University of Southern California Dissertations and Theses

Acoustic predictors of Type A behavior.

(USC Thesis Other)

Acoustic predictors of Type A behavior.

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content ACOUSTIC PREDICTORS O F TYPE A BEHAVIOR
by
Beverly M eyers M ischer
A Dissertation Presented to the
FACULTY O F THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillm ent of the
Requirem ents for the Degree
D OCTO R O F PHILOSOPHY
(Communication Arts and Sciences)
December 1989
C opyright 1989 Beverly M eyers M ischer
UMI Number: DP22451
All rights reserved
INFORMATION TO ALL USERS
The quality of this reproduction is dependent upon the quality of the copy submitted.
In the unlikely event that the author did not send a complete manuscript
and there are missing pages, these will be noted. Also, if material had to be removed,
a note will indicate the deletion.
Published by ProQuest LLC (2014). Copyright in the Dissertation held by the Author.
Dissertation Publishing
UMI DP22451
Microform Edition © ProQuest LLC.
All rights reserved. This work is protected against
unauthorized copying under Title 17, United States Code
ProQuest LLC.
789 East Eisenhower Parkway
P.O. Box 1346
Ann Arbor, Ml 4 8 1 0 6 -1 3 4 6
UNIVERSITY OF SOUTHERN CALIFORNIA
THE GRADUATE SCHOOL
UNIVERSITY PARK 0.
LOS ANGELES, CALIFORNIA 90089 I A »Wi
m
b
HUS
This dissertation, written by
Beverly _ Me^ers _Mischer............
under the direction of h.zv. Dissertation
Committee, and approved by all its members,
has been presented to and accepted by The
Graduate School, in partial fulfillment of re
quirements for the degree of
DOCTOR OF PHILOSOPHY
Dean of Graduate Studies
Date .............. Novemb er # ,1989
DISSERTATION COMMITTEE
'ftairperson
%
DEDICATION
This dissertation is dedicated to my daughters,
Jen n ife r and H eather
They have shared th eir bountiful enthusiasm for all of life with
me and they have been my teachers all the years we have
know n each other.
j ACKNOWLEDGEMENTS
j M any friends and advisors have persevered w ith me in the
| completion of this project. Sincere thanks to everyone who helped,
I
| I am blessed w ith a committee w ith a sense of hum or. June
i Shoup believed in this research w hen almost no one else did. She
; has gently reconfirm ed that curiosity is essential in life and in
j
scientific research; that inspiration and struggle are essential in life
^ and in scientific research; that w riting is re-w riting; and that
i
! researching new ideas requires looking at the roots of the science.
f
i
i Ed Purcell introduced the research er to "industrial stren g th ”
I
statistics. His guidance fostered an expanded perspective of
research, and an appreciation of the findings in this study, Jerry
| Davison graciously accepted this m averick on his psychology
research team. He m anaged his ra th e r disparate team w ith spirited
humor, thoughtful counsel, and excellent scholarship.
; Sigma Xi provided a G rant-in-A id of Research award. USC
, provided a m atching grant. V incent DeQuattro and Jerry Davison
!
provided support th ro u g h th e ir NIH g ra n t # 1R001HL31090-01.
\ Sandra Seagal's work w ith voice and personality originally
! inspired me to this work. Betty McMicken at W estern Medical
!
Center provided technical equipm ent. Don Cooper shared his
lib rary and h is ideas generously. Greg Stratton at Perspective
; Systems saved the data files from the Mac abyss.
Don M ischer has been supportive in every phase of this
project, from its inception to its completion.
Suzanne Roberts was the itin e ran t gram m arian of this
{dissertation. She made me laugh w hen the scene was bleak and the
I
j data w ere gone. Karen Beard graciously shared h e r new computer,
[ all the software in the world, the view of the Pacific, h e r laughter,
| and good nature. Sharon Grinage has a h ea rt of gold and has been
: supportive in a m ultitude of large and small ways th ro u g h all the
i
, years of my academic pursuits. The friendship of all m entioned
! above is an enduring treasure
TABLE OF CONTENTS
PAGE
LIST O F TABLES...................................................................................vii
LIST O F FIGURES..................................................................................viii
CHAPTER ONE
LITERATURE REVIEW................................................................ 1
The Type A Behavior P attern................................................. 2
Type A Assessment.........................................................6
Speech Analysis in Type A Assessment.................. 11
Speech Acoustics.......................................................................17
The Flow of Speech........................................................18
Fundamental Frequency.............................................. 22
Formants..........................................................................24
Form antFrequencies.........................................25
Formant Bandwidths...........................................30
Power................................................................................ 32
Timing and Duration.....................................................34
The Challenge in Type A Assessment..................................37
CHAPTER TWO
STUDY O F ACOUSTIC VARIABLES........................................... 39
D eterm ining Relevant Acoustic Speech Variables........... 39
The Hypotheses..........................................................................40
CHAPTER THREE
RESEARCH METHODS AND PROCEDURES.................................42
NIH Psyche Hypertension Study............................................42
Description of Subjects.............................................................44
The Structured Interview ........................................................45
Structured Interview s for the Acoustic Study 46
M ilitary Commander Monologue...........................................46
Recording................................................................................... 48
Filters..........................................................................................49
TABLE OF CONTENTS - c o n tin u e d
PAGE
I Hum Filter. ,,......................................................................49
| Frequency Devices Series 901F................................... 50
i Digital Analysis of the Speech Wave.................................... 50
M easuring Acoustic Speech Variables................................ 51
Duration M easures....................................................... 51
! Power................................................................................ 54
i Fundamental Frequency............................................... 55
| Form ant Measures...........................................................55
J Speech Units and M easures.......................................... 55
Phrase Measures.................................. 55
j Word M easures......................................................56
i Vowel M easures....................................................56
I The 20 Variable M easures.............................................57
; Phrase Measures................................................... 57
i Word M easures...................................................... 57
! Vowel Measures.....................................................57
I CHAPTER FOUR
| RESULTS.......................................................................................58
| Factor Analysis...........................................................................58
j Statistical Analysis of the Data................................................61
I Discriminant Analysis...............................................................70
! Five Predictor Variables................................................ 73
{ Factor 1 "Duration"................................................ 73
! Factor 2 "Power"................................................... 73
! Formant 1............................................................... 74
Formant 3..............................................................74
WordFO V ariability............................................ 74
CHAPTER FIVE
DISCUSSION.................................................................................75
I Summary of Results.................................................................. 75
Influences of the Five Predictors......................................... 76
! Factor 1 Duration.............................................................78
; Factor 2 Power..................................................................78
! Formant 1......................................................................... 78
TABLE OF CONTENTS - co n tin u e d
PAGE
Form ant3........................................................................ 79
Word FO Frequency V ariability..................................80
Possible Connections and Future Studies............................ 80
Acoustic Predictors of Type A Behavior
Continued................ 80
Speech, the Vagus Nerve and the H eart.................. 81
‘ Emotion, Cognition, FO, and Power V ariability 86
j The Five Coefficients and the Limen Scale .....90
I ATSS and the A nger Video............................................ 91
Acoustic Analysis of Interview er V ariability 92
j Acoustic Speech Analysis and Other Constructs...92
, Acoustic M easures Above Form ant.............................92
| Conclusion................................................................. 93
i
I References..........................................................................................94
i
1 Appendix A
| The Structured Interview .......................................................101
j Interview er Notes.....................................................................106
! Appendix B
Protocol for the Monolo gue.................................................. I ll
M ilitary Commander Monologue ................................. 113
I
LIST O F TABLES
TABLE PAGE
1. Correlation Matrix of 20 Acoustic V ariables................ .62
2. Rotated Factor Pattern M atrix............................................69
3. Distribution of A and B Predictions.................................. 72
4. Standardized Canonical Coefficients.................................72
LIST OF FIGURES
FIGURE PAGE
1. Graph of Type A Non-Type A Hostility........................... 7
2. Interrelationship of Speech Dimensions........................21
3. Spectrum Analysis of Form ants........................................ 26
4. Acoustic Vowel Chart............................................................28
5 Spectrogram of the Word ''Scorch"................................... 52
6. Scree Plot of Eigenvalues..................................................... 68
7, Histogram of A and B Distribution.....................................71
ABSTRACT
This study investigates the acoustic correlates of the speech of
Type A behavior patterns. It has been established th a t trained
j interview ers who rate Type A and B subjects depend in large
; m easure on speech characteristics and stylistics to evaluate the
j subjects. The challenge, therefore, in this study was to identify the
: acoustic features w hich discrim inate Type A and B behavior
1 patterns.
i
j Three Type A subjects and th ree Type B subjects w ere
i
j selected from an on-going investigation of behavior patterns.
Recordings of a monologue w ere collected following a specially
j developed interactive interview designed to challenge subjects and
i to elicit Type A behavior.
j Based on perceptual observations of the speech of Type A and
B subjects, the follow ing hypotheses w ere posited: th a t Type A and
| B speech can be differentiated th ro u g h acoustic speech measures;
th a t Type A and B speech will be differentiated more distinctly in
stressed speech th an unstressed; th at Type A speakers w ill
dem onstrate more variability in fundam ental frequency th an Type
B speakers; th a t Type A speakers w ill show stro n g er power
m easures and m ore power variability than Type B speakers; th at
Type A speakers w ill speak faster th an Type B speakers; and that
form ants will be analyzed as they are critical in com m unicating
speech sounds, although it is not expected that form ants will
differentiate A and B behavior types.
Acoustic m easures w ere taken for those features w hich would
most directly relate to the above hypotheses The follow ing results
j supported the original hypotheses. Acoustic speech m easures
! derived from stressed speech samples differentiated Type A and
i Type B successfully in four distinct areas. Type A speakers exhibit
; more fundam ental frequency variability than Type B. Type A
J speakers dem onstrate m ore variability in power m easures th an
l
' Type B. Direct power com parisons were not included in the
j analysis. Type A subjects speak only slightly faster th an Type B
speakers.
i
| Contrary to expectations, the first form ant freq u en cy is the
j best predictor of Type A behavior. The th ird form ant and the th ree
j form ant bandw idths also participated successfully as predictors.
I
j The value of the form ant m easures in discrim inating Type A and B
behavior patterns, discovered in this investigation, is not only a
m ajor contribution to the field of speech science, but prom ises to be
i
im portant in future investigations relating speech and personality
behaviors th a t may prove to be equally significant.
CHAPTER ONE: LITERATURE REVIEW
, i
j A continuing need exists in psychology and speech science for
j research w hich clarifies the relationship between psychological j
I behaviors and perception of speech. Psychologists and medical
; doctors are increasingly interested in speech as a tool in diagnosing |
and ra tin g medical conditions and psychological states and as an aid
in directing and assessing effective treatm ent techniques (Sanford,
i 1942; Pfau,1978; Hecker, Chesney, Black, 6cRosenman,1981; Rice 6 c
1
S Koke, 1981, K rahulec, Urbanova, 6cRafajdusova, 1985) The re cen t
r
v
; research history of Type A behavior pattern, a v e ry im portant
; psychological-behavioral construct, dem onstrates in a somewhat
dramatic m an n er the im portance of investigating the essential
j interplay of psychological patterns, speech inform ation, and disease.
An overview of Type A research literature, specifically the speech-
I
related studies, indicates the im portance of exploring acoustic
, m easures w hich relate to Type A speech behaviors assessment. The
t
present investigation proposes acoustic speech analysis of Type A
, and Type B speech to determ ine the best acoustic predictors of
j
I Type A behavior.
4
I
i
j The Type A B ehavior P attern
The description of Type A pattern behavior has been in the
| process of re fin em en t during the past 30 years. Over 25 differen t
characteristics and behaviors have been m entioned in the literature
as constituting Type A behavior. Although specific characteristics
i
I of Type A behavior have been refined and expanded, the general
I description has changed v ery little.
j Type A Behavior P attern is an action-em otion complex th at
can be observed in anyone who is aggressively involved in a
! chronic, incessant struggle to achieve more and more in less
j and less time, and if required to do so, against the opposing
j efforts of other th in g s or other persons. . . It is a socially
| acceptable— indeed often praised— form of conflict. Persons
j possessing this p attern also are quite prone to exhibit a free-
: floating but extraordinarily w ell-rationalized hostility.
(Friedman 6c Rosenman, 1 9 7 4 )
; Type A pattern is consistently characterized as th ree basic
I
| interrelated behaviors w hich are provoked in response to
i
j situations or events. They are as follows: (1) explosive vocal
intonation, (2) competitiveness, and (3) latent hostility (Friedman 6c
I Rosenman, 1 9 5 9 ,1 9 7 4 ; Friedm an, Brown 6cRosenman, 1 9 6 9 ) . The
I Type A response is evoked w hen ad e fin in g elem ent of the behavior
■ p attern is engaged. For example. Type A individuals perceive an
t
external incentive to accom plish som ething and at the same time
j
confront a probability of failure. The thw arting elem ents in the
j situation may be a time constraint, a h ig h ly com petent competitor,
| or a persistent fru stratio n of any kind. Any of these events m ay be t
i
i perceived as th rea te n in g to Type A individuals and may arouse
i
feelings of fe ar or hostility It is suggested th at Type A individuals
; fear failure more th at Type B s and subsequently th ey strive to
I
avoid failure more vigorously. Type A behaviors may em erge in i
3
| response to a need to gain and m aintain control of environm ental j
| events, as Type A persons are m ore reactive to th reats to th e ir
! 1
1 sense of control th an Type B individuals (Houston, 1983)
Type A behavior is described as aggressive, dom inant, quick- j
thinking, self-confident, autonomous, extroverted, changeable, and <
1 1
I adventurous. In com parision, Type B behavior is described as
i
1 significantly more deferent, inhibited, introverted, conscientious,
l
1 trusting, and submissive. Type B is commonly defined as a relative
absence of the Type A behaviors (Chesney, Black, Chadwick, 6c
I Rosenman, 1981; Matthews, 1982).
I
j Research literature and clinical experience during the past 30 ,
| years suggest an association between Type A behavior, coronary
h e a rt disease (CHD), coronary atherosclerosis, and re c u rre n t
: m yocardial in farctio n (Friedman & Rosenman, 1959; Rosenman etal.,
I
! 1 9 S 4 ; Rosenman, Brand, Sholtz, 6c Friedm an, 1 9 7 6 ; Williams, Haney,
Lee, Blum enthal, 6c W halen, 1 9 8 0 ) . The W estern Collaborative
i
! Group Study, a m ajor longitudinal study of 3,154 volunteers in good
J health, ages 39 to 59, indicated th a t Type A behavior, as assessed
| by the Structured Interview (SI), constitutes a significant additional
j risk factor for CHD. The Type A risk factor is independent of the
i
: traditionally acknowledged risk factors of age, blood pressure,
| serum lipids and lipoproteins, cigarette smoking, and heredity.
| Twice the num ber of asymptom atic subjects, classified as Type A,
! w ithout m easurable coronary damage at onset, developed CHD
i
; during the first eig h t years of the study Indications are th at the j
j behavior pattern precedes the disease; however, the behavior is not
I confirm ed as a causal factor (Rosenman et al.,1976).
|
I An enduring question concerns the effect of Type A behavior
j on overall health beyond coronary h e a rt disease. In response to
j biochemical and environm ental challenges, Type A subjects
i dem onstrate heightened sym pathetic nervous system reactivity
evidenced in elevated m easures of systolic blood pressure,
, epinephrine or norepinephrine, h eart rate, and serum triglyceride
♦
p rio r to and follow ing a fat m eal (Matthews, 1982; Matthews,
i Krantz, Dembroski, MacDougall, 1982; Baker, Hastings, &c Hart,
i
; 1984). A 22-year follow up study was conducted of the o rginal
W estern Collaborative Group Study (WCGS) using 2,365 men, aged
61-71. The men are 93 p ercen t of the survivors of the original
: W CGS who responded to a h ealth status questionnaire. The results
indicate th a t Type A behavior is positively associated w ith general
m orbidity The m orbidity index is constructed from questions
i
| about chronic disease conditions w hich w ere reported to physicians,
diagnosed and treated. The associated m orbidity p attern is not due
f
to the presence of coronary h e a rt disease or any o ther particular
medical condition. Furtherm ore, extreme Type A personalities
t
| had the h ig h est level of m orbidity w hile those w ith the fewest
j Type A characteristics had the lowest level of m orbidity (Shoham-
Yakubovich, Ragland, Brand 6c Syme, 1988).
I
I Component behaviors and characteristics of the Type A
i
behavior p attern may be individualized somewhat for each person
t
5 I
!
j The com ponents of the behavior m ay be differentially predictive, as j
not all Type A individuals develop h e a rt disease. In a W estern '
I J
j Collaborative Group follow-up study of CHD risk factors questions |
| w ere raised about Type A behavior as a continuing risk factor in
i
| coronary disease, since in the fourth in terv al of the study Type A
i ;
* showed no association w ith CHD m ortality w hile traditional risk
factors of systolic blood pressure, serum cholesterol and age showed
consistent associations. In the analysis of the continuing series of
t
four m ajor studies of the W CGS, it is reported th at Type A and B
i behavior patterns were positively, but n o t significantly, associated
w ith CHD in the first and th ird analysis periods of the study. The
behavior patterns w ere significantly negatively associated in the
second analysis period. In the most re cen t studies, d uring the
22-year in terv al study, Type A and B behavior pattern s w ere not
■ associated w ith CHD. T herefore, Type A and B behavior p attern s
I
are questioned as a risk factor for coronary h e a rt disease m ortality
i (Ragland 6c Brand, 1988a; 1988b). It is im portant to keep in m ind a
corollary study of W CGS in the same 22-year time interval, w hich
! concluded th a t Type A behavior p attern was positively associated
with h ig h e r rates of general morbidity.
i
, There is increasing interest in clinical investigations and
research literature in establishing anger, hostility and aggression as
; an im portant set of underlying com ponents linking Type A to
pathologies (C arver 6c Glass, 1 9 7 8 ; Diamond, 1 9 8 2 ; W einstein,
i Davison, DeQuattro, 6c Allen, 1 9 8 6 ; Chesney 6 cRosenman, 1 9 8 5 )
i Hostility was dem onstrated as an additional risk factor w ith Type A
i
j
I behavior pattern in an investigation w hich focused on the possible
{ relationship of Type A behavior pattern and hostility to the medical
i
i diagnosis of significant coronary occlusions.
Figure 1, show n below, illustrates the effect of hostility,
j Although Type A behavior alone proved to be a g reater risk factor
; than hostility alone, significant occlusions w ere present in high
! percentages for those subjects who w ere "Non A" and who w ere
■ evaluated as h av in g hostility (Williams et ai.,1980). In a 25-year
; follow up of 255 medical students, using MMPI (M innesota
i
| M ultiphasic Personality Inventory) scores, hostility scores emerge
j as more stro n g ly related to atherosclerosis than Type A behavior
j
' p attern (Barefoot, Dahlstrom, & Williams, 1983)
I Type A Assessment
j
R esearchers are continuously developing and refin in g
methods of identifying individuals who exhibit the behavior
pattern. Clearly, the curren tly preferred, although not perfect,
l
| assessm ent technique for determ ining Type A behavior is a
personal interview conducted by trained personnel who pace the
j interview briskly and emphasize key words (Herm an, Blumenthal,
J Black, &Chesney, 1981).
i
The interview , slightly modified since its in itial use in I960,
I covers four topic areas: (1) ambition and drive, (2) competitiveness,
i
' (3) emotional expression, and (4) im patience and time u rgency The
standardized questions in the Structured Interview (SI) w hich was
Percent W ith A t Least O ne Occlusion O f M o re T han 75%
90
80
70
60
50
40
30
20
10
0
Non A A Non A A
(7) (28) (61) (211)
Low Hostility High Hostility
MALES
Figure 1. Comparison of the role of hostility in Type A and Non-A
in significant coronary occlusion incidence.
8
developed from early studies of Type A behavior specifically for
use in the W estern Collaborative Group Study, is shorter, req u irin g
about 10 m inutes to adm inister The revised SI appears to be more
effective th an the original (Friedman &c Rosenman, 1959; Rosenman
& c Friedm an, 1961; Chesney, Eagleston & Rosenman, 1980,
' Rosenman 1978).
I D uring the Structured Interview subjects respond to a
i num ber of questions concerning typical reactions to a variety of
i
! fam iliar situations, SI questions are designed to engage the subject
| and to provoke the Type A behavior complex by finding areas of
I
i discussion in w hich individuals can express in n e r feelings.
Interview ers are trained to in terac t w ith and challenge subjects in a
m an n er th a t is designed to provoke subjects to expose covert
i
hostility, im patience, competitiveness, and other emotional
i
1 responses (Rosenman, 1978). The Structured Interview is discussed
l
j in detail in Chapter Three.
j The Structured Interview is considered more effective th an
i
I paper-and-pencil measures in identifying individuals
| psychologically and behaviorally at risk for developing CHD. SI is
! the Type A m easure w hich best relates to speech characteristics as
i
i predictors of the behavior. Comparative analyses of Type A
i
i assessm ent methods indicate that paper-and-pencil, self-rep o rt
i assessments, such as the Jenkins Activity Survey (JAS),
! Fram ingham Type A Scale, and the Adjective Check List, define
i areas of overlap between research-derived descriptions of the
|Type A behavior pattern and subject self-description. Comparisons
of self-m easures and ra te r assessm ents point out the areas of
Type A behavior th a t Type A subjects deny Rosenm an (1978 ) has
rem arked, "Many Type A subjects believe th ey lack the v ery
Type A traits from w hich they already su ffe r a surfeit." It is clear
