
DISORDERS
53
12 (19791, 53-61
THE PERCEPTUAL IDENTITY AND CHARACTERISTICS OF AGING IN NORMAL MALE ADULT SPEAKERS DAVID E. HARTMAN Department
of Communicative
Disorders,
College of Saint Teresa,
Winona, Minnesota 55987
Twenty untrained male and 20 untrained female listeners were requested to judge the age of 46 male speakers from samples of conversational speech. Thirty-seven samples were judged similarly by the two groups and were therefore provided as listening stimuli for ten male and ten female speech pathologists. Through an a posteriori schema, speech pathologists were required to specify the salient features of speech judged as being characteristic of each speaker’s perceived age. The results of the study indicated that untrained listeners used similar perceptual sets to judge age until after the fiftieth year of life. Regardless of sex, listeners tended to underestimate the age of the speakers. Male and female speech pathologists viewed low pitch and hoarseness as being the most salient features of speech for the speakers. Qualitative analysis revealed that the most prominent features indicated by speech pathologists could be classified according to pitch, quality, articulation, and rate of speech.
Introduction In recent years, increasing attention has been directed toward the aging process. Coinciding with this research, investigators have noted anatomical and physiological changes in the speaking apparatus that are related to perceptual and acoustic changes in speech (Mysak, 1959; Levin, 1962; McGlone and Hollien, 1963; Greene, 1964; Luchsinger and Arnold, 1965; Ptacek et al., 1966; Zemlin, 1968; Kaplan, 1971; Hollien and Shipp, 1972; Ryan, 1972). Moreover, it has been observed that listeners are able to utilize perceptual characteristics of aging speech in order to make an age judgment of a speaker. Ptacek and Sander (1966) found that listeners were able to differentiate the voices of younger adults [under age 35) from older subjects (over 65) on the basis of a prolonged vowel and oral readings. Shipp and Hollien (I 969) used a tape-recorded passage of reading as a stimulus for 25 listeners to make direct age estimations of 175 male speakers between the age of 20 and 89 yr. The results indicated that untrained listeners could accurately estimate age along a 70-yr continuum. Ryan and Burk (1974) had naive listeners make direct age estimates of male speakers between 40 and 80 yr who had recorded a specified reading passage. Perceptual judgments were made as to the presence or absence of designated speech characteristics. In addition, selected samples were subjected to acoustic analysis. The data revealed that five perceptual features were strong predictors of perceived age.
The author’s present address is: Section of Speech Pathology, Clinic, Rochester, MN 55901.
o Elsevier North Holland Publishing Company, Inc., 1979
Department
of Neurology,
Mayo
0021.9924/79/001053-09$01.75
54
DAVID
E. HARTMAN
From these investigations it can be inferred that the speech signal contains perceptual features that can be used by a listener in making a judgment of a speaker’s age. This observation implies that a listener utilizes variations in acoustic cues, which are a direct or indirect result of anatomical and physiological changes associated with the aging process. Although the inferences set forth by previous research may prove tenable, several important factors have not been considered in the conduct of the studies. First, the “closed response” and a priori format utilized in these investigations may have excluded critical perceptual features from analysis; that is, listeners have been told the upper and lower age limits of the speakers prior to the experiment, and. limits have been placed on the number of perceptual features from which a listener is allowed to choose. The sole attempt at an a posteriori description of the perceptual features of aging speech was presented by Ptacek and Sander (1966), who used an informal method to secure information from listeners. Indeed, they reported that, “The preceding comments from listeners were obtained in a casual manner and are presented primarily for their suggestive value” (p. 276). Second, previous research has used a reading passage or a prolonged vowel as the speaking stimulus. It has been shown that the “unnaturalness” of such situations may influence the way a speaker operates and therefore how he is perceived (Snidecor, 1943; Bricker and Pruzansky , 1966; Tiffany and Hollien, 1968; Michel and Wendahl, 1971; Faircloth et al., 1974; Singh et al., 1974). Because of the nature of the task, the speaker may feel pressed to maximize his speaking behavior, leading to overarticulation, changes in habitual pitch, and the like. Third, the sex of the listener has not been regarded as an active variable. Investigators have reported either using a “group of listeners” without respect to sex, or a homogeneous population of listeners (Ryan and Burk, 1974). It is possible that males use different perceptual features than females in identifying age and therefore identify it differently. The purposes of this investigation were to determine (1) if there was a significant difference in the judgments of untrained male and female listeners in identifying the age of male speakers from spontaneous conversation, (2) the perceptual features of speech that male and female speech pathologists identify as being characteristic of a judged speaker’s age, and (3) how male and female speech pathologists rank these perceptual features on the basis of similarity and frequency of occurrence. Method and Results In order to accomplish the objectives of this study, it was divided into two parts. The character of the second part was dependent on the findings of the first.
