AGE-RELATED DIFFERENCES IN ARTICULATORY
PHYSIOLOGY AMONG ADULT FEMALES
RICHARD JACK MORRIS
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN
PARTIAL FULFILLMENT OF THE REQUIREMENTS
FOR THE DEGREE OF DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
There are so many who have helped make this project
possible that it is difficult to enumerate the responsible
parties. First, I would like to acknowledge the guidance
and support that I have received from the members of my
supervisory committee. Dr. Douglas Hicks, Dr. Harry
Hollien, Dr. Lynne Webb, and Dr. Mark Yang have all offered
their time and knowledge. I also wish to offer my thanks
and appreciation to my mentor and chairperson, Dr. W. S.
Brown, Jr. He helped me by prodding me onward, joking with
me, and keeping his door open for me when I have needed
help. Even when I have been at my clumsiest, I have felt I
My fellow students in IASCP and the Speech department
have been a source of solace and ideas. People who deserve
individual mention are Ruth Huntley, Donald Goldberg, Karen
Massey, Janet Harrison, Brian Klepper, Tom Sawallis, Ernie
Hepler, and Lynne Harshman. To this group I must add Juan
Talavera-Martin, a man who helped me immensely in many ways.
I also need to acknowledge my religious community, the
people who share my foundation: the special corner of
Judaism, B'nai Or, and its members Dennis Shuman, Renee
Hoffinger, Shelly Isenberg, Bahira Sugarman, Rick Chess,
Michael Koszegi, Gayle Mann, Elsa Kula, and John Simpson;
also, the congregation at B'nai Israel, particularly Rabbi
Allan Lehmann and Howard Rothman.
The support and encouragement that I have received from
my family, even though we are separated by miles, have been a
constant buoy. My brother, Fred Morris, my sister, Linda
Thornton, and my parents Jack and Ruth Morris have given me
more than I can express.
The daily sharing of the trials and tribulations of
this project and unconditional, unselfish love provided by
my wife, Jamie Wulkan, is truly the source of energy from
which this project came. May I live long enough to share as
much with her as she has shared with me.
At the very last, as at the very first, this and all
things are a gift from the holy one, blessed be the name.
May we see the blessing in all that we do.
TABLE OF CONTENTS
ACKNOWLEDGMENTS . . . . . . . . . .
ABSTRACT . . . . . . . . . .
I INTRODUCTION . . . . . . . . .
II LITERATURE REVIEW . . . . . . .
Increased Number of Older People . . . .
Age-Related Physiologic Differences . . .
Measurement of Speech Timing, Intraoral Air
Pressure and Speech Intensity . . . .
Purpose . . . . . . . . . .
Hypotheses . . . . . . . . . .
III METHODS . . . . . . . .. ..
Subjects . . . . . . . . . .
Equipment . . . . . . . . .
Procedures . . . . . . . . . .
Analyses . . . . . . . . . .
IV RESULTS . . . . . . . . . ..
Introduction . . . . . . . . .
Research Findings . . . . . . .
Summary of Results . . . . . . . .
V DISCUSSION . . . . . . . . . .
Introduction . . . . . . . . .
Speech Production Similarities Between
the Two Groups . . . .
Speech Production Differences Between
the Two Groups . . . .
Implications for Future Research . .
Conclusion . . . . . . .
REFERENCES . . . . . . . . . . .
. . 99
f I I f
A HEALTH STATUS QUESTIONNAIRE. . . . . ... 127
B DIAGRAM OF TIMING AND PRESSURE RECORDING
EQUIPMENT . . . . . . . . 129
C RESPONSE FORM FOR PERCEPTUAL TASK. . . .. .131
BIOGRAPHICAL SKETCH . . . . . ... . . .. 135
Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy
AGE-RELATED DIFFERENCES IN THE ARTICULATORY
PHYSIOLOGY OF ADULT WOMEN
Richard Jack Morris
Chairman: W. Samuel Brown, Jr.
Major Department: Speech
The purpose of this study was to examine differences
between the speech of women in two age groups, 20-30 years
old and over 75 years old. Age-related differences in peak
intraoral air pressure (PIO), consonant duration (CD), vowel
duration (VD), voice onset time (VOT), and vocal intensity
were studied. The intensity task consisted of three
repetitions of sustained /a/ at three loudness levels. For
the other speech tasks, consonant-vowel (CV), vowel-
consonant-vowel (VCV), and vowel-consonant (VC) syllables
were embedded in the carrier phrase "Speak __ again."
The consonants were /p, t, b, d, s, and z/ combined with
/a/. Short samples from the syllable task were played to
listeners who estimated the ages of the speakers. The
listeners consistently estimated ages that closely
approximated the chronological age of the speakers, r=.88.
Peak PIO magnitudes did not differ significantly
between the young and non-denture wearing older speakers.
The denture wearing older speakers exhibited higher peak
PIO values for the voiceless consonants. CD values were
longer among the older speakers for the test consonants
except for /t/. The older speakers exhibited significant
increases in VD. The older group also exhibited consis-
tently shorter VOT values than the younger speakers, but not
enough to change normal perceptual identification. For the
intensity tasks the older speakers exhibited narrower
intensity ranges, the range being restricted at both the
minimum and maximum effort levels. All of the dependent
variables exhibited greater variability among the older
The conversational intensity and PIO results indicate
that age-related physiologic differences do not signifi-
cantly affect the pressure levels associated with normal
speech. Similarly, the VOT data indicated that this
parameter was not meaningfully affected by the aging
process. Increased consonant and vowel durations may be
related to the percept of a slower speech rate for older
speakers, while the greater variability for all of the
dependent variables for the older speakers may be related to
the percept of less precise articulation among geriatric
One of the outstanding demographic phenomena of the
twentieth century is the aging of populations in developed
nations. Extended life expectancy and reduced birth rate
have resulted in those over age 65 being an increasing
proportion of the population. As the population has aged,
increasing numbers of health related professionals have
investigated the sociologic, psychologic, anatomic, and
physiologic ramifications of a larger aged population.
The aging process includes organismic alterations to
neurotransmission, muscle strength, and mucosal elasticity.
These alterations of the entire human organism influence
speech production. The speech produced is so altered that
people can determine the age of speakers from taped speech
samples (Hartman, 1979; Hartman & Danhauer, 1976; Ptacek &
Sander, 1966; E. Ryan & Capadano, 1978; Shipp & Hoilien,
1969). The relationship between perceived age differences
among speakers and the physiologic changes in the speech
production mechanism has not been investigated. Addi-
tionally, few extant studies have examined the effect of
known age-related physiologic changes on speech production.
The physiologic changes within the speech production
mechanism that occur with aging may alter certain measurable
parameters of speech. Research indicates that one para-
meter, speech timing, is affected by aging (Mysak, 1959;
Mysak & Hanley, 1959; W. Ryan, 1972). Two other parameters
that may be affected by aging include the ability to gener-
ate typical values of intraoral air pressure and vocal
intensity. These three parameters are among the most
important factors in creating and maintaining intelligible
speech production (Lisker & Abramson, 1964; Muller & Brown,
1980; Peterson & Lehiste, 1960). Age-based changes within
any of these parameters may implicate specific structural
changes in the speech production mechanism.
Previous studies have found that geriatric subjects
speak more slowly than young adult subjects (Mysak, 1959;
Mysak & Hanley, 1959; W. Ryan, 1972). W. Ryan (1972) found
that when pauses were removed from speakers' discourse, the
older subjects still exhibited a slower speaking rate than
younger subjects. This indicates that some segment of the
speech signal is slower among older speakers. The slowed
speech rate may be due to differences in consonant closure
duration. This could imply that older speakers are slower
in initiating and completing consonants, or that more time
is needed to impound adequate intraoral air pressure. A
second possibility is that the slowed speech rate is due to
a difference in vowel duration. This could imply that the
smooth transitions between the consonants are sustained
longer to be certain that they are perceived correctly or
that the oral motor system is slower in completing
A final aspect of speech timing that may be involved in
perceived age differences is voice onset time. Previous
studies of aging and voice onset time have yielded disparate
results that include reduced voice onset times among older
subjects (Benjamin, 1982), no age differences in voice onset
time (Neiman, Klich, & Shuey, 1983), and increased varia-
bility in voice onset times among older speakers (Sweeting
and Baken, 1982) finding of increased variability among
older speakers. Further investigation of voice onset time
may clarify how aging influences the timing interaction
between articulation and phonation.
In addition to speech timing, intraoral air pressure
and sound pressure level for speech may be altered by aging.
Intraoral pressure is an important aspect of stop and
fricative consonants, as sound pressure level is with
vowels. Both of these pressures depend upon the air flow
from the respiratory system.
While older subjects exhibit reduced vital capacity
(Alexander & Kountz, 1932; Pierce & Ebert, 1965; Ptacek,
Sander, Maloney, & Jackson, 1966; Pump, 1971), the effect of
this reduction on speech has not been clearly documented.
Ptacek et al. (1966) stated that the reduction of vital
capacity is related to reduced maximum levels of intraoral
air pressure in older subjects. However, this relationship
has not been tested for conversational utterances.
Age differences in vocal intensity have been examined
at both maximum and comfortable effort levels. Ptacek et
al. (1966) found that older subjects exhibited lower
intensity than younger subjects at maximum effort levels,
while W. Ryan (1972) reported that older subjects' speech
intensity was greater than that of younger subjects at
comfortable effort levels. It could be hypothesized that if
the vocal intensity of older subjects is reduced at all
effort levels, this would indicate that reduced vital
capacity affects vocal intensity (Ptacek et al., 1966).
However, if the older people show increased vocal intensity,
then other physiologic changes, such as reduced oral
proprioception would be implied. An investigation of vocal
intensity at maximum, minimum, and comfortable levels should
indicate which aging changes are salient to vocal intensity.
This research project investigated the age-related
physiologic changes within the speech production mechanism
as demonstrated by speech timing, intraoral air pressure,
and vocal intensity. Specifically, it involved comparing
two age groups along the parameters phoneme duration, voice
onset time, intraoral pressure, and sound pressure level.
Increased Number of Older People
Since the turn of the century, an increasingly larger
proportion of the population of the United States is living
past age 65. Hayflick (1981) stated that in the developed
countries, the reduction of mortality from early childhood
diseases has resulted in the survival of a greater propor-
tion of the population to adulthood and old age. In 1900
one of every six infants died in the first 12 months of
life, by 1982 this was reduced to one of every 88 (Yin &
Shine, 1985). At the other end of the life span, reduction
of diseases prevalent among older people has added only two
to three years to life expectancy (Hayflick, 1981). In
other words, the decreased rate of premature deaths has been
the major factor in the increase of life expectancy at birth
from 47 years in 1900 to 75 years in 1982 (Yin & Shine,
1985). The trend of an increasing proportion of older
people in the population is predicted to continue into the
The increasing number of older people within the popu-
lation has prompted many health related professionals to
investigate the effect that age-related sociologic, psycho-
logic, acoustic, anatomic, and physiologic changes have on
normal functioning. These changes may create new problems,
exacerbate existing problems, and affect appropriate
treatment of problems. Practitioners in the health related
professions must differentiate between normal aging changes
and abnormal changes in order to correctly diagnose and
treat problems. Professionals in communication have begun
to probe the effects age-related physiologic changes may
have on communication skills.
Age-Related Physiologic Differences
Humans over 70 years old exhibit reduced physiologic
ability when compared to younger populations. Physiologic
differences include reduced speed of neurologic transmission
(LaFratta & Canestrari, 1966; Norris, Shock, & Wagman,
1953), reduced maximum muscle strength (Welford, 1977),
reduced speed of muscle action (Welford, 1977), and reduced
elasticity of extracellular connective tissue (Hall, 1976).
These aging changes occur at differing rates in various body
systems, with greater reductions in systems involving a
number of connections between nerves and other nerves or
muscles (Weg, 1975).
Speech Production Mechanism Differences
Review articles by Kahane (1981) and Meyerson (1976)
have described how the age-related organismic physiologic
differences alter each of the speech production systems:
respiration, phonation, and articulation. The present
discussion will focus on the aging differences found in these
three speech systems as they may relate to speech timing,
magnitudes of intraoral air pressure, and speech intensity.
The lower respiratory system of older people exhibits
decrements in its component parts (Kahane, 1981; Meyerson,
1976). It consists of the rib cage and connected vertebrae,
the diaphragm and other thoracic musculature, the lungs, the
bronchial tubes, and the trachea below the vocal folds
(Zemlin, 1968). Studies of aging differences within the
respiratory system have tended to focus on either the
thoracic wall or the tissues within the lungs.