from com parative studies th at self-reports m easure Type A factors
| w hich relate to different aspects of the behavior pattern than those
1 most re le v an t to the SI, although the Aggression scale of the
! Adjective Check List correlates best with SI typing of subjects
i
j (Chesney et al.,1981; Herm an et al.,1981; M atthews, 1982).
i Trained ra te r evaluations of interview audio tapes produce a
| subjective global assessm ent of 20 behavioral characteristics w hich
I
I rate subjects on a dim inishing scale: Al, fully developed pattern;
! A2, m any Type A characteristics present, but not the complete
i
j pattern; B3, some Type A characteristics; and B4, relative absence of
! Type A characteristics. In ra re instances subjects are rated Type X ,
i
j an even mix of Type A and B behaviors (Rosenman 1978, Chesney
I etal., 1980; M atthews, 1982).
; In addition to global ratings, SI can be scored reliably for
! individual questions and for Type A speech behaviors Factor
j
analysis of individual answ ers on the interview revealed four
1 underlying independent factors. They w ere (1) ra te r perceptions of
speech behavior, (2) self-reports of pressured drive, (3) anger, and
(4) com petitiveness (M atthews et al.,1982)
I
J There are num erous hypotheses w hich seek to identify the
‘source, or cause, of Type A and B behavior patterns. In
' environm ents in w hich Type A behaviors are rew arded, Type A
behaviors are consequently associated w ith success (Chesney 6c
I Rosenman, 1985) Type A individuals may learn to be h ard driving,
I
I excessively involved in work, im patient with an overriding sense of
, time urgency, com petitive, and hostile. (Matthews, 1982;
■ B lum enthai, O'Toole, 6c Haney, 1984).
i
; It has been proposed th at cognitive processes contribute to
| Type A behavior. Type A individuals hold fears and distorted
j beliefs w hich cause them to perceive events and the environm ent
; in such a way th a t they respond w ith "Type A behavior.” The
i
behaviors impact, and perhaps modify, the en v iro n m en t and other
individuals in a dynam ic way, so th a t the Type-A -response-
| behaviors activate sets of responses. Type A subjects respond to
I
experim ental tasks as more im portant and dem onstrate more
I physiological arousal th an Type B subjects. Type A individuals may
habitually respond to stress w ith more denial and repression than
Type B (P ittn er 6c Houston, 1980; Price, 1982; Houston, 1983;
W einstein, Davison, DeQuattro, 6c Allen, 1986; B ushnell, 1987). Type
A pattern individuals may perceive the environm ent differently
th an Type B pattern and may respond w ith behaviors th ey believe
! prom ote control and provide safety.
j Research suggests speech behaviors, particularly those
1 associated w ith Type A, may serve as indicators of o th er more
j m ajor aspects of behavior, such as com petitiveness, time urgency,
! simulated anger, contempt, dominance, confidence, extravagance,
j boldness, en erg y level and aggressiveness. Studies of the accuracy
I of these speech and psychological associations w ith Type A
| behavior p attern are mixed and generally inconclusive (Williams &
i Stevens,1972; S cherer 6t S cherer, 1981; S cherer, 1981). The
!
j relationship between speech factors and psychological factors
: underlying Type A behavioral characteristics have not been
I satisfactorily identified, n o r have environm ental factors w hich
provoke Type A behavior been exam ined sufficiently to offer
i
j conclusive explanations of the causes of distinctive Type A
. behaviors.
' Speech Analysis in Type A Assessment
1 Evaluation of speech characteristics and stylistics is in teg ral in
scoring the SI as most of the SI scale characteristics relate to speech
behaviors Raters are specifically instructed to focus on the m an n er
j of speech and to disregard content (Rosenman, 1978; Chesney etal.,
; 1980). Studies confirm th at SI interview ers and ra te rs in scoring
! speech factors actually rely upon perception of, in preferential
j order; (1) voice em phasis, (2) speed of speaking, (3) speed of
I
j answ ering, and, despite instructions to the contrary, (4) answ er
j content (Scherwitz, Berton,6c Leventhal,1977; Howland & c Siegman,
1982; B lum enthal et al , 1984).
: In an early speech analysis study, focused on speech
j em phasis, subjects Typed A and B by personal interview , read a
two p arag rap h diatribe of a m ilitary com m ander exhorting his
| troops p rio r to battle. The reading is constructed to induce
I .
exhibition of two Type A characteristics, eagerness to excel at any
challenging activity, and laten t hostility. Experim enters converted
j am plified recordings of the simple and hortatory ren d erin g s of the
| readings into d irect-cu rren t voltage. A resistor capacitor netw ork
| was designed w hich perm itted identification of the loudness w ith
: w hich subjects modulated speech and the duration of loudness
I
■ above a set level. The voice analysis test segregated the m ajority of
i Type A from Type B subjects only in the h o rtato ry reading. Type A ■
i '
| subjects recorded num erous h arsh , explosive speech rhy th m s on
J graphic p rin ts of voltage levels. The investigators suggest the test
I failed to detect an y time u rg en cy in Type A sam ples as average
j
times did not v ary significantly between simple and hortatory
i
| readings (Friedm an et al,, 1969). It is possible th a t the reading did
! not induce a sense of time u rg en cy in subjects.
I
i Publication of the W estern Collaborative Group Study findings
! of Type A behavior as predictive of CHD (Rosenm an etal.,1976),
I renew ed in terest in investigation of the speech characteristics
I
j w hich are in teg ral to Type A assessment. Major findings of one
| study conclude th a t valid and reliable Type A assessm ent can be
I made by a simple voice analysis of counting and ratin g , allowing the
t
! use of relatively u n train ed coders. Coders w ere trained to quantify
»
i eig h t speech stylistics w hich are accepted as key identifiers of Type
A behavior by perform ing frequency counts of plosive, clipped and
! repeated words, in terru p tio n s and talkovers; m easuring silences
1 and delays w ith a stopwatch; and m aking simple ra tin g s of volume
j and speed of voice. The m ajor fin d in g asserted th a t relatively
i
j u n train ed coders can produce valid and reliable behavior pattern
assessm ent by simple speech analysis of counting and ratin g
(S chucker & c Jacobs, 1977). Ratings of volume and speed, as defined
I
i in the study, are dependent upon individual ra te r perception of
I speech and are considered relative measures.
Speech data from the SI provide the p rim ary source of
i Type A speech research. Statistical analysis of SI data of four topic
’ areas w hich included (1) am bition and drive, (2) com petitiveness,
J
(3) em otional expression, and (4) im patience and time urg en cy
; established expressive speech characteristics as the best
discrim inators of behavior typing. Greater interaction with
i interv iew ers was associated also w ith Type A subjects. C orrelation
1 of interview ers' speech characteristics w ith respondents' post
interview feelings indicated Type A subjects felt more rushed and
i
annoyed w hen the interview er increased speech emphasis. The
i
! re se arc h ers suggest Type A em otional reactions may be im portant
i
j m ediators fo r lin k in g interview ers' behavior to Type A covert
i
■ feelings and reactions and overt behavioral responses (Scherwitz et
■ al., 1977). It is contended th at some interv iew ers m ay use the four
; speech characteristics in varying priority and degree w hich
i
indicates to the investigators th a t Type A behaviors are not
necessarily an interlocking complex. Interview ers spoke and asked
j questions more quickly of Type A th an Type B subjects, suggesting '
j th at interview ers are influenced by subjects' speech behavior !
(Scherwitz, Graham, Grandits, 6 c Billings, 1987). This response by ,
j interview ers is considered by some research ers to be a significant
I
aid in id entifying Type A subjects (Rosenman, 1978). The data for
this study are from the M ultiple Risk Factor T rial (MRFIT) studies of I
! 1979 in w h ich Type A, as assessed by MRFIT interview ers, failed to
i
' predict CHD.
; Mildly h y p ertensive subjects w ere typed more precisely on a
j scale of Al, A2, B and X, by trained voice coders in a study derived
i
i
I from e a rlie r w ork ( Schucker 6c Jacobs, 1977), w hich had quantified
! eig h t speech stylistics. Statistical correlation of audio tape codings
confirmed th at the eight speech stylistics differentiate types in
; th ree interview techniques: the Structured Interview , a non-
i provocative interview , and casual conversation. Type Al and
I
{ B rated at opposite extrem es in each of the eight speech stylistics.
’ Type X resem bled Type A in speed and delays, and Type B in
}
speech volume levels and frequency of occurrence of plosive words
(Jacobs 6 c Schucker, 1981).
As the process of quantifying Type A speech characteristics
develops, a m ajor goal is to determ ine w h eth e r objective,
automated procedures for m easuring speech variables are efficient
and effective in predicting Type A and B classifications from SI
| data. The advantage of automated procedures in objectivity and
j efficiency, as well as the potential for mass screening, is compelling.
A regression equation w hich calculated voice volume with
reaction-tim e as m easured by sem i-autom atic voice analysis
j classified Type A and B subjects at 89 p ercen t ag reem en t w ith
| Structured Interview classifications by Rosenman. The equation
i dem onstrated better agreem ent w ith Rosenman th an the study's
i
! two best raters. The study used audio tapes of the the o rig in al 66
1
Rosenman interview s in both evaluations. Significant negative
correlations w ere produced betw een Type A and B ra tin g s and
j
semi-auto mated reaction-tim e scores. S ignificant positive
' correlations w ere produced between behavior type ratin g s and
j voice volume and duration of voicing (Howland Sc Siegm an, 1982).
j
' Negative correlation w ith reaction-tim e may relate to differences
betw een perception of rate and actual m easures of rate (Feldstein Sc
i Bond, 1981). It is suggested th a t sem i-autom atic analysis predicts
j interview er ratin g s better th an paper-and-pencil tests because
i vocal behavior analysis focuses on some of the cues interview ers
j use in assessing Type A behavior, w hile paper-and-pencil tests rely
j on quite d iffe ren t data.
I Automatic Vocal Transaction Analysis (AVTA), a com puter
| scored index of frequency of occurrences of simultaneous speech
w ith 5-point ra te r judgem ent of loudness levels, accounted for 38
j p ercen t of the variance in severity of coronary a rte ry disease as
! scored by the Gensini method in an under-sixty subject group. The
!
■ Gensini scale is a weighted index of occlusion severity and location
I
i of occlusion in the coronary arteries. In the over-sixty age group,
! 16
I
| pauses in speech and accelerated speech rate related to severity of
| occlusion. The research ers, who at the time of the study did not
i have access to objective com puter scoring of loudness, commented
I
I that in order to compensate for confusion in the perception of
loudness, speech rate was always used as a covariate for volume
I ratings. Type A and B prediction of AVTA and loudness, as judged
| by raters, correlated well with traditional global assessm ents of the
j interview s in both interview styles, challenge and non-challenge,
suggesting th at interview style is not critical. There was no
relationship between global SI scores and Gensini occlusion scores
(Siegman, Feldstein, Tomasso, Ringel, & Lating, 1987) Acoustic
inform ation indicates th at loudness influences perception of speed
| of speech, not th at loudness influences com puter m easurem ent of
rate of speech (Feldstein &cBond,1981).
Methods used by interview ers and ra te rs in evaluating
subjects from speech inform ation in the SI are documented and
analyzed in the research literature. The synthesis of SI scoring
inform ation confirm s the superiority of speech characteristics and
i
J stylistics in evaluating the behavior pattern. A m ajor challenge in
| Type A research is to define more precisely and scientifically the
j speech behavioral cues w hich interview ers and raters perceive as
I re le v a n t to Type A and B p attern s and to relate the inform ation to
j physiological indicators of CHD. Voice em phasis is variously
i described as: loudness, intensity, power, volume, h ig h -en e rg y
speech, tense speech, plosive speech, abrupt speech, explosive
■ conversational intonation. These descriptions are considered in
'
Type A literatu re to be sim ilar to each other. Speech studies
| associate loudness w ith simulated anger, contem pt, dom inance,
I confidence, extraversion, boldness, en erg y level, and aggressiveness
i
! (Natale, Entin, 6 c Jaffe,1979; Sanford, 1942). Investigations of the
relationship of the psychological variables to loudness and to Type
! A and B constructs have been inconclusive
I
Research in Type A speech behavior is a process of refin in g ,
discarding, and redefining the vocal characteristics and stylistics
associated with the pattern. The direction of research in the
investigation of speech characteristics is toward automated analysis
of speech inform ation. The search for objective methods of
, assessing the behavior pattern indicates computerized acoustic
analysis of Type A and B speech sam ples as a logical area to
! continue exploration. The validity of the acoustic wave in speech, as
a measure of psychological and physiological behaviors, depends
I
| upon a complex indirect interrelationship between the psychological
and physiological actions required to produce speech sounds and
the speech signal w hich is transm itted and perceived. The phonetic
identity of elem ents of a speech message is strongly correlated w ith
i
j certain acoustic features. However, th ere is not a one-to-one
I correspondence w hich facilitates easy separation of speaker and
j message com ponents of speech signals (Peterson 6 c B arney, 1952),
I Speech Acoustics
+
j It is intuitively obvious th a t the speech signal contains
| inform ation about both the speaker and the message Speech
i
i movements, largely the result of muscle contractions th at affect the
i
| a ir column for speech, are the response to nerve impulses,
i controlled in the nervous system, and directed by the brain. The
i
, b rain initiates and controls all motor events w hich take place
! during speech (Zemlin, 1981). It is reasonable to assume dynam ic
interactions among physiological and perceptual dim ensions of
! speech, psychological and em otional states, habitual behaviors,
; health, and the acoustic speech signal. It is the complexity of the
: speech system itself and the complexity of the interactions w ithin
the speech system w hich have contributed to the difficulty in
j
! describing the sequential actions of speech production and
i perception. It is not possible to trace precisely the nerve im pulses
! ■
from the brain th ro u g h the body n o r is it possible to trace precisely
! the p attern s of influence w ithin the speech system. It is not known
! w h eth er the shape of the vocal tract im pacts the resp irato ry system
i
| and the subglottal pressure of the breath stream for speech or if the
f
! sequence begins w ith actions w ithin the respiratory system w hich
I
i serve as signals to vocal tract shaping, or if all events occur
I
j sim ultaneously.
I The p resen t investigation aims to identify acoustic speech
variables w hich are indicated as useful in predicting Type A and B
; speech behavior patterns. It is im portant, therefore, to discuss
■ briefly four speech areas critical in acoustic investigations of
i
speech. These areas are projected to be sign ifican t in acoustic
analysis based on inform ation reported in the perception of Type A
and Type B speech,
j The Flow of Speech
t
; Speech is a flow of m eaningful sound p attern s produced,
transm itted and perceived over a sequence of time. The flow of
; speech sound p atterns is basic in its description, and in trin sic to
i
understanding elem ents of speech comm unication. Speech acoustic
! technologies are designed only to a m inor degree to facilitate the
I
| study of the flow of speech. The lim ited ability to study speech in
m ovem ent is not determ ined by deficiencies in acoustic science, but
j because precise knowledge of how speech elem ents are organized,
’ physiologically and perceptually, is not cu rren tly available (Pickett,
; 1980).
j The continuously changing flow of speech sounds is the result
| of the body's regulation of the b reath as it passes from the lung s to
i
j the atm osphere. The breath is the media by w hich speech is
j
created As breath passes th ro u g h the resp irato ry system, it is
altered to form a ir flow p attern s of language. Nerve im pulses move
and control muscles of the pharynx, soft palate, tongue, jaw, and
! lips, changing the shape of the vocal tract. The balance of muscle
; tension and a ir flow th ro u g h the vocal folds into the vocal tract
produce specific phonemes, the smallest units of language w hich
signify a difference in m eaning.
Speech involves sim ultaneous action in m any parts of the
brain. Body parts w hich are im portant in speech processing have
among the h ig h est representation of associated gray m atter along
the motor and sensory strips of both cerebral hem ispheres. The
density of th e ir representation is an index of the sensitivity of the
relationship of speech production and perception m echanism s
i
; w ithin the neuro-physiological system. The process begins in the
i brain w ith n erv e im pulses to muscles w hich contract and extend to
i
| alter the shape of the vocal cavity. Changes in vocal cavity shape
! cause a ir pressure changes The air pressure changes alter the
sounds produced by the a ir flow from the lungs (Lieberm an, 1977;
Pickett, 1980; Zemlin, 1981). D uring speech m ultiple actions occur
sim ultaneously throughout speech production and perception
{ m echanism s. The sim ultaneous and interdependent n atu re of
speech makes it difficult to describe sequentially th e physiological
functioning of speech. It is, how ever, possible to m easure the
acoustic variables in the transm itted speech signal
Interrelations of the psychological, physiological, and acoustic
j dim ensions of speech production are illustrated in Figure 2, below
| (Perkins, 1971). A lthough speech is traditionally analyzed as an
’ an acoustic or physiological interaction, psychological dynamics are
a sig n ific a n t p art of the speech process. Speech comm unications are
generally conceived psychologically, produced physiologically,
transm itted acoustically, received physiologically, and perceived
i psychologically (Denes S c Pinson, 1973) Each are a depicted in the
P S Y C H O L O G I C A L D I M E N S I O N S ! P R O D U C T I V E
P H Y S I O L O G I C A L D I M E N S I O N S
i n t o q u o t ie n t
A C O U S T I C A L D I M E N S I O N S
P S Y C H O L O G I C A L D I M E N S I O N S
R E C E P T I V E
« * » O J O O IC -l F U N D A M E N T A L F R E Q U E N C E
PHONETIC AN 0
Figure 2. Psychological, physiological and acoustic dimensions of speech interact in a
dynamic process.
N >
figure produces inform ation about speech specific to itself, and is
j dynam ically interdependent w ith the others. The acoustic area is
j defined by principles of physics, specifically the properties of
j speech sound waves It is a fa irly recen t developm ent th at acoustic
j speech analysis has become m eaningful to behavioral psychologists
j and medical re se arc h ers outside the speech pathology field.
i Acoustic analysis of psychological functions of speech requires the
; separation of speech production and speech perception as indicated
! in Figure 2 by P erkin's as Psychological Dimensions: Productive"
I and "Psychological Dimensions: Receptive."
! Acoustic m easurem ents of speech sounds are specified by
four m ajor param eters: (1) fundam ental frequency, (2) form ant
frequencies, (3) power and (4) tim ing. The flow of a ir from the
j lungs provides e n erg y fo r speech sound production. As the flow of
| a ir passes from the lungs to the lips, the a ir is shaped and modified
| by physiological speech m echanism s w hich respond to nerve
I im pulses and dynam ic m ovem ents to differentiate the a ir flow into
I
| p attern s of sound recognized as speech. The modified speech wave
i can be acoustically m easured in those areas th at are perceptually
I
| relev an t for Type A Behavior.
t
i
! Fun dam ental F requen cv
1
) The speech function most directly associated w ith the larynx
!
i is the generation of quasijperiodic series of puffs of air w hich
I constitute the source of acoustic energy th at characterizes speech
j phonation, th at is, voicing. Traditionally vowels, and some
j consonants, in English are voiced (Lieberm an, 1977; Pickett, 1980;
j Zemlin, 1981).
The vocal folds are the prim ary m echanism in phonation. The
! folds vibrate w hen the laryngeal muscles move the m ajor cartilages
! of the lary n x to cause the folds to adduct (Lieberm an, 1977,
i Pickett, 1980; Zemlin, 1981) Basically the folds are composed of
| the mucous m em brane and the vocalis muscle, coupled by the
I
; in terv en in g vocal ligam ent (Hirano 6 c Kakita, 1985). Negative and
! positive a ir pressures below and above the glottis and the Bernoulli
I
! principle, ra th e r th an nerve impulses alone, set the folds into
| vibration. V ibratory cycles of the vocal folds are determ ined by
I aerodynam ic p ressu res/relatin g to the principles of physics. These
. v ibratory cycles are the basis of the validity of acoustic analysis of
| the speech wave Frequency of vocal fold vibration, th a t is, num ber
; of openings and closings p er u n it of time, determ ine the
j fundam ental frequency (F0) of the speech wave, a p rim ary acoustic
feature in speech analysis (Lieberm an, 1977). Duration of open
and closed phases of glottal vibration during phonation are not, in
general, equal (House 6c Stevens, 1958) Muscles controlling the
folds allow a wide ran g e of adjustm ents in the th ick n ess and
I tension of the folds to alter fundam ental frequency (Lieberm an,
i
1 1977; Zemlin, 1981). Vocal folds of men, women and children
i
| produce v ibrations of 80 to 500 cycles p er second (CPS), also called
24
Hertz (Hz). Fundam ental frequency in men generally is m easured
I betw een 80Hz and 240Hz.
1
I Listeners perceive the lowest freq u en cy of the complex
i
| speech wave as speaker's pitch. V ariations in fundam ental
| frequency contribute to distinguishing individual voices. The
fundam ental frequency provides inform ation about the sex, age,
height, w eight, and em otional state of the speaker (Williams &c
I Stevens, 1972; S cherer, 1981; Fowler, 1985). Intonation p attern s
i
; w hich convey m eaning in phrases and sentences are produced by
l
j changing the fundam ental freq u en cy (Pickett, 1980). Fundam ental
| frequency changes affect form ant frequencies, such th at increasing
! the F0 produces speech th at sounds as if it w ere produced by a
j shortened vocal tract Decreasing frequency has the effect of a
j
i length ened vocal tract. The perception of loudness also seems to be
i
1 related to fundam ental frequency, as a low er F0 is often perceived
i
1 as louder, even though th ere may be no m easurable attending
i increase in in ten sity (Feldstein 6c Bond, 1981). Hoarseness,
i
i breathiness, abru p t voice, evenness of speech may be affected with
| ;
j m odifications in vocal fold adduction, tension, and the v ibratory
I ,
| interval. Any or all of the fundam ental frequency acoustic j
variables m ay be determ inants of Type A behavior. !
i
i
Form ants
*
A form ant is a resonance of the vocal tract. Adjustments in
I
the shape and len g th of the supralaryngeal air passages cause them i
to respond as acoustic filters. The resonances modify the a ir flow, |
i
i
25
| am plify certain harm onics of the speech wave and attenuate others,
} form ing differentiated form ant frequencies and associated
|
( bandwidth s. T ransfer functions that correspond to particu lar
I
! supraiaryngeal vocal tract shapes are complex and difficult to
follow. In general, form ant frequencies are raised and lowered as
the len g th of the vocal trac t is decreased and increased,
!
j respectively. The vocal tra c t is lengthened by protruding the lips
j (about 1 cm maximum), or by low ering the lary n x (approxim ately 2
: cm maximum); it is shortened by com pressing the lips and raising
j the lary n x (Fant, 1970; Lieberm an, 1977). The acoustic effect at the
1 lips has the same harm onics as the acoustic effect at the source
; (vocal fold vibration). The am plitudes of the harm onics, as #
| modified by the tra n sfe r function of the vocal tract, change the
i
quality of the sound. Voice quality or tim bre is determ ined by the
j harm onics of th e sound. A speech m annerism of Type A speakers
i
i has been described as "pursing lips." The associated physiological
j and acoustic significance of this behavior has not been well studied, j
Pressing the lips together shortens the laryngeal tract and affects
i