PERCEPTUAL
IDENTITY
OF AGING
IN MALE
SPEAKERS
55
First Procedure Forty-six white males, aged 25-70 yr and with normal speech and Speakers. hearing, served as speakers. The lower age limit was chosen to insure that pubescent voice change had occurred (Hollien and Malcik, 1967). The upper age limit was selected to insure that speech involution had or was occurring (Greene, 1964; Luchsinger and Arnold, 196.5; Zemlin, 1968). Five minutes of conversational speech was tape-recorded for each of the 46 speakers during a discussion of a neutral topic with the examiner. Conversations took place in a sound-treated room, and all discourse was recorded on a Tandberg Model 9141X tape recorder and an Advent Noise Reduction Unit using the Dolby system. From each 5-min conversational sample, 30 set of continuous speech containing less than 9 set of pausing (extrapolated from Singh et al., 1974) was randomly selected and spliced to form the experimental listening tape. Six additional 30-set samples were randomly added to the tape in order to test intrajudge reliability. Listeners. Twenty males and 20 females with normal hearing binaurally were selected as listeners. None had prior knowledge of the research. Procedures. Listeners were seated in a listening laboratory and speech samples were presented binaurally through TDH-49 earphones at a comfortable loudness level. The instrumentation used for playback of the experimental tape was the same as that used for recording of speech samples. Listeners were allowed 5 set to make direct age estimations by year for each 30-set sample. Prior to the task, verbal and written instructions were provided to the listeners. Results of First Procedure For the listening task, the Pearson Product-Moment Correlation Coefficient (Ferguson, 1971) yielded correlations of 0.93 for males and 0.94 for females for mean perceived age estimations, indicating high intra- and interjudge reliability. Multivariate F-tests and a discriminant analyses (Bruning and Kin&, 1968) revealed a nonsignificant difference in the age estimations of the male and female listeners (df = 23, 16; F = 0.70; p < 0.7913; df = 23, 16; F = 0.71; p < 0.7758). However, both groups of listeners tended generally to underestimate the age of the speakers. Furthermore, univariate F values were determined for the perceived age of each speaker. Thirty-seven speakers whose individual F ratios were at or exceeded the 0.10 level of confidence were selected for further analysis in the second part of the study. Those speakers whose ages showed a significant difference in age judgments between the male and female listeners (ages 50,51,58,60,64,67,68,69,70) were at or beyond the fifth decade of life and reflected lower mean perceived age estimations by the male listeners. Second Procedure Listeners. Ten male and 10 female speech pathologists with normal hearing binaurally served as listeners. Each possessed a masters degree in speech pathol-
56
DAVID E. HARTMAN
ogy and the Certificate of Clinical Competence in speech pathology or eligibility for the Clinical Fellowship Year. Procedures. The listening environment was the same as that indicated for the first procedure. Prior to listening to each speech sample, listeners were told the mean perceived age of each speaker. Their task was to orthographically describe, as specifically as possible, the salient characteristics of each speaker’s speech in reference to his perceived age. Listeners were allowed 45 set to write their descriptions. Verbal and written instructions were given to the listeners before the listening task. Results of Second Procedure Immediately following the listening task, listeners were asked to verbally clarify any descriptions considered ambiguous by the examiner. The responses were then assembled according to similarity and frequency of occurrence. Table 1 presents the incidence with which male and female speech pathologists specified the same features for the 37 male speakers. For purposes of comparison, the chronological age (CA) and the grand mean perceived age (GYPA) of each speaker is provided. The percent of listeners who indicated the presence of a feature for each speaker is also presented. Table 1 reveals that of the 25 features used by both male and female listeners, low pitch (23%) was used most frequently. Hoarseness (15%), rapid rate (14%)) precise articulation (13%), slow rate (13%)) high pitch (12%), and clear quality (11%) were specified at least 10% of the time. Analysis of features by perceived age reveals that rapid rate, high pitch, precise articulation, and clear quality appeared to be the best discriminators of speakers who were judged to be below 30 yr. For speakers who were perceived to be between 30 and 40 yr, low pitch, precise articulation, clear quality, and moderate pitch appeared to be the dominant indices for describing the speech of speakers in this decade. The initial appearance of hoarseness, glottal fry, breathiness, and long pauses is also recognized during this period. For speakers who were judged to be between 40 and 50 yr of age, low pitch, hoarseness, imprecise articulation, and moderate rate appeared to be the most representative features of their speech. The dominant features of speech for speakers perceived to be within the 5@60 yr age range appeared to be low pitch, hoarseness, slow rate, imprecise articulation, breathiness, and long pauses. Examination of the most prominent features indicated by the listeners reveals that they may be classified according to pitch, quality, articulation, and rate of speech. From the judgments of this population, it can be postulated that these four parameters of speech serve as the primary perceptual bases for describing the speech of males ranging in age from 25 to 66 yr.