Age-related changes to the thoracic wall. The
vertebral column and rib cage exhibit structural changes
with aging. The vertebral column of older people undergoes
kyphosis, or a change in curvature (Alexander & Kountz,
1932). The range of motion of ribs connected to these
vertebrae is restricted by the kyphosis as well as decalci-
fication of the ribs (Alexander & Kountz, 1932; Dhar,
Shastri, & Lenora, 1976). Bates and Christie (1955) and
Dhar et al. (1976) reported that their older subjects
exhibited a "barrel chest" configuration of the costal
cartilages, leaving the rib cage in an elevated position.
An elevated resting position of the rib cage reduces the
volume of air which may be forced out of the lungs, thereby
reducing vital capacity (Alexander & Kountz, 1932; Bates &
Similarly, age-related weakening of the thoracic
musculature may reduce mobility of the thoracic wall.
Several authors include thoracic muscle weakening as a
factor in the reduced vital capacity of older subjects in
comparison to younger subjects (Dhar et al., 1976; Kahane,
1981; Meyerson, 1976; Thurlbeck, 1979); however, a paucity
of data exists to quantify the role of the weakening of the
thoracic wall musculature in reduced vital capacity among
older subjects. Researchers have cited previous studies
based on clinical impressions or their own clinical
impressions in reporting the weakening of the thoracic wall
musculature. Regardless of the status of the thoracic wall
musculature, reduction of mobility of the bones of the
thoracic wall is a significant factor in the reduction in
Age-related changes within the lung. In addition to
the changes in the thoracic wall, structural changes within
the aged lung operate to reduce vital capacity. A breakdown
of the structure of the elastic fibers occurs within the
lungs, with a concomitant increase in the mass of elastic
tissue in the pleura and septa of the lungs (Pierce & Ebert,
1965; Pump, 1971; Thurlbeck, 1979). These changes reduce
the vital capacity of older people by reducing the elastic
recoil of the lung (Pierce & Ebert, 1965; Pump, 1971; Shock,
1962; Thurlbeck, 1979; Turner, Mead, & Wohl, 1968). The
diminished elastic recoil of older subjects results in
reduced pulmonary pressure at given lung volumes (Mead,
Turner, Macklem, & Little, 1967). Reduction in pulmonary
pressure means that less air needs to be driven out of the
lungs during exhalation in order to achieve a balance with
Reduced elastic recoil and pulmonary pressure within
the lung do not affect total lung volume, which is corre-
lated with the height rather than the age of the person
(Bates & Christie, 1955; Dhar et al., 1976; Pierce & Ebert,
1965; Turner et al., 1968). Increased functional residual
capacity allows total lung volume to remain constant when
the elastic tissue changes reduce vital capacity
(Greifenstein et al., 1952; Mead et al., 1967; Pierce &
Ebert, 1965; Turner et al., 1968). This arrangement of
increased functional residual capacity and decreased vital
capacity results in a reduction of the volume of air
available for speech.
Reduced vital capacity and air volume may result in
lower intensity levels for the speech produced (Kahane,
1981). Ptacek et al. (1966) found significantly lower vital
capacities, peak intraoral air pressure, and peak intensity
levels for their older subjects than for their younger
subjects. These investigators concluded that a relation-
ship existed between the lower vital capacity and the
lowered levels of both peak intraoral air pressure and
intensity of the speech produced by their older speakers.
However, W. Ryan (1972) found that older speakers spoke with
a higher intensity level than younger speakers. This
difference indicates that aging changes in the respiratory
system may not make a crucial difference in speech
The phonatory system consists of the cartilages,
muscles, and mucosa of the larynx (Zemlin, 1968). Each of
these tissues exhibits an aging pattern that may affect
speech timing, intraoral air pressure magnitudes, and speech
Cartilage. The cartilaginous framework of the larynx
undergoes ossification with aging (Kahane, 1983; Kahane,
Stadlan, & Bell, 1979; Noback, 1949; Zemlin, 1968).
Ossification begins during the second decade of life and
progresses with age, beginning low on the cartilage and
progressing caudocranially (Kahane, 1983; Kahane et al.,
1979; Noback, 1949). This process is more pronounced for
the hyaline cartilages, thyroid, cricoid, and arytenoid,
than for the elastic cartilage of the epiglottis (Kahane,
1983; Kahane et al., 1979; Zemlin, 1968). Ossification of
the hyaline cartilages is significant because these carti-
lages anchor the intrinsic laryngeal muscles that are
involved in phonation. Cartilage ossification may reduce
the nonmuscular forces in the larynx that depend on the
Spring action of the thyroid/cartilage by compressing the
thyroid cartilage (Kahane, 1981, 1983). This process could
slow vocal fold movement.
Slowing laryngeal activity can alter speech timing,
intraoral air pressure magnitudes, and vocal intensity.
These effects would result from changes in the relationship
between the phonatory and either the articulatory or the
respiratory system. Varying the temporal relationship
between the phonatory and articulatory systems would affect
voice onset time and voiced phonemes. These changes could
affect the meaning of the speech produced. Varying the
temporal relationship between the phonatory and respiratory
systems could result in loss of air and inefficient use of
the available air. This air loss could reduce magnitudes
of both intraoral air pressure and vocal intensity, since
the expiratory air is the driving force behind both of these
parameters. Similar effects on speech timing, intraoral
air pressure magnitudes, and vocal intensity may be noted
for age-related differences in the muscles and mucosa of the
Muscles. Aging changes within the intrinsic laryngeal
musculature include atrophy (Bach, Lederer, & Dinolt, 1941;
Ferreri, 1959; Hirano, Kurita, & Nakashima, 1983), fiber
breakdown (Bach et al., 1941; Segre, 1971), and reduction of
blood supply (Bach et al., 1941; Ferreri, 1959). The effect
of these changes on the laryngeal muscles has been a subject
of much debate. Ferreri (1959) reported that the atrophied
vocal fold mass would vibrate more quickly and produce a
higher fundamental frequency. In contrast, Segre (1971)
focused on the fibrous breakdown of the thyroarytenoid
muscle and stated that the predominant effect of age on the
larynx is increased vocal fold flaccidity that results in a
lower fundamental frequency.
Cross-sectional studies of the fundamental frequency of
adult females have revealed two trends: one of relative
stability across age groups (Gilbert & Weismer, 1974;
McGlone & Hollien, 1963) and one of lower speaking funda-
mental frequency among older women (Benjamin, 1981; de Pinto
& Hollien, 1982; Stoicheff, 1981). This discrepancy of
results would best be resolved with more longitudinal data,
which may indicate the effects of age-related anatomic
changes within the larynx on the voice.
Reduced mass or increased flaccidity of the intrinsic
laryngeal muscles could result in significant temporal
changes in the speech signal. Either of these variations in
the vocal fold musculature could affect the timing of vocal
fold closing, which could result in a variation in
Perhaps more important to the integrity of the speech
signal are the effects of the potential loss of synchron-
icity between the phonatory system and the other speech
systems. Structural alterations observed in laryngeal
muscles from older subjects, such as atrophy (Bach et al.,
1941; Ferreri, 1959; Hirano et al., 1983) and fiber breakdown
(Bach et al., 1941; Segre, 1971), could cause the glottis to
be open for longer periods of time. This could result in
inefficient use of the air from the respiratory system.
Less efficient air usage would reduce the amount of air
available as the driving force behind intraoral air pressure
and speech intensity. Altering the timing relationship
between the phonatory and articulatory systems could affect
all phonemes that require voicing. This would most notably
affect voice onset time and phoneme duration, which are
paramount in the accurate production of speech (Lisker &
Abramson, 1964; Muller & Brown, 1980; Umeda, 1977).
Mucosa. The mucosa consists of the epithelium and
lamina propria that line the lumen of the larynx (Hirano et
al., 1983). These authors defined the vocal ligament as the
thickening of the intermediate and deep layers of the lamina
propria at the glottal margin. Elastic fibers comprise the
intermediate layer and collagen fibers comprise the deep
layer of the laimina propria. Histological studies of the
structure of the vocal ligament have revealed age-related
differences in adults (Ferreri, 1959; Hirano et al., 1983).
Ferreri (1959) reported reduced elastin in the extracellular
network of the vocal folds and an increase in fatty deposits
in vocal folds from subjects over 70 years old. Hirano et
al. (1983) found that the pattern of reducing elastic fibers
combined with increasing collagen tissue resulted in a
reduction in extracellualr connective tissue in the vocal
folds from the second through sixth decade of life followed
by an increase in extracellular fibers in the vocal folds
from the sixth through ninth decade of life.
The reduced number of more pliant elastic fibers
combined with an increase in the stiffer collagen fibers
creates a stiffer vocal ligament. In addition, Kahane
(1983) reported that the increased proportion of collagen
fibers results in a "wavy" medial border of the vocal fold.
This finding is consistent with reports that the vocal folds
of older subjects are bowed at midline (Honjo & Isshiki,
1980; Segre, 1971). The wavy, bowed, and stiff appearance
of the vocal ligaments may also be affected by a general
drying of the mucosa.
The outer, epithelial layer of the mucosa is signifi-
cantly drier in older subjects than in younger ones
(Ferreri, 1959). Drier mucosa presents a thin, yellowish
appearance when viewed laryngoscopically (Segre, 1971). The
drying and thinning of the epithelial layer of the vocal
folds in older subjects may stiffen the structures and may
alter the midline of the vocal folds (Kahane, 1981, 1983).
These mucosal alterations could result in the bowing of the
vocal folds described by Honjo and Isshiki (1980) and Segre
(1971). Bowing at the midline of the vocal folds may cause
inefficient vocal fold closure and reduced subglottal
pressure (Kahane, 1983). Inefficient air usage at the
glottis could result in reduced air pressure in the oral
cavity and less sound pressure in the air.
In a similar manner, the indices of speech timing may
be affected by the structural changes within the laryngeal
mucosa. The stiffening of the mucosa may alter the timing
between the larynx and the articulators by increasing the
time it takes for the vocal folds to close and thus affect
speech timing. This prolongation of voicing time may
cascade into prolongation of other phonemes. Thus, slowed
timing between the phonatory and articulatory systems could
result in prolongations of phoneme duration.
The articulatory system consists of three connected
tubes: the laryngopharynx, oropharynx, and nasopharynx.
Valving points of articulation within these tubes are the
velopharyngeal port, the back, blade, and tip of the tongue,
the teeth, the lips, and the jaw (Zemlin, 1968). Aging
changes of the latter list of structures consist mainly of
involution and atrophy (Kahane, 1981).
Teeth and mandible. The most noticeable change to the
oral structure for many older people is the loss of teeth.
Kaplan (1971) reported that 50% of all Americans have
lost their teeth by age 65, and that tooth loss increases to
67% by age 75. The main cause of tooth loss among
the elderly is periodontal disease (Kaplan, 1971; Klein,
1980). Klein (1980) reported that 97% of people over
45 years old have some degree of periodontal disease.
Loss of natural teeth has been associated with involu-
tion of the toothless, or edentulous, jaw (Goldstein, 1936;
Kaplan, 1971; Klein, 1980). Goldstein (1936) found that his
older subjects, 80% of whom were edentulous, had
reduced mandibular height from alveolar resorption. The
subjects also had an increased angle to the ramus that moved
the entire mandible forward relative to the face. In con-
trast, the results of a longitudinal study by Israel (1973)
indicated that the mandible, along with the rest of the
craniofacial bones, continues to grow slowly throughout
adulthood. Israel (1973) found that the mandible increases
in size by approximately 5% between the thirtieth
year and sixtieth year. He also stated that aging itself
was not a factor in mandible resorption. However, he did
not include subjects who were over age 65, leaving his
conclusions concerning aging and mandible resorption
The reduction of mandibular mass of the edentulous
person could alter the shape and size of the oral cavity.
Such an alteration of the cavity is likely to have an effect
on tongue carriage and the ability of the tongue to make
correct articulatory contact. These physiologic changes may
cause slower oral motor activity and prolonged phonemes.
Tongue. Studies of the tongue have revealed age-related
differences in both sensory and motor functioning. The
tongue does not atrophy with aging as most striated muscles
do (Balogh & Lelkes, 1961; Kaplan, 1971; Klein, 1980; Price
& Darvell, 1981). In fact, the size of the tongue increases
in edentulous oral cavities (Balogh & Lelkes, 1961; Kaplan,
1971; Klein, 1980; Price & Darvell, 1981). While Kaplan
(1971) and Klein (1980) reported clinical impressions that
the tongue loses muscle tone with aging, Price and Darvell
(1981) found no evidence of this among specimens at autopsy.