j h ig h e r form ant frequencies w hile puckering the lips would extent
i
( the vocal tract and affect lower form ant frequencies.
Form ant Frequencies ;
Since form ants are the n atu ral resonance characteristics of the
vocal cavities, th ere are associated frequencies for all form ants.
! Figure 3 , below, shows an analysis of a speech wave by spectrum
! analysis. Form ant frequencies are center frequencies w hich perm it
I
CDB)
110
100 :
90
80
70
60
50
40
30
20
10
2000 3000
FREQUENCY (HZ)
1000 4000 5000
s
Figure 3. Spectrum analysis of the first five formants in the vowel of the word "scorch," j
i
t\>
O '
! maximum sound en erg y to pass th ro u g h the vocal tract filter
j systems. Form ants are num bered from low to h ig h frequency: FI,
i
! F2, F3, etc The first five form ants are shown on Figure 3 for the
! vowel in the word "scorch ." Form ants reflect the physical
; properties of the vocal trac t w hich produce the spectra. Form ant
I
! frequencies ra n g e from about 80 Hz to m ore th an 20,000 Hz.
; Controlled variation of form ant frequencies is perhaps the single
j most im portant factor in speech. Form ant resonances are im portant
i cues in perceptual identifications, particularly of vowels, as form ant
I frequencies are often held in steady state w hich provides longer
! pattern m atching time.
1 Phonetic vowel charts, as illustrated in Figure 4 below,
approxim ate the acoustic realities of speech waves more closely
; th an they do X-rays of tongue positions fo r vowel production
; (Peterson 6c B arney, 1952). “Vowel height" is more closely
determ ined by the first form ant freq u en cy than by the h eig h t of
i the tongue. The point of maximum constriction and the len g th of
I the trac t from the glottis to the point of constriction are more
' critical in acoustic m easurem ent of speech th an the highest point of
J the tongue (Fant, 1970). The fro n t to back dim ension in the oral
I
i cavity is more sim ply expressed by referen ce to the difference
i
| between the first and second form ant frequencies than to any
j
I m easure of the actual position of th e tongue (Ladefoged, 1982)
Form ants for vowels are affected by the length of the pharyngeal-
j oral tract, the location of constrictions in the tract, and the degree of
FREQUENCY OF SECOND FORMANT IN CYCLES PER SECOND
2500 2000 1500 1000 500 100 20
20
100
HIGH
43
109
74
BACK RONT
500
52
34
115
LOW
1000
Figure 4. The phonetic vowel chart approximates the acoustic realities of speech waves
more closely than actual positions of the tongue in vowel formations.
ro
FREQUENCY OF FIRST FORMANT
IN CYCLES PER SECOND
, 29
>
! narrow ness of the various constrictions. Speech scientists have
j studied form ant frequency influences Six of the ru les of form ant
J m ovem ent are given as follows.
i
! 1. Frequency of FI is lowered by a constriction in the fro n t
h alf of the oral cavity The greater the constriction, the more
! FI is lowered.
i
i
2. Frequency of FI is raised by constriction of the p h ary n x
The g reater the constriction, the more FI is raised.
3. F requency of F2 tends to be lowered by a back tongue
constriction. The greater the constriction, the m ore F2 is
lowered.
i
I
I 4. F requency of F2 is raised by a fro n t tongue constriction.
The g reater the constriction, the more F2 is raised.
i
5. Frequencies of all form ants are lowered by lip rounding
and low ering the larynx. The more rounding and lowering,
i the m ore form ant frequencies are lowered.
1 6. Freqencies of all form ants are raised by com pressing the
; lips and raisin g the larynx. The more the lips are compressed
j and the larynx raised, the more form ant frequencies are
! raised.
j (Pickett, 1980)
I
; Type A speech has not been described in a m an n er th a t would
I
j app ear to relate to form ant frequencies. Form ant m easures are
| critical in m any aspects of speech research; therefore, study of
I
; form ant freq u en cy m easures is indicated as a m eans for
i investigating the vocal qualities and characteristics perceived by
i
i raters as differentiating Type A and B speech. T here are form ant
j rules for perception of speech, but they are not directly relevant to
J this study.
I
| Form ant Bandwidths
| R eferring to Figure 3 above, note th at on the first form ant a
j horizontal line has been draw n 3dB down from the peak The
' m easure of the horizontal line is the bandwidth (BW1) of the first
i
i form ant, the m easure of frequency across points on the rising and
I
| falling sides of the form ant curve. The am plitude of a form ant peak
' is proportional to form ant frequency. If a form ant frequency is
I
i
1 doubled, th a t fo rm an t peak is increased by 6dB The presence of
I
j other resonance form ants, especially w hen neighboring form ants
i
i are close in frequency, may affect individual bandwidths. W ithin
■ 3dB below the peak th e influence, although p resen t in some cases,
j is minimized (Dunn, 1961). M easurem ent of form ant frequencies
J associated w ith d iffe ren t vowels is well know n and accurate, but
, the m easurem ent of bandw idths has been a later developm ent in
I
| acoustic speech analysis and it is not as w ell studied (Bogert, 1953)
1 ;
I Fletcher, in 1929, is noted as possibly the firs t au th o r to
| m ention dam ping of n atu ra l vowels (Dunn, 1961). Bandwidth
i
j m easurem ents provide inform ation about dam ping of the energy
I (rate of decay) of the speech wave. There are several sources of
I
j en erg y loss in th e vocal tract: (1) radiation of sound en erg y from
; the m outh, (2) absorption of sound by walls of the vocal tract,
(3) loss of sound en erg y at the glottal opening, (4) absorption of
j sound in the nasal cavity, and (5) loss due to viscosity at the walls
31
of the vocal tract, particularly w hen there are narrow constrictions
in the trac t (House 6c Stevens, 1958).
Bandwidths are difficult to derive from analysis of n atu ral
j speech because of irreg u larities in the glottal source spectrum
| (Klatt, 1980). Various experim ental methods have been used to
| investigate bandwidths, including a direct acoustic m ethod of
I
1 im pulse excitation to parts of the vocal tra c t and a sinusoidal
\
j sw ept-tone sound source stim ulus (Fujim ura 6c Lindqvist, 1970).
' The sinusoidal m ethod is dependent upon skillful articulation by
; the subject and control of glottal conditions. Estim ating bandwidths
i
: th ro u g h experim entation is ham pered by the h ig h dissipation of the
first form ant in most of the experim ents.
t
: It is possible to some degree to determ ine sources of vocal
| tract e n erg y loss in individual vowel form ant production by
j
j separating form ant resonances w ith band filters, tuned to a single
| resonance. Separating resonances provides for analysis w hich
delineates the relation betw een the rate of decay of the sound
i
! radiated by the vocal tra c t and the fin ely tuned bandw idth of the
I
: resonance (House 6c Stevens, 1 9 5 8 ) . The complete transm ission
j characteristic, th a t is, the ratio of volume velocity am plitude at the
: lips to th a t at the glottis, is the product of the tra n sfe r im pedance of
j individual resonances. The source spectrum cannot be separated
i from the tra n sfe r function and must be considered in deriving
i
; bandwidth (Dunn, 1961) From D unn’s original study on form ant
! 3 2 ;
bandw idths th e re have been efforts to form alize the rules th a t
| !
explain the associated areas of absorption for d ifferen t sounds. |
i
; There are five rules w hich explain bandwidth m easurem ents as
I
| they relate to particular energ y absorption areas
| 1. R elatively open vowels, w ith closed glottis and presum ably
I closed velum, lose en erg y principally through radiation from
the m outh and absorption by the walls of the vocal tract.
2 Sm aller resonance bandw idths are characteristic of closed-
glottis conditions.
! 3. Radiation loss is negligible at low frequencies and increases
! sh arp ly as F2 increases.
4, The contribution from w all losses is greatest at low
frequencies and decreases w ith increasing frequency.
5 Increasingly larg er bandwidths are m easured for h ig h e r
frequency resonances.
I (House & Stevens, 1958)
|
1 In speech synthesis experim entation with bandwidths, the prim ary
perceptual effect of bandwidth change seems to be an increase or
! decrease in the effective intensity of a form ant energy
j concentration (Klatt, 1980). The association of Type A behavior 1
, pattern w ith h ig h and sustained en erg y levels in the perception of
j Type of A speech suggests bandw idth m easures of form ants m ay be
j relevant in acoustic analysis.
i
Power
Power is the capacity of an in stru m en t to m agnify. In speech
' the various vocal m echanism s, including breath flow from the lungs.
| 33
I vibrations of the vocal folds, and the filter resonances of the vocal
| tract, all discussed above, constitute the in stru m en t w hich
j capacitates the power of the speech signal. Power in speech signals
j is re fe rre d to as am plitude, intensity, or volume. It is perceived in
■ the psychological domain as loudness. T here are five traditional
j
systems of pow er m easurem ents: dynes p er square centim eter, RMS
i (Root Mean Square) w hich is a com putation of the standard
I deviation of a wave from the m ean, decibels, sones, and phons.
i
i Loudness is commonly discussed in decibels, a logarithm ic scale
w hich relates to auditory increm ents of perception. Each decibel on
[
the scale approxim ates a step increase in perception of loudness.
An increase of lOdB in in ten sity of a stim ulus sound doubles the
sensation of loudness. The perception of loudness increases more
i
j slowly th an the actual increases in in ten sity It is in terestin g to
' com pare a list of sound p ressure levels, most m easured a few feet
from the source.
■ 0 dB threshold of h e a rin g
20 dB ru stlin g of leaves
30 dB w hisper
45 dB quiet speech
65 dB conversation (3 feet away)
75 dB sin g in g (3 feet away)
85 dB loud shouting
90 dB pneum atic drill (10 feet away)
120 dB jet plane from runw ay;
am plified rock and roll (6 feet away)
130 dB p ain fu lly loud music
1 (adapted from Denes 6c Pinson, 1973)
t Certain phones are intrinsically of greater power than others
i
j Lehiste (1976) reported low er vowels produce g reater in ten sity
j th an h ig h e r ones w ith narrow er constrictions Power levels reflect
t
j coordination in tension and m ovem ent of the speech production
i
! m echanism s. This im pacts the breath flow, the complicated tim ing
| of phonem es, and acoustic resonances, all of w hich may be
sig n ifican t in Type A behavior. The intricate but im portant
relationship of in ten sity to Type A behaviors must be explained
more fully,
1 Timing and Duration
i
! All speech is ordered in time. Durational variations in silences
I
: before and during speech responses and phonetic segm ents
j contribute im p o rtan t inform ation about the message and about the
' em otional state of the speaker, Response time to questions and
| comments, rate of speaking, and variability in speech rate are
: considered significant in Type A speech . They are a significant p art
i
of the Structured Interview ratin g scale Studies indicate th a t even
v ery small changes in segm ental duration can be detected. These
j small changes are often perceived as deviant from the rh y th m
| Duration is a critical feature in speech cognition and perception as
listeners often depend upon durational contrasts to cue phonetic
segm ent changes (Klatt, 1974).
In the past 40 y ears speech scientists have experim ented
j w ith developing a tim ing model of speech production. An early
i
associated chain model, w hich held th a t the stim ulus of one
m ovem ent is required to cue the next m ovem ent, was discredited
by Lashley, who proposed a constantly interactiv e system. In
Lashley s model the tem poral ordering device is a syntax, an
in teg ratin g schema, w hich ordered the words and ordered the
m otor actions. A m athem atical model designed by Ohman views
tem poral ordering as the resu lt of the speaker m oving from vowel
to vowel. The releasing and arrestin g of consonants is
I
! superim posed upon the vocalic stream . A com puter model
; developed by Henke supports a scan ahead m echanism fo r m otor
i
i
j control (Borden S c H arris, 1981).
! Speech tim ing models later proposed relatin g tem poral
i
j ordering to p attern s of relative tim ing of the segm ents in a phrase.
; In such models, the tim ing of stressed item s is planned first and
i
j given prim ary articulatory em phasis by the speaker. The tim ing
i
!
! and articulation of the less accented p art of the phrase receive
! secondary consideration (M artin, 1972). Stress-tim ing is the
j tendency for stress to occur at equal intervals or rhythm ically.
! L isteners seem to sense the rh y th m of speech and use it to help
i.
I predict the re st of the message. R hythm in speech is illusive,
how ever, and as described by M artin, does not seem to be an
acoustic param eter. It is suggested th at rh y th m exists in the mind
! of the speaker, but is tem porarily b lurred as it is transform ed into
! the acoustic stream of speech. The listener perceives the rh y th m
i despite speaker rate changes and o th er factors th at make the
I rh y th m difficult to specify objectively. Speech rh y th m in M artin's
I
36 ■
I I
study is evaluated by perception of speech T here does not seem to
j be a practical m ethod of m easuring Type A and B speech rh y th m s ;
!
j at th is point in speech investigations, although the theory may be !
i '
j relev an t to Type A speech behavior. P attern-m atching analysis
i
1 w hich may provide useful inform ation about Type A speech
i
\ rh y th m s is n o t proposed for this acoustic investigation.
I Durational m easures require determ ining u n it divisions in
i
i
j speech production. Isolating discrete phrase, word, vowel, phonem e
| and silence units is challenging work: p a rt science, part perceptual
| skill, and p art art. Separate, serially ordered phonological or
i
phonetic segm ents from an acoustic signal in w hich acoustic
i
j correlates of d ifferen t segm ents are interw oven can be m easured
by draw ing segm entation lines perpen d icu lar to the time axis that
serve as the rig h t edge of one segm ent and the left edge of another,
j In digital analysis systems it is possible to analyze the speech wave
fram e by fram e to determ ine segm ent b eginnings and endings,
; acoustically by wave form and perceptually by listening. Type A
J speech may be characterized by such durational variability.
| Duration m easures in speech are often given as relative measures.
Duration m easures in the p resen t study are focused on com paring
' the rate of speech in phrases, words, vowels, and the accom panying
i
! silences.
i
I
; There are m ultiple complex theories of the source of
I
! durational effects in speech. It has been proposed, for example,
th a t stressed vowels or vowels in fin al position are long in duration
in o rd er to accommodate fundam ental freq u en cy (FO) m ovem ents
th at m ark m ain stress or sentence finality. D urational effects at the
■
! phrase level m ay be explained in th is m anner. Changes created by
i
l the interaction of phonem es, shortened vowel duration as preceding
I
; or following consonants are added to a syllable, may be sig n ifican t
i
; in accounting fo r variability in duration (Fowler, 1985) Type A
tim ing behaviors possibly relate to some of these theories,
1 p articularly those w hich influence stressed speech effects. The
: natu re of the speech text for the present investigation requires
^ focus on durational measures, ra th e r than phonem ic tim ing and
response tim ing measures.
! The C hallenge in Type A Assessm ent
From the discussion above, speech perception seems to be an
i
J adaptation of generalized p attern m atching abilities in the brain.
I
| The sound flow is decoded by the listener th ro u g h speech
i
; perception m echanism s w hich are designed to cue im portant
! features in the acoustic p attern s of the speech signal. The
production and transm ission of speech are m atched to the ability to
receive speech. The vocal folds and oral cavity produce sounds
, w hich include freq u en cies from 80 Hz to w ell over 20,000 Hz. The
l auditory system is most responsive to freq u en cies of 80 to 6,000
i
; Hz. A difficulty in approaching analysis of perception of speech is
i
the fleeting and tra n sie n t n atu re of sound. As words are spoken,
they cease to exist, although they may be recalled th ro u g h verbal
repetition, or recordings. In both instances the ev en t is a new
event, a copy of the original, not the sound itself. Even during the
fleeting existence of sound, it is decidedly inaccessible, an illusive
lin k betw een speaker and listen er (Ladefoged, 1962)
Emphasis in speech, speed of speaking, speed of answ ering,
| and occasionally evenness of speech are areas commonly delineated
* as im portant in ra te r perception of Type A and B speech. They are
j considered the prim ary speech characteristics active in predicting
b ehav ior type (S chucker 6c Jacobs, 1977; Scherw itz et al.,1977;
i
■ Jacobs 6c Schucker, 1981; Howland 6c Siegm an, 1982; H ecker etal.,
\ 1981; B lum enthal et al., 1984). Vocal em phasis, in terch a n g ed w ith
plosive words and volume, is the single speech feature popularly
! accepted in the literatu re as critical in identifying Type A behavior.
!
j It is sig n ifican t th a t Type X, an even mix of p attern A and B
| behaviors, resem bles Type A in rate of speech, and is like Type B in
! plosive word freq u en cy and volum e (Jacobs 6c Schucker, 1981).
! Type X subjects are occasionally grouped w ith Type B in statistical
i
>
analysis of data. T here is no rep o rtin g of com bining Type X w ith
! Type A subjects, p erh ap s a default-acknow ledgem ent th a t lack of
| th e em phasis feature in the speech p attern categorizes speech as
"not-A."
i
I
I
39
| CHAPTER TWO: STUDY O F ACOUSTIC VARIABLES
i
D eterm ining R elevant Acoustic Speech V ariables
The descriptions of perceptual speech characteristics are not
standardized by SI raters. It is not possible to relate the v ariety of
subjective adjectives found in Type A literatu re to specific acoustic
m easures. T herefore, the investigator determ ined it is im portant to
I.
produce m easures of num erous acoustic variables in order to assure
consideration of the m ajor speech features indicated by the
| literature as im portant in predicting the behavior type SI
I
i assessments of Type A speech differentiation predict significance in
i
! greater variability in pitch, greater range in power stress areas and
i
J g reater variab ility in rate of speech in Type A speakers. These
i
| prosodic, acoustic param eters, together w ith phonetic form ant
; inform ation, are used in th is investigation to study Type A speech
behaviors.
Acoustic speech m easures selected fo r analysis in th is study
are: fundam ental frequency, form ant frequencies and bandwidths,
power, and tim ing, specifically duration of speech, duration of
silences, and derived ratios. A problem in relating acoustic analysis
of speech to Type A and B behavior p attern s is to reduce the
num ber of variables w ithout losing im portant differentiating
inform ation. The p resen t study investigates all m easurable acoustic
i
variables w hich are thought to be re le v an t to speech behaviors
I considered im portant by Structured Interview ra te rs The acoustic
| data are subm itted to statistical analysis program s designed to
i
; reveal individual acoustic speech variables and com binations of
; variables w hich best predict the behavior pattern.
! The Hypotheses
! It is assumed th a t Type A and B speech can be differentiated
■ th ro u g h acoustic speech measures. This is consistent with Friedman
and R osenm ans o rig in al conceptualization (1974). It is assumed
th a t Type A and B speech w ill be differentiated m ore distinctly in
i
stressed speech th a n in unstressed. In terview ers' reports of the
| perception of Type A and Type B speech behaviors indicate a
relationship between the perception of Type A and B speech
; p attern s and the acoustic m easures of Type A and B speech wave
; patterns
J It is assumed th a t Type A speakers w ill dem onstrate more
I
»
j variability in fundam ental freq u en cy (FO) m easures of longer
i speech units. Fundam ental freq u en cy is a m ajor source of
j inform ation in speech perception as it is related to pitch judgem ent
■ Pitch is a speech perception feature w hich is w ell understood by SI
! ra te rs and is one of the few speech perception features w hich
I
seems to relate in a reasonably direct m an n er to a single acoustic
; param eter. Type A speakers have been characterized as showing
i
m ore expression in speech. It is assumed th a t p a rt of the
I dim ension of expression is m easured in FO variability.
It is assumed th a t Type A speech w ill dem onstrate more
v ariab ility in pow er m easures and th a t Type A speech will m easure
j h ig h e r average power. It is assumed th a t the perceptions of Type A
speech as loud and Type B speech as quiet relate to the acoustic
m easure of power. Type A speech has been described as stressed
and as dem onstrating g reater em phasis and en erg y th an Type B
, speech. These characteristics m ay also relate to pow er m easures.
i
Power m easures w ill be studied across phrases, words and vowels.
I It is assumed th a t Type A speech w ill be spoken at a faster
| rate th an Type B speech. D urational m easures for Type A speech
j w ill be sh o rter th an Type B. Type A speakers are perceived during
the Structured Interview as speaking faster th an Type B speakers.
| Type A speakers are p articu larly described as accelerating the rate
1 of speech at the ends of phrases. Silences preceding and following
§
j words and p h rases w ill be m easured. Duration of phrases, words
I
; and vowels w ill be m easured.
i
1 It is assumed th at although Type A and Type B speech w ill
not be differentiated by form ant frequencies and form ant
bandw idths, it is im portant to m ake these acoustic m easures as
form ants are critical in speech communications, particularly in
differentiating speech sounds. Form ant frequencies and
|
I bandw idths of vowel steady states, th erefo re, w ill be analyzed.
i
CHAPTER THREE: RESEARCH METHODS AND PROCEDURES
i
i
i
NIH Psyche-H ypertension Study
The data selected for the investigation of acoustic variables in
, Type A speech are from a th ree y e a r study. H ypertension: The
Psyche, N orepinephrine, and Relaxation, funded by a National
, Institute of H ealth g ra n t (NIH g ra n t # 1R001HL31090-01) to
i
j V incent DeQuattro, M.D. and Gerald C . Davison, PhD., U niversity of
! Southern California. The psyche-hypertension study was conducted
i
j to investigate interrelationships in th ree areas: (1) the psychological
j profile, (2) n eu ral reactivity, and (3) response to relaxation training.
A v ariety of psychological assessm ents was adm inistered to
I
j subjects. N eural reactivity was derived from blood level m easures
! of epinephrine and norepinephrine. Psychological profile and
j n eu ral reactivity w ere investigated to determ ine th e ir role in
predicting successful response to a relaxation train in g program
designed to reduce h y p erten sio n The hypothesis of the study was
th at borderline hypertensive m en who secrete h ig h levels of
n o rep in ep h rin e w ill be more responsive to relaxation train in g as a
treatm en t m easure.
The NIH p sy che-hypertension study provides a ric h base of
inform ation for investigation of medical-psychological-speech
| constructs and relationships. Subjects w ere medically and
^psychologically screened before acceptance, including evaluation of
; 43
i
jthe MMPI, plethysm ography, echocardiogram , and systolic time
j intervals. Subjects w ere carefully m onitored m edically and
, psychologically during the pre-treatm ent, post-treatm ent, and one-
|year-follow -up phases of the investigation. All medical m easures
;were repeated at subject term ination in the study. Blood pressure
I readings w ere recorded for all medical, psychological, and
' relaxation train in g visits. E pinephrine and n o rep in ep h rin e plasma
:
| secretions w ere m easured d u rin g the an g er behavioral assessm ents
! Structured Interview s w ere conducted w ith subjects d u rin g the
i
i th ree phases of the investigation. Additional psychological data
‘ w ere recorded for n in e paper-and-pencil measures: Beck
I
Depression Inventory, Fear of Negative Evaluation Scale, Duke
! Hostility and Cynicism Questionnaire, Spielberger State-Trait
: Anxiety In ventory and Spielberger State-Trait A nger and A nger
i
Expression Scales, Buss-Durkee Hostility Inventory, H arburg
; Hostility and A nger Expression Scale, MMPI, and Jenkins Activity
! Survey (JAS) w hich provided supplem ental inform ation to
I
! Structured Interview ratings.
! Recorded speech data are available for about 82 m en from
|
i Structured Interview s and two new areas of behavioral assessm ent,
i
: A rticulated Thoughts in Simulated Situations (ATSS), and subject
; response to specially produced video tape dram atizations, designed
to provoke an g e r responses (A nger Video). ATSS and A nger Video
assessm ents w ere conceptualized by Davison and produced by