PERCEPTUAL
IDENTITY
OF AGING
IN MALE SPEAKERS
57
Discussion First Procedure
The finding that there was a nonsignificant difference in the overall age judgments of untrained male and female listeners, and that female listeners were consistently more accurate in judging age after the fifth decade, represents original observations. It could be postulated that male and female listeners use similar perceptual sets for judging age until after the fiftieth year. From this point to the beginning of the seventh decade, male perceptions of aging speech change, possibly because of egocentric involvement associated with aging. This premise is substantiated, in part, by studying those ages that reflected a significant difference in mean perceived age estimations. All were within or beyond the fifth decade and exemplified lower age judgments on the part of the male listeners. The trend for untrained listeners to underestimate the age of male speakers was congruent with earlier research (Shipp and Hollien, 1969). This phenomenon, which has been discussed by these authors, may be a reflection of the age of the listeners themselves. In their study and in the present investigation, listeners were “college age” students-under 30 yr of age. The above findings have important experimental implications for future research that utilizes untrained male listeners for making age judgments. First, in order to either substantiate or negate the premise that listeners alter their perceptual sets when making an age judgment of someone their own sex, the present study should be replicated using female speakers. Second, listeners should be matched with speakers in terms of chronological age in order to determine if more accurate age estimations are possible when persons in both populations are of similar age. Second Procedure
For the 2.5 features of speech identified by speech pathologists as being characteristic of male speech across four perceived (age decades, the finding that low pitch and hoarseness were the most recurring was not unexpected. Analysis of individual perceived ages reveals that 76% of the speakers were judged to be older than 30 yr of age. Examination of the third, fourth, and fifth perceived age decades shows that low pitch and hoarseness initially appeared during the third decade, and were dominant features for the latter two decades. Ptacek and Sander (1966) found that some of the perceptual features characteristic of “old” speakers (67-87 yr) were low pitch, hoarseness, reduced intensity, hesitancy, and less pitch variability. Comparable features indicated in the present investigation were specified for speakers whose chronological age ranged from 25 to 66 yr. The majority of these speakers (86%) were younger than 60 yr
58
TABLE 1 Features Indicated by 10 Male and 10 Female Speech Pathologists Features and incidence
of occurrence
for 37 Male Speakers
(%)
CA 27.60 25.00 33.00 26.00 37.00 29.00 38.00 28.00 56.00 30.00 57.00 35.00 44.00 42.00 49.00 31.00 53.00 40.00 41.00 36.00 48.00 39.00 32.00 45.00 43.00 46.00 47.00 55.00 62.00 63.00 54.00 61.00 34.00 52.00 65.00 59.00 66.00
22.88 23.63 24.30 24.65 26.80 27.76 27.95 28.48 30.85 31.16 31.38 31.43 31.56 31.95 32.13 34.00 34.66 34.75 38.15 38.23 38.45 39.43 39.50 41.80 42.40 42.86 45.30 47.40 48.10 48.66 48.82 50.95 51.20 54.13 55.15 55.58 56.73
CA = chronological
5 5
5 5 15 10 25 20 5 10 30 45 30 50 5 40 5 15 35 20 15 65 10 80 25 60 35 5 25 55 40 15 25 15
15 5
65 40 25 65 20 40 15 55 45 5 35 15
15 5 5 20 5 IO 15 15 15 15 5 20 15 25 50 10 40 10 45 45 40 60 30 30
30
35 10 5
20 20 30 10 30 20 10 20 15 15 5 40 15 10 10 30 10 15 20 30 35 20 25 5 5 5 5 5 5
5 5 5 10 20 25 10 30 10 15 15 25 5 25 15 10 15 5 30 25 50
15
35 20 50 40 60 25 55 10 15
10 10 5
5 25 5
20 30 10 15 5 20 35 15 10 10 10 20 10 15 5 25 5 30 20 10 10 15 10
10 15 5 15 5
5
20
10
5 5
45
10 15 20 20 45 15
age; GE = grand mean perceived
5 age.
5 10
5
10 15 20 5 10
5 25 10 15
25 25 5 10 10 15 35 5 5 10 30
5 5 10 5 5 5 15 5 10 3.5 20 20 25 20 5 15 30 25
10 10 20 15 10 15 15 30 5 30 15 5 20 5 5 10
30 25 10 5 5 5
5 5
5 5 25
5 25 10 5 25 10 30 15 20 5 15 10 15 5 10 5 20 10 10 10 5 5 5 10
59 TABLE
1 (continued)
Features and incidence
20 25 15 30
5 5
15
10 5 10 5 10 30 20 10 20 5 10 15 15 5 10 5 10 15 10
10
10 5 10 10
5 15 5 5 15 5 20
15
10 5 5 25 5
10
5 5
20 10 20 5 25 30 10 5 5
20 10
5
5 5 5
20 5 25
5 10
5 15 5 5 10 5
5
10 5 5 5
5 5 5 5
5 5 10 10
5 10 35
5
5
25 5 5 15 5 45
15 5 10
5 5 5 5
50 10 5 25 10 45
15 20 20 5 20 5
15 10 5
5 5
5 10
5 5 5 5 10 5
15 20 15
5
5
5
5
15 5
10 20 5
5 5
5 5
10
5 5 5
5 5
5 10
IO
5 30
15 15 5 15
(%)
15
10 10
10 5
5
of occurrence
5 5 5 5 10 5 5 10 5 5 5
10
5 10 15 5 5
20 5
10 5
5 5 5 10 5 5 10
5
15
10
5
5
5
5
10
5
20 15
10 10
5
10
5 5 5
5
20