The maximum suction created by older subjects with their
natural teeth, or dentate older subjects, was significantly
less than that created by the dentate younger subjects.
However, the edentulous older subjects generated signifi-
cantly more suction than either dentate group (Price &
Darvell, 1981). Thus, the edentulous older subjects that
Kaplan (1971) and Klein (1980) described as exhibiting
reduced muscle tone were found to have increased tongue
force by Price and Darvell (1981). These authors reasoned
that the edentulous person uses the tongue to greater and
more varied purposes during mastication and deglutition than
does the dentate person.
Older dentate and edentulous subjects are equally
susceptible to reduced sensitivity to taste (Balogh &
Lelkes, 1961; Corso, 1975; Kaplan, 1971; Klein, 1980). The
reduced sensitivity occurs among all four of the basic
tastes: sweet, sour, bitter, and saline (Balogh & Lelkes,
1961; Corso, 1975; Kaplan, 1971; Klein, 1980). One might
assume this would imply a general reduction of perception on
the tongue. While Balogh and Lelkes (1961) found that to be
true for some older subjects, others reported an increased
sensitivity to touch and pain. The increased sensitivity
was related to increased friction during tongue movement
caused by a reduced moisture level (Balogh & Lelkes, 1961).
The physiologic differences caused by aging of the
tongue may influence speech production. In a study of
maximum tongue excursion for phonating /a/, /i/,and /k/,
Sonies, Baum, and Shawker (1984) found that the tongues of
dentate older subjects exhibited significantly less
posterior movement in articulating /a/ than the tongues of
younger subjects. No significant differences in tongue
movement were noted between the groups for production of /i/
or /k/. Once the tongue achieves the phonemic position,
age-related muscular changes may be responsible for the
reduced ability to hold maximum intraoral air pressure
(Ptacek et al., 1966). In addition, the muscular changes
may also affect the timing of speech.
Diadokokinetic rate is an indicator of speech timing
that is related to phoneme duration (Fletcher, 1972). The
faster a person moves the tongue to repeat a phoneme, the
shorter the minimum duration of that phoneme for that per-
son. While some investigators have found that both male and
female older subjects exhibit slower diadokokinetic rates
than younger subjects (Price & Darvell, 1981; Ptacek et al.,
1966), Shanks (1970) found no significant differences in
diadokokinetic rates among three age groups of women.
The differing results for diadokokinetic rate among
older speakers may be related to several factors. One
possible factor is the varying ability of the experimenters
to motivate older subjects to perform maximally on the task
(Ptacek et al., 1966). A second source of variability may
be the use of different syllables by each experimenter.
Varying the stimulus syllables meant that different tongue
movements were involved, making direct comparisons
difficult. Finally, the finding among older subjects of a
slower diadokokinetic rate (Price & Darvell, 1981; Ptacek et
al., 1966) may be related to the findings of reduced tongue
excursion (Sonies et al., 1984) and slower speaking rate
(Hollien & Shipp, 1972; Mysak, 1959; Mysak & Hanley, 1959;
Ptacek & Sander, 1966; W. Ryan, 1972; W. Ryan & Burk, 1974).
These behaviors could all indicate reduced tongue mobility
among older subjects that could affect speech timing,
intraoral air pressure magnitudes, and vocal intensity.
Oral mucosa. The oral mucosa exhibits atrophy and loss
of elasticity in older subjects, primarily because of a
reduction of saliva (Corso, 1975; Kaplan, 1971; Klein,
1980). Klein (1980) reported that the drying of the mucosa
would increase the friction during intraoral movements.
Increased friction during intraoral movement would tend to
reduce the speaking rate and increase phoneme duration.
Lips. The lips become wrinkled and drawn on older
subjects. Two factors have been reported to account for
these changes. Kaplan (1971) noted that the lips become
drier with age, particularly for edentulous people, while
Klein (1980) stated that the lip musculature becomes weaker
with age. Of these aging effects, the weaker musclature
would affect speech to a greater degree. Weakening of the
lip musculature would reduce the ability to maintain intra-
oral pressure during the production of /p/ and /b/. Lip
muscle weakness would also slow lip movements, prolonging
phoneme duration and reduce voice onset time.
Oral sensation. Kahane (1981) reported that while
several studies describe oral sensation and perception of
young adults, a paucity of research exists on the oral
sensitivity of older subjects. He stated that one should be
able to safely generalize findings of perception and sensa-
tion declines from similar body tissues to the oral struc-
tures. These declines would be consistent with the reduced
speed of receptive neural transmission described by LaFratta
and Canestrari (1966). Age-related declines in oral
perception and sensation may reduce articulatory feedback
Oral perception and sensation changes could alter
speech timing, intraoral air pressure magnitudes, and speech
intensity. Reduced oral perception and sensitivity may
alter speech timing by increasing the duration needed to
recognize achievement of phonetic placement. A timing
change would increase phoneme duration as well as alter the
relationship with the larynx as indicated by voice onset
time. The ability to perceive pressure levels could be
reduced by the change in oral perception and sensation.
Reduced perception may result in an increased level of
intraoral air pressure and speech intensity during conver-
sational speech as was found by W. Ryan (1972).
Effects of Age-Related Physiologic Differences on Speech
Timing, Intraoral Air Pressure Magnitudes, and Vocal
The previous discussion described the age-related
physiologic differences that occur in the respiratory,
phonatory, and articulatory systems of the speech production
mechanism. Within each system, these differences were shown
to have a potential effect on speech timing, intraoral air
pressure magnitudes, and vocal intensity. The degree to
which aging affects these three speech parameters is not
If the aging changes within the speech production
mechanism significantly reduce the ability to correctly time
the speech, such differences probably would alter the
intelligibility of the speech. When alterations among the
stream of sounds occur, there is a breakdown between the
speaker's intended message and the actual message sent.
This breakdown may be the basis for the percept of age
differences among speakers. Aging changes in the speech
production mechanism may create such a breakdown.
Aging changes within the speech production mechanism may
significantly reduce the ability to develop pressure during
speech; this reduction would alter the projection of the
speech signal. Since the air stream from the respiratory
system is the power source that drives speech production,
reductions in that air stream would diminish the effective
radius of the speech signal. Thus, investigation of the
age-related differences within the speech production mech-
anism may reveal significant differences in speech timing,
intraoral air pressure magnitudes, and vocal intensity.
Alterations of speech timing, intraoral air pressure
magnitudes, and vocal intensity associated with aging have
been investigated using mainly perceptual and acoustic
variables. These studies have been limited in number and
vary in methodology. The varying methodologies may have
been a factor in the differing results among some of the
Timing differences between older and younger adult
speakers have been consistently reported in both perceptual
and acoustic studies. Listeners report that a reduced rate
is among the principal perceptual features that mark the
speech of older adults (Hartman, 1979; Hartman & Danhauer,
1976; Ptacek & Sander, 1966; E. Ryan & Capadano, 1978; W.
Ryan & Burk, 1974). Acoustic studies that measured the
ratio of phonation time to total speaking time (Mysak, 1959)
and words per minute (Mysak, 1959; Mysak & Hanley, 1959; W.
Ryan, 1972) found a slower speaking rate among geriatric
subjects. The specific variables of speech timing that are
involved in the slower speech rate have yet to be fully
A few studies have investigated the effect of specific
timing variables on the slower speech rate of older people.
Benjamin (1982) reported that the speech of older speakers
is marked by longer vowel durations, longer stop consonant
durations, and shorter voice onset times. The vowel dura-
tion information was appropriately separated into stressed
and unstressed vowels, but the consonant information was
not divided into voiced and voiceless consonants. Since
consonant timing differs by voicing, these relative differ-
ences may change with age. Thus, the information available
from dividing the consonants into voiced and voiceless
categories may provide valuable information concerning age
differences in consonant timing.
The voice onset time information provided by Benjamin
(1982) differs from that previously reported. While she
reported reduced voice onset times among older speakers,
other studies reported no significant duration differences
between older and younger adult speakers (Neiman et al.,
1983; Sweeting & Baken, 1982). However, a closer analysis
of the data of Neiman et al. (1983) revealed that when the
different vowel conditions were separated, the older speakers
had significantly shorter voice onset times than did the
younger speakers. Since Benjamin (1982) did not report
which stop consonants she measured, direct comparisons of
the results of the studies are difficult. Sweeting and
Baken's (1982) finding of increased variability of voice
onset time among older speakers has not been corroborated by
other studies of voice onset time among older subjects.
Thus, the three studies of the voice onset time of older
speakers have differing patterns of results. Benjamin
(1982) and Neiman et al. (1983) reported shorter durations
among older subjects, while Sweeting and Baken (1982)
reported no differences. In contrast, Sweeting and Baken
(1982) reported increased voice onset time variability while
the others did not. The results of all three studies imply
that voice onset time may be a component of the perceived
difference between older and younger adult speakers.
Further investigation of the effect of aging on voice onset
time is needed to help clarify this relationship.
Stability of voice onset time durations across adult
age groups is counter-intuitive, given the reduced motoric
and neurologic rates that have been reported for older
adults (Birren, 1974; Welford, 1977). However, if the voice
onset time durations are found to be stable across age
groups, then the neuromotor function for the production of
voice onset time may be relatively impervious to the slower
motoric and neurologic rates reported by Birren (1974) and
In general, the reports from acoustic studies of age-
related differences in vowel duration, consonant duration,
and voice onset time may be related to the perceptual reports
of less precise articulation among older speakers (Hartman,
1979; Hartman & Danhauer, 1976; E. Ryan & Capadano, 1978; W.
Ryan & Burk, 1974).
Vocal Intensity and Intraoral Air Pressure Magnitudes
Less agreement exists between the perceptual and
acoustic studies of aging differences in vocal intensity
than was found for speech timing. Perceptual studies reveal
that listeners report reduced (Ptacek & Sander, 1966; E.
Ryan & Capadano, 1978) or unchanged (Ramig & Ringel, 1983)
intensity for older speakers. An acoustic study by W. Ryan
(1972) revealed increased intensity among older speakers.
In a study that measured intraoral air pressure among older
subjects, acoustic and physiologic measurement of maximum
effort revealed a reduction in both speech intensity and
intraoral air pressure (Ptacek et al., 1966). However,
difficulty exists in comparing studies of maximum and com-
fortable effort, so the maximum effort results do not alter
the discrepancy between the acoustic and perceptual studies
of comfortable loudness.
A possible factor in the discrepancy between the
acoustic and perceptual findings is that the age estimates of
the voices in the perceptual studies were under 60, while the
acoustic study described speakers over age 60. Thus, the
listeners could have been matching features to a stereotype
for middle aged rather than geriatric speakers (E. Ryan &
Capadano, 1978). Since a similar discrepancy between
acoustic and perceptual studies was not found for speech
timing, the pattern of aging differences may vary among the
speech parameters. An investigation that relates perceived
speaker age to speech timing, intraoral air pressure magni-
tudes, and vocal intensity may provide more complete infor-
mation concerning the effect of the age-related physiologic
differences within the speech production mechanism on the
A study of the age-related physiologic differences in
speech should include measures that will be indices of the
timing and pressure levels of the speech produced. The
intricate timing of the coordination of respiratory,
phonatory, and articulatory events to create speech may be
altered by aging of the component structures. The pressure
levels of speech may be compromised by the reduced vital
capacity of the lungs (Pierce & Ebert, 1965; Turner et al.,
1968) as well as the reduced muscular ability found among
older subjects (Dhar et al., 1976; Kaplan, 1971). The
variables determined as salient in age-related differences
of speech timing, intraoral air pressure magnitudes, and
vocal intensity from a physiologic study could then be
tested using perceptual and acoustic studies.
Measurement of Speech Timing, Intraoral Air Pressure,
and Vocal Intensity
Two effective indices of the timing of speech are
phoneme duration and voice onset time. These variables
represent different aspects of the temporal dynamics of
speech. Phoneme duration provides an index of the ongoing
interphonemic relationship of coarticulation. The temporal
pattern of articulatory constriction is indicated by conso-
nant duration and the timing of the transitions between the
constrictions is indicated by vowel duration. Aging effects
within either category of this index may reveal an age-
related discoordination among the articulators. Addition-
ally, increased variability of phoneme duration in repeated
syllables among the older speakers may indicate a reduced
ability to consistently position the articulators.