Davison and the psychology staff of the NIH study. The investigator
decided to use a m ore traditional Type A and Type B data base, but
theses data are available for fu tu re studies.
i
I
i
D escription of Subjects
! Six subjects from a subset of 25 m en who recorded th e data
i
base w ere scheduled for a pilot study. The investigation was
j intended to expand to 20 subjects, including the o rig in al six. The
| study was eventually lim ited to th re e Type A subjects and th ree
| Type B subjects, because of recording difficulties th at w ere
l
• encountered during the interview s. Subjects are Caucasian males,
; college graduates, five with graduate o r professional degrees, ages
40, 47, 47, 47, 48, and 52 All speak English as th e ir native
language P articipants w ere unm edicated borderline essential
i
; h y p erten siv es whose diastolic blood pressures ran g e from 90 to
1 105 mmHg, None has a h isto ry of alcohol or drug abuse, secondary
hypertension, diabetes, previous h e a rt attack, stroke, or clinically
! significant arhythm ias. W eight of subjects was w ithin actuarial
averages fo r age and heig ht, All w ere non-sm okers. Subjects w ere
i
I recruited th ro u g h new spaper announcem ents and physician
1 re fe rra ls. The Type A concept was n o t presented as p a rt of the
i
study to subjects during the time of th e ir participation.
I The Structured Interview
i
1 The S tructured Interview (SI), (Friedm an ficRosenman, 1959;
I ^
j Rosenm an 6 c Friedm an, 1961, Rosenm an, 1978, Chesney e ta i., I\||j0)
I 45
! is a standardized interactive interview designed to elicit and scale
jType A behaviors. The SI is a perform ance based m easure of Type
j A, as observed behaviors are its m ajor focus (Price, 1982).
: Interview ers are train ed to use long, awkw ard pauses, to falter in
speech, and to fum ble w ith the interview m aterials. Key w ords are
em phasized in questions Leading questions and in terru p tio n s to
the responses are an im portant p art of the procedure. Interview ers
; specifically ask questions designed to provoke the Type A
behaviors in subjects.
i A lthough questions are fa irly standardized, the interview er's
style is determ ined personally. If a subject becomes im patient,
! interview ers behave in a m an n er designed to provoke fu rth e r
, im patience in the subject, dragging questions, perhaps in terru p tin g
| questions to rep eat a phrase. H urried questions are delivered to a
J slow responder in an effo rt to h u rry the subject, or uncover
i
, controlled im patience, Subjects are not perm itted to ram ble.
| Interview ers m anipulate the interview and the m aterial v ery
cleverly and subtly so th a t subjects are denied control in the
exchange. The SI dem ands th a t interview ers conduct the interview
in a m an n er th a t challenges and provokes the subject. Interview ers
are trained to be firm , but not unfriendly, and to stay w ithin the
i
bounds of civilized, norm al conversation. Balance and m anagem ent
; of the interview are essential to its success (Chesney et al., 1980;
j H ecker et al., 1981). A copy of the interview and in stru ctio n s to
; the interview ers are included in Appendix A.
46
S tructured Interview s for the Acoustic Study
In the p re se n t study assessm ents of subjects w ere completed
! independently of the interview . Subjects w ere video taped and
audio taped sim ultaneously d u rin g th e interview Interview ers
w rote sh o rt com m ents and made b rie f evaluations of reactions to
subjects’ overt behaviors im m ediately follow ing th e interview . At a
j later time two raters, train ed by Chesney at Stanford Research
; Institute (SRI), viewed video tapes; rated subjects independently;
and assigned global ra tin g s of Type Al, A2, B3, B4, and X , an even
| mix of both behavior patterns. Global scores w ere determ ined by
j evaluating all inform ation from the interview , since the global
i scoring of SI is associated w ith increased incidence of coronary
h e a rt disease. Occasionally a th ird ra te r, trained by Rosenm an at
I
I SRI, evaluated subjects w hen the two p rim ary ra te rs did not agree.
i
M ilitary Commander Monologue
The p resen t study focuses exclusively on speech behavior in
i
Type A and B patterns. It was determ ined th at standardized
speech readings fo r subjects would provide more versatility in
investigating acoustic inform ation. Review of Type A speech
lite ratu re led to the selection of the M ilitary Commander Monologue
to troops used in Friedm an's Voice Analysis Test fo r Detection of
1
' B ehavior P attern (Friedm an etal.,1969), p erh ap s the earliest Type
!
’ A speech analysis study. The m onologue w hich was constructed
i
: specifically fo r the 1969-study, as it was considered suitable
! 47 :
I i
j dem ographically to the m ajority of m en in the study. It is believed
J the monologue offers Type A subjects challenging activity th a t
| inspires striv in g to excel and perhaps uncovers covert hostility.
I Simple and h o rtato ry readings w ere suggested by Friedm an as
i
offering validating com parisons of norm al and provoked states of
Type A speakers w ith in subjects and betw een subject types.
I
I Copies of the M ilitary Commander Monologue and in stru ctio n s to
j subjects and in terv iew ers are included in Appendix B.
j
! Simple and h ortatory readings of the two p arag rap h
J
i monologue constitute the speech sample for acoustic analysis in the
p re se n t study. Standardized readings w ere selected in the belief
they provide raw speech data w hich facilitate direct com parison of
; acoustic m easures betw een and w ith in subjects. Speech sam ples
w hich are standardized are considered p articu larly im portant in
i m easures w hich involve duration of speech, duration of silences and
I
j derived calculations w hich relate to ratio and rate The plan was to
i com pare simple and hortatory readings w ithin and across subjects.
i
J Direct com parisons of the same phonem es across speakers offer
! maximum opportunity to explore and discover v ariab ility and
j
; sim ilarity between speakers and circum stances Comparisons of
I
! simple and h o rtato ry readings w ithin speakers provides
I [
inform ation concerning speaker variability in quiet and emotional
I situations. The sim ple readings w ere dropped from the study
because of difficulties w ith the recordings, as described below.
i
I
Recording
Cassette tapes w ere recorded on a M arantz Model PMD 340
jrecorder using Shure Model 5453D m icrophones during the visit for
I
ithe firs t S tructured Interview . Efforts w ere made to m aintain
i
speaker distance from the m icrophone and m aintain constant
i volum e levels on the equipm ent. Only the in terv iew er was p resen t
during subjects' monologue recordings. Tim ing of the monologue
'v a ries from 45 seconds to one m inute, depending upon individual
I \
t sty le.
j Recording sessions w ere conducted by psyche-hypertension
i staff m em bers in sub-optimal acoustic conditions, a concrete
stairw ell room at the USC Medical School. Carpet and o th er acoustic
!
! treatm ents supplied to the room w ere in su fficien t to elim inate the
| in terfe ren ce of stairw ell activity and large motors in the basem ent
^ of the building. Monologue recordings w ere completed following
, th e first SI appointm ents during the first assessment, pre-
relaxation train in g , phase of the study. The audio recording
difficulties severely limited the fin al selection of subjects for the
study. Noise in te rfe re n c e and equipm ent failure obscured the
speech data in some of th e recordings.
Circum stances of the original recordings necessitated
abandoning plans to use direct power m easures, in favor of relative
m easures, w ith in subjects. Extrem ely low volume on some
| recordings eventually led to dropping the simple readings of the
■ monologue and contributed to reducing the data base to six
subjects. R esearchers generally agree it is extrem ely difficult to
I
| provide subjects a sem blance of n atu raln ess w hile recording data
| and m aintain m icrophone placem ent adequately, calibrate
i equipm ent precisely, and m onitor recording conditions sufficiently
to produce accurate absolute m easurem ents of intensity.
Investigators frequently accept experim ental conditions w hich
provide relative and relational, ra th e r th a n absolute, m easures
j (Hollien, 1981). The investigator employed a v ariety of electronic
i
filtering techniques and com puter tra n sfe r functions, designed to
i
| produce better analysis data by im proving the quality and boosting
i
j the volume of the original recordings. The filterin g devices w ere
!
; successful in increasing volume, but w ere not en tirely successful in
■ producing clean speech data.
i
! Filters
! Hum Filter
j The recorded data w ere passed th ro u g h a band pass filter
■ designed by Lars Neilson. The specifications of the filte r w ere as
t
j follows: Cl) fre q u en cy response: 20 Hz to 10kHz w ith 4 notches at
1 60, 120, 180, and 240 Hz; (2) maximum attenuation: 59.2Hz: -84dB,
' 119.1Hz: -66dB, 177.1Hz: -74dB, 238 1Hz: -78dB; (3) signal to noise
j ratio: -82 dB (re 5Vp); (4) maximum in p u t level: 5 Vp (=3-5 Vrms).
| F requency Devices Series 90IF
J Before the speech data w ere stored into p rim ary files of a
speech analysis program , the recordings w ere processed through a
50
i
i F requency Devices Series 90IF filter. Four subjects w ere recorded
at lOdB gain. Two subjects w ere recorded at 20 dB gain. The two
I subjects recorded at 20dB gain w ere Type B subjects. The volume
i
| on the o rig in al tapes was too low to produce good digital analysis
w ithout the 20dB gain.
Digital A nalysis of the Speech Wave
i
! Digital rep resen tatio n of the speech wave is dependent upon a
| basic principle developed by A lexander Graham Bell. The principle
‘ functions to control electrical c u rre n t so that the c u rre n t of
l
! electricity varies its intensity w hen sound is moving th ro u g h it, in
i
th e same m a n n e r as air v aries its density w hen sound is m oving
th ro u g h it. The prin cip le is fundam ental to all systems of recording,
1
i transm ission, or processing of speech signals in w hich the speech
i
j signal is represented by reproducing the amplitude fluctuations of
j
th e speech wave, In digital system s speech w aves are rep resen ted
by sequences of num bers w hich specify the p attern of am plitude
fluctuation. The speech wave, represen ted as a continuous time
function variable, is sampled periodically in time to produce a
sequence of m easures.
The investigator used the In teractiv e Laboratory System (ILS)
software package w hich offers com prehensive signal processing for
data acquisition, file m anagem ent, data m anipulation, and graphics.
Support features of ILS include data display and editing, digital
j filtering, spectral analysis, speech processing, and pattern
1 classification.
f
i
I
! M easuring Acoustic Speech Variables
i
[ Duration M easures
i
1 Duration m easures w ere firs t grossly determ ined from the
*
j speech wave display. Since m uch of th e Type A and B speech
| construct is based on speech perception, speech units for final
i
' analysis w ere determ ined by speech perception. P relim inary
\ m easurem ents of speech u n it duration began w ith the final
] phonem e of the syllable preceding the u n it of speech to be
m easured and ended w ith the in itial phonem e of the following
; syllable. This m ethod was employed to in su re th a t all silences and
all speech units w ere complete w ithin the utterance. To determ ine
i
| speech and silence units the investigator listened to final phonem es
i
until the fin al release of the breath flow of the phonem e. Initial
j phonem es w ere studied in the same m a n n e r to determ ine the
: beginning of breath flow for the phonem e.
D uring the process of determ ining duration measures, the
investigator studied the spectrographic analysis of the speech wave
movements, fram e by fram e, as an aid in determ ining beginning
and ending fram es An ILS spectrogram is represented below in
j Figure 5 of the word "scorch." The speech wave is displayed on the
j u pper portion of the screen as it is on the printed spectrogram . The
! vertical axis re p resen ts the am plitude of the wave. The horizontal
K 1.12E+04 MSEC
I I I I I I I I I I I I I I
l l l l l l I I I i I I l "T I I I i i i i i i i i i i i r i i [ r
i m i ' llt" ♦ W l i f A m in ■ i jf
...................I I I I
I I I 1 I I I I I I I J J I -1. J I
h I | ' 11
' 1 I I | I
" 1 1 : 111
- i ' 1
11
Mil
• • i i
<>,|
I I I M
I
I , 1 " 1 I
I I 1 I I ' ,
l |
I | , l | | I III I
| | | I I | M I l I I I I M 1 I I
I I
1 | I ' I 1,11 i I 1 L
I 1 ' I |
I I
■ l
II II | | I | |
I
l I ' I I
I I-
II 1 1 * I
i I I
I I I > * I i I | I I
I I
I > * I 1 1 | ■ I | | I I
l .
I . ■ • ' ' 1 . , i 1 1 1 1 ' 1 i I ,
' 1 ' 1 ' ‘ • • 1 .............. | i 1 1 . I | | | ' I
E:\BEUERLV\UDi800. FLD1= ID= 0, SF= 10000
N= 200j M=12j PR= 98j HftM=V, SDFR= 5548, STFR= 1758, NFR= 163, CTX= 64
Figure Spectrographic analysis of the word "scorch'1 displays the speech wave;
power and fundamental frequency ; and analysis of form ants and bandwidths.
I 53
i
j axis is time in milliseconds. Power and fundam ental freq u en cy are
' tracked below the speech wave in the middle section coordinated in
l
j time w ith the wave above. Analysis of fo rm an t freq u en cies and
j bandw idths are displayed in the low er portion of the analysis
i
J display. It is possible to study details of the spectrographic analysis
' fram e by fram e w hile playing back the digitized speech segm ent
1 represen ted by the spectrogram .
j Precise fin al releases and precise beginnings of the breath
: flow w ere determ ined by listening to the critical phonem es in each
I direction in increm ents of one context point. Speech sam ples were
; digitized at 10,000 sam ples p e r second. The context of 64 sample
j
: points for display and analysis increm ents of the speech sample
■ i was determ ined before the digitization process. In a context of 64
I
■ the speech wave is m arked to display, analyze and playback the
: wave in in crem en ts of 64 sam ple points (64 x 1/10,000 =6.4 ms of
I speech). In the p re se n t study the analysis fram es w ere set to 128
| points p e r fram e w hich in a context of 64 sample points provided a
| 64 sample point overlap for each analysis fram e. The investigator
t
; changed the context for playback and display of critical fram es
!
| from 64 sample points to one sam ple point. Reducing th e context
| allowed the investigator to display and listen to the changes in the
1 speech wave in sm aller increm ents so th a t a more precise point of
i
j perceptual change could be identified. As a check of accuracy the
investigator listened to the critical sets of fram es from a point in
clear silence backward to the v ery beginnings of the fin al breath
I 54
i
j flow of the preceding phonem e. A sim ilar backward technique was
used to determ ine the precise b eg in n in g s of phonem es. This
procedure was repeated until a sample point was selected as the
beginning o r the ending of a phonem e. Both speech and silence
j
j durational m easures w ere determ ined using this analysis method
1 The technique allowed the investigator to perceive
I m ovem ents in the speech breath flow th at are n o t perceptible in a
! 64 sample point in crem en t of listen in g Listening in the context of
I
i one' has the effect of separating' the speech sample, so th at the
I m ovem ent of the breath flow can be exam ined in more detail. It is
' sim ilar in perceptual effect to w atching film o r video tape in slow
motion. Sample points, ra th e r th an fram es or m illiseconds are the
i
j units for duration m easures in the p re se n t study. Speech units for
j the six phrases, 11 words and 14 vowels fo r all six subjects w ere
■ ! separated from the la rg e r speech sam ples using this fram e by
i
J fram e and point by point speech perception method.
! Power
j RMS (Root Mean Square) m easures of the wave form for each
j analysis fram e (context 64 sample points, 128 points p er fram e) j
! w ere produced by the ILS program . Speech u n it power variability
was determ ined by subtracting the h ig h e st and lowest RMS
readings w ith in the speech unit. It is recognized th a t RMS is not
| traditionally the most common m easure of speech power, but it is
I
: the only d irect m easure available in the ILS program . It is possible ,
to co n v ert RMS to an y of the m ore common m easures
55
Fundam ental F requency ;
The sampled data w ere analyzed using L inear Prediction ■
A utocorrelation (LPC) to determ ine fundam ental freq u en cy (FO),
! m easured in Hertz. The FO averages for word and vowel segm ents
. . I
! w ere th en calculated by the investigator. V ariability was i
! j
j determ ined by subtracting the h ig h e st and lowest readings w ithin
1 th e speech u n it G unnar Fant recom mended to the investigator in
; 1983 th a t lo n g er speech units m ay provide m ore useful acoustic
inform ation about the relationship of speech behavior patterns to
! o th er constructs of behavior.
!
1 Form ant M easures
i
Spectrographic analysis perform ed a Fast F ourier Transform
(FFT) to compute a freq u en cy spectrum (form ant peaks) fo r each
J
j analysis fram e of the vowel. Four form ant peaks w ere specified in
*
' the analysis. F orm ant freq u en cy and bandw idth m easures were
j calculated for the first th ree peaks.
I
| Speech Units and M easures
i P hrase M easures
! Sis phrases w ere selected from the M ilitary Commander
j Monologue The criterion for selecting the six phrases was to
I
j provide phonetic sets of stressed, unstressed and n e u tra l words and
) '
j vowels. T hree judges evaluated the p h rases of the monologue in
j o rder to determ ine w h ich phrases, read by norm al speakers, would
f produce stressed, unstressed, and n e u tra l sets Four p h rases n ea r
j the b eg in n in g w ere selected as capable of producing good examples ,
of n atu ral speech em phasis. One phrase from the middle of the
m onologue and the fin al sentence complete the phrase set. The six
p h rases w ere analyzed p rim arily fo r duration of speech and
silences and the ratios of the two m easures to each other. Power
v ariab ility was computed across the phrase,
j Word M easures
i
j From the set of six p h rases 11 rep resen tativ e w ords w ere
i
i isolated w hich fit the criterio n of stressed, unstressed and n eu tral
' sets. Four stressed vowels, fo u r unstressed vowels and th re e
j n e u tra l vowels w ere selected. Words w ere analyzed fo r duration of
I
, speech and duration of silences before and a fte r the occurrence of
I the word in the phrase. Power variability was calculated w ith in the
I
word. Fundam ental fre q u en cy m easures w ere derived across the
duration of the word. Fundam ental freq u en cy variab ility and
j averages w ere computed fo r each word.
' Vowel M easures
!
j The vowels of the eleven w ords form the vowel analysis data.
I
| T hree of the 11 vowels are diphthongs, each w ith two m easurable
portions, m aking a total of 14 vowels for the study. Vowels w ere
i
| m easured for duration. Fundam ental freq u en cy variability and
averages w ere computed. Vowel steady states w ere analyzed for
form ant freq u en cy and bandw idth m easures As d iphthongs are
j often in a state of transition th ro u g h o u t th e ir production, th ey w ere
t analyzed for the ’ ’steadiest state" form ant m easures of each portion.
; 5 7
I The 20 Variable M easures
I
j 'P hrase M easures
j 1. speech duration (PSPD)
1 2. silence duration preceding phrase and w ith in phrase (PSID)
3. ratio of speech to silence CPSPR)
4. ratio of silence to speech (PSIR)
i 5 pow er v ariab ility (PPWV)
|Word M easures
| 6 speech duration (WDUR)
7 silence duration preceding word (WSIL)
8 pow er v ariab ility (WPWV)
9. fundam ental freq u en cy v ariab ility (WFOV)
10. fundam ental fre q u en cy average (WFOA)
1 Vowel M easures
, 11. voicing duration (VDUR)
i 12. pow er v ariab ility (VPWV)
j 13. fundam ental v ariab ility (VF0V)
; 14. fundam ental average (VFOA)
15 fo rm an t 1 (FI)
16. fo rm an t 2 (F2)
1 17. fo rm an t 3 (F3)
j 18. bandw idth 1 (Bwl)
I 19. bandw idth 2 (Bw2)
I 20. bandw idth 3 (Bw3)
I
58
CHAPTER FOUR: RESULTS
Factor Analysis
! A p rim ary objective in science is to sum m arize data so th at
' em pirical relationships can be grasped by the hum an m ind. Factor
analysis provides an im portant tool in defining critical variables by
sum m arizing in terrelatio n sh ip s am ong them (Gorsuch, 1983) Type
A behavior is a constellation of physical, em otional and intellectual
' behaviors. The speech process is a complex in terrelatio n sh ip of
m ultiple physiological, psychological intellectual and sociological
behaviors. Factor analysis provides an appropriate research
. m ethod to describe the in terrelatio n sh ip s of acoustic variables
j
i am ong them selves.
I Factor analysis reduces a la rg e r num ber of acoustic speech
J - r
j variables to a few u sefu l com prehensible, lin ear com binations of
j variables called "factors." The complexities of speech production
and the w ealth of inform ation in the speech wave dictate an
acoustic analysis program w h ich provides access to num erous
variables. T here are referen ces in Type A literatu re w hich describe
!
| ra te r perception of speech characteristics and stylistics as they
r
! relate to Type A b ehavior (H ecker et al., 1981; C hesney eta l., 1981).
I
i It is not possible, how ever, to determ ine from such subjective
l descriptions w hich acoustic v ariables relate to the ra te r descriptions
! of features perceived in speech behaviors. Therefore, it is
f
f
i in h e re n tly necessary to consider a m ultitude of acoustic m easures.
I In the p re se n t study of Type A speech patterns, a total of six
phrases, 11 words, 14 vowels, m easured for 20 speech variables
from six speakers form the data base w h ich is a total of 1,536
acoustic m easures of duration, power, fundam ental frequency,
form ant frequencies and form ant bandwidths. All the correlations
: am ong all the variables w ere calculated and subm itted to factor
i
analysis.
j Factor analysis reveals substantial relationships am ong some
i variables and m inim al relationships am ong others. Both results
; provide valuable inform ation. Each factor represents an are a of
■ generalization w ithin w hich data can be summarized th a t is
i qualitatively distinct from th at represented by o th e r factors. Factor
; analytic methods reveal quantitative distinctions w ithin variable
: groupings as well, and they recognize th at an y relationship is
j
; limited to a p articu lar are a of applicability. The quantitative
I differences are reflected in factor loadings w hich are calculated for
; each variable w ithin a factor A large factor s loading fo r a variable
indicates a strong relationship betw een th a t variable and the factor.
t
j Variables w hich qualitatively differ from o ther variables and from
produced factors are processed statistically as singleton variable
i
i factors (K erlinger, 1979; Gorsuch.,1983).
j Factor analysis proves particularly useful in exploratory
j
( research and searches h ig h volum e data for possible qualitative and
60
: quantitative distinctions (Gorsuch, 1983). T hrough factor analysis
I
I the maximum am ount of inform ation from the original acoustic
I
j variables is included in derived variables, or factors. The original
set of 20 variables is reduced to a sm aller set of five variables and
variable com binations th at account for most of the reliable variance
of the in itial m ultiple variable set. Factor analysis was essential in
the p resen t study in order to establish the appropriate balance
; betw een variables and observations. The recom m ended
I
; relationship is five observations to each predictor variable. The
i
exploratory n atu re of the investigation required analyzing the data
1 for 20 d iffe ren t predictor variables. The num ber of speakers
m ultiplied by 14 vowels produced 84 observations. It was
i
1 essential to reduce the 20 predictor variables to 14 or less to satisfy
I
! the recom m ended ratio.
I
, Delineating critical acoustic variables benefits fu rth e r
(
research by offering the re search er the possibility of reducing the
i
, num ber of variables considered fo r acoustic analysis. This results
I
; in increased efficiency, economy and flexibility in collecting and
j analyzing data, w hile maximizing the am ount of relev an t
inform ation provided for fu rth e r studies in acoustic analysis of
Type A speech. The eventual aim of such re se arc h is to build
; constructs th a t are conceptually clearer th an a p rio ri ideas, and to
> develop theories from derived constructs.
61
Statistical Analysis of the Data
For best results in discrim inate analysis procedures it is
| essential to reduce the num ber of predictor variables in the p resen t
study from 20 to 14 or less in o rder to accommodate the
recom m ended ratio of five observations fo r each predictor variable,
i A correlation m atrix, 20 v ariables by 20 variables, sorted by
: speaker type, was calculated fo r all of the variables in the study
and appears in the follow ing five pages as Table 1 .
A lthough th e correlation of the 20 acoustic variables in the