5
1.5
5
5 5 10 5 5
5 5
10 5
10 20 15 15 5 5 10 5 5
15 10 5 5 5 15 10 15 5 5 5
5
5
5 5
60
DAVID
E. HARTMAN
of age. Ryan (1972) found that men between 60 and 80 yr displayed a significant decrease in reading rate when compared to those in the age range 40-50 yr. In the present investigation, slow speaking rate was a prominent feature of speakers who ranged in age from 34 to 66 yr (x = 56.16 yr). Additionally, Ryan and Burk (1974) observed that five perceptual features (air loss, laryngeal tension, voice tremor, imprecise consonants, and slow rate of articulation) were effective predictors of perceived age of males 40-80 yr of age. With the exception of voice tremor, which was not specified in this investigation, comparable features were considered prominent for speakers who ranged in age from 34 to 66 yr. The implications of these results are that the features previously indicated as being characteristic of older speakers may be, in reality, salient features of younger speakers also. However, when entertaining this conclusion, consideration should be given to differences in research design and the ages of the speakers utilized in earlier studies. The most tangible similarity between this and earlier research is that the majority of features indicated by all listeners can be classified according to pitch, quality, articulation, and rate of speech. In view of this finding, these parameters could be considered the primary perceptual characteristics of aging speech in the male population. This research was based on a doctoral dissertation completed at Bowling Green State University, Bowling Green, Ohio. Special thanks are given to Melvin Hyman, who served as dissertation advisor, and Jeffrey Danhauer, William Hinkle, and Raymond Tucker, who assisted in the design of the study. Portions of this investigation were presented at the Annual Convention of the Acoustical Society of America, San Francisco, November 1975, and appeared in the Journal of the Acoustical Society of America, (1976), 59: 713-715. References Bricker, P. D., and Pruzansky, S. (1966). Effects of stimulus content and duration on talker identification. J. Acoust. Sot. Am. 40: 1441-1449. Bruning, I. L., and Kintz, B. L. (1968). Computationa/ handbook of statistics. Glenview: Scott Foresman. Faircloth, M. A., Faircloth, S. R., and Singh, S. (1974). Phonologic proficiency in the spontaneous speech of trainable mentally retarded subjects. Unpublished manuscript. Ferguson, G. A. (1971). Sraristical analysis in psychology and education. New York: McGraw-Hill. Greene, M. C. (1964). The voice and ifs disorders. Philadelphia: Lippincott. Hollien, H., and Malcik, E. (1967). Evaluation of cross-sectional studies of adolescent voice change in males. Speech Monogr. 34: G-84. Hollien, H., and Shipp, T. (1972). Speaking fundamental frequency and chronologic age in males. J. Speech Hear. Res. 15: 155-159. Kaplan, H. M. (1971). Anatomy and physiology of speech. New York: McGraw-Hill.
PERCEPTUAL
IDENTITY
OF AGING IN MALE SPEAKERS
61
Levin, N. M. (1962). Voice and speech disorders: medial aspects. Springfield: Charles C Thomas. Luchsinger, R., and Arnold, G. E. (1965). Voice-speech-language. Belmont: Wadsworth. McGlone, R. E., and Hollien, H. (1963). Vocal pitch characteristics of aged women. J. Speech Hear.
Michel,
Res. 6: 16+170.
J. F., and Wendahl,
Handbook
Mysak,
R. (1971).
of speech pathology
Correlates
and audiology.
E. D. (1959). Pitch and duration
of voice production.
In L. E. Travis (ed.),
New York: Appleton-Century-Crofts.
characteristics
of older males. /. Speech Hear.
Res. 2:
46-54.
Ptacek,
P. H., and Sander,
E. K. (1966).
Age recognition
from voice. J. Speech Hear.
Res. 9:
27>277.
Ptacek, P. H., Sander, E. K., Maloney. W. H., and Jackson, C. R. (1966). Phonatory and related changes with advanced age. J. Speech Hear. Res. 9: 353-360. Ryan, W. J. (1972). Acoustic aspects of the aging voice. /. Gerontol. 27: 265-268. Ryan, W. J., and Burk, K. W. (1974). Perceptual and acoustic correlates of aging in the speech of males. J. Commun. Dis. 7: 181-192. Shipp, T., and Hollien, H. (1969). Perception of the aging male voice. J. Speech Hear. Res. 12: 703-710. Singh, S., Faircloth, M. A., and Faircloth, S. R. (1974). Articulatory proficiency in the conversational speech of mentally retarded subjects. Unpublished manuscript Snidecor, J. C. (1943). A comparative study of the pitch and duration characteristics of impromptu speaking and oral reading. Speech Monogr. IO: 5%56. Tiffany, W. R., and Hollien, H. (1968). Speaking fundamental frequency as a function of type of speech production. Unpublished manuscript. Zemlin, W. R. (1968). Speech and hearing science: anatomy and physiology. Englewood Cliffs, N.J.: Prentice-Hall.