While phoneme duration is an index of temporal factors
among the articulators, voice onset time is an index of the
temporal factors between the phonatory and articulatory
systems (Lisker & Abramson, 1964, 1967; Zlatin, 1974).
Increased or decreased voice onset time would indicate a
change in the interaction of the phonatory and articulatory
mechanisms. Similarly, an increased variability in voice
onset time among older adults may indicate a reduced ability
to consistently maintain the fine degree of inter-systemic
interaction required for speech. Thus, the temporal indices
of phoneme duration and voice onset time may provide strong
indications of a temporal component in age-related speech
Phoneme duration has been examined in two categories:
consonant duration and vowel duration. For this investi-
gation consonant closure duration was operationally defined
as the time interval beginning with the increase of intra-
oral pressure from baseline and ending with the drop of
intraoral pressure at consonant release.
The operational definition of vowel duration for this
investigation was the time interval beginning with the onset
of voicing after the consonant release and ending with the
offset of voicing. Both consonant closure duration and
vowel duration are strongly affected by the contiguous
phonemes, as well as such factors as stress and place in
word, phrase, and utterance.
Consonant duration. Consonant duration in connected
speech is inextricably bound to the relationship between the
consonant and its phonetic environment (Flege & Brown, 1982;
Klatt, 1973; Lehiste, 1972; Malecot, 1968; Miller, 1981;
Oller, 1973; Port, 1981; Schwartz, 1972; Stathopoulos &
Weismer, 1983; Umeda, 1977; Van Hattum & Worth, 1967).
Location within a word is an important variable for stop
consonant durations, with word initial and word final conso-
nants being more affected by their location within the word
than by their vowel environment (Umeda, 1977). Oller (1973)
reported that, given similar vowel environments, word
initial stop consonants have a greater duration than word
medial stop consonants and word final stop consonants have a
greater duration than word initial stop consonants. This
pattern has been found at the phrase level also (Umeda,
1977; Van Hattum & Worth, 1967). In contrast, Stathopoulos
and Weismer (1983) found word initial stop consonants to
have the greatest durations followed by medial and final
Consonant duration varies systemically by place of
articulation (Lehiste, 1972; Repp, 1984; Stathapoulos &
Weismer, 1983; Umeda, 1977), with labials having the longest
duration, then apicals and the shortest durations among the
velars. However, in the intervocalic position, the apical
consonants exhibit the shortest duration (Lehiste, 1972;
Umeda, 1977). Closure duration is generally longer for
voiceless stop consonants than for their voiced cognates
(Brown, 1979; Lisker, 1957; Muller & Brown, 1980; Sharf,
1962; Subtelny et al., 1966). The reported exceptions are
intervocalic /t/ and /d/ which are both taps (Muller &
Brown, 1980) and word initial stop consonants before a
stressed vowel (Stathopoulos & Weismer, 1983).
The pattern of stop consonant durations in words and
sentences is similar, with initial and final stop consonant
durations being greater than those for medial stop conso-
nants (Schwartz, 1972; Umeda, 1977, Van Hattum & Worth,
1967). Temporal variables, such as stress, contiguous vowel
duration, word length, and phrase length alter consonant
duration (Port, 1981). For example, stop consonants are
longer both before and after stressed vowels in comparison to
unstressed vowels (Malecot, 1968; Stathopoulos & Weismer,
1983; Subtelny et al., 1966). Additionally, closure duration
of initial stop consonants decreases with increased word
length and with increased phrase length (Schwartz, 1972;
Umeda, 1977). While many variables may affect consonant
duration, vowels are responsible for most of the difference
in total utterance duration that occurs during a change in
speaking rate (Gay, 1978; Miller, 1981).
Vowel duration. As shown for consonants, several
factors influence vowel duration for young adult speakers.
The principal factors are vowel tenseness, phonetic context,
stress, utterance length, and speaking rate (House, 1961;
House & Fairbanks, 1953; Lehiste, 1972; Klatt, 1973; Miller,
1981; Peterson & Lehiste, 1960; Sharf, 1964; Umeda, 1975).
While realizing that none of these factors operates
independently during running speech, the contribution of
each to vowel duration will be described individually.
In English, vowels can be separated into two categories,
tense and lax. The tense vowels, /i/, /u/, /o/, /e/, /a/,
/s/, and /o/, have relatively long durations and the lax
vowels, /I/, /n/, /A/, and /e/, have relatively short
durations (House, 1961; Sharf, 1964). When Verbugge and
Isenberg (1978) created complex waves with vowel formant
values between those for /e/ and /e/ and then manipulated
the durations of the formants, they found that their
listeners heard the short durations as /s/ and the long
durations as / /. This demonstrates the importance of vowel
duration in the recognition of vowels.
Post-vowel consonants affect vowel duration in English.
Principally, vowels preceding voiceless consonants are of
shorter duration than vowels preceding voiced consonants
(House, 1961; House & Fairbanks, 1953; Klatt, 1973; Peterson
& Lehiste, 1960; Sharf, 1962, 1964). A second factor
affecting vowel duration is that vowels preceding fricative
consonants have longer durations than those preceding stop
consonants (House, 1961; House & Fairbanks, 1953; Klatt,
1973; Peterson & Lehiste, 1960; Sharf, 1964). In contrast,
the place of articulation of the consonant following the
vowel has no significant effect on the vowel duration
(Sharf, 1964). Additionally, consonants that precede vowels
have no consistent effect on vowel duration (Peterson &
Vowels in stressed syllables are longer than those in
unstressed syllables (Lehiste, 1972; Umeda, 1975). At the
sentence level, vowels in stressed words have longer
durations than those in unstressed words (Peterson &
Lehiste, 1960; Port, 1981). These vowel duration effects
from stress are closely related to those from length of
utterance and rate of speech (Gay, 1978).
As word length increases, the duration of the vowels
within the word decreases; the same effect was found at the
sentence level (Klatt, 1973; Lehiste, 1972; Port, 1981).
Lehiste (1972) also reported that vowel durations are
shorter in syllables at the beginning of phrases than at the
end of phrases. A limiting factor of these variations is
that each vowel has a minimum duration, beyond which there
can be no further reduction (Klatt, 1973; Port, 1981). The
minimum phoneme duration is consistent across increased
utterance length and increased speaking rate (Port, 1981).
When speaking rate is altered, the most variable portion
is the vowel duration (Gay, 1978; Miller, 1981). Miller
(1981) stated that increased speech rate not only reduces
absolute vowel duration but also reduces relative vowel
duration. He found a greater reduction in the duration of
tense vowels than in the duration of lax vowels, thus
reducing the difference between tense and lax vowels. When
the speech rate is altered, the listener alters the duration
parameters for perception of vowels (Verbugge & Shankweiler,
1977). These authors reported that for a given rate of
speech there are established duration parameters for vowels.
If those parameters are violated, the vowel will be
As the preceding discussion indicates, phoneme duration
is a complex index of speech timing. In order to effec-
tively utilize phoneme duration as an index of speech
timing, several controls are needed. Since Umeda (1977)
reported that word position is important for consonants,
this will need to be systematically varied while maintaining
a similar vowel environment (Oller, 1973). Different places
of articulation (Lehiste, 1972; Umeda, 1977) and both voiced
and voiceless consonants (Brown, 1979; House, 1961; House &
Fairbanks, 1953; Klatt, 1973 Lisker, 1957; Muller & Brown,
1980; Sharf, 1962, 1964; Subtelny et al., 1966) are also
needed in order to effectively measure phoneme durations.
By investigating consonant and vowel durations under these
conditions, specific speech timing information can be
Voice Onset Time
The other index of speech timing, voice onset time,
reveals different speech timing information than phoneme
duration. While phoneme duration involves primarily
articulatory events, voice onset time involves the inter-
action of articulatory and phonatory events. By using voice
onset time as an index of speech timing, the age differences
in two components of the speech production mechanism may be
examined during stop consonant production.
Voice onset time is the difference between the onset of
quasi-periodic oscillation of the vocal folds and oral
release of the consonant (Lisker & Abramson, 1964; Zlatin,
1974). Significant differences of voice onset time occur
with changes of voicing and place of articulation
(Abramson, Lisker, & Cooper, 1965; Flege, 1982; Klatt, 1975;
Lisker & Abramson, 1964, 1967; Smith, 1978). The differ-
ences between the voice onset times of voiced stop conson-
ants and their voiceless cognates are consistently signifi-
cant in English and in other languages (Lisker & Abramson,
1964). The consonants /b/, /d/, and /g/ are characterized
by voice onset times that range from prevoicing, or vocal
fold oscillation before consonant release, to voicing that
lags shortly behind the release of the consonant. Their
homorganic cognate consonants /p/, /t/, and /k/ are charac-
terized by voice onset times having a considerably greater
lag of voicing after the consonant release (Abramson et al.,
1965; Flege, 1982; Klatt, 1975; Lisker, 1957; Lisker &
Abramson, 1964, 1967; Smith, 1978).
Besides separating stop consonants by voicing, voice
onset time differences can consistently separate stop
consonants by place of articulation (Klatt, 1975; Lisker &
Abramson, 1964, 1967; Zlatin, 1974). Voice onset times are
shortest for labial stops, then apical stops, and are
longest for velar stops (Lisker & Abramson, 1964, 1967;
In measuring voice onset time, the same controls are
relevant as they are for phoneme duration. For both indices
of speech timing the voicing contrast and place of articula-
tion are important variables. The two indices should reveal
subtle variations in speech timing that result from the
age-related physiologic differences in the speech production
Intraoral Air Pressure
Investigation of the phonetic differences in intraoral
air pressure have been reported as long ago as Rousselot
(1897). Phonetic importance was ascribed to intraoral air
pressure pulses by Stetson (1951), who stated that the
syllable was the basic unit of speech production. The
pressure pulse was further investigated by Black (1950) who
reported that voiceless stop consonants exhibit greater peak
intraoral pressure values when compared to their voiced
cognates. Over the past thirty years the equipment used to
determine intraoral air pressure has been improved and shown
to be more reliable (i.e., Arkebauer et al., 1967; Brown,
1979; Subtelny et al., 1966).
Intraoral air pressure has proven to be an effective
means for measuring physiological differences among
phonemes. Information gathered from measuring intraoral air
pressure includes peak magnitudes of intraoral air pressure,
consonant durations, and vowel durations (i.e., Arkebauer et
al., 1967; Brown, 1979; Brown & McGlone, 1969a; Brown &
McGlone, 1969b; Subtelny et al., 1966).
Phonetic differences between stop consonants have been
reported in intraoral air pressure studies. The most fre-
quently reported finding is greater peak amplitudes of the
pressure pulse for voiceless consonants than for their
voiced cognates (Arkebauer et al., 1967; Bernthal &
Beukelman, 1978; Black, 1950; Brown, 1979; Brown & McGlone,
1969a, 1979; Brown, McGlone, Tarlow, & Shipp, 1970; Karnel &
Willis, 1982; Klich, 1982; Lisker, 1957; Malecot, 1955,
1966a, 1966b, 1968; Netsell, 1969; Subtelny, Worth, &
Sakuda, 1966; Tatham & Morton, 1973; Warren, 1964; Warren &
Hall, 1973). Several investigators (Arkebauer et al., 1967;
Black, 1950; Brown, 1979; Malecot, 1955; Subtelny et al.,
1966) also reported that the peak intraoral pressure assoc-
iated with the production of stop consonants is greater than
that for fricative consonants. Word initial and word medial
stop consonants exhibit greater peak pressure values than
word final stop consonants (Arkebauer et al., 1967; Brown,
1979; Brown et al., 1970; Dixit & Brown, 1978; Malecot, 1968;
Subtelny et al., 1966;). Additionally, peak intraoral air
pressure values increase with increases in speech intensity
(Arkebauer et al., 1967; Brown & Brandt, 1971; Brown et al.,
1973; Hixon, 1966). Brown et al. (1973) reported that the
peak intraoral pressure associated with consonants is sensi-
tive to the vowel context, with peak pressure values being
greater for consonants preceding /i/ than for those preced-
ing /u/, which are greater than those preceding /a/. Peak
intraoral pressure values for consonants have been reported
to be stabile between presentations in syllables and in
words as well as across repeated productions (Brown &
McGlone, 1969b; Brown & Shearer, 1970; Murry & Brown, 1976).