p re se n t study of Type A and B speech is n o t a p rim ary goal of the
t
: investigation, correlation of the variables is an im portant step in
factor analysis. The correlation m atrix is used to derive factor
, loadings w hich are th en rotated to sim ple structure. Factor analysis
i
i also produces standardized scoring coefficients w hich are used to
calculate factor scores that can be used as in p u t fo r a discrim inate
. analysis. The m atrix is presented w ith speaker type fo r each of
1 th e v ariables so th a t it is possible to m ake relativ ely simple
I
1 com parisons of correlations of variables to Type A and B. The
i
j correlation m atrix provides inform ation about variables w hich is
! useful in determ ining singleton variables for inclusion in the
d iscrim in an t analysis. Words, p h rases and vowels are coded in the
data base so th at it is possible to ru n correlations w h ich are focused
i
| on the stressed and unstressed acoustic m easures.
i
I
i
I
j
I
PEARSO N CORRELATION COEFFICIENTS / P R O B >' |R| U NDER H0:RHO=0 / N = 78
STYP PD U R PSIL PSIR PSPR P P W V WSIL W S P W FO V W F O A
STYP
SPEAKER TYPE
1.00000
0.0000
-0.01495
0.8967
-0.03960
0.7307
-0.12181
0.2881
0.09670
0.3997
0.12163
0.2888
-0.13337
0.2444
-0.05794
0.6143
0.27883
0.0134
0.15544
0.1742
PD U R
PHRASE DURATION
-0.01495
0.8967
1.00000
0.0000
0.95503
0.0001
0.74733
0.0001
-0.79159
0.0001
0.02608
0.8207
0.35875
0.0013
-0.16163
0.1574
-0.14606
0.2020
-0.17610
0.1230
PSIL
PHRASE SILENCE DURATION
-0.03960
0.7307
0.95503
0.0001
1.00000
0.0000
0.87482
0.0001
-0.89701
0.0001
-0.00087
0.9940
0.36497
0.0010
-0.25282
0.0255
-0.09972
0.3850
-0.17748
0.1201
PSIR
PHRASE SILENCE TO TOTAL DURATION RATIO
-0.12181
0.2881
0.74733
0.0001
0.87482
0.0001
1.00000
0.0000
-0.96989
0.0001
-0.03619
0.7531
0.37299
0.0008
-0.26632
0.0184
-0.08977
0.4345
-0.21640
0.0570
PSPR
PHRASE SPEECH TO TOTAL DURATION RATIO
0.09670
0.3997
-0.79159
0.0001
-0.89701
0.0001
-0.96989
0.0001
1.00000
0.0000
0.01902
0.8687
-0.38668
0.0005
0.26658
0.0183
0.08363
0.4666
0.21540
0.0582
P P W V
PHRASE P O W E R VARIABILITY
0.12163
0.2888
0.02608
0.8207
-0.00087
0.9940
-0.03619
0.7531
0.01902
0.8687
1.00000
0.0000
0.12681
0.2686
0.14899
0.1930
-0.11686
0.3082
0.19413
0.0886
USIL
W O R D SILENCE DURATION
-0.13337
0.2444
0.35875
0.0013
0.36497
0.0010
0.37299
0.0008
-0.38668
0.0005
0.12681
0.2686
1.00000
0.0000
0.01087
0.9248
0.00367
0.9746
0.11552
0.3138
W S P
W O R D SPEECH DURATION
-0.05794
0.6143
-0.16163
0.1574
-0.25282
0.0255
-0.26632
0.0184
0.26658
0.0183
0.14899
0.1930
0.01087
0.9248
1.00000
0.0000
0.12053
0.2932
0.18913
0.0972
W FO V 0.27883
0.0134
-0.14606
0.2020
-0.09972
0.3850
-0.08977
0.4345
0.08363
0.4666
-0.11686
0.3082
0.00367
0.9746
0.12053
0.2932
1.00000
0.0000
0.36066
0.0012
W F O A 0.15544
0.1742
-0.17610
0.1230
-0.17748
0.1201
-0.21640
0.0570
0.21540
0.0582
0.19413
0.0886
0.11552
0.3138
0.18913
0.0972
0.36066
0.0012
1.00000
0.0000
W P W V
W O R D P O W E R VARIABILITY
0.14602
0.2021
-0.05477
0.6339
-0.05415
0.6377
-0.07236
0.5290
0.04747
0.6798
0.76034
0.0001
0.31394
0.0051
0.25544
0.0240
0.03961
0.7306
0.30029
0.0076
VDUR
V O W E L DURATION
-0.05777
0.6154
0.01744
0.8795
-0.04251
0.7117
-0.04333
0.7064
0.05132
0.6554
0.01394
0.9036
0.00317
0.9780
0.42176
0.0001
-0.01151
0.9203
0.07264
0.5273
Table 1
Correlation m atrix of 20 acoustic v ariab les
PSIL
PHRASE SILENCE DURATION
-0.05415
0.6377
-0.04251
0.7117
-0.14416
0.2080
-0.16965
0.1376
-0.03126
0.7859
-0.13344
0.2441
-0.06570
0.5677
0.12125
0.2903
-0.18385
0.1071
-0.03812
0.7404
PSIR
PHRASE SILENCE TO TOTAL DURATION RATIO
-0.07236
0.5290
-0.04333
0.7064
-0.11734
0.3062
-0.25391
0.0249
-0.03147
0.7845
-0.13355
0.2438
-0.08807
0.4433
0.07686
0.5036
-0.20247
0.0754
-0.04302
0.7084
PSPR
PHRASE SPEECH TO TOTAL DURATION RATIO
0.04747
0.6798
0.05132
0.6554
0.17337
0.1290
0.26602
0.0186
0.01212
0.9161
0.12680
0.2686
0.11754
0.3054
-0.09881
0.3894
0.18018
0.1144
-0.03285
0.7753
PP W V
PHRASE P O W E R VARIABILITY
0.76034
0.0001
0.01394
0.9036
-0.22371
0.0490
0.12362
0.2809
0.57519
0.0001
0.48514
0.0001
0.40533
0.0002
0.40198
0.0003
0.09281
0.4190
-0.24172
0.0330
WSIL
W O R D SILENCE DURATION
0.31394
0.0051
0.00317
0.9780
-0.07747
0.5002
0.06172
0.5914
0.33224
0.0030
0.16245
0.1553
0.11316
0.3239
0.31547
0.0049
-0.36628
0.0010
0.15151
0.1854
W S P
W O R D SPEECH DURATION
0.25544
0.0240
0.42176
0.0001
-0.09162
0.4250
0.05346
0.6420
0.21193
0.0625
0.14570
0.2031
0.09356
0.4152
0.11740
0.3060
-0.00618
0.9571
0.01848
0.8724
W FO V 0.03961
0.7306
-0.01151
0.9203
0.49532
0.0001
0.18323
0.1083
0.00903
0.9375
0.08343
0.4677
0.17982
0.1152
0.09290
0.4185
0.24511
0.0305
0.25882
0.0221
W FO A 0.30029
0.0076
0.07264
0.5273
0.00159
0.9890
0.68964
0.0001
0.30787
0.0061
0.31057
0.0057
0.33204
0.0030
0.33912
0.0024
0.37278
0.0008
-0.07548
0.5113
W P W V
W O R D P O W E R VARIABILITY
1.00000
0.0000
0.13048
0.2548
-0.22248
0.0503
0.22202
0.0507
0.81617
0.0001
0.63494
0.0001
0.48435
0.0001
0.53454
0.0001
0.06072
0.5975
-0.10322
0.3685
VDUR
VO W E L DURATION
0.13048
0.2548
1.00000
0.0000
0.13493
0.2389
0.07042
0.5401
0.29764
0.0081
0.08996
0.4335
-0.01042
0.9279
-0.00874
0.9395
-0.11062
0.3350
0.16535
0.1480
VFOV -0.22248
0.0503
0.13493
0.2389
1.00000
0.0000
0.07871
0.4933
-0.14549
0.2038
-0.15707
0.1697
-0.09892
0.3889
-0.31121
0.0055
-0.07007
0.5421
0.20889
0.0664
W P W V VD U R VFOV VFOA V P W V B W 1 B W 2 B W 3 F1 F2
VFOA 0.22202
0.0507
0.07042
0.5401
0.07871
0.4933
1.00000
0.0000
0.31602
0.0048
0.22322
0.0495
0.34130
0.0022
0.35148
0.0016
0.32455
0.0037
-0.06557
0.5684
Table 1
C orrelation m atrix of 20 acoustic variables
as
Os)
VFOV 0.06120 -0.17777 -0.14416 -0.11734 0.17337 -0.22371 -0.07747 -0.09162 0.49532 0.00159
0.5945 0.1194 0.2080 0.3062 0.1290 0.0490 0.5002 0.4250 0.0001 0.9890
VFOA 0.13952 -0.15724 -0.16965 -0.25391 0.26602 0.12362 0.06172 0.05346 0.18323 0.68964
0.2231 0.1692 0.1376 0.0249 0.0186 0.2809 0.5914 0.6420 0.1083 0.0001
VPW V 0.06936 -0.01525 -0.03126 -0.03147 0.01212 0.57519 0.33224 0.21193 0.00903 0.30787
V O W E L P O W E R VARIABILITY
VARIABILITY
0.5463 0.8946 0.7859 0.7845 0.9161 0.0001 0.0030 0.0625 0.9375 0.0061
B W 1 0.08749 -0.14320 -0.13344 -0.13355 0.12680 0.48514 0.16245 0.14570 0.08343 0.31057
BANDWIDTH 1 0.4462 0.2110 0.2441 0.2438 0.2686 0.0001 0.1553 0.2031 0.4677 0.0057
B W 2 0.21374 -0.05598 -0.06570 -0.08807 0.11754 0.40533 0.11316 0.09356 0.17982 0.33204
BANDWIDTH 2 0.0602 0.6264 0.5677 0.4433 0.3054 0.0002 0.3239 0.4152 0.1152 0.0030
STYP PD U R PSIL PSIR PSPR P P W V WSIL W S P W F O V W FO A
B W 3 0.14209 0.14503 0.12125 0.07686 -0.09881 0.40198 0.31547 0.11740 0.09290 0.33912
BANDWIDTH 3 0.2146 0.2052 0.2903 0.5036 0.3894 0.0003 0.0049 0.3060 0.4185 0.0024
F1 0.25458 -0.23532 -0.18385 -0.20247 0.18018 0.09281 -0.36628 -0.00618 0.24511 0.37278
FO RM A N T 1 0.0245 0.0381 0.1071 0.0754 0.1144 0.4190 0.0010 0.9571 0.0305 0.0008
F2 0.04984 -0.06027 -0.03812 -0.04302 -0.03285 -0.24172 0.15151 0.01848 0.25882 -0.07548
FO RM A N T 2 0.6648 0.6001 0.7404 0.7084 0.7753 0.0330 0.1854 0.8724 0.0221 0.5113
F3 0.15681 -0.22464 -0.22042 -0.19163 0.14479 -0.02190 0.01971 0.04707 0.18457 0.07142
FO RM A N T 3 0.1704 0.0480 0.0525 0.0928 0.2060 0.8491 0.8640 0.6823 0.1057 0.5343
W P W V V D U R VFOV VFOA V P W V B W 1 B W 2 B W 3 F 1 F2
STYP 0.14602 -0.05777 0.06120 0.13952 0.06936 0.08749 0.21374 0.14209 0.25458 0.04984
SPEAKER TYPE 0.2021 0.6154 0.5945 0.2231 0.5463 0.4462 0.0602 0.2146 0.0245 0.6648
PDUR -0.05477 0.01744 -0.17777 -0.15724 -0.01525 -0.14320 -0.05598 0.14503 -0.23532 -0.06027
PHRASE DURATION 0.6339 0.8795 0.1194 0.1692 0.8946 0.2110 0.6264 0.2052 0.0381 0.6001
Table 1
C orrelation m atrix of 20 acoustic v ariab les
VP W V 0.81617 0.29764 -0.14549 0.31602 1.00000 0.51260 0.36065 0.48987 -0.04403 -0.02666
VO W E L P O W E R VARIABILITY 0.0001 0.0081 0.2038 0.0048 0.0000 0.0001 0.0012 0.0001 0.7019 0.8168
B W 1 0.63494 0.08996 -0.15707 0.22322 0.51260 1.00000 0.54267 0.43678 0.26822 0.10773
BANDWIDTH 1 0.0001 0.4335 0.1697 0.0495 0.0001 0.0000 0.0001 0.0001 0.0176 0.3478
B W 2 0.48435 -0.01042 -0.09892 0.34130 0.36065 0.54267 1.00000 0.74914 0.22464 -0.33784
BANDWIDTH 2 0.0001 0.9279 0.3889 0.0022 0.0012 0.0001 0.0000 0.0001 0.0480 0.0025
B W 3 0.53454 -0.00874 -0.31121 0.35148 0.48987 0.43678 0.74914 1.00000 0.10585 -0.19595
BANDWIDTH 3 0.0001 0.9395 0.0055 0.0016 0.0001 0.0001 0.0001 0.0000 0.3563 0.0855
F1 0.06072 -0.11062 -0.07007 0.32455 -0.04403 0.26822 0.22464 0.10585 1.00000 -0.03277
FO R M A N T 1 0.5975 0.3350 0.5421 0.0037 0.7019 0.0176 0.0480 0.3563 0.0000 0.7758
F2 -0.10322 0.16535 0.20889 -0.06557 -0.02666 0.10773 -0.33784 -0.19595 -0.03277 1.00000
FO RM A N T 2 0.3685 0.1480 0.0664 0.5684 0.8168 0.3478 0.0025 0.0855 0.7758 0.0000
F3 -0.02258 -0.01413 0.23232 -0.10339 -0.06864 0.13039 -0.28156 -0.42425 0.06625 0.53785
FO RM A N T 3 0.8444 0.9023 0.0407 0.3677 0.5504 0.2552 0.0125 0.0001 0.5644 0.0001
F3
STYP 0.15681
SPEAKER TYPE 0.1704
PDUR -0.22464
PHRASE DURATION 0.0480
PSIL -0.22042
PHRASE SILENCE DURATION 0.0525
PSIR -0.19163
PHRASE SILENCE TO TOTAL DURATION RATIO 0.0928
PSPR 0.14479
PHRASE SPEECH TO TOTAL DURATION RATIO 0.2060
PPW V -0.02190
PHRASE P O W E R VARIABILITY 0.8491
WSIL 0.01971
Table 1 W 0R D S1LENCE DURATION 0.8640
C orrelation m atrix of 20 acoustic variables
W S P 0.04707
W O R D SPEECH DURATION 0.6823
W FO V 0.18457
0.1057
F3
W FO A 0.07142
0.5343
UPUV -0.02258
W O R D P O W E R VARIABILITY 0.8444
VDUR -0.01413
VO W E L DURATION 0.9023
VFOV 0.23232
0.0407
VFOA -0.10339
0.3677
VP W V -0.06864
VO W E L P O W E R VARIABILITY 0.5504
B W 1 0.13039
BANDWIDTH 1 0.2552
B W 2 -0.28156
BANDW IDTH 2 0.0125
B W 3 -0.42425
BANDWIDTH 3 0.0001
F1 0.06625
FO R M A N T 1 0.5644
F2 0.53785
FO R M A N T 2 0.0001
F3 1.00000
FO R M A N T 3 0.0000
Table 1
C orrelation m atrix of 20 acoustic v ariab les i
O N t
;
A scree plot of 16 Eigenvalues resu ltin g from the factor analysis is
prin ted below as F igure 6. It clearly indicates th a t th e re are two
factors, shown separately h ig h on the v ertical axis,
j Table 2, show n below, p resen ts the rotated factor p attern
'm atrix for the acoustic speech variables. Factor 1 is composed
i p rim arily of (1) phrase speech duration, (2) phrase silence duration,
j (3) phrase silence to speech ratio, (4) phrase speech to silence ratio
,and (5) silence duration preceding word. V ariables in Factor 1
relate to each o th e r as m easures of duration.
t
I Factor 2 is composed of (1) p h rase pow er variability, (2) word
1 pow er variability, (3) vowel pow er variability, (4) bandw idth 1, (5)
bandw idth 2, and (6) bandw idth 3 Factor 2 variables relate as
j pow er values.
! Once the two factors w ere derived from factor analysis, the
i
next step was to construct factor scores using the m ethod suggested
[ by Gorsuch (1983). The w eights from the factor analysis w ere
I
1 incorporated in a statistical ru n to derive the new factor scores for
i
j Factor 1 and Factor 2 These two factors w ere com prised of 11 of
I
| th e o rig in al 20 variables. Next, the investigator exam ined the
! correlations of the o th er nin e predictors w ith speaker type and
1 determ ined th a t FI, F3 and WFOV had the h ig h e st zero order
!
correlations w ith speaker type, yielding th ree additional singleton
! predictor variables. These procedures resulted in five predictor
I
| variables: two factors and th ree singleton predictor variables.
5 +
0 +
+ -----------
0
FACTOR
8 9 0 1
2 3
4 5 6
- - + ...............................+
14 16 10 12
Figure 6. Scree plot of Eigenvalues resulting from factor analysis.
ROTATED FACTOR PA T T ER N
VARIABLES FACTOR1 FACTOR2
PD U R 0.85464 -0.17960
PSIL 0.95386 -0.20366
PSIR 0.89423 -0.21990
PSPR -0.91996 0.21554
PP W V 0.14107 0.68399
WSIL 0.48849 0.19101
W S P -0.16698 0.23962
W P W V 0.15717 0.90928
VDUR -0.01171 0.12224
VPW V 0.21969 0.77289
F1 -0.20102 0.22043
B W 1 0.02101 0.70518
F2 -0.08032 -0.18751
B W 2 0.10196 0.70902
F3 -0.25687 -0.10669
B W 3 0.28591 0.68681
Table 2
Rotated p attern m atrix of acoustic v ariab les
PHRASE DURATION
PHRASE SILENCE DURATION
PHRASE SILENCE TO TOTAL DURATION RATIO
PHRASE SPEECH TO TOTAL DURATION RATIO
PHRASE P O W E R VARIABILITY
W O R D SILENCE DURATION
W O R D SPEECH DURATION
W O R D P O W E R VARIABILITY
V O W E L DURATION
V O W E L P O W E R VARIABILITY
FO R M A N T 1
BANDW IDTH 1
FO R M A N T 2
BANDW IDTH 2
FO R M A N T 3
BANDW IDTH 3
J
h
1 70
i The five pred icto r variables w ere entered as predictors fo r a
i
| discrim inant analysis w ith speaker type as the criterion.
i
I
I
i
| D iscrim inant A nalysis
j D iscrim inant analysis is a h ig h ly useful form of m ultivariate
i
j analysis w h ich predicts group m em bership. It is a special
! adaptation of m ultiple regression in w hich the criterion is a
! categorical variable. D iscrim inant analysis revealed th at the two
factor and th ree singleton predictor variables of th e p resen t
; investigation accounted for about 18 p ercen t of the varian ce in
!
Type A and B subjects. The investigation was m ore accurate in
! predicting Type B subjects th an in predicting Type A. Figure 7
| show n below is a histogram of the distribution of predicted Type A
i or B categories. Type A speakers are coded ”1" and Type B
speakers are coded "0." Table 3, presented below, re p resen ts the
I
! distribution of correct and in co rrect type predictions using the
acoustic speech variables of the p re se n t study. The two factors and
, th ree singleton variables w ere accurate in predicting Type B 76
t
j p e rc e n t of the time. Type A speakers w ere predicted accurately 62
J p e rc e n t of the time.
J Five P redictor V ariables
| Inform ation con cern in g the acoustic p attern s of Type A and B
\
speech can be derived from studying the discrim inant function
coefficients. Table 4 presents the standardized canonical
discrim inant function coefficients of the five predictor variables.
I
8 +
1
1
1
1
1 0010 1 111
1 0010 1 111
0 0000 11 0 111 1
0 0000 11 0 111 1
0 0000 011 01 110 0
0 0000 011 01 110 0
1000000 000001001000 1
1000000 000001001000 1
1
1
111
111
1 1
1 1
1110 1
1110 1
11
11
X................... +...................+.................. +................. +.................... +..................+.................. +.................. X
OUT -3 .0 -2 .0 -1 .0 .0 1.0 2.0 3.0 O U T
CLASS 000000000000000000000000000000000000000001111111111111111111111111111111111111111
CENTROIDS
CLASSIFICATION 0 = TYPE B TYPE A
Figure 7. Histogram of the distribution of predicted Type A or B categories.
72
Table 3
i
! Predicted groups for Type A and Type B in discrim in an t analysis
ACTUAL GROUP NO. O F CASES PREDICTED GROUP MEMBERSHIP
___________ 0_____________ L
;Group 0
Type B
i Group
j Type A
1
38
42
29
76.3%
9
237%
16 26
38.1% 61.9%
. PERCENT OF •GROUPED’ ' CASES CORRECTLY CLASSIFIED: 68.75%
* Table 4
r
i
Standardized canonical coefficients of the five predictor variables
i
; F actorl -0 00448 (PSPD, PSID, PSPR, PSIR,WSIL)
Factor2 0.48055 (PPWV, WPWV, VPWV, BW1, BW2, BW3)
F orm antl 0 54390
Form antS 0.35953
WFOV 0.53368
73
T he predictor v ariables listed above account fo r over 18 p e rc e n t of
the v ariab ility in Type A and Type B speakers. F ourteen o f the
o rig in al 20 acoustic speech variables are represented in the five
l
i predictors.
? Factorl "Duration"
(PSPD, PSID, PSPR, PSIR,WSIL)
; The sign and th e m agnitude of the coefficient of Factor 1
! (coefficient -0.00448) indicate a v e ry small negative value w hich is
' in terp reted to m ean th a t Type A speakers speak faster th a n Type B
I speakers, but the difference is m inor.
i
! Fjs r t n r llP n \jjp t'"
; (PPWV, WPWV, VPWV, BW1, BW2, BW3)
: The sign and the m agnitude of the coefficient of Factor 2
i (coefficient 0.48055) reveal an U nderlying pow er factor in the data.
| All th ree m easurem ent levels of duration (phrases, words and
j
i vowels) participate in this factor. The positive sign of the
i coefficient fo r Factor 2 indicates th at Type A speech shows greater
! pow er variability th a n Type B in phrases, words, and vowels. The
j sign also indicates th at Type A bandwidth m easures are la rg e r than
j Type B bandw idths.
1 Form ant 1
! Form ant 1 (C oefficient 0.54390) is a singleton predictor
| variable w hich is related positively to Type A speech behaviors.
■ The sign indicates th at first form ant frequency is h ig h e r for Type A
t
i speakers. The first form ant frequency is the strongest predictor of
I
speaker type.
7 4 ;
i
Form ant 3 i
i
Sim ilarly to Form ant 1, F orm ant 3 (C oefficient 0.35953) is a
i
singleton predictor variable related positively to Type A speech
I
behaviors. Third form ant freq u en cy is h ig h e r fo r Type A speakers, :
though the d ifferen ce betw een Type A and B speakers is not as
dram atic as FI.
Word F0 V ariability
I Word F0 V ariability (C oefficient 0.53368) is the singleton
I predictor w hich has the second h ig h est coefficient of the five
i
1 predictors. WFOV is the acoustic m easure of the difference betw een
i
! h ig h est and lowest fundam ental fre q u en cy m easures w ithin a word
j The sign and the m agnitude of the regression coefficient indicate
I th a t Type A speech produces m ore v ariab ility in fundam ental
r
freq u en cy across a word.
L.
J 75
i
!
i
i
I
I
CHAPTER FIVE: DISCUSSION
Sum m ary of Results
The th re e Type A and th re e Type B speakers are best
| differentiated (in order of efficiency) by acoustic m easures of first
; form ant frequency; fundam ental freq u en cy variability at the word
| level; power v ariab ility at phrase, word, and vowel levels; the first
! th ree form ant bandw idths and th ird form ant frequency. The
d ifferen tiatio n betw een Type A and Type B firs t fo rm an t m easures
i indicates th a t Type A speech is produced w ith m ore constriction in
' the back p h ary n g eal cavity. H igher Type A fundam ental frequency
| v ariab ility and pow er v ariab ility m easures suggest th a t Type A
! speakers habitually alter these acoustic variables, and therefore,
, th ey provide w ider and more varied ra n g e s of pow er and
i
: fundam ental frequency in th e ir speech p atterns The analysis of
i
j acoustic m easures indicates th a t in th is investigation Type A
| speakers spoke only sligh tly faster th an Type B speakers. The
! results of the investigation indicate th at acoustic speech analysis is
{ a prom ising tool in Type A and B beh av io r p attern assessm ent,
j Acoustic data from discrim inant analysis predict Type B
| m em bership m ore accurately th a n Type A. Type B m em bership is
predicted accu rately 76 p e rc e n t of the time, w h ereas Type A
| m em bership is predicted accurately 62 p e rc e n t of the time Of the
r 76
i
{ o rig in al 20 acoustic speech variables, 14 are represented in the five
j predictors used in the discrim inant analysis. The two factors and
th e th ree singleton predictor variables account for 18 p e rc e n t of the
variab ility in Type A and Type B subjects. Factor analysis is an
i
| efficient and pow erful method of organizing the m ultitude of speech
| acoustic variables necessary in such exploratory analyses.
i
i Influ en ces of the Five Predictors
i
Factor 1 'D u ratio n ''
; F actorl (PDUR, PSIL, PSIR, PSPR, WSIL [Coefficient -0 004481)
j does n o t discrim inate Type A and B speech sam ples well. The
j lite ratu re about Type A rate of speech would have indicated a
1 g re ater differen tiatio n in Type A and Type B on the Factor 1
i
I duration measures. The sm all value of the coefficient may reflect
I an u ncertainty, as reported by Friedm an and referen ced in Chapter
l
One, th at the M ilitary Commander Monologue, provokes a sense of
| time urgency. Since Type A behaviors are considered response-
| oriented behaviors, if time u rg en cy is not perceived, the behavior
I
p attern m ay n o t be elicited from Type A subjects.
It is possible th a t Type A speech is perceived as m uch faster
th an Type B speech, but is in fact only slightly faster. Vocal
in ten sity and vocal frequency, both of w hich are distinguished as
I acoustic features of Type A speech in the p resen t study, influence
th e perception of faster speech rate, together and separately. The
I '
i findings of Feldstein and Bond as reported in C hapter One are
I 77
j confirm ed in th is sample. The durational differentiation between
■ Type A and B speech perceived by in terv iew ers may occur only in
| interactive conversation.
I Factor 2 ‘ Power"
l
j G reater Type A pow er v ariab ility (PPWV,WPWV, VPWV,
BW1, BW2, BW3 [Coefficient 0.4805]) confirm s ra te r perception of
| Type A speech as affecting the listen er in a m ore pow erful, more
!
energetic and more em phatic m anner. The perception of pow er in
Type A speech is reflected stro n g ly in acoustic power m easures.
Power levels reflect coordination in the physiological production of
' speech, w hich affects the acoustic resonances of the vocal tract.
Loudness, w h ich is related to the perception of power, im pacts the
listen er and the speaker in an immediate m anner, typical of the
Type A response-oriented behavior pattern.
i F orm ant bandw idths are a m easure of the rate of decay of a
i
sound radiated by the vocal tra c t as discussed by Klatt and
t
refere n ced in Chapter One. The speech wave of Type A speakers
suffers less decay from en e rg y loss in the vocal tra c t wall. This
decreased e n e rg y loss in vocal tra c t walls m ay indicate th a t Type A
| speakers a lter the vocal tra c t w alls so th a t the walls absorb less of
j the en e rg y of the speech wave. The m odifications m ay be the
■ resu lt of tension in the walls w h ich increase tissue density, or they
| m ay occur in some o th er m an n er. Type A speakers may modify
I vocal tract wall viscosity through changes in the surface of the
| m uscles or in the mucous lin in g so th a t the w alls absorb less of the
sound wave The source e n e rg y may be of g re ater m agnitude for
Type A speakers, w h ich relates to Klatt's fin d in g s as reported in
| Chapter One. It is not possible to know if th e re is a direct
relationship between physiological modifications and the changes in
form ants u n til fu rth e r studies have been conducted.
The perceptual effect of larg er bandwidth conditions is an
j
, increase in the effective intensity of form ant en erg y concentration.
| Acoustic bandw idth m easures confirm the p erceptual data
' concerning the h ig h e r en erg y descriptions of Type A speech. The
! fact th a t bandw idth m easures contribute to the perception of h ig h e r
| en erg y in speech supports Klatt's claim as m entioned in Chapter
! One. Type A speech may be characteristically produced w ith more
! open glottis conditions.
!
I F orm ant 1
]
i The freq u en cy of FI (Coefficient 0.54390), the more pow erful
of the two fo rm an t predictor variables, is raised by constriction of
; resonance areas of the vocal tract, particularly the pharynx. The
' acoustic data suggest th a t Type A speakers constrict the p h ary n g eal
| walls more th an Type B speakers. Increased constriction of the
i
i
' p h ary n g eal walls relates to the findings of h ig h e r bandwidths for
I Type A speakers as well. Type A bandw idth m easures indicate
i
| decreased e n e rg y loss from the walls of the vocal tract. Increased
i
constriction of the p h ary n g eal walls may alter the flow of speech in
1 such a m an n er th a t the alteration causes the w all to resonate the
i
1 speech wave more efficien tly and attenuate the wave less. The
investigator suggests th at increasing constriction may in some
I
jm anner increase the density of the tissue of the p h ary n g eal are a so
(that the speech wave is am plified by th e walls in a d iffe re n t
m anner. Form ant frequencies are often held in steady state for
vowel production. This suggests th a t Type A speakers may
(regularly m aintain increased pharyngeal wall constriction during
I
: steady state vowel periods. The firs t fo rm an t fre q u e n c y is lowered
i by a constriction in the fro n t h a lf of the oral cavity. It would seem
! reasonable th a t h ig h e r FI frequencies indicate a less constricted
i (open) fro n t h alf oral cavity.
i
I According to the rules of form ant frequencies presented in
j Chapter One, the freq u en cy of all form ants, including Form ant 1 and
; Form ant 3, is raised by com pressing the lips and raisin g the larynx.
The m ore lips are com pressed and the more the lary n x is elevated,
j th e more fo rm an t frequencies are raised.
i
j Form ant 3
!
! The h ig h e r freq u en cy of F3 (Coefficient 0.35953) for Type A
I
; speakers may be related to the g en eral elevation of all form ant
I frequencies by the com pressing of the lips and raising of the larynx
j as discussed above. Not m uch is know n w hich distinguishes the
I
th ird form ant vocal resonance areas.
F orm ant 2 is noticeable by its absence in contributing to the
j d ifferen tiatio n of Type A and Type B speech. This may be partially
| explained by the essential role of F2 in carry in g the linguistic
j message. The second form ant conveys precise inform ation
|
I
I 80
i
necessary fo r the differentiation of speech sounds at th e phonem ic
level. It may be th a t the ta rg e t focus of the second fo rm an t does
notallow the speaker to m ake m odifications w hich reflect the
j individuality of the speaker w ithout the risk of obscuring the
i
j message The freq u en cy of F2 is controlled by constriction of
| various parts of the tongue.
I
j Word F0 V ariability
