Title: Children's judgments of personality on the basis of voice quality
Full Citation
Permanent Link: http://ufdc.ufl.edu/UF00097773/00001
 Material Information
Title: Children's judgments of personality on the basis of voice quality
Physical Description: ix, 84 leaves : illus. ; 28 cm.
Language: English
Creator: Phillis, Judith Ann, 1942-
Publication Date: 1969
Copyright Date: 1969
Subject: Child psychology   ( lcsh )
Personality   ( lcsh )
Psychology thesis Ph. D   ( lcsh )
Dissertations, Academic -- Psychology -- UF   ( lcsh )
Genre: bibliography   ( marcgt )
non-fiction   ( marcgt )
Thesis: Thesis--University of Florida, 1969.
Bibliography: Bibliography: leaves 82-84.
Additional Physical Form: Also available on World Wide Web
General Note: Manuscript copy.
 Record Information
Bibliographic ID: UF00097773
Volume ID: VID00001
Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
Resource Identifier: alephbibnum - 000559210
oclc - 13447569
notis - ACY4658


This item has the following downloads:

PDF ( 3 MBs ) ( PDF )

Full Text






Copyright by
Judith Ann Dhillis
19 69

To my parents

Ackn owr le dgmo nts

The author wishes to express her sincere gratitude

for the assistance and guidance so willingly provided by

Dr. James C. Dixon, Dr. Norman N. Markel, Dr. Henry S.

Pcnnypacker, Pr. Madelaine Carey Ramey, Dr. Audrey S.

Schumacher, and Dr. Vernon D. Van De Piet, the members of

her doctoral committee.

The author is particularly grateful to the following

cor.uaittee riembers:

Dr. Norman NI. Markc-l provided extensive assistance

and cncourargement with regard to both theoretical and practical


Dr. Ma&delaine Carey .eamey's advice concerning

matters of statistical design was most valuable.

Dr. James C. Dixon was a patient mentor through

four years of graduate study.

The author also gratefully acknowledges the assistance

of the following;:

;.Mrs. Ruth Duncan arranged for the use of the facilities

of the P. K. YonIg Laboratory School, and the teachers and

students were extremely cooperative.

Mrs. Will Selfrige devoted maany hours to the

pr-paration of the ex-erimental tapes.

The: facilities of the Univerit, CC::puting Center

and Mrs. Mary Lynch, consultant, made the analysis of the

results a much simpler task.

Miss Kathy Wright provided much appreciated

assistance in the tabulation of the data.

Miss Pat Powers carefully proofread the manuscript

and Mrs. Charlotte Gregory typed it professionally despite

the pressures of time.


List of Tables...........

List of Figures..........

Chapter I Introduction.

Chapter II -- ethod.....

Chapter III Results....

Chapter IV Discussion..

Chapter V SuFJrary......

Appcndi.-, A. .............
Appendix B.. ............

List of References.......


















Table 1 Frequencies and Major Factor Loadings
of the Adjective Scales .......................

Table 2 Original and Direct Magnitude Estimation
T-scores ........... . .............. ..........

Table 3 Means and Standard Deviationr of
Percentile Placenent on IAbility Tests
for the Twelve Subject Groups.................

Table 4 Means and Standard Deviations of Subject
Age in Mionths for the Twelve Subject
Groups ........... ... ............ .............

Table 5 Subjects Discarded from the Twelve Subject
Groups.......... ... ...... .. .................

Table 6 Analysis of Variance of Mean Evaluative
Ratings .....................................

Table 7 Mean Evaluative Ratings of the Truelve
Voices b,' the Males and the Females and
by the Three Age Groups.... .................

Table 8 Mean Evaluative Ratings of the Six Voice
Types- by the Males and the Fenales and
by the Three Age Groups........................

Table 9 Mean IEvaluctive Ratings of the Three Voice
Quality Profiles by the Maleis and the
Females and by the Three .ge Groups............

.able 10 Analysis of Variance of Mean Dynamisn
Ratings ............. ................. ......

.able 11 Mean Dynvr-aism Ratings of the Twelve Voices
by the Malj.es and the Females and by the
Three Age Grcups ............. ..........

able 12 Significance Levels for the Newman-Keuls Tests
on the Dynatmisrn leans of the Voices of the
Same Voice Type...............................

?able 13 Mean Dynamiisn Ratings of the T.ix Voice Types
by the Males and the Females and by the
Three Age Groups ..............................

















Table 14 Mean Dynamism Ratings of the Three Voice
Quality Profiles by the Males and the
Females and by the Three Age Groups ............ 56



Figure I The Sex X Voice Interaction of the
Evaluative Analysis ............... . . ..

Figure 2 The Age X Voice Interaction of the
Evaluative Analysis. ........................

Figure 3 Schematic Presentation of the Newman-
Keuls Tests on the Ordered Voice Evaluative
Me ?ins ........... .. . . ... ......*.

Figure 4 Plot of the Voice Type Means at the Two
Levels of A of the Evaluative Analysis......

Figure 5 Plot of the Voice Type Means at the Three
Levels of B of the Evaluative Analysis......

Figure 6 Plot of the Voice Quality Profile .leans at
the T\'o Levels of A of the Evaluative
Analysis ................. . ... ...... .. .

Figure 7 Plot of the Voice Quality Profile Means; at