The phonetic differences that are evident from peak intra-
oral air pressure are found among children as young as six
years old and remain relatively stable into adulthood
(Arkebauer et al., 1967; Brown, 1979).
A natural extension of the above studies is to find if
the peak intraoral pressure values are different for older
subjects. The known changes in perioral mucosa, muscula-
ture, and innervation (Kaplan, 1971; Klein, 1980) may affect
both the ability to generate peak intraoral pressures and
the ability to sense the intraoral pressures that have been
generated. Age-related alterations of perioral sensation
may be qualitatively different from the anesthetically
induced loss of perioral sensation previously investigated
(Prosek & House, 1975; Ringel & Steer, 1963; Scott & Ringel,
1971). The use of anesthetic to block perioral receptors
did not result in significant differences in peak intraoral
air pressure or consonant durations. Perhaps the age-
related reductions in neural behavior combined with the age
related changes to the oral musculature and mucosal systems
result in significant reductions in peak intraoral air
pressure values among older adult speakers.
For a nonspeech task, Ptacek et al. (1966) reported
that maximum intraoral pressure as measured by a U-tube
manometer was significantly reduced for older speakers in
comparison to younger speakers. However, the nonspeech task
of maximum intraoral pressure requires none of the precision
or short duration involved in normal articulation.
Leeper and Noll (1972) found intraoral air pressure and
speech intensity varied directly with effort level. Simi-
larly, Brown and Brandt (1971, 1972) reported that, under
masking, their subjects increased intraoral air pressure
concomitant to increases in sound pressure level. Results
of these studies indicate a strong relationship between
speech effort and intraoral air pressure and vocal inten-
sity. Together, the two variables provide a physiologic
measure that is closely related to an acoustic measure.
Studies of vocal intensity frequently relate intensity
to a physiologic variable, possibly because of the
difficulty in producing consistent vocal intensity levels
(Leeper & Noll, 1972). These authors found that even when
subjects attempt to control their effort level, vocal
intensity level exhibits great variability. In a series of
studies on "comfortable" loudness levels, Brown and his
associates (Brown, Murry, & Hughes, 1976; Murry & Brown,
1983) have found within session vocal intensity to be more
consistent than between session vocal intensity. They
found that between session vocal intensity varied as much
as 25 dB over five days within a single subject. Addition-
ally, Rubin et al. (1967) reported that intersubject
variability of vocal intensity is greater than intrasubject
The physical properties of the radiating source of the
speech as well as the proximity of the microphone affect the
recording of vocal intensity (Hixon, 1966; Rubin et al.,
1967). The greater the cross-sectional area of the open
mouth, the greater is the intensity (Hixon, 1966; Isshiki,
1965), and reducing the mouth-to-microphone distance by
half increases the intensity by 6 dB (Rubin et al., 1967).
While controlling the mouth opening is difficult during
running speech, a need for controlling the mouth-to-
microphone distance when recording vocal intensity is
Vocal intensity has been compared to physiologic
variables such as subglottal air pressure, intraoral air
pressure, and air flow. The most consistent finding is that
increases in subglottal air pressure are accompanied by
increases in vocal intensity (Hixon, 1966; Isshiki, 1964,
1965; Ladefoged & McKinney, 1963; Rubin et al., 1967).
Hixon (1966) also reported that changes in vocal intensity
are always accompanied by changes in intraoral air pressure.
In contrast to the findings for the two air pressure mea-
sures, no clear relationship between air flow rate and
vocal intensity has been found in the modal register
(Isshiki, 1964, 1965; Rubin et al., 1967). Since no clear
relationship between air flow and vocal intensity has been
found and since measurement of subglottal air pressure is
highly invasive (Isshiki, 1964, 1965), intraoral air pres-
sure is a preferred variable to be used in conjunction with
vocal intensity. When intraoral air pressure and vocal
intensity are used in combination, an effective physiologic
and acoustic combination is achieved for measuring
age-related differences in speech.
As previously expressed, there is an increased number
of older people in the population. As the number of older
people has increased, interest in investigating age-related
changes has increased also. In the field of phonetics, the
effects of age differences on the perception and acoustics
of speech have been studied.
Perceived aging effects upon speech production may
occur because of anatomic and physiologic changes within the
bony and cartilaginous skeleton, the muscular body, and the
protective mucosal covering (Kahane, 1981; Meyerson, 1976).
The alteration of each of these tissues with aging may cause
variations in speech timing, peak intraoral air pressure
magnitudes, and speech intensity, parameters which may
affect the effectiveness of communication.
Stiffening of the skeletal framework, atrophy of the
musculature, and drying of the mucosa of the speech
production mechanism are likely to alter the velocity of
these structures. Alterations in velocity can be measured
by indices of speech timing. Since older speakers exhibit
slower rate of speech (Mysak, 1959; Mysak & Hanley, 1959;
W. Ryan, 1972; Shipp & Hollien, 1971), investigation of
specific components of the speech may determine the nature
of the slower speech. By measuring phoneme duration and
voice onset time, temporal variations of the component
phonemes and the relationship between consonants and the
contiguous vowels may be revealed. This information may
indicate if the slower rate of speech among older speakers is
related to slower velocities of the articulators or possibly
inertia in stopping and starting articulatory movements.
The voice onset times will also provide information
concerning the interaction of the articulatory movements and
the onset of vocal fold oscillation.
Similarly, the stiffening of the skeletal framework,
atrophy of the musculature, and breakdown and alteration in
the extracellular fibrous network within the speech
production mechanism could limit the pressures that may be
generated for speech. In fact, Ptacek et al. (1966) found
reduced maximum speech intensity levels among their older
adult subjects. However, W. Ryan (1972) reported increased
intensity for speech among his older subjects. These
divergent findings implicate different systems as being
important in determining the speech intensity of older
speakers. The discrepancy of these findings needs to be
resolved. This can be accomplished by measuring speech
intensity at comfortable, minimum, and maximum loudness
levels. The close relationship between intraoral air
pressure and speech intensity establishes a check of
reliabiliy between the physiologic and acoustic measures.
In this manner a valid indication of age-related differences
along these parameters will be provided.
As established throughout the previous discussion, the
age-related physiologic differences in the speech production
mechanism may affect speech timing, intraoral air pressure
magnitudes, and speech intensity. Phoneme duration and
voice onset time may reveal aspects of aging variations in
speech timing. These variables may be related to perceptual
features that listeners utilize when determining speakers
Listeners have demonstrated the ability to separate
taped voice samples of young adult speakers from those of
old adult speakers, and to discriminate adult speakers by
age decade (Hartman, 1979; Hartman & Danhauer, 1976; Ptacek
& Sander, 1966; E. Ryan & Capadano, 1978; Shipp & Hollien,
1969). While age differences are perceived in adult voices,
the specific features that vary across age groups are not
clear. The determination of the presence of age-related
differences in speech timing, intraoral air pressure magni-
tudes, and speech intensity may provide baseline data for
isolating features relevant to the perception of a speaker's
Information concerning speech timing, intraoral air
pressure magnitudes, and speech intensity may be useful in
determining the relevant variables for future perceptual
studies of age-related speech differences. In addition,
the physiologic data will increase the knowledge concerning
the normal changes that occur in speech over the course of
adult aging. This information will be useful for those
generating models of the effect of adult aging on the
The purpose of this investigation was to determine if
age-related physiologically based differences exist in the
speech of adult females. A perceptual study was completed
to determine if listeners could perceive the age differences
between the two populations.
The following competing research hypotheses were tested:
1. Listeners can differentiate the two age groups of
speakers, but no age-related differences are found for the
relevant physiologic and acoustic variables.
2. Listeners can differentiate the two age groups of
speakers, and age-related differences are noted for the
relevant physiologic and acoustic variables.
3. Listeners cannot differentiate the two age groups of
speakers, and no age-related differences are found for the
relevant physiologic and acoustic variables.
4. Listeners cannot differentiate the two age groups of
speakers, but age-related differences are noted for the
relevant physiologic and acoustic variables.
The increasing proportion of older adults in the popu-
lation has resulted in certain health related professions
exhibiting increased interest in age-related physiologic
differences in the speech mechanism. These physiologic
differences may be related to listeners' ability to deter-
mine the age of voices that they hear. The perceived age
differences may a result of age-related changes in the
respiratory, phonatory, and articulatory components of the
speech production mechanism. The reported physiologic
differences within the speech production mechanism may
significantly affect speech timing, intraoral air pressure,
and vocal intensity. Relevant indices of speech timing
include phoneme duration and voice onset time. By measuring
phoneme duration, voice onset time, intraoral air pressure,
and vocal intensity, it may be determined if the presence
of age-related physiologic differences within the speech
production mechanism affect the speech produced. This
information may be important for isolating the parameters
which are relevant to the perceived speech differences
between older and younger speakers.
The speakers were 50 ambulatory, Caucasian women
divided by age into two equal size groups consisting of 25
each. The first group included women aged 20 to 35 years,
who represented young adults. The second group included
women over 75 years of age, who represented old adults.
Since physiologic aging is evident by 70 years of age even
among healthy individuals (Hayflick, 1981), the women over
age 75 should exhibit the normal physiologic changes that
accompany advanced age. These two groups represent the
end-points of age-related differences in speech physiology.
By testing women at the end-points, this author assumes that
any possible age differences should be evident, without the
potential masking of the differences that may occur by using
women across the age span.
In order for health status to be held as constant as
possible across age groups, a short questionnaire (Appendix
1) and a hearing screening were used to determine if the
women meet the following criteria:
1) The women were capable of maintaining an independent
household, or were independent living.
2) They reported no history of bronchial asthma,
emphysema, or other respiratory disease.
3) The women had no reported history of aphasia or
other neurologic disorders affecting the speech or
4) They reported no history of senile dementia or
5) They had no reported structural abnormalities of the
6) They exhibited no articulation errors as determined
by a speech-language clinician during the reading of a
These criteria were chosen to help the investigator
minimize possible health variations between the two groups.
The first criterion refers to whether the subjects were
independent living. Lubinski (1981) reported reduced
language usage and reduced variety of utterances among older
people in nursing care facilities in comparison to those who
were independent living. The questions about the respira-
tory system were to reveal if that system was generally
intact. Since the respiratory system is the pump that
provides the air to drive the rest of the speech production
mechanism, information concerning the health status of this
system was needed. The remaining criteria addressed by the
questionnaire relate to the status of the phonatory and
articulatory portions of the speech production mechanism and
the relevant portions of the nervous system. These criteria
include appropriate voice, resonance, and articulation
The final criterion for the selection of subjects was
that they exhibit no hearing loss greater than 35 dB SPL in
at least one ear for the frequencies of 500 Hz, 1000 Hz, and
2000 Hz. An audiologic pure-tone screening was used to
determine if the potential subjects meet this criterion.
The selection of a 35 dB hearing loss in the better ear was
based on the National Health Survey of 1962, as reported by
Bergman (1980). This survey found that the median hearing
level for the better ear of women aged 75 through 79 years
was between 20 and 30 dB for the frequencies that were
screened during this investigation. In addition, Shanks
(1970) reported that the average hearing loss among 60 women
aged 60 through 80 years was 22 dB in the better ear and 28
dB in the worse ear. Shanks (1970) findings agree with
those summarized by Bergman (1980) from studies of the
hearing acuity of older women.
The listeners were two groups of 20 women each drawn
from the same age groups as the speakers. Women aged 20 to
35 years comprised the young group and women over age 75
comprised the old group.
In order to be included in the study, listeners had to
pass a multiple-choice rhyme-type speech discrimination test
(Griffiths, 1967) at 97% correct for the younger group and
84% correct for the older group. The lower speech discrimi-
nation criterion for the older listening group compensated
for the lower speech discrimination scores exhibited by
older people (Bergman, 1971; Goetzinger & Rousey, 1959;
Harbert, Young, & Menduke, 1966; Kasden, 1970; Punch &
McConnell, 1969). At the stated levels of performance, the
listeners in both groups exhibited normal speech discrimi-
nation for their age level. The speech discrimination of
listeners aged 20 to 35 years has ranged between 94 and 100%
(Bergman, 1971; Goetzinger & Rousey, 1959; Kasden, 1970),
while that for listeners aged 75 and over has ranged between
79 and 90% (Bergman, 1971; Goetzinger & Rousey, 1959;
Harbert, Young, & Menduke, 1966; Kasden, 1970; Punch &
McConnell, 1969). This selection criterion permitted the
investigator to infer that the listeners possessed adequate
hearing acuity for both the test instructions and the
The system of measurement used while recording the
speakers in this study was particularly appropriate for the
parameters measured and the populations investigated. Brown
and his associates have demonstrated that intraoral air
pressure recordings provide an accurate method for measure-
ment of phoneme duration (Brown & Murry, 1976; Flege &
Brown, 1982; Muller & Brown, 1980), and when combined with a
laryngeal microphone, voice onset time (Flege & Brown,
1982). The intraoral air pressure and voicing signals were
recorded on separate channels of a system capable of storing
and presenting oscillographic displays of the two signals.