; The acoustic data confirm th e perception of Type A speech as
! exhibiting more pitch variability. The greater variability in
j fundam ental fre q u en cy (C oefficient 0.53368) of Type A speakers
i
. indicates th a t they v a ry vocal fold vibration more th an Type B
j speakers. Word level v ariab ility seems to be m ore efficien t th an
I
! vowel level variability in identifying Type A speech. Fant's claim
m entioned in C hapter Three proves to be valid
i
i
i
j Possible Connections and Future Studies
1
Acoustic Predictors of Type A B ehavior
The p re se n t investigation of th ree Type A speakers and th ree
i
i
I Type B speakers constitutes a p io n eerin g study of these previously
uninvestigated acoustic speech m easures and th e ir relationship to
the behavior type. It would seem a logical immediate step to
develop the acoustic inform ation reported from the p re se n t study
by designing a sim ilar investigation w hich draws from an expanded
data base of Type A and Type B speakers and replicates c u rre n t
| m ethodology. The data base m ay include the M ilitary Commander
! Monologue and o ther speech m aterials specifically developed to suit
! the dem ographics of the subjects. It would be im p o rtan t to include
j raw data w h ich provide for analysis of speaker speech rate m ore
effectively and w hich provide a ric h source of m aterial particularly
j suitable fo r analyzing the acoustic features th a t are most efficien t
i as predictors in the p resen t study. The categorical c rite ria for this
| study would be expanded to include A1, A2, B3, B4, and X .
!
j Speech, the Vagus N erve and th e H eart
j The Type A speakers in th e p re se n t study show h ig h e r
j form ant freq u en cy averages th an Type B speakers, fo r both
: form ant 1 and form ant 3, but p articu larly fo r fo rm an t 1. Form ant
bandw idths are produced as p redictor variables on Factor 2, the
' pow er factor As reported in C hapter One, form ant freq u en cies and
j bandw idths are produced by the in teractio n s of com plicated
! resonance filters in the vocal tract and, as such, are intim ately
, associated w ith the physiological responses of the body.
' From the Form ant Rules in C hapter One and the discussion of
| the results above, it is know n th a t FI is raised by constriction of the
1 p h ary n x and o th er resonance areas. The m usculature of the
| p h ary n x is in p art controlled by two ram i of the vagus nerv e, the
\ Xth cranial nerve. The vagus, w hich has its origins in the nucleus
i am biguous and is derived from the h in d b rain, w anders th ro u g h the
! body. It is extensively involved in m uscles of the speech
i
: m echanism s in both the p h ary n x and the larynx. The p h ary n g eal
82
j branch, of the vagus contains both sensory and m otor fibers w hich
| supply the m uscles and mucous m em brane of the p h ary n x .
| The vagus is of p articu lar in te re st in the p re se n t study of
j Type A speech because of the course of the re c u rre n t n erv e w hich
| arises from the vagus at a point considerably below the larynx. The
i left re c u rre n t n erv e loops u nder and behind the aortic arch, serves
i
; th e pericardium of the h eart, and th e n trav els down to the kidneys
j
| and the adrenal glands. It has b ran ch es w h ich eventually lead to
; the re sp irato ry system, the digestive system and th e sym pathetic
nervous system.
The in teractio n s of physiological system s served by the vagus
i
|
j n erv e are n o t well understood by speech physiologists. However,
the interrelatio n sh ip of speech, h e a rt reactivity, and catecholoamine
' reactivity w ithin Type A subjects m erits investigation of the
; connections. A single effective stimulus to the rig h t vagus nerve
i
; exerts a m easurable in h ib ito ry effect upon the h e a rt rate. The
I likely supposition is th a t it alters the electrical behavior of the cell
i
j m em brane. This phenom enon is called the vaso-vagal reaction
l
| The reaction can occur w ith almost any shock stim ulus, such as the
| p rick of the needle w hen blood is being draw n. It is in terestin g to
i
<
I note th a t Fulton (1955) reports, from an old experim ent of Gaskell,
I th a t the in ju ry potential, as recorded betw een an in ju red point and
; an u n in ju red p o in t on the surface of the tortoise h eart, was
■ dim inished w hen the vagus was stimulated.
83
U nder norm al circum stances the h e a rt is kept under constant
vagal restrain t. This is called a state of vagal "tone" w hich varies
from individual to individual. The m ajor regulation of vagus n erv e
j activity occurs th ro u g h its participation in the depressor reflexes of
I the a rc h of th e aorta and carotid sinus. These and associated
i
j vasopressor receptors are in constant activity, w hich results in a
! constant discharge of the vagus fibers to the h eart. Almost all of
- the increase in h e a rt rate th a t takes place w ith exercise m ay be
; ascribed to a cen tral inhibition of vagal tone during exercise. The
j rem arkably slow pulse in some h ig h ly train ed athletes is also
! apparen tly a reflection of tonic vagal control. Vagal tone is at a
; maximum during adolescence and early adulthood. In in fan c y and
old age it is at a m inim um .
i
i The second im portant receptive zone for vagal reflex is the
i resp irato ry tra c t The reflex is best dem onstrated w ith a sig n ifican t
i
response to lary n g eal pressure. Holding the breath and exerting
I
i p ressu re in the lary n x is a treatm en t com m only recom m ended by
j
j doctors for stopping arhythm ic tachycardia (extrem e accelerated
| h e a rt rate). The vagal reflex serves to reduce circulation and lower
: oxygen requirem ents during extrem e physical exertion, such as
w eight liftin g and some form s of the m artial arts. These extrem e
physical stress activities are often followed w ith carefully timed
i
| release of sounds. The investigator w onders if the release of sounds
I from w ith in the vocal tra c t stim ulates the vagus and contributes to
i
j relieving the affects of potentially harm fu l physical stress.
j The response assum es some im portance d u rin g operations on
|the neck and pharynx, and th ere is an association of the vagii w ith
l
control of sym pathetic release of ep in ep h rin e (Fulton, 1955)
| According to Nicolas Yankopoulis, M. D, a practicing
j cardiologist, it is know n th ro u g h clinical experience that during
i
vaso-vagal reaction the voice is characterized as "lower" and it
[Sometimes becomes hoarse. It is not clear th a t pitch and in ten sity
jof voice are altered, but it would seem a likely associated response.
i
I W hat makes the investigation in terestin g from aT ype A and
.acoustic speech perspective are the association of the vagus nerve
, in controlling laryngeal and pharyngeal m usculature and the
contribution of the p h ary n g eal resonance area to elevated first
1 form an t m easures fo r Type A speakers
The results of the p resen t study suggest investigating the
association of acoustic features and the physiological functioning of
the speech production m echanism s w hich are indicated as
im portant in producing significant individual variations in form ant
j freq u en cy and bandw idth m easures. Acoustic analysis of p articu lar
j speech features in v ary in g states of physical stress as well as
psychological stress may provide inform ation about the relationship
j between acoustic features and accom panying physiological
responses. It would be valuable to explore an y association betw een
the vagal response in controlling h ea rt rate and the role of the
t
i
; vagus in speech production. The variability in Type A association
i w ith coronary h e a rt disease th ro u g h the four phases of the W estern
; 85
Collaborative Group Study, as analyzed by Ragland and Brand and
referen ced in C hapter One, may indicate th at som ething in the Type
; A behavior p attern becomes a m ediating factor for dim inishing a
; continuing risk of coronary h e a rt disease. Speech behaviors as they
| relate to the vagus m ay contribute to the possible p attern of
^ m ediation.
j A study could be constructed w ith the participation of speech
scientists, cardiologists, neurologists, and psychologists w hich tracks
i
! the association of changes in acoustic speech features, neurological
j responses, and cardiac responses in various conditions of physical
i
| and psychological stress. Of p articu lar in te re st would be how these
; changes relate to the vagus n erv e in speech production and the
l
| vagal response in the heart. Acoustic analysis com parisons could be
i
! made w ith in and betw een subjects for all test conditions.
P lethysm ographic equipm ent m ay be adapted to m easure
! m uscular m ovem ent in the speech m echanism s. If th is is possible,
; response of some of the speech laryngeal and p h ary n g eal
| m usculature could be m easured w hile producing speech sam ples for
j acoustic analysis in some experim ental settings. It is also possible
i in some degree to m easure n erv e responsivity feed back to the
i
I brain from various stimulus. Although it is not certain th ere is any
1 direct, indirect, or telegraphic stim ulus from the vagus during or
i
, following its activity in the speech m usculature or to the a ffe re n t
branch of the vagus w hile it is involved in m aintaining vagal tone",
i it m ay be possible to m easure the com m unication and response to
the b rain w ith modified neurological equipm ent.
Emotion. Cognition. FO. and Pow er V ariability
It is possible to propose an association of Type A increased
em otional arousal and Type A cognitive processes w ith Type A
acoustic speech m easures of increased fundam ental freq u en cy
v ariab ility and increased pow er v ariability. Studies of Type A
j interactions w ith both interview ers durin g the Structured
| Interview and w ith subjects in competitive experim ents confirm
! th a t Type A subjects are m ore em otionally reactive th a n Type B
' subjects d u rin g the experiences. Type A subjects im pact
' in terv iew er behavior and the style of the interview m ore than
j Type B subjects during the SI The im pact upon in terv iew e rs is
| considered im portant assessm ent inform ation for differentiating
; types by some raters. A th eo ry about Type A behaviors proposes
| th a t Type A persons suppress and deny stress w hich leads to
| increased sym pathetic nervous system arousal. Sym pathetic
' nervous system arousal, the fig h t or flig h t response, contributes to
] h eig h ten ed reactivity fo r Type A subjects in stressful situations as
j discussed in C hapter One.
I Studies of the cognitive elem ent in Type A behavior p attern
assert th a t cognitive processes contribute to Type A behavior,
i
' including the em otional reactivity of Type A. Type A individuals
' hold fears and distorted beliefs w h ich cause them to perceive
, events and people d ifferen tly th a n Type B persons Type A
87
subjects respond to experim ental tasks as m ore im p o rtan t and
dem onstrate more physiological arousal th a n Type B subjects.
i
I Em otionally reactive Type A individuals respond to th re a ts in the
t
| en v iro n m en t w ith habitual overt Type A behaviors. The behaviors
I
| are designed to establish personal control in situations and to
. provide safety fo r them selves. In fact, Type A behaviors im pact
I th e ir en v iro n m en t dynam ically As suggested by the re se arc h ers in
i cognition for Type A behavior p attern and reported in C hapter One,
the situation and others involved in th e experience are in reality
, altered, som etimes dram atically, by the Type A behaviors
i
i As reported in C hapter One, fundam ental freq u en cy and
pow er are the two acoustic features w h ich are related in a more
direct w ay in speech perception th a n most o th er acoustic m easures.
1 Type A speech in the p re se n t study was differentiated in the
: variability of fundam ental frequency (pitch) and in the variability
: of power (loudness). These two acoustic features are g en erally
| understood and comm only accepted as im portant in speech
com m unications and as such are the acoustic speech variables w ith
; the "highest profile." As referenced in Chapter One, speakers
i produce intonation patterns to convey m eaning in phrases and
!
! sentences by changes in the fundam ental frequency; th at pitch
i
! provides inform ation about the speaker s age, sex, health, size and
’
em otional state; and th a t loudness provides inform ation about
| em otional states and the degree of speaker involvem ent in
; experiences. Elem ents of pitch and loudness are on occasion and to
j v ary in g degrees consciously controlled by speakers. Speakers are
! capable of elicitin g overt and subtle response behaviors from
{listeners by controlling pitch and loudness.
{
Speakers are im m ediately aw are of the im pact of th is
behavior on the en v iro n m en t and on them selves The behaviors
m ay assure speakers in some way th a t the situation is controlled by
i them . These speech behaviors at times have been useful, necessary
and rew arding Type A speakers may develop m ore conscious
! control of these two relatively overt speech behaviors in an effo rt
to control and im pact situations m ore im m ediately and m ore
dram atically.
| R esearch indicates th at Type A persons m ay become addicted
i
1 to the catecholoam ines produced by the sym pathetic nervous
: system d uring such em otional and physiological behavior responses,
j as discussed in C hapter One. It is possible th a t the addictive n atu re
I
j of the behavior p attern propels Type A individuals into creating the
i
! set of speech response behaviors in them selves, and it subsequently
; leads them into creating confirm ing responses in th e ir environm ent.
! The speech behaviors form p atterns of relating and com m unicating
i
! w hich are activated on occasions w hen the behavior is n e ith e r
i
useful n o r necessary,
i Speech behaviors are obviously recognized and controlled by
! Type A speakers on some occasions T here is the possibility th a t on
| o th er occasions speech behaviors are n o t acknowledged in eith er
j th e ir more extrem e or m ore subtle form s At times the speech
j p attern s of Type A persons m ay be so habitual as to be unnoticed,
unconscious and uncontrolled. Rosenman and o ther research ers
i
i
| have found th a t Type A subjects are not p articu larly successful in
i
i
recognizing th e ir own behaviors in experim ental situations. Such
com parisons of Type A self-m easures and ra te r assessm ents usually
| p o in t out th e areas of Type A b ehavior th a t Type A subjects deny.
The cycle of Type A speech behaviors may in some way become
: autom atic and self-perpetuating by the addictive n atu re of the ru sh
, of the catecholoam ines, as w ell as by the immediate response th a t
!
| the Type A speech behaviors, in particular, elicit from norm al life
| en counters w ith people and situations.
! The account of the p re se n t investigation provides valuable
I
inform ation w hich enhances any future investigations in acoustic
I speech analysis. Data for a new study can be designed to focus on
I
1 those acoustic m easures w hich are indicated as most effective in
d iffe ren tiatin g Type A and B speech. It may be possible to use
1 available data or to g a th e r data w h ich allow the speech scientist to
, analyze form ant frequencies and bandw idths across vowels in the
■
i flow of ru n n in g speech. The speech sam ples can be from both
J
j n atu ra l and sim ulated settings. In e ith e r n atu ra l or sim ulated
I speech settings the results of the p re se n t study provide im petus for
i
investigations in acoustic speech to expand the focus of speech
; featu res targeted fo r study.
!
Acoustic analysis m ay provide inform ation about the
effectiveness of Type A speakers' control and influence in various
90
speech situations. Data from acoustic analysis of speech sam ples
j from experim ental situations of in teractio n s of Type A and B
j subjects in w ith cohorts, in Solitude, or o th er situations w here the
j
! potential fo r influencing cohorts through vocal affect is minimized,
i
I o r frustrated by o v ert non-response, may provide inform ation
about an y association of Type A acoustic speech m easures and a
: proposed need on th e ir p a rt to im pact the en v iro n m en t th ro u g h
i
Type A speech behaviors. The study m ay make a contribution
toward understanding the probable role of cognitive processes in
i
! m otivating speech behavior in differen t speech situations.
i
1 The Five Coefficients and the Limen Scale
i
The coefficients from the p resen t study could be used in a
i
statistical analysis of the lim en of psycho-acoustic speech. Limen
)
! m easure the m inim al discernible differences in acoustic speech as
j perceived by the ear. It is possible to re se arc h the acoustic
. literature to find lim en fo r fundam ental frequency, form ant
frequencies, power, and p erh ap s bandw idths. It m ay be possible to
| scale the five acoustic coefficients produced by statistical analysis of
| these data to the lim en scale of speech perception, The five acoustic
! predictor variables w hich w ere the most m eaningful in the p resen t
study could be analyzed to determ ine the th resh o ld of
1 d ifferen tiatio n fo r Type A and Type B sp ee ch . It would be
in terestin g to know to w h at degree Type A and Type B speech
i
: behaviors meet, exceed, o r fail to m eet the m inim al scale of lim en
j
; for the acoustic variables w hich w ere im portant in this study.
I 91
i An investigation could be conducted w ith m inim al additional
i
| statistical analysis and w ithout any new speech data. The challenge
i
; in such a study would be to develop a lim en scale w h ich relates to
i
J th e data of the p re se n t investigation and w hich could be adapted as
| an index fo r com parison w ith o th er studies. A lim en scale w hich
: would be developed exclusively fo r Type A could be useful in the
, cognitive study proposed above, as an index of th e relatio n sh ip of
! perception of speech to th e acoustic m easures of the differentiation
i of Type A speech in various experim ental environm ents,
j ATSS and the A nger Video
• Speech data from th e A nger Video assessm ent of the Psyche-
4
1
: h y p erten sio n Study, associated w ith the p resen t investigation and
j referen ced in C hapter Three, could be analyzed to determ ine
j acoustic interaction w ith psychological stressors and em otional
, reactivity of Type A persons. The ATSS (A rticulated T houghts in
!
. Simulated Situations) data from the same NIH study, also
l referenced in Chapter Three, have been coded by B ushnell for
; linguistic content to cover several areas of psychological behavior
; w hich includes degree of self-support and degree of self-criticism .
[ It may be possible from the data to determ ine acoustic features
; w hich relate to these seldom verbalized personal attitudes in
1 p articu lar psychological states w hich are explored in the ATSS
' concept. The data from th e ATSS and the A nger Video assessm ent
w hich are p a rt of the data base of the P syche-hypertension Study
1 are available fo r acoustic analysis. The technical quality of the
! recordings is unknow n to the investigator. However, this study has
i
show n th at sig n ifican t acoustic m easures can be obtained w ith
i relatively poor recordings.
I
| Acoustic A nalysis of In terv iew er V ariability
j In terv ie w er speech sam ples could be analyzed to determ ine
i
i
■ any acoustic changes in th e ir speech w hile conducting interview s
| w ith Type A and Type B subjects. Since in flu en ce on interv iew ers
j speech p attern s is considered an im portant cue in assessing Type A
| behavior, it m ay be possible to m easure acoustic changes in speech
I
behaviors of in terv iew ers from subject type to subject type.
i
| Acoustic Speech A nalysis and O ther Constructs
The analytic procedures and th e acoustic variables of this
: study may prove to be useful in investigating o th er personality and
■ behavioral constructs It is not en tirely clear w hat is being
’ differentiated in the behavior p attern s of the two personality types.
i
j It may be th a t w h at has been addressed in the acoustic analysis of
i
; th is study is a m easure of hostility in speech behavior, more th an
| identification of Type A and B behavior p attern s exclusively.
| Acoustic M easures Above Form ant 3
\
A nalysis of form ants above the th ird fo rm an t m ay be
considered in an y of the above studies Speech scientists
traditionally focus on the first th ree form ants of speech as
■ investigations in speech are generally focused on accurate
’ production of phones and the analysis of th e ir production. For
' example, in the p re se n t study the investigator analyzed four
I
1 93
I
form ant peaks w ith the intention of using only th ree of the
form ants in the data base. Inform ation concerning personality
j
: differentiations may be be available in the h ig h e r form ant
j freq u en cies and bandwidths.
i
I
t
Conclusion
I The value of the form ant m easures in discrim inating Type A
■ and B behavior patterns, discovered in this investigation, is a m ajor
i contribution to the field of speech science since norm ally these
i
i m easures are p rim arily associated w ith identification of sound
t
I
1 segm ents. The results of the p re se n t investigation indicate th a t
I
| acoustic speech variables are a prom ising tool in Type A and B
| beh av io r p a tte rn assessm ent and may become even m ore im portant
| in studying o th er aspects of behavior as th ey relate to speech
! patterns. The p resen t investigation is im portant in both speech
f
! science and personality studies w ithin psychology, as it h as taken
I
j th e first step in experim entally show ing a direct relation between
i
; acoustic speech and Type A behaviors. Finally, the study leads to
i
! m any possible future investigations involving physiology, acoustics,
I and the perception of speech production and th e ir relationship to
j psychological perso n ality constructs beyond Type A and Type B .
i
!
r
REFERENCES
i Baker, L. J., Hastings, J. E , 6c Hart, J. D . (1984). E nhanced
; psychophysiological responses of Type A coronary patients during
• Type A -relevant im agery. Journal of B ehavioral M edicine. 7, 287-
' 306 .
t
1 Barefoot, J, C., Dahlstrom, W .G ., 6c Williams, R. B. ( 1 9 8 3 ) . Hostility,
J CHD incidence, and total m ortality: A 25-year follow -up study of
. 255 p h y sician s Psychosom atic M edicine. 45., 59-63-
i
i
1 B lum enthal, J. A., O’ Toole, L. C., 6c Haney, T. (1984). B ehavioral
! assessm ent of the Type A b ehavior p attern . Psychosom atic
! Medicine. 46, 415-423
I
Bogert, B. P. (1953 ). On the band w idth of vowel form ants [Letter to
j the editor]. The Journal of the Acoustical Society of A m erica. 25.
! 791-792.
i
! Borden, G. J., 6cHarris, K. S. (1981) Speech science p rim er (pp. 47-
I 1 6 0 ) Baltimore: W illiams 6c W ilkins.
i
! B ushnell, M E. (1987). Cognition of Type A and B individuals.
Unpublished m aster's thesis, U niversity of Southern California,
j Los Angeles.
i
| Carver, C . S., 6*Glass, D, C . (1978). C oronary-prone behavior p attern
' and interpersonal aggression. Journal of Personality and Social
j Psychology. 36. 361-366.
i
j Chesney, M. A., Black, G. W., Chadwick, J. H., 6cRosenm an, R. H.
j (1981). Psychological correlates of th e Type A behavior pattern,
j Journal of B ehavioral M edicine. 4. 217-229.
j Chesney, M. A., Eagleston, J. R., 6c Rosenman, R. H. (1980). The Type
: A Structured Interview : A behavioral assessm ent in the rough.
| Journal of B ehavioral A ssessm ent. Z, 255-272
i 95
i
Chesney, M A., 6c Rosenm an, R. H. ( 1 9 8 5 ) A nger and hostility in
! cardiovascular behavioral disorders. New York: H em isphere
| Publishing Corporation.
i
Denes, P B., 6c Pinson, E (19-73). The speech chain: The physics and
biology of spoken lang u ag e. New York: Doubleday.
; Diamond, E. L. (1982) The role of a n g e r and hostility in essential
j H ypertension and Coronary H eart Disease. Psychological B ulletin. 92.
1410-433
] Dunn, H K. (1961) Methods of m easuring vowel fo rm an t
j bandwidths. The Journal of the Acoustical Society of A m erica. 33.
; 1737-1746.
i
I
j Fant, G. (1970). Acoustic th eo ry of speech production: With
\ calculation based on X-rav studies of Russian articu latio n s. The
; Hague: Mouton
I
■ Feldstein, S., & Bond, R. (1981). P erception of speech rate as a
! function of vocal intensity and frequency. Language and Speech. 24.
I 387-395.
i
j Fowler, C . A. (1985) C u rren t perspectives on language and speech
I production: A critical overview. In Raymond G . D aniloff (Ed ),
, Speech science: R ecent advances (pp. 193-278). San Diego, CA:
. College-Hill Press.
i
j Friedm an, M., Brown, A. E., 6cRosenman, R. H. (1969). Voice
analysis test fo r detection of behavior p attern. Journal of the
! A m erica Medical Society. 208. 828-836.
Friedm an, M., 6c Rosenm an, R. H. (1959). Association of specific
■ overt behavior p attern w ith blood and cardiovascular findings.
Journal of the A m erican Medical Society. 169. 1286-1296.
Friedm an, M., 6c Rosenm an, R. H. (1974). Type A B ehavior and Your
! Heart. New York: Fawcett Columbine.
96
iFujim ura, 0., 6c Lindqvist, J. (1970). Sw eep-tone m easurem ents of