OF COMMUNICATION
DISORDERS
53
12 (19791, 53-61
THE PERCEPTUAL IDENTITY AND CHARACTERISTICS OF AGING IN NORMAL MALE ADULT SPEAKERS DAVID E. HARTMAN Department
of Communicative
Disorders,
College of Saint Teresa,
Winona, Minnesota 55987
Twenty untrained male and 20 untrained female listeners were requested to judge the age of 46 male speakers from samples of conversational speech. Thirty-seven samples were judged similarly by the two groups and were therefore provided as listening stimuli for ten male and ten female speech pathologists. Through an a posteriori schema, speech pathologists were required to specify the salient features of speech judged as being characteristic of each speaker’s perceived age. The results of the study indicated that untrained listeners used similar perceptual sets to judge age until after the fiftieth year of life. Regardless of sex, listeners tended to underestimate the age of the speakers. Male and female speech pathologists viewed low pitch and hoarseness as being the most salient features of speech for the speakers. Qualitative analysis revealed that the most prominent features indicated by speech pathologists could be classified according to pitch, quality, articulation, and rate of speech.
Introduction In recent years, increasing attention has been directed toward the aging process. Coinciding with this research, investigators have noted anatomical and physiological changes in the speaking apparatus that are related to perceptual and acoustic changes in speech (Mysak, 1959; Levin, 1962; McGlone and Hollien, 1963; Greene, 1964; Luchsinger and Arnold, 1965; Ptacek et al., 1966; Zemlin, 1968; Kaplan, 1971; Hollien and Shipp, 1972; Ryan, 1972). Moreover, it has been observed that listeners are able to utilize perceptual characteristics of aging speech in order to make an age judgment of a speaker. Ptacek and Sander (1966) found that listeners were able to differentiate the voices of younger adults [under age 35) from older subjects (over 65) on the basis of a prolonged vowel and oral readings. Shipp and Hollien (I 969) used a tape-recorded passage of reading as a stimulus for 25 listeners to make direct age estimations of 175 male speakers between the age of 20 and 89 yr. The results indicated that untrained listeners could accurately estimate age along a 70-yr continuum. Ryan and Burk (1974) had naive listeners make direct age estimates of male speakers between 40 and 80 yr who had recorded a specified reading passage. Perceptual judgments were made as to the presence or absence of designated speech characteristics. In addition, selected samples were subjected to acoustic analysis. The data revealed that five perceptual features were strong predictors of perceived age.
The author’s present address is: Section of Speech Pathology, Clinic, Rochester, MN 55901.
o Elsevier North Holland Publishing Company, Inc., 1979
Department
of Neurology,
Mayo
0021.9924/79/001053-09$01.75
54
DAVID
E. HARTMAN
From these investigations it can be inferred that the speech signal contains perceptual features that can be used by a listener in making a judgment of a speaker’s age. This observation implies that a listener utilizes variations in acoustic cues, which are a direct or indirect result of anatomical and physiological changes associated with the aging process. Although the inferences set forth by previous research may prove tenable, several important factors have not been considered in the conduct of the studies. First, the “closed response” and a priori format utilized in these investigations may have excluded critical perceptual features from analysis; that is, listeners have been told the upper and lower age limits of the speakers prior to the experiment, and. limits have been placed on the number of perceptual features from which a listener is allowed to choose. The sole attempt at an a posteriori description of the perceptual features of aging speech was presented by Ptacek and Sander (1966), who used an informal method to secure information from listeners. Indeed, they reported that, “The preceding comments from listeners were obtained in a casual manner and are presented primarily for their suggestive value” (p. 276). Second, previous research has used a reading passage or a prolonged vowel as the speaking stimulus. It has been shown that the “unnaturalness” of such situations may influence the way a speaker operates and therefore how he is perceived (Snidecor, 1943; Bricker and Pruzansky , 1966; Tiffany and Hollien, 1968; Michel and Wendahl, 1971; Faircloth et al., 1974; Singh et al., 1974). Because of the nature of the task, the speaker may feel pressed to maximize his speaking behavior, leading to overarticulation, changes in habitual pitch, and the like. Third, the sex of the listener has not been regarded as an active variable. Investigators have reported either using a “group of listeners” without respect to sex, or a homogeneous population of listeners (Ryan and Burk, 1974). It is possible that males use different perceptual features than females in identifying age and therefore identify it differently. The purposes of this investigation were to determine (1) if there was a significant difference in the judgments of untrained male and female listeners in identifying the age of male speakers from spontaneous conversation, (2) the perceptual features of speech that male and female speech pathologists identify as being characteristic of a judged speaker’s age, and (3) how male and female speech pathologists rank these perceptual features on the basis of similarity and frequency of occurrence. Method and Results In order to accomplish the objectives of this study, it was divided into two parts. The character of the second part was dependent on the findings of the first.