While the traditional method for measuring speech timing has
been spectrography, the oscillographic display has some
advantages over the spectrogram.
Brown (1985) discussed the advantages and disadvantages
of sound spectrography and oscillographic display of intra-
oral air pressure and voicing in the temporal measurement of
speech samples. Disadvantages that he reported for spectro-
graphy included a) the 2.4 second duration available for
each spectrogram is brief, b) the sample needs to be
recorded in a quiet environment, c) the phoneme boundaries
are often difficult to determine from the acoustic signal,
and d) accurate measurement of the spectrogram requires
experience. In contrast, the printed oscillographic display
has no duration constraint, does not require a quiet envir-
onment, provides clear examples of phoneme boundaries, and
is relatively easy to measure. Brown (1985) reported that
the principal advantage of the printed oscillographic
display is that the clarity of the phoneme boundaries makes
them easily measurable (Figure 1); in contrast, spectrogram
have potential difficulties in determining the delineation
of the onset and offset of phonemes (Figure 2).
In addition to providing a clear and accurate presen-
tation of the data, the methodology used was relatively
noninvasive. The insertion of a thin tube (outside diameter
2.82 mm) into the palatal vault to measure intraoral air
pressure was the most invasive aspect of the data collec-
tion. This method is not as invasive as the tracheal
puncture required for accurate determination of subglottal
pressure, the alternative method for measuring air pressure
Figure 1. Representation of an oral air pressure trace and
voice signal for the syllable /pa/. The duration of conson-
ant /p/ is represented by points (a) and (b). A sudden
marked increase in air pressure denotes the onset of conson-
ant closure (a), while the consonant release is signaled by
a sudden drop in pressure (b). The duration of the vowel
/a/ is depicted by the points (c) and (d). The relationship
between the release of the consonant gesture and the onset
of voicing, voice onset time, is depicted by points (b) and
(c). Finally, the magnitude of the peak intraoral air
pressure is depicted by points (e) and (f).
TYPE B/65 SONAGRAM KAY ELEMETRICS CO. PINE BROOK. N. J.
Figure 2. Spectrogram of the phrase "Speak /pa/ again."
Stop consonant release is at point (a) and the onset of
voicing is at point (b).
within the vocal tract. The researcher deemed the rela-
tively noninvasive status to be of particular importance in
insuring the participation of the older subjects. In
addition, the intraoral pressure pulse provided information
on phoneme duration that would not have been available from
subglottal air pressure measures. Thus, the less invasive
technology also provided more information along the para-
menters relevant to this study.
The equipment used in this experiment recorded the
intraoral air pressure signal, the voicing signal, and the
acoustic signal. A schematic drawing of the equipment array
for each of these signals is shown in Appendix B.
Intraoral Air Pressure. Intraoral pressure was
measured by a system beginning with a polyethlene tube
(inside diameter 1.79 mm, outside diameter 2.82 mm, and 25
cm long) that was inserted into the corner of the subject's
mouth. This tube was directed high into the palatal vault
in a manner such that the end of the tube was on a coronal
plane with the posterior margin of the hard palate and free
of contact with the surrounding oral structures. The tube
had holes bored near the end placed in the oral cavity, in
order to help prevent spurious pressure recordings (Hardy,
1965). The opposite end of this tube was connected to a
Statham PM6TC differential pressure transducer that sent
the signal to an Accudata 105 gage, and an Accudata 109 DC
amplifier. The amplified signal passed through a low-pass
filter with a cut off at 100 Hz to eliminate phonation
vibrations that would add quasiperiodic variability to the
pressure signal. The amplified and filtered signal was
directed to one channel of a Honeywell 1508A Visicorder.
The Visicorder is a light recording oscillograph that
creates permanent recordings by directing a light beam
reflected from a galvonometer onto light sensitive paper.
Intraoral pressure is an effective method for measuring
peak intraoral pressure (Arkebauer et al., 1967; Brown,
1979; Brown and Brandt, 1971; Brown and McGlone, 1969a;
Brown et al., 1970, 1973; Brown and Shearer, 1970; Malecot,
1955; 1966a; 1966b; 1968; Subtelny, 1966), consonant dura-
tion (Brown, 1979; Flege and Brown, 1982; Muller and Brown,
1980), and vowel duration (Flege, 1982). When intraoral air
pressure measurements are combined with laryngeal vibration,
voice onset time information is made available (Flege and
Voicing signal. The voicing signal was detected by a
system beginning with a contact microphone. The contact
microphone was a unidirectional microphone from a hearing
aid. Since the contact microphone was unidirectional, it
generated a signal only during vocalization and did not
respond to environmental noise. The contact microphone was
held in place against the skin over one plate of the thyroid
lamina by an elastic band that went around the neck and
attached to itself by velcro strips at both ends. The
signal was amplified by a 5x AC amplifier and directed to a
second channel on the Visicorder, where it was simulta-
neously recorded with the air pressure signal.
Acoustic signal. The acoustic signal was detected by a
system beginning with a Quest 1.125 inch PZT ceramic omni-
directinal microphone at a mouth to microphone distance of
28.3 cm. The signal from the microphone passed through a
Quest model 215 sound level meter. This signal was then
amplified and recorded half track on an Akai GX-77 tape
recorder at a tape speed of 19 cm per second. The recorded
signal was directed through a rectifier-integrator circuit.
Then the speech signal envelope was processed through an A/D
converter and sent to a PDP-11/23 computer. The speech
signal envelope was sampled at a rate of 50 Hz, then
processed and analyzed.
Calibration. All equipment was calibrated before and
after each session. The intraoral air pressure system was
calibrated using a U-tube manometer. One inch increments of
pressure were induced into the system and the detected
changes in the fluid of the U-tube manometer caused by this
pressure were recorded in centimeters of water (cmH20) and
checked against previous recordings for reliability. A 1000
Hz tone at 110 dB SPL (re 0.0002 microbar) generated by a
Quest C-12 piston phone set over the microphone was recorded
at the beginning of each taping session to calibrate the
intensity measurement system. In this study the voicing
signal was a relevant variable only for its absence or
presence and not for its amplitude. The contact microphone
was tested for functioning by noting the presence of a
signal on the oscillographic trace during voicing and no
signal at other times.
The Visicorder also was used to observe the presence of
appropriate movement for the two signals displayed by the
galvonometers. This was most relevant for the pressure
signal, since the pressure sensing tube was susceptible to
being occluded by saliva. If the pressure sensing tube
became occluded, the experimenter disconnected it from the
pressure transducer and forced air through the tube from the
end that had been connected to the transducer. In this
manner any saliva was ejected from the tube, which was then
reconnected and the procedures continued or repeated if
Speech sample. The consonants of interest were /p/,
/b/, /t/, /d/, /s/, and /z/. These consonants were com-
bined with the vowel /a/ in a series of CV, VCV, and VC
syllables. These syllables were imbedded in the carrier
phrase "Speak __ again." If the subject voiced contin-
uously through the consonants /b/, /d/, and /z/, the
investigator would have difficulty defining the limits of
prevoicing. So "speak" was used as the initial word in the
carrier phrase to minimize the occurrences of continuous
voicing, since the final /k/ is a voiceless phoneme. The
utterances were said at comfortable intensity and frequency
The subject was directed to repeat each phrase five
times. Each phrase was produced on a separate breath to
avoid variability caused by speaking on a reduced volume of
air. The middle three of the five phrase productions were
averaged to give the mean production value. In addition,
the middle three utterances were compared in order to
determine the constancy of production of the independent
variables. Use of the middle three utterances and not the
initial and final ones controlled for the series effect.
Malecot (1966b) defined the series effect as the systemic
variation in multiphasic utterances with utterance initial
phrases exhibiting greater speech effort and utterances
final phrases exhibiting lesser speech effort than the
central portion of the utterance.
The set of consonants used provided examples of stop
consonants at the labial and alveolar positions and frica-
tive consonants at the alveolar position in an established
phonetic environment. Place of articulation is a signifi-
cant variable for indices of speech timing as demonstrated
for phoneme duration (Lehiste, 1972; Stathopoulos & Weismer,
1983; Umeda, 1977) and voice onset time (Klatt, 1975; Lisker
and Abramson, 1964, 1967; Zlatin, 1974). Significant
variables for phoneme duration include manner of articu-
lation (Lehiste, 1972; Umeda, 1977), intraoral air pressure
(Arkebauer et al., 1967; Black, 1950; Brown, 1979; Brown &
McGlone, 1969a, 1969b; Malecot, 1955; Netsell, 1969),
consonantal voicing (Brown, 1979; Lisker, 1957 Muller &
Brown, 1980; Sharf, 1962), and voice onset time (Flege,
1982; Klatt, 1957; Lisker, 1957; Lisker & Abramson, 1964,
1967). By measuring the indices at both positions and in
both manners at the alveolar position, possible differences
by age were determined.
The phonetic environment was designed to maximize the
phonetic stability of the test consonant environment. A
constant vowel controlled for nonrelevant consonant varia-
tion caused by vowel differences (Oller, 1973). In
addition, using a carrier phrase avoided the phonetic
variability caused by utterance initial or final placement
(Muller and Brown, 1980).
The speech samples used for the listening task were the
second and third productions of the phrase "Speak /aba/
again." The second and third repetitions of the phrase are
relatively stable productions, since these repetition are
neither at the beginning nor end of the phrase repetitions.
The speech samples were duplicated from an Akai GX-77
two-track tape recorder onto a Panasonic RX-F11 cassette
tape recorder. During the listening sessions, the Panasonic
tape recorder was used. The listening environment was
structured such that the speech samples were at least 75 dB
at the heads of the listeners, as measured on a Quest 215
sound level meter.
Subject selection. All subjects were volunteers, with
the younger group consisting of students from the University
of Florda and residents of the Gainesville community, and
the older group consisting of individuals belonging to sev-
eral Gainesville senior citizens organizations. All sub-
jects completed the procedures during a five month period.
During a preliminary session each subject had her
hearing screened at 35 dB SPL for the frequencies 500 Hz,
1000 Hz, and 2000 Hz. The hearing screening task was first
in order to encourage participation. If she successfully
passed the hearing screening, she then completed the
questionnaire (Appendix A). If the potential subject met
the criteria specified by the questionnaire, then she read
the first paragraph of the "Rainbow Passage" (Fairbanks,
1954). If her articulation was within normal limits,
she was scheduled for an experimental session.
Preliminary procedures. When each subject arrived for
the experimental session, she was seated in an IAC booth.
The contact microphone was placed on the skin over one plate
of the thyroid lamina and kept in place by an elastic strap
with velcro at the ends that was positioned around the
The acoustic microphone was positioned at a constant
microphone to mouth distance of 28.3 cm and at an angle of
25 degrees to the left of the mouth, relative to the front
of the face. The angle of the microphone and mouth to
microphone distance were maintained by having the subjects
place the front of their chin midway between their lower lip
and the bottom of their chin against a preset wooden scale.
Every subject was directed to keep her chin in contact with
the wooden scale.
Intensity procedures. The subject was directed to
produce the vowel /a/ three times for five seconds each at a
comfortable loudness level. For this task the sound level
meter was set at the dB level that most closely approx-
imates a 0 VU reading. Then she was asked to repeat the
vowel /a/ for five seconds at her maximum loudness level
three times. Again, the sound level meter was set using the
dB level that most closely approximates a 0 VU reading.
Each time she repeated the vowel /a/, the experimenter
urged her to produce it louder than she had the previous
time. Then the subject was asked to produce the vowel /a/
for five seconds at her minimum loudness level three times.