I voeal-tract characteristics. The Journal of the Acoustical Society of
I A m erica. 49. 541-558.
i
I Fulton, J F (1955) A Textbook of Physiology (pp 686-692)
Philadelphia: W. B. Saunders and Company
Gorsuch, R L. (1983) Factor analysis. New Jersey: Law rence
j Erlbaum Associates, Publishers.
I Hecker, M. H. L„ Chesney, M argaret A., Black, G . W., 6c Rosenm an, R.
: H. (1981). Speech analysis of Type A behavior In J. K. Darby, Jr
j (Ed ), Speech evaluation in m edicine (385-397). New York: Grune
j and S tratto n .
j FEerman, S., Blum enthal, J. A., Black, G. M., 6c Chesney, M. A.
i (1981). S elf-rating of Type A (co ro n ary p ro n e) adults: Do type A s
j know th ey are type A's? Psychosom atic M edicine. 43., 405-413
i
I
I H irano, M., ScKakita, Y. (1985) Cover-body th eo ry of vocal fold
! vibration In R G Daniloff (Ed ), Speech science: Recent advances
■ (pp. 1-46). San Diego: College-Hill Press.
Hollien, H (1981). Analog instrum entation for acoustic speech
i analysis. In J. K. Darby, Jr. (Ed ), Speech evaluation in p sy ch iatry .
(79-103). New York: G rune and Stratton.
House, A., 6c Stevens, K. N. ( 1 9 5 8 ) . Estimation of fo rm an t band
w idths from m easurem ents of tra n sie n t response of the vocal tract.
! Journal of Speech and H earing R esearch 1, 309-315
I
f
Houston, B. K (1983). Psychological responsivity and the Type A
behavior pattern. Journal of R esearch in P ersonality. 17. 22-39.
Howland, E. W , 6cSiegman, A. W. (1982). Toward the automated
j m easurem ent of the Type-A behavior p attern . Journal of
' B ehavioral M edicine 5. 37-55
i Jacobs, D . R. Jr., 6c Schucker, B. (1981). Type A behavior pattern,
j speech, and coronary h e a rt disease. In J. K. Darby, Jr. (Ed ), Speech
I evaluation in m edicine (pp. 363-383). New York: G rune and
j Stratton.
i
K erlinger, F N. (1979). Behavioral re se a rc h . New York: Holt,
{R inehart and Winston.
Klatt, D H. (1974). The duration of [s] in English words. Journal of
: Speech and H earing A ssociation. 17. 51-63.
i
I Klatt, D . H. (1980). Software for a cascade/parallel fo rm an t
synthesizer. Journal of th e Acoustical Society of A m erica. 67. 971-
! 995.
K rahulec, I , Urbanova, 0., 6cRafajdusova, I. (1985). D isturbances of
! the static and dynam ics of the spine and neuroticism in functional
voice disorders. Ceskoslovenska Oto logvngologie. 34. 92-95
Ladefoged, P (1962). Elem ents of acoustic p h o n etics. Chicago: The
U niversity of Chicago Press.
LadefogedT-P— (-1982-)— A-course in-phonetics^-New Y ork: H areourt_
Brace Jovanovich, Inc.
Lehiste, I. (1976). Suprasegm ental features of speech.
Contem porary issues in experim ental p h o n etics. New York:
Academic Press
Lieberm an, P. (1977). Speech physiology and acoustic phonetics.
New York: Macmillan Publishing Co., Inc.
M artin, J. G . (1972). R hythm ic versus serial stru ctu res in speech
and o th er behaviors. Psychological Review . 79, 487-509.
M atthews, K. A. (1982). Psychological perspective on the Type A
behavior p attern Psychological B ulletin. 91. 293-323.
! 98
j M atthews, K. A., Krantz, D S., Dembroski, T. M., 6 c MacDougall, J. M.
j(1982). U nique and common v arian ce in S tructured Interview and
j Jenkins Activity Survey m easures of the Type A behavior pattern.
' Journal of P ersonality and Social Psychology. 42. 303-313
i
j Natale, M., Entin, E. 6 c Jaffe, J. (1979). Vocal in te rru p tio n s in dyadic
! com m unication as a function of speech and pathology and
j audiology. Journal of Personality and Social Psychology. 37. 865-
878.
» P erkins, W H (1971). Vocal function: A behavioral analysis in Lee
Edward T ravis (Ed ), Handbook of speech pathology and audiology.
i Englewoods Cliffs: Prentice-H alf Inc.
i
Peterson, G. L., 6c B arney, H. L. (1952). Control m ethods used in a
study of the identification of vowr els. Journal of The Acoustical
Society of A m erica. 24. 175-184.
Pfau, E. M (1978) Personality test procedures for d ifferen tial
! diagnosis in phoniatry. Folia P h o n iatrica. 30. 43-49.
i
1 Pickett, J M. (1980). The sound of speech com m unication: A prim er
j of acoustic phonetics and speech perception. Baltimore: U niversity
j P ark Press
! P ittn er, M S., 6cHouston, B. K. (1980). Response to stress, cognitive
f coping strategies, and the Type A behavior pattern. Journal of
P ersonality and Social Psychology. 39. 147-157.
J Price, V. A. (1982). Type A behavior p attern A model for research
I and p ra c tic e . New York: Academic Press.
I
j Ragland, D. R. 6c Brand, R. J ( 1988a). C oronary h e a rt disease
j m ortality in the W estern Collaborative Group Study A m erican
j lournal of Epidemiology . 127. 462-75.
Ragland, D R 6c Brand, R. J. ( 1 9 8 8 b ) . Type A behavior and
■ m ortality from coronary h e a rt disease, New England Journal of
. M edicine. 3 1 8 . 6 5 - 6 9 .
Rice, L. N. 6c Koke, C J. ( 1 9 8 1 ) . Vocal style and the process of
psychotherapy. In J. K. Darby, Jr. (Ed ), Speech evaluation in
p sy ch iatry ( 1 5 1 - 1 6 8 ) . N e w York: Grune and Stratton.
Rosenm an, R. H. (1978). The in terv iew m ethod of assessm ent of the
coronary-prone behavior p attern . In T. M. Dembroski, S. M Weiss,
J. L. Shields, S. G Haynes, 6c M. Feinleib (Eds.) C oronary-prone
behavior (pp. 55-69) New York: Springer-V erlag.
Rosenman, R. H., Brand, R J , Sholtz, R. I., 6c Friedm an, M ( 1 9 7 6 ) .
| M ultivariate prediction of coronary h e a rt disease d u rin g 8.5 y ear
: follow-up in the W estern Collaborative Group Study The A m erican
j Journal of Cardiology 37, 903-909.
; Rosenman, R. H. 6c Friedm an, M . ( 1 9 6 1 ) . Association of specific
behavior p attern in women w ith blood and cardiovascular findings.
| C irculation. 24. 1173-1184.
j Rosenman, R. H„ Friedm an, M., Straus, R., Wurm, M., Kositchek, R.,
j H ahn, W , 6c W ethessen, N. ( 1 9 6 4 ) A predictive study of coronary
h e a rt disease: The W estern Collaborative Group Study. Journal of
' the Am erican Medical A ssociation. 189. 15-22.
*
Sanford, F. H. (1942), Speech and personality. Psychological Bulletin.
39, 811-845.
S cherer, K. R (1981). Speech and em otional state. In J. K. Darby, Jr.
(Ed ) Speech evaluation in p sy ch iatry (pp. 189-220). New York:
Grune and Stratton.
S cherer, K. R. 6c Scherer, U. ( 1 9 8 1 ) . Speech behavior and
personality, In J. K. Darby, Jr. (Ed ), Speech evaluation in psychiatry
(pp. 1 1 5 - 1 3 5 ) New York: G rune and Stratton.
Scherwit2, L., Berton, K., 6cLeventhal, H. (1977). Type A assessm ent
and interaction in the behavior p attern interview . Psychosomatic
M edicine. 39. 229-240.
r
; 100
;Scherwit2, L , Graham, L, E., Grandits, G , 6 c Billings, J. (1987)
(Speech characteristics and behavior-type assessm ent in the
(multiple risk factor trial (MRFIT) Structured Interview s. Journal of
(B ehavioral M edicine 10. 173-195
f
Schucker, B , 6 c Jacobs, D , R. (1977). Assessm ent of behavioral risk
I for coronary disease by voice characteristics. Psychosomatic
' M edicine. 39, 219-228.
[
|Shoham -Y akubovich, I., Ragland, D . R., Brand, R., J., 6cSyme, S. L.
I (1988). Type A behavior p attern and h ealth status afte r 22 y ears of
j follow-up in the W estern Collaborative Group Study. A m erican
| Journal of Epidemiology. 128. 597-588.
j
Siegman, A. W., Feldstein, S., Tomasso, C . T., Ringel, N., 6 c Lating, J.
(1987). Expressive vocal behavior and the severity of coronary
j a rte ry disease. Psychosom atic M edicine. 49. 545-561
, W einstein, K. M, Davison, G. C , 6 c DeQuattro, V , 6 c Allen, J. W. (1986).
I Type A behavior and cognitions: Is hostility the bad ac to r? P aper
: presented at the A nnual Convention of the A m erican Psychological
I Association, W ashington, D C
I
; Williams, R B., Haney, T. L , Lee. K.L., Blum enthal, J. A., 6 c W halen,
R E. (1980). Type A behavior, hostility, and coronary
{ atherosclerosis. Psychosom atic M edicine. 42. 539-549.
| Williams, C . 6c Stevens, K. (1981). Vocal correlates of em otional
| states. In J. K Darby, Jr (Ed ) Speech evaluation in psychiatry (pp.
{ 221-240) New York: Grune and Stratton.
f
i
! Zemlin, W. R. (1981) Speech and h e a rin g science: Anatomy and
! physiology (2nd ed ). New Jersey : Prentice-H all.
101
I APPENDIX A
I
! STRUCTURED INTERVIEW PROTOCOL
j THE PSYCHE-HYPERTENSION STUDY
i V ersion 1
, "I would appreciate it if you would answ er the follow ing questions
I to the best of y o u r ability. Your answ ers w ill be kept in the
, strictest confidence. Most of the questions are concerned w ith yo u r
, superficial habits and none of them, w ill em barrass you." (Begin
j tape now.)
i
I Your code n u m b er i s ______ _
1
! 1 May I ask you age?
i
!
. 2. W hat is y o u r jo b ________________ ?
' (a) How long have you been in this type of w ork?
(b) Tell me m ore about y o u r job.
i
1 3- Are you SATISFIED w ith y o u r job level?
1 (a) W hy? Why not?
j (b) If n o t -- Have you done a n y th in g about it?
i 4. Does yo u r job carry HEAVY responsibility?
(a) Is th e re an y time w hen you feel p articu larly RUSHED or
under pressure?
(b) W hen you are u n d er PRESSURE does it b o th er you?
I 5. Would you describe y o u rself as a HARD-DRIVING, AMBITIOUS
I type of person in accom plishing the th in g s you w ant, OR would you
j describe y o u rself as a relativ ely RELAXED and EASY-GOING person?
i (a) Are you m arried?
j (b) (If m arried) How would you WIFE/HUSBAND describe you
j in those term s — as HARD-DRIVING and AMBITIOUS or as
relaxed and easy going?
j (c) (If no) NEVER?
(d) Has s h e /h e ev er asked you to slow down in y o u r w ork?
| Speed up?
! 6. W hen you get ANGRY or UPSET, do people around you know
| about it?
I (a) How do you show it?
(b) Do you ev er pound on y our desk? Slam a door? Throw
! things?
I
j 7 Do you th in k you drive HARDER to ACCOMPLISH th in g s th a n most
! of you associates?
i
8, Do you take w ork hom e w ith you?
(a) How often?
| (b) Do you really do it?
j
I 9. Do you have children?
: (If no children) Have you ever played w ith sm all children?
! With you ch ild ren , w h en th ey w ere around the ages of 6 and 8, did
; you EVER play com petitive games w ith them , like cards, checkers,
Monopoly?
1 (a) Did you always allow them to WIN on PURPOSE?
| (b) Why o r w hy not?
10, W hen you play games w ith people YOUR OWN age, do you play
1 for the FUN of it, or are you really in th e re to WIN?
| 11. Is th e re COMPETITION in you job?
| (A) Do you enjoy this?
j 12, W hen you are in you automobile, and th e re is a ca r in you lane
. going FAR TOO SLOWLY fo r you, w h at do you do about it?
(a) Would you MUTTER and COMPLAIN to yourself? Honk you
! h o rn ? Flash y our lights?
| (b) Would anyone riding w ith you know th a t you w ere
! ANNOYED?
13. Most people w ho w ork, have to get up fa irly ea rly in the
I m orning, in you p articu lar case, uh-w hat-tim e-uhhh-do-you-uhh,
, ordinarily u h -u h -u h -g et up?
103
14. If you make a DATE w ith someone for, oh, two o'clock in the
afternoon, would you BE THERE on TIME?
(a) Always? N ever?
(b) If YOU are kept w aiting, do you RESENT it?
j (c) Would you SAY a n y th in g about it? How would you say it?
| (d) W hy? or Why not?
i
- 15 Do you OFTEN fin d th a t w hile you are listen in g to ONE th in g you
1 are THINKING about som ething ELSE?
(a) N ever? Always?
i
J 16. if you see someone doing a job ra th e r SLOWLY and you KNOW
I th a t you cold do it faster and better yourself, does it make you
■ RESTLESS to w atch him ?
j (a) Would you be tem pted to STEP IN and D O IT y ourself?
(b) Have you ev er done th at?
(c) What would you do if someone did th a t to you?
j 17 Do you OFTEN do two th in g s a THE SAME TIME — like reading
I while w atching TV, reading w hile eating, w riting or reading w hile
J talking on the telephone?
j (a) N ever? Always?
i
! 18 W hat IRRITATES you most about y o u r w ork, o r the people w ith
{ whom you w ork?
j (a) Why is th is so bad?
i (b) W hat do you do about it?
J 19. Do you EAT rapidly? Do you WALK RAPIDLY? A fter you've
i FINISHED eating, do you like to GET UP and GET GOING?
20. W hen you go out in the evening to a re sta u ra n t and you find
e ig h t o r ten people WAITING AHEAD OF YOU fo r a table, w ill you
! w ait?
| (a) Most of the time, how long w ill you w ait?
(b) W hat w ill ;you do w hile you are w aiting?
| Cc) Are you im patient w hile you are w aiting?
21. W hat would you do if you had made a reserv atio n at a
| re sta u ra n t and upon a rriv in g the hostess tell you th a t th ere will be
' a 20-m inute w ait?
I (a) W hat if a fte r w aiting 20 m inutes the hostess says th a t it
w ill be a n o th e r 20 m inutes?
22. Would you EVER ask a n o th e r person in a re sta u ra n t to stop
sm oking?
! (a) W hat would you say? How would you do it?
1 (b) (If no) W hat if you com panion asked you to ask a m an
sm oking a cigar to stop? How would you do it?
(c) (If no) Why not?
1 23. Do you HURRY in doing most things?
24 How do you feel about WAITING in lines — bank lines?
superm arket line? post office lines?
■ 25 Do you ALWAYS feel anxious to GET GOING and FINISH
! w h atev er you have to do?
I (a) Always? N ever?
t
i
26. Do you have the feeling th a t TIME is passing too RAPIDLY for
i you to ACCOMPLISH all the th in g s th a t you THINK you should GET
i DONE in one day?
! (a) Do you OFTEN feel a sense of TIME URGENCY o r TIME
j PRESSURE?
T hat completes the Interview . T hank you v e ry m uch.
Closure: T h is completes the interview of Subject (give code
numbers)."
I
i
I
105
SCORING
i TAPE NO_____
j READING_____
i SIDE: count:
HANDSHAKE:
, ATTITUDE:
i
! GENERAL
i APPEARANCE
i
I
i MOTOR
PACE:
SPEECH
, HURRYING:
i
VOICE
QUALITY:
RHYTHMIC
MOVEMENTS:
■ HANDS/FEET
!
I
I FACIAL
' EXPRESSION:
LAUGHTER:
FIST
CLENCHING:
P articip an t ID _
Date__________
BEHAVIOR PATTERN INTERVIEW
"limp wrist"
moist 6c cold
friendly
calm
slow
none
subdued
none
flat
none
w eak average strong
average w arm 6c dry
n eutral
nervous
average
none
average
occasional
expressive
"round
mouth"
rare
hostile
tense
h u rrie d
a le rt
occasional fre q u e n t
strong
fre q u e n t
lateral
m outh
speech
lateral
smile
clenched
jaw
occasional fre q u e n t
, SIGHING: none occasional freq u en t
106
TYPE A BEHAVIOR STRUCTURED INTERVIEWER NOTES
SETTING and EQUIPMENT
It is suggested th a t the in terv iew be conducted in a quiet
setting w ith the interview er and subject sitting across the corner
of a table or a desk from each other, approxim ately 2 1/2 feet
apart, in a business-like setting The attitude of the in terv iew er
IS business-like ra th e r th a n therapist-attentive.
A rem ote lapel m icrophone [Sony ECM 150 condensor w ith
plug] and a quiet portable tape reco rd er [Sony TCM 5000 EV] are
recom m ended. M ichael H ecker suggested th a t the equipm ent be
obviously visible to the subject.
Interview s are recorded on 30 m inute cassette table, one
in terv iew to each side [TDK DC6 c s s 99], Advance th e tape by
hand th ro u g h the leader before insertin g in the m achine. Tapes
m ay be pre-labeled, or labeled im m ediately follow ing interview .
Tapes should begin and end w ith subject's code num ber.
INTERVIEWER
The in terv iew e r is to m aintain a "Type A 2“ b ehavior d u rin g
the interview , focusing during the initial questions on establishing
rap p o rt w ith the subject, p rim arily th ro u g h the use of the voice,
supplem ented by eye contact and responsive facial expression.
Establishing rap p o rt in the early p art of the interview is im portant
as some of the questions are v ery personal and re q u ire a degree
of confidence in the interview er.
The in terv iew er may have to v ary style slightly in the initial
stages of the in terv iew in order to establish ra p p o rt w ith a
subject.
An am bient, n a tu ra l stressor is provided by the in terv iew e r
in the situation, by response to the subject w ith a le rt Type A2
behavior, consistently picking up on the inform ation provided by
th e subject.
The in terv iew e r w orks to m aintain a balance betw een
following the script and m aintaining rap p o rt w ith the subject
th ro u g h follow up responses to the subject's statem ents
The scrip t is a guideline, a point of departure fo r spontaneous
questioning and probing w hen the subject has taken "bait.''
However, it is im p o rtan t to cover all points in the script, and to
stay w ithin the time limit.
107 |
The in terv iew er creates a v ariety of events re q u irin g response ;
from the subject: mild challenges [bait] w hich will show variations j
in A and B type responses
The design is to elicit exam ples of the b ehavior p attern as
I reflected in the inform ation the subject reveals in describing
| h is /h e r responses in habitual situations in the subject s daily j
I activities. The interview er
1 rem ains a le rt to the evidence of Type A behavior, is ready to
i capitalize on the inform ation, and to develop the subject s
1 description of the behavior response through probing, relevant,
spontaneous follow up questions concerning the particular
j situation w hich elicited the response in the non-interview
environm ent. The in terv iew er needs to 'open' questions to elicit
! m ore A o r B type behaviors [generally speech/audio behaviors],
j The Type A 2 b ehavior of the in terv iew e r m aintains the pace
| and controls the progress of the interview , and aids in sustaining
j volume control in in terv iew er and subject,
| Some subjects may feel sufficiently challenged by the m an n er
j of the interview and the mildly challenging dem eanor of the
| interv iew er to exhibit Type A behavior in interactio n w ith the
! interview er. It is im portant th at the challenge of the interview
f situation, and the in terv iew er's challenges not cause the subject to
w ithdraw from the interview The in terv iew er guards against
, fostering a Type A ch allenger/opponent situation,
I
| THE INTERVIEW
I The W estern Collaborative Study is the "gold standard" for the
i Structured Interview .
The in terv iew er listens actively, w ith a "third ear."
The sc rip t is a guideline, a point of departure fo r spontaneous
i questioning and probing w hen subject has taken "bait.” However,
j it is im portant to cover all points in the script, and to stay w ithin
the time limit.
j The goal is to elicit examples of behavioral response in
I particular environm ents, to provide as ric h interp retatio n of
' behavior as possible in 12-15 m inutes.
( The in terv iew er establishes rap p o rt in th e early questions,
j T hen moves to m ore patterned' behavior. Questions 3 & 4 are
I im portant in establishing relationships and rapport.
108
"Looking for work, is w o rk '. V olunteer work, is w o rk '."
Activities su rrounding most purposeful reg u larly scheduled
activity can provide w ork-related inform ation.
A nger-in and an g er-o u t assessm ents [a p articu larly im portant
area of Type A behavior research ] are facilitated by questions: 3 >
6, 18, 22 The in terv iew e r needs to be aw are of the im portance of
these questions w hen conducting the interview in order to in su re
th a t answ ers are probed, to rev eal the subject s coping and
responding styles.
TYPE A BEHAVIOR and ASSESSMENT
The dichotom y of Type A/Type B b ehavior reflects m edical
authoritarianism . A lthough Type A behavior patterns are
expressed on a continuum [Al, A2, X, B3, B4], CHD-proneness may
be dichotomous. Hostility and im patience m ay be m ore im portant
in the cluster of Type A behaviors. However, hostility is not an
essential com ponent of even Type Al behavior; a subject may be
Al o r 2, and n o t be hostile.
It is stressed th a t Type A behavior is a response to perceived
environm ental events.
In assessing Type A b ehavior it is im p o rtan t to separate
anxiety from tru e Type A behavior. Anxiety may give the illusion
of Type A response. It is im p o rtan t to "peel away" the
"externalized" anxiety to get to the real behavior.
TYPE A TYPICAL BEHAVIOR IN CHALLENGE SITUATIONS AND
DURING THE STRUCTURED INTERVIEW:
Vigorous motor activity, especially speech:
h y p er alertness
loud speaking voice
energetic syllable em phasis
ab ru p t single word responses
dynam ic rate of speech
acceleration of speech rate at the ends of breath groups
special alertness w hen listening: eager to "come in"
v ery fast response time in answ ering questions
rapid b reath in g before speech
109
C ontent-related behavior:
com petitiveness fo r control of situations, others, or of interview
e g. rev erses role by asking questions in response to questions,
talks excessively and does not allow in terv iew er to in te rru p t in
order to continue the interview , subject in te rru p ts in terv iew er
hostility and aggressiveness toward others or in terv iew er and
interview situation
e.g. asks in terv iew er to "define that", fault-finding w ith the
questions, disagrees w ith word choice of questions
evidence of h y p e r alertness [during interview ] and w ithin
inform ation given during interview concerning behaviors in
w ork and daily activities
note w h at kinds of events cause arousal in the subject
condescending attitude toward others or interview er: discount
others' opinions, beliefs, behaviors
time urgency: m aximal use of time: tig h tly scheduled days,
m ultiphasic activities, economy of word usage in answ ers:
e.g.INTERVIEWER: Do you have children?" SUBJECT: "Two. Four
and six."
em otionally disconnected
complaints: large, or sm all and profuse
The protocol in the pages above are notes the investigator took
during tra in in g fo r the Structured Interview at SRI w ith M argaret
Chesney and M ichael Hecker.
110
i
I
A PPEN D IX B
INSTRUCTIONS FOR THE M ILITARY COMMANDER MONOLOGUE
i
| Following are the instructions to th e interv iew er used during
j the recording sessions of the M ilitary Commander Monologue
i Instructions are included for taking blood pressure readings w hich
| w ere not used in the p re se n t study
; Instructions to the in terv iew er:
I Please check all recording settings, m icrophones, etc.
! Instructions to the subject:
j The in terv iew er says;
I "Please read the monologue to yo u rself silently, u n til you are
I sure you can read it aloud w ithout stum bling Your reading of the
j m onologue w ill be recorded two separate times. W hen you are
« ready to read it aloud let me know."
i
I Instructions to the in terv iew er:
Using the m anual control, take th e subject s first blood
i pressure reading at the beginning of the silent reading. The
) m onologue reading time is about 45 seconds w hich m eans the blood
pressure reading needs to be taken v e ry n e a r the beg in n in g of the
silen t reading. If we compute two 45 second readings and a little
! ex tra study time, th e two m inute in te rv a l w ill be covered by the
| end of the first oral reading of the monologue.
I
! Instructions to the in terv iew e r:
! For the firs t reading start the tape recorder before the subject
; begins reading.
i
! Instructions to the subject:
; The in terv iew er says:
Please read the monologue aloud, preten d in g th a t you are in
J your own home, alone."
! I l l
Instructions to the in terv iew e r:
Using the m anual control, take the subject's second blood
pressure reading at the fin ish of the first reading. Make sure th a t a
two m inute in te rv a l has passed since the previous blood pressure
reading. Hopefully, the time in terv al w ill have passed w ithout
fu rth e r waiting.
♦
j Instructions to the subject for th e second monologue re ad in g :
! The interview er says:
| "For the second reading I'd like you to im agine th a t you are
; the officer on the battlefield talking to y our troops. If you'd like,
you may read the monologue silen tly again. I w ill tell you w hen to
, begin reading aloud."
I Instructions to the in terv iew er:
! Wait u n til 1 m inute and 15 to 30 seconds have passed since
| the previous blood pressure reading before giving the signal fo r the
i subject to begin the second oral reading.
i
; Instructions to the subject:
' The interview er says:
j "You may begin the second reading."
r
\
! Instructions to the in terv iew e r:
j Using the m anual control, take the third blood pressure
| reading at the fin ish of the second oral monologue reading th an k
the subject for h is participation in the readings, etc. T urn off the
j audio recording equipm ent, etc.
\
t
r
112 i
M ilitary Commander Monologue I
l
The copy of the reading given to subjects during the recording
j sessions was in la rg e r p rin t and was not titled. :
i '
; This is the w ay th a t you and me, ev ery God dam ned one of us
i a re going to lick the h ell out of w hoever stands in our way. And I
I don't give a damn w h e th e r you like w h at I'm telling you o r not.
I This is the way I say it's got to be done. First, w e're going to sm ack
them h ard w ith m o rtar fire, understand? I w ant you to p o u r it on
them! Let the bastards feel it get hot, re ally h o t around them .
Singe the h e ll out of them! Scorch the bastards, fry them , burn
, th e ir guts out. Make ashes out of them .
I A fter the m ortars, I’ll tell you w h en to advance. And w h en I
give the signal, don't crawl, you ru n forward! Remember, it's y our
! skin or theirs! All rig h t, enough talk, now let's get the lead out of
| o u r p ants and get going. Hey! One m ore th in g , good luck!
I i
THE SIX PHRASES SELECTED FOR ANALYSIS '
i 1. Singe the h ell out of them!
[ 2. Scorch the bastards
3- Fry them
4. B urn th e ir guts out j
5. Don't craw l
6. Hey! One m ore th in g , good luck!