PERCEPTUAL
IDENTITY
OF AGING
IN MALE
SPEAKERS
55
First Procedure Forty-six white males, aged 25-70 yr and with normal speech and Speakers. hearing, served as speakers. The lower age limit was chosen to insure that pubescent voice change had occurred (Hollien and Malcik, 1967). The upper age limit was selected to insure that speech involution had or was occurring (Greene, 1964; Luchsinger and Arnold, 196.5; Zemlin, 1968). Five minutes of conversational speech was tape-recorded for each of the 46 speakers during a discussion of a neutral topic with the examiner. Conversations took place in a sound-treated room, and all discourse was recorded on a Tandberg Model 9141X tape recorder and an Advent Noise Reduction Unit using the Dolby system. From each 5-min conversational sample, 30 set of continuous speech containing less than 9 set of pausing (extrapolated from Singh et al., 1974) was randomly selected and spliced to form the experimental listening tape. Six additional 30-set samples were randomly added to the tape in order to test intrajudge reliability. Listeners. Twenty males and 20 females with normal hearing binaurally were selected as listeners. None had prior knowledge of the research. Procedures. Listeners were seated in a listening laboratory and speech samples were presented binaurally through TDH-49 earphones at a comfortable loudness level. The instrumentation used for playback of the experimental tape was the same as that used for recording of speech samples. Listeners were allowed 5 set to make direct age estimations by year for each 30-set sample. Prior to the task, verbal and written instructions were provided to the listeners. Results of First Procedure For the listening task, the Pearson Product-Moment Correlation Coefficient (Ferguson, 1971) yielded correlations of 0.93 for males and 0.94 for females for mean perceived age estimations, indicating high intra- and interjudge reliability. Multivariate F-tests and a discriminant analyses (Bruning and Kin&, 1968) revealed a nonsignificant difference in the age estimations of the male and female listeners (df = 23, 16; F = 0.70; p < 0.7913; df = 23, 16; F = 0.71; p < 0.7758). However, both groups of listeners tended generally to underestimate the age of the speakers. Furthermore, univariate F values were determined for the perceived age of each speaker. Thirty-seven speakers whose individual F ratios were at or exceeded the 0.10 level of confidence were selected for further analysis in the second part of the study. Those speakers whose ages showed a significant difference in age judgments between the male and female listeners (ages 50,51,58,60,64,67,68,69,70) were at or beyond the fifth decade of life and reflected lower mean perceived age estimations by the male listeners. Second Procedure Listeners. Ten male and 10 female speech pathologists with normal hearing binaurally served as listeners. Each possessed a masters degree in speech pathol-
56
DAVID E. HARTMAN
ogy and the Certificate of Clinical Competence in speech pathology or eligibility for the Clinical Fellowship Year. Procedures. The listening environment was the same as that indicated for the first procedure. Prior to listening to each speech sample, listeners were told the mean perceived age of each speaker. Their task was to orthographically describe, as specifically as possible, the salient characteristics of each speaker’s speech in reference to his perceived age. Listeners were allowed 45 set to write their descriptions. Verbal and written instructions were given to the listeners before the listening task. Results of Second Procedure Immediately following the listening task, listeners were asked to verbally clarify any descriptions considered ambiguous by the examiner. The responses were then assembled according to similarity and frequency of occurrence. Table 1 presents the incidence with which male and female speech pathologists specified the same features for the 37 male speakers. For purposes of comparison, the chronological age (CA) and the grand mean perceived age (GYPA) of each speaker is provided. The percent of listeners who indicated the presence of a feature for each speaker is also presented. Table 1 reveals that of the 25 features used by both male and female listeners, low pitch (23%) was used most frequently. Hoarseness (15%), rapid rate (14%)) precise articulation (13%), slow rate (13%)) high pitch (12%), and clear quality (11%) were specified at least 10% of the time. Analysis of features by perceived age reveals that rapid rate, high pitch, precise articulation, and clear quality appeared to be the best discriminators of speakers who were judged to be below 30 yr. For speakers who were perceived to be between 30 and 40 yr, low pitch, precise articulation, clear quality, and moderate pitch appeared to be the dominant indices for describing the speech of speakers in this decade. The initial appearance of hoarseness, glottal fry, breathiness, and long pauses is also recognized during this period. For speakers who were judged to be between 40 and 50 yr of age, low pitch, hoarseness, imprecise articulation, and moderate rate appeared to be the most representative features of their speech. The dominant features of speech for speakers perceived to be within the 5@60 yr age range appeared to be low pitch, hoarseness, slow rate, imprecise articulation, breathiness, and long pauses. Examination of the most prominent features indicated by the listeners reveals that they may be classified according to pitch, quality, articulation, and rate of speech. From the judgments of this population, it can be postulated that these four parameters of speech serve as the primary perceptual bases for describing the speech of males ranging in age from 25 to 66 yr.
PERCEPTUAL
IDENTITY
OF AGING
IN MALE SPEAKERS
57
Discussion First Procedure
The finding that there was a nonsignificant difference in the overall age judgments of untrained male and female listeners, and that female listeners were consistently more accurate in judging age after the fifth decade, represents original observations. It could be postulated that male and female listeners use similar perceptual sets for judging age until after the fiftieth year. From this point to the beginning of the seventh decade, male perceptions of aging speech change, possibly because of egocentric involvement associated with aging. This premise is substantiated, in part, by studying those ages that reflected a significant difference in mean perceived age estimations. All were within or beyond the fifth decade and exemplified lower age judgments on the part of the male listeners. The trend for untrained listeners to underestimate the age of male speakers was congruent with earlier research (Shipp and Hollien, 1969). This phenomenon, which has been discussed by these authors, may be a reflection of the age of the listeners themselves. In their study and in the present investigation, listeners were “college age” students-under 30 yr of age. The above findings have important experimental implications for future research that utilizes untrained male listeners for making age judgments. First, in order to either substantiate or negate the premise that listeners alter their perceptual sets when making an age judgment of someone their own sex, the present study should be replicated using female speakers. Second, listeners should be matched with speakers in terms of chronological age in order to determine if more accurate age estimations are possible when persons in both populations are of similar age. Second Procedure
For the 2.5 features of speech identified by speech pathologists as being characteristic of male speech across four perceived (age decades, the finding that low pitch and hoarseness were the most recurring was not unexpected. Analysis of individual perceived ages reveals that 76% of the speakers were judged to be older than 30 yr of age. Examination of the third, fourth, and fifth perceived age decades shows that low pitch and hoarseness initially appeared during the third decade, and were dominant features for the latter two decades. Ptacek and Sander (1966) found that some of the perceptual features characteristic of “old” speakers (67-87 yr) were low pitch, hoarseness, reduced intensity, hesitancy, and less pitch variability. Comparable features indicated in the present investigation were specified for speakers whose chronological age ranged from 25 to 66 yr. The majority of these speakers (86%) were younger than 60 yr
58
TABLE 1 Features Indicated by 10 Male and 10 Female Speech Pathologists Features and incidence
of occurrence
for 37 Male Speakers
(%)
CA 27.60 25.00 33.00 26.00 37.00 29.00 38.00 28.00 56.00 30.00 57.00 35.00 44.00 42.00 49.00 31.00 53.00 40.00 41.00 36.00 48.00 39.00 32.00 45.00 43.00 46.00 47.00 55.00 62.00 63.00 54.00 61.00 34.00 52.00 65.00 59.00 66.00
22.88 23.63 24.30 24.65 26.80 27.76 27.95 28.48 30.85 31.16 31.38 31.43 31.56 31.95 32.13 34.00 34.66 34.75 38.15 38.23 38.45 39.43 39.50 41.80 42.40 42.86 45.30 47.40 48.10 48.66 48.82 50.95 51.20 54.13 55.15 55.58 56.73
CA = chronological
5 5
5 5 15 10 25 20 5 10 30 45 30 50 5 40 5 15 35 20 15 65 10 80 25 60 35 5 25 55 40 15 25 15
15 5
65 40 25 65 20 40 15 55 45 5 35 15
15 5 5 20 5 IO 15 15 15 15 5 20 15 25 50 10 40 10 45 45 40 60 30 30
30
35 10 5
20 20 30 10 30 20 10 20 15 15 5 40 15 10 10 30 10 15 20 30 35 20 25 5 5 5 5 5 5
5 5 5 10 20 25 10 30 10 15 15 25 5 25 15 10 15 5 30 25 50
15
35 20 50 40 60 25 55 10 15
10 10 5
5 25 5
20 30 10 15 5 20 35 15 10 10 10 20 10 15 5 25 5 30 20 10 10 15 10
10 15 5 15 5
5
20
10
5 5
45
10 15 20 20 45 15
age; GE = grand mean perceived
5 age.
5 10
5
10 15 20 5 10
5 25 10 15
25 25 5 10 10 15 35 5 5 10 30
5 5 10 5 5 5 15 5 10 3.5 20 20 25 20 5 15 30 25
10 10 20 15 10 15 15 30 5 30 15 5 20 5 5 10
30 25 10 5 5 5
5 5
5 5 25
5 25 10 5 25 10 30 15 20 5 15 10 15 5 10 5 20 10 10 10 5 5 5 10
59 TABLE
1 (continued)
Features and incidence
20 25 15 30
5 5
15
10 5 10 5 10 30 20 10 20 5 10 15 15 5 10 5 10 15 10
10
10 5 10 10
5 15 5 5 15 5 20
15
10 5 5 25 5
10
5 5
20 10 20 5 25 30 10 5 5
20 10
5
5 5 5
20 5 25
5 10
5 15 5 5 10 5
5
10 5 5 5
5 5 5 5
5 5 10 10
5 10 35
5
5
25 5 5 15 5 45
15 5 10
5 5 5 5
50 10 5 25 10 45
15 20 20 5 20 5
15 10 5
5 5
5 10
5 5 5 5 10 5
15 20 15
5
5
5
5
15 5
10 20 5
5 5
5 5
10
5 5 5
5 5
5 10
IO
5 30
15 15 5 15
(%)
15
10 10
10 5
5
of occurrence
5 5 5 5 10 5 5 10 5 5 5
10
5 10 15 5 5
20 5
10 5
5 5 5 10 5 5 10
5
15
10
5
5
5
5
10
5
20 15
10 10
5
10
5 5 5
5
20
5
1.5
5
5 5 10 5 5
5 5
10 5
10 20 15 15 5 5 10 5 5
15 10 5 5 5 15 10 15 5 5 5
5
5
5 5
60
DAVID
E. HARTMAN
of age. Ryan (1972) found that men between 60 and 80 yr displayed a significant decrease in reading rate when compared to those in the age range 40-50 yr. In the present investigation, slow speaking rate was a prominent feature of speakers who ranged in age from 34 to 66 yr (x = 56.16 yr). Additionally, Ryan and Burk (1974) observed that five perceptual features (air loss, laryngeal tension, voice tremor, imprecise consonants, and slow rate of articulation) were effective predictors of perceived age of males 40-80 yr of age. With the exception of voice tremor, which was not specified in this investigation, comparable features were considered prominent for speakers who ranged in age from 34 to 66 yr. The implications of these results are that the features previously indicated as being characteristic of older speakers may be, in reality, salient features of younger speakers also. However, when entertaining this conclusion, consideration should be given to differences in research design and the ages of the speakers utilized in earlier studies. The most tangible similarity between this and earlier research is that the majority of features indicated by all listeners can be classified according to pitch, quality, articulation, and rate of speech. In view of this finding, these parameters could be considered the primary perceptual characteristics of aging speech in the male population. This research was based on a doctoral dissertation completed at Bowling Green State University, Bowling Green, Ohio. Special thanks are given to Melvin Hyman, who served as dissertation advisor, and Jeffrey Danhauer, William Hinkle, and Raymond Tucker, who assisted in the design of the study. Portions of this investigation were presented at the Annual Convention of the Acoustical Society of America, San Francisco, November 1975, and appeared in the Journal of the Acoustical Society of America, (1976), 59: 713-715. References Bricker, P. D., and Pruzansky, S. (1966). Effects of stimulus content and duration on talker identification. J. Acoust. Sot. Am. 40: 1441-1449. Bruning, I. L., and Kintz, B. L. (1968). Computationa/ handbook of statistics. Glenview: Scott Foresman. Faircloth, M. A., Faircloth, S. R., and Singh, S. (1974). Phonologic proficiency in the spontaneous speech of trainable mentally retarded subjects. Unpublished manuscript. Ferguson, G. A. (1971). Sraristical analysis in psychology and education. New York: McGraw-Hill. Greene, M. C. (1964). The voice and ifs disorders. Philadelphia: Lippincott. Hollien, H., and Malcik, E. (1967). Evaluation of cross-sectional studies of adolescent voice change in males. Speech Monogr. 34: G-84. Hollien, H., and Shipp, T. (1972). Speaking fundamental frequency and chronologic age in males. J. Speech Hear. Res. 15: 155-159. Kaplan, H. M. (1971). Anatomy and physiology of speech. New York: McGraw-Hill.
PERCEPTUAL
IDENTITY
OF AGING IN MALE SPEAKERS
61
Levin, N. M. (1962). Voice and speech disorders: medial aspects. Springfield: Charles C Thomas. Luchsinger, R., and Arnold, G. E. (1965). Voice-speech-language. Belmont: Wadsworth. McGlone, R. E., and Hollien, H. (1963). Vocal pitch characteristics of aged women. J. Speech Hear.
Michel,
Res. 6: 16+170.
J. F., and Wendahl,
Handbook
Mysak,
R. (1971).
of speech pathology
Correlates
and audiology.
E. D. (1959). Pitch and duration
of voice production.
In L. E. Travis (ed.),
New York: Appleton-Century-Crofts.
characteristics
of older males. /. Speech Hear.
Res. 2:
46-54.
Ptacek,
P. H., and Sander,
E. K. (1966).
Age recognition
from voice. J. Speech Hear.
Res. 9:
27>277.
Ptacek, P. H., Sander, E. K., Maloney. W. H., and Jackson, C. R. (1966). Phonatory and related changes with advanced age. J. Speech Hear. Res. 9: 353-360. Ryan, W. J. (1972). Acoustic aspects of the aging voice. /. Gerontol. 27: 265-268. Ryan, W. J., and Burk, K. W. (1974). Perceptual and acoustic correlates of aging in the speech of males. J. Commun. Dis. 7: 181-192. Shipp, T., and Hollien, H. (1969). Perception of the aging male voice. J. Speech Hear. Res. 12: 703-710. Singh, S., Faircloth, M. A., and Faircloth, S. R. (1974). Articulatory proficiency in the conversational speech of mentally retarded subjects. Unpublished manuscript Snidecor, J. C. (1943). A comparative study of the pitch and duration characteristics of impromptu speaking and oral reading. Speech Monogr. IO: 5%56. Tiffany, W. R., and Hollien, H. (1968). Speaking fundamental frequency as a function of type of speech production. Unpublished manuscript. Zemlin, W. R. (1968). Speech and hearing science: anatomy and physiology. Englewood Cliffs, N.J.: Prentice-Hall.