The sound level meter was again set to most closely
approximate a 0 VU reading. Each time she repeated the
vowel /a/, the experimenter encouraged her to produce it at
a softer level than the previous time. The maximum and
minimum loudness level productions were repeated three times
each so that the subjects were able to overcome normal
inhibitions against speaking at the limits of their dynamic
The maximum volume task may be considered to be
subjective; however, it seemed that by encouraging the
subjects throughout the task, the loudest possible vocal-
ization samples were at least attempted. This task preceded
the timing and intraoral air pressure tasks so that the
voice was least fatigued when attempting the maximum
Timing and intraoral pressure procedures. For this
task, the subject had the pressure sensing tube positioned
in her mouth. Once the tube was correctly positioned, she
was directed to say a series of CV, VCV, and VC syllables
using the bilabial and alveolar stop and alveolar fricative
consonants /p/, /b/, /t/, /d/, /s/, /z/ and vowel /a/ for a
total of 18 test syllables. These syllables were inbedded
in the carrier phrase "Speak again." The subject was
instructed to repeat the utterances at comfortable frequency
and intensity levels. Each utterance was repeated until it
was judged by the experimenter to be produced at a comfort-
able level. Then, the subject was directed to repeat each
phrase five times, with each phrase production on a separate
Preliminary procedures. The 50 recordings were
duplicated and the second and third productions of "Speak
/aba/ again" were extracted. Each speaker was presented
twice in random sequence on the 100-item experimental tape.
Each item on the experimental tape included the item number,
followed by a one second pause, which was followed by the
experimental phrase, which was followed by a five second
pause. The experimental tape was played over a loudspeaker
to a group of listeners in a quiet room. Seating for all
listeners was adjusted to provide at least a 75 dB sound
pressure level at the head of each listener. These inten-
sity levels were calibrated before each listening session
using a Quest 215 sound level meter.
Experimental procedures. A multiple-choice rhyme-type
speech discrimination screening test (Griffiths, 1967) was
administered to all subjects at the beginning of each
listening session. To exhibit adequate listening skill, the
older subjects had to achieve a discrimination score of 84%
or better and the younger subjects had to achieve a
discrimination score of 97% or better.
After the speech discrimination test, the listeners
were informed that they were about to hear a tape of voices,
and that they were to estimate the age of each speaker.
Then, the listeners received a response sheet (Appendix C)
on which to write their age estimates. Once all the
listeners had their response sheets, they were told that the
voices they were about to hear came from two groups of
speakers, one young and one old. The listeners were
instructed to make the best estimate that they could as to
the age of the speaker for each voice sample. Additionally,
the listeners were told that the speech samples could be
repeated in order for them to make the best possible estima-
tion of speaker age. Five example items were provided at
the beginning of the experimental tape so that the listeners
could adapt to the task.
Measurement of the peak intraoral air pressure and the
indices of speech timing, phoneme duration and voice onset
time, was accomplished by marking the oscillographic trace.
Vocal intensity was measured by a computer program (Hicks,
1980). Schematic examples of the oscillographic signal
and the points of measurement are depicted in Figure 1
Phoneme duration. Consonant duration (figure 1) was
determined from the point where the intraoral air pressure
trace departed from baseline (point a) until the drop in the
pressure tracing indicating the articulatory release of the
consonant (point b). The departure of the intraoral air
pressure signal from baseline (point a) indicated the moment
of articulatory closure, and the sudden drop in intraoral
air pressure (point b) indicated the moment of articulatory
release. Vowel duration was the time elapsed from the onset
of oscillation of the voicing trace (point c) until either
the final glottal pulse registered by the throat microphone
(point d) or the departure of the intraoral pressure tracing
from baseline for the following phoneme.
Voice onset time. Voice onset time (figure 1) was
defined as the time elapsed from the sudden drop of the
intraoral pressure trace indicating the articulatory release
of the consonant (point b) to the onset of oscillation of
the voicing signal (point c).
Intraoral air pressure. Peak intraoral air pressure
(figure 1) was indicated by the point of maximum departure
(point e) from baseline (point f) of the intraoral pressure
Vocal intensity. Vocal intensity was analyzed using
a computer software package developed by Hicks (1980). The
analysis procedure involved directing the taped speech signal
through a rectifier-integrator circuit. Then the speech
signal envelope was processed through an A/D converter and
sent to a PDP-11/23 computer. The speech signal envelope
was sampled at a rate of 50 Hz, and then processed and
analyzed. The printout from this program provided the
following information: A) mean intensity, B) standard
deviation, C) intensity distribution in two forms: 1) table
and 2) histogram (Figure 3). Information from the computer
printout was then converted to absolute intensity values by
determining the relationship between the intensity value of
the calibration tone generated by the piston phone and the
level at which the sound level meter had been set during the
initial recording. This procedure, which is depicted at
Figure 3 part c, converted the constant voltage output of
the sound level meter back to the original decibel values.
a. DISTRIBUTION TABLE
dB # %
65 5 1.94
66 32 12.40
67 25 9.69
68 31 12.02
69 19 7.36
70 20 7.75
71 52 20.16
72 56 21.71
73 17 6.59
74 1 0.39
c. sound level meter
setting = 110 dB
calibration tone = 74 dB
dB in distribution table
65 = 101
66 = 102
67 = 103
68 = 104
69 = 105
70 = 106
71 = 107
72 = 108
73 = 109
74 = 110
dB 5 5 6 6 7 7 8 8 9 9
0 5 0 5 0 5 0 5 0 5
Figure 3. Vocal intensity information was presented in a
distribution table (a) and a histogram (b) by the computer.
Conversion from the constant voltage level of the sound
level meter involved knowing the setting of the sound level
meter and intensity of the calibration tone (c).
Constancy of phoneme production. Constancy of phoneme
production across repetitions was analyzed for the middle
three productions of the test syllables. This analysis
involved comparing the standard deviations of the phoneme
durations, voice onset times, peak intraoral air pressure,
and vocal intensity using the F-max test for homogeneity of
variance (Winer, 1971).
For the speaking tasks, both intra- and interjudge
reliability was determined. Random selections of 30% of the
tracings from the Visicorder were remeasured by the princi-
pal investigator to determine intrajudge reliability.
Similarly, random selections of 33% of the tracings were
measured by a second investigator to determine interjudge
reliability. Intrajudge reliability was 98% and interjudge
reliability was 97%. Ten percent of the intensity produc-
tions were analyzed twice using the computer program; inter-
run reliability was 96%. In addition, 10% of the intensity
productions were measured using a Bruel & Kjaer Graphic
Level Recorder. When the intensity levels measured from the
level recorder tracings were compared with those from the
computer printouts, the reliability between the two machines
was 92%. Lower reliability figures for this procedure may
be because of different sampling rates. The computer pro-
gram sampled the intensity productions at 50 Hz, while the
sampling of the level recorder tracings was at 10 Hz.
For the listening task, intrajudge reliability was
determined by correlating each of the judges first and
second age estimates for each of the speakers. This corre-
lation showed how consistent each of the judges was in
making age estimations.
Independent variables in this study were a) the age
group of the speaker: young (20 to 35 years) and old (75
years and older); b) the consonants: stops /p/, /b/, /t/,
and /d/ and fricatives /s/ and /z/; c) the word place: word
initial, word medial, and word final; d) the vocal effort
level: comfortable, maximum, and minimum; and e) the age
group of the listener: young (20 to 35 years) and old (75
years and older). Dependent variables for the speaking
tasks of this study were 1) peak intraoral air pressure, 2)
consonant duration, 3) vowel duration, 4) voice onset time,
and 5) vocal intensity. The dependent variable for the
listening task was the estimated age of the speakers.
The primary method of statistical analysis for the
speaking tasks was analysis of variance (ANOVA). A 2 x 6 x
3 ANOVA was conducted for peak intraoral air pressure,
consonant duration, vowel duration, and voice onset time. A
2 x 3 ANOVA was conducted for vocal intensity. All ANOVAs
were performed with an alpha level of p=.05. Any signifi-
cant findings were explored pairwise using the Tukey LSD
method (Winer, 1971). Constancy of the productions within
speakers was tested using an F-max test for the homogeneity
of variances (Winer, 1971).
The primary method of statistical analysis for the
listening task was the Pearson product-moment correla-
tion. The mean age estimation by the listeners were
calculated for each speech sample. These means were
correlated with the chronological age of each speaker.
Previous researchers consistently have found high corre-
lations, r=.76 to r=.93, between perceived speaker age and
chronological age (Hartman, 1979; Hartman & Danhauer, 1976;
Horii & W. Ryan, 1981; E. Ryan & Capadano, 1978; W. Ryan &
Burk, 1974; Shipp & Hollien, 1969).
The purpose of this study was to investigate the
possible presence of age-related physiological differences
between the speech of two groups of adult females. These
differences were investigated for the variables of (1)
speech timing, (2) intraoral air pressure, and (3) vocal
Speech timing and intraoral air pressure recordings
were obtained from a system that began with (1) a throat
microphone and (2) a sensing tube attached to a pressure
transducer. These signals were amplified and displayed as
tracings from two channels on the print out of an oscillo-
graphic recorder. Tracings on these print outs were
measured to obtain information on the dependent variables
(1) peak intraoral air pressure, (2) consonant duration, (3)
vowel duration, and (4) voice onset time.
Vocal intensity data were obtained through a system
that included a condenser microphone, a sound level meter,
and an audio tape recorder. The recorded signals were
directed through a rectifier-integrator circuit, an A/D
converter at a sampling rate of 50 Hz, and a PDP-11/23
computer. The output from the computer was converted
from the sound level meter constant voltage levels to the
decibel levels of the original signal.
Perceptual data were obtained by playing short utter-
ances from each speaker to two groups of listeners, one
group age matched with the younger speakers and the with the
older speakers. The listeners recorded their estimation of
the age of each speaker that they heard.
All the dependent variables were analyzed for differ-
ences between the two age groups using ANOVA statistical
techniques. The variables of (1) peak intraoral air
pressure, (2) consonant duration, (3) vowel duration, and
(4) voice onset time were analyzed for the effect of the
consonants and, with the exception of voice onset time,
syllable position. When appropriate, ANOVAs of the
interactions of age group, consonant, and syllable position
Intraoral Air Pressure
Peak intraoral air pressure was operationally defined
for this study as the distance from the baseline of the
pressure trace on the oscillograph to the greatest deflec-
tion from that baseline. Variations in peak intraoral
pressure magnitudes were analyzed for both age groups by
consonant and by syllable position. The comparisons were
completed both within and between the age groups.
Significant differences in peak intraoral air pressure
were found between the two age groups (Table 1). Signifi-
cant differences were found for the interactions of (1) age
with consonant and (2) age, consonant, and syllable posi-
tion. The interactions of age and syllable position and did
not exhibit significant differences between the two age
Consonants. Both age groups exhibited greater peak
intraoral air pressure values for the voiceless consonants
/p/, /t/, and /s/ in comparison to the voiced consonants
/b/, /d/, and /z/. Within group comparisons, using Tukey's
Studentized Range (Winer, 1971), indicated that the peak
intraoral pressure values from both groups of speakers had
significant differences between /p/, /t/, and /s/ and the
voiced stops, /b/ and /d/ (Table 2). Intraoral pressure
values for /z/ did not differ significantly from the
voiceless consonants for the older speakers.
Between group comparisons of peak intraoral pressure
magnitudes revealed significantly greater intraoral pressure
values among the older speakers than among the younger
speakers for the consonants /p/, /t/, /s/, and /z/ (figure
6). The two voiced stop consonants, /b/ and /d/, exhibited
no significant differences in peak intraoral pressure
between the two age groups.
Table 1. Within and between age group ANOVAs of intraoral
air pressure for the experimental conditions.
df SS F p
Age x Cons.
Age x Syl. Position
Cons. x Syl. Position
Age x Cons. x Syl. Pos.
Consonant x Syllable
Consonant x Syllable
* indicates a significant difference within the condition
Table 2. Within age group comparisons of intraoral air
pressure (PIO) levels for each consonant.
Mean PIO Tukey
Consonant in cmH20 Grouping
/p/ 8.42 A
/s/ 8.21 A
/t/ 7.42 A
/z/ 7.10 A
/b/ 4.38 B
/d/ 4.15 B
/s/ 6.59 A
/p/ 6.53 A
/t/ 6.28 A
/z/ 5.29 B
/b/ 4.74 B C
/d/ 4.39 C
Means with the same letter under Tukey Grouping do not
The between age group difference in peak intraoral air
pressure was examined for the effect of dental status. The
women who wore dentures produced consonants with signifi-
cantly greater intraoral air pressure than either the
younger or older women with natural teeth, the dentate
speakers (Table 3). No significant differences were found
between the younger and older dentate women.