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Cognitive predictors of intention -behavior consistency in safer sex practices

PDF

A behavioral comparison of the nonverbal and verbal communication of distressed and nondistressed marital couples

PDF

Correlates of deceit: A cross-cultural examination

PDF

Communication rules and perceived outcomes within and between domestic culture groups: A comparison of intracultural and intercultural conversations among acquaintances

PDF

A structural equivalence and contingency theory perspective on media usage and communication performance: The case of voice messaging

PDF

Beyond skill development: The effects of training and development on the attitudes and retention of employees

PDF

A descriptive study of the dramatic function and significance of the clown during Hopi Indian public ceremony

PDF

Construction validation of the Vocal Behavior Inventory

PDF

A ship in a bottle: How communication builds the lifeboats of the world's largest A.A. group

PDF

Attachment as a predictor of self -monitoring and sexual behavior

PDF

Contemporary campaign communication: a case study of a Nevada state senate race

PDF

A social cognitive approach to the nature of input processes in reception of nonverbal messages

PDF

A study of the "eastern" writer in Hollywood in the 1930's

PDF

A description of the development, administration, production process, and production environment of the public television dramatic series: "Hollywood Television Theater"

PDF

Communication and the implementation of information systems: The case of word processing in state and local governments

PDF

A critical study of the screwball comedy film, 1934-1941

PDF

Communication strategies in barricade-hostage confrontations: Theory, research, and police experience

PDF

Communication, consensus and commitment in collective action organizations

PDF

A system of participation and allocation preferences in organizations

PDF

An analytical study of the dramatic criticism of Joseph Wood Krutch as published in "The Nation", 1924-1952

Asset Metadata

Creator Mischer, Beverly Meyers (author)

Core Title Acoustic predictors of Type A behavior.

Degree Doctor of Philosophy

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag OAI-PMH Harvest,psychology, behavioral,Speech Communication

Language English

Contributor Digitized by ProQuest (provenance)

Advisor Shoup, June (committee chair), Davison, Jerry C. (committee member), Purcell, Edward (committee member)

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c17-719501

Unique identifier UC11344164

Identifier DP22451.pdf (filename),usctheses-c17-719501 (legacy record id)

Legacy Identifier DP22451.pdf

Dmrecord 719501

Document Type Dissertation

Rights Mischer, Beverly Meyers

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the au...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus, Los Angeles, California 90089, USA