Syllable position. Within age group comparisons of
the peak intraoral pressure values revealed significant
differences by syllable position for the consonants /t/ and
/d/ among the older subjects, and /t/, /d/, and /b/ among
the younger subjects (Table 4). When significant differ-
ences by syllable position occurred, syllable initial
consonants exhibited greater peak intraoral air pressure
values than did syllable final consonants.
The three way interaction of age, consonant, and
syllable position also was significant for peak intraoral
air pressure (Table 1). When the data were more closely
examined (Table 5), the consonants /t/, /d/, and /b/ again
differed from the other test consonants. While the other
test consonants consistently exhibited significant differ-
ences between the two age groups, /t/, /d/, and /b/
exhibited individual patterns. For the /t/ productions, the
CV and VCV syllables exhibited significant differences
between the age groups, while the VC syllables did not. For
the /d/ productions, the CV syllables exhibited significant
differences between the age groups, while the VCV and VC
syllables did not. Finally, for the /b/ productions, none
of the syllable positions exhibited significant differences
between the two age groups.
/p/ /b/ It/ /d/ Is/ /z/
Between group comparisons of peak intraoral
pressure (PIO) by consonant.
Table 3. Comparison of peak intraoral air pressure (PIO) by
age group and dental status.
Voiceless Cons. Voiced Cons.
Mean PIO Mean PIO
Subjects n in cmH20 Tukey in cmH20 Tukey
Younger 25 6.47 A 4.80 A
Older 10 7.02 A 4.21 A
Older 15 8.69 B 5.83 B
different letter under Tukey Group indicates a significant
Table 4. Within age group comaprisons of peak intraoral air
pressure (PIO) by syllable position.
Mean PIO Tukey
in cmH20 Group
Mean PIO Tukey
in cmH20 Group
under Tukey group represents a significant
/a t /
Table 5. Between age
group comparisons of intraoral air
consonant at each syllable position.
* Indicates a significant difference between the age groups
Consonant duration was operationally defined as the
difference between the onset of the air pressure pulse rise
(consonant closure) on the oscillograph trace to the sudden
drop of pressure (consonant release). Variations in
consonant duration were analyzed for both age groups by
consonant and by syllable position. The comparisons were
completed both within and between the age groups.
Significant main effects were found for each of the
three experimental conditions: age group, consonant, and
syllable position (Table 6). Significant interactions also
were found among the consonants by syllable positions.
Within group comparisons (Table 6) revealed that for
both the older and younger subjects, significant differences
were found for the main effects of consonant and syllable
position. Significant interaction effects also were found
for both age groups for the consonant by syllable
Consonants. The voiceless consonants were produced
with greater consonant duration than the voiced consonants
for both age groups (Table 7). The older speakers exhibited
greater consonant durations than the younger speakers for
both the voiced and voiceless consonants. The data in Table
6 reveal that consonants produced by the younger subjects
exhibited significant differences for each cognate pair: /s/
greater than /z/, /p/ greater than /b/, and /t/ greater than
/d/. The older subjects, however, exhibited significant
differences only for the alveolar fricatives, /s/ and /z/,
and bilabial stops, /p/ and /b/.
Table 6. Between and within group ANOVAs for consonant
duration for the experimental conditions.
Age x Cons.
Age x Syl. Pos.
Cons. x Syl. Position
Age x Cons. x Syl. Pos.
Consonant x Syllable
Consonant x Syllable
df SS F p
* indicates a significant difference within the condition
Table 7. Consonant durations (CD) for each consonant by
Mean CD Tukey
Consonant in msec Grouping
/s/ 144 A
/p/ 126 B
/z/ 115 BC
/b/ 106 C
/t/ 104 C D
/d/ 88 D
/s/ 125 A
/p/ 101 B
/z/ 97 B
/t/ 92 BC
/b/ 83 C
/d/ 72 D
Means with the same letter under Tukey Grouping are not
Significant differences exist between the age groups
for all of the consonant durations, except for the consonant
/t/ (Figure 5). For the other test consonants, the older
subjects exhibited significantly greater consonant durations
than did the younger subjects.
Syllable position. The within age group comparisons
of consonant durations by syllable position revealed similar
patterns of significant differences for the younger and
older groups (Table 8). For most of the consonants, the
CV syllables exhibited significantly longer consonant dura-
tions than the VC syllables. The consonant /t/ was the
exception to this pattern, exhibiting insignificant differ-
ences in consonant duration among the three syllable
/pl /b/ It/ Id/ /s/ /zl
Figure 5. Between age group comparisons for consonant
duration (CD) by consonant.
Table 8. Within age group comparisons of consonant
duration by syllable position.
Mean CD Tukey
in msec Group
0.122 A B
Mean CD Tukey
in msec Group
0.100 A B
Different letters under Tukey
difference among the syllable
Group indicate significant
Vowel duration was operationally defined as the differ-
ence between the onset of the oscillation of the voicing
trace (voicing onset) on the oscillograph print out and the
return of the voicing trace to baseline (voicing offset).
Variability in vowel duration was analyzed for both age
groups by consonant and by syllable position. The compari-
sons were completed both within and between the age groups.
Vowel duration variability was examined within and
between the two age groups for each consonant context in the
VCV and VC syllable conditions. Table 9 indicates that
significant differences were found between the age groups
for each of the three experimental conditions: age group,
consonant, and syllable position. In addition, significant
interactions were found (1) between the age groups by
consonant, (2) between the age groups by syllable position,
and (3) among the consonant by syllable position. However,
no significant three way interaction was found for vowel
duration among the three experimental conditions. The older
speakers exhibited greater vowel durations in all consonant
conditions (Figure 6).
Within age group comparisons indicated that both age
groups exhibited significant effects for the experimental
conditions of consonant, syllable position, and the
interaction of consonant and syllable position (Table 9).
Table 9. Between and within ANOVAs for vowel duration (VD)
for the experimental conditions.
df SS F
Age x Cons.
Age x Syl.
Cons. x Syl.
Age x Cons. x Syl.
Consonant x Syllable
Consonant x Syllable
* indicates a significant difference within the
/p/ /b/ It/ /d/ /s/ /z/
Figure 6. Between age group comparisons of vowel durations
(VD) by consonant context.
Consonants. The within group comparisons of vowel
duration for each consonant context are displayed in Table
10. For both age groups, the vowel durations were greatest
in the alveolar fricative context and least in the bilabial
stop context. The vowel durations in the context of voiced
consonants were significantly greater than those in the
context of voiceless consonants (Table 10). Additionally,
the data in Table 10 demonstrate that the older speakers
exhibited longer vowel durations for both the voiced and
voiceless consonant contexts.
Table 10. Within group comparisons of vowel duration (VD)
for each consonant by age group.
Consonant in msec
Consonants sharing the same
not differ significantly
letter under Tukey Grouping do
Syllable Position. Within age group comparisons
revealed that both the older and younger speakers exhibited
greater vowel durations in the VC condition than they did
for the VCV condition (Figure 7). Thus, while the numeric
values differed for the vowel durations between the two age
groups, they exhibited the same pattern of vowel duration
differences between the syllable conditions.
Between age group comparisons of vowel duration
revealed significant differences between the two groups by
syllable position and by consonant (Table 11). The older
subjects consistently exhibited significantly greater vowel
durations when compared to the younger speakers. Differ-
ences in vowel duration were found for both syllable
positions in every consonant context.
Figure 7. Within and between age group comparisons of
vowel duration (VD) by syllable position.
Table 11. Between age group and between
comparisons of vowel duration
(VD) by consonant.
df SS F p
* indicates a significant difference in vowel duration for
the consonant under the stated condition
Voice Onset Time
Voice onset time was operationally defined as the
difference between the sudden drop of the pressure pulse
trace on the oscillograph print out (consonant release) and
the onset of oscillation of the voicing trace (vowel onset).
Variability in voice onset time was analyzed for both age
groups by consonant in the CV syllable position. The
comparisons were completed both within and between the age
Voice onset time varied significantly between the age
groups, among the consonants, and by consonant between the
age groups (Table 12). Within age group analyses revealed
significant differences among the consonants in both the
older and younger age groups.
Table 12. Between and within group ANOVAs for voice onset
time for the experimental conditions.
Condition df SS F p
Age 1 0.043 19.36 0.0001*
Error 48 0.105
Consonant 5 0.517 95.31 0.0001*
Age x Consonant 5 0.031 5.77 0.0002*
Error 239 0.260
Consonants 5 0.293 38.04 0.0001*
Error 119 0.183
Consonants 5 0.259 34.25 0.0001*
Error 120 0.182
* indicates a significant
difference within the
For both age groups, the voiceless stop consonants /p/
and /t/ had significantly greater voice onset time values
than the voiced consonants /z/, /b/, and /d/ (Table 13).
For the older speakers, the consonants /p/ and /t/ had
significantly longer voice onset times than /s/, and /s/ had
significantly longer voice onset times than /b/, /d/, and
/z/. For the younger speakers, the voice onset times of
/s/, /d/, and /b/ did not differ significantly. However,
the /z/ exhibited significantly shorter voice onset times
than the other voiced consonants produced by the younger
Table 13. Within age group comparisons of voice onset time
(VOT) by consonant.
Consonants with the
same letter under Tukey
Grouping do not
Between age group comparisons of voice onset time
revealed that the younger women produced significantly
longer voice onset times than the older women for the
consonants /p/, /b/, /t/, /d/, and /s/ (Figure 8). Only
the consonant /z/ did not exhibit voice onset times that
differed significantly between the two groups.
/p/ /b/ It/ Id/
Figure 8. Between age group comparisons of voice onset time
(VOT) by consonant.
A factor that may have influenced the voice onset time
data was the occurrence of continuous voicing. Continuous
voicing indicated that a speaker did not discontinue voicing
during the /k/ sound of the word 'speak' in the carrier
phrase, but voiced the vowel /i/ into the next syllable.
This phenomenon was more prevalent among the older speakers,
occurring 56 times on 29 different consonants in comparison
to 38 occurrences on 14 different consonants among the
For measurement purposes, the voice onset time for the
continuously voiced consonants was labeled as prevoiced with
a magnitude equal to the value of the consonant duration.
These voice onset times were operationally defined as the
difference between the consonant closure indicated on the
pressure trace and the onset of the vowel on the voicing
Intensity levels of the speakers' utterances of
sustained /a/ were recorded at three effort levels: minimum,
comfortable, and maximum. The intensity data were analyzed
for significant differences with a 2 x 3 ANOVA. Variations
in intensity level were analyzed for both age groups by
Significant differences in vocal intensity were found
for the main effect of effort level and the interaction of
effort level and age group (Table 14). No significant
differences in vocal intensity were found for the main
effect of age group.
Table 14. Between and within group ANOVAs for vocal
intensity for the experimental conditions.
Condition df SS F p
Age 1 15.59 0.52 0.476
Error 48 1444.00
Effort level 2 60776.88 1333.36 0.0001*
Age x Effort level 2 1631.64 35.80 0.0001*
Error 95 2165.13
*indicates a significant difference for this condition
For both age groups the vocal intensity varied directly
with effort level; the greatest intensities were generated
during the maximum efforts and the least intensities were
generated during the minimum efforts. The interaction of
effort level with the age groups was significant for the
maximum level and the minimum level, but not for the
comfortable level (Figure 9). The data in Figure 9 indicate
that the younger speakers exhibited greater maximum effort
intensity levels and lower minimum effort intensity levels
when compared to the older speakers.
Figure 9. Between age group comparisons of vocal intensity
(SPL) by effort level.
Variability of Speech Production
The variability of the repeated utterances was analyzed
between the groups for the production of the dependent
variables (1) peak intraoral air pressure, (2) consonant
duration, (3) vowel duration, (4) voice onset time, and (5)
vocal intensity. The F-max test for homogeniety of
variance was used to compare the variability between the two
age groups. Significant differences in variability were
found between the two age groups for all the dependent
variables (Table 15). Since no significant age effects were
exhibited for the vocal intensity task, variability was
analyzed between age groups by effort level. This revealed
significantly different levels of variability between the
two age groups for the minimum effort level only.
Table 15. Between age group comparisons of variability by
Variable Age Group df F P
Intraoral Air Pressure Older 446 1.88 0.0001*
Consonant Duration Older 446 1.72 0.0001*
Vowel Duration Older 296 2.01 0.0001*
Voice Onset Time Older 145 2.66 0.0001*
Minimum Older 74 2.91 0.011*
Comfortable Older 74 1.12 0.782
Maximum Older 74 1.34 0.477
*indicates a significant difference between the age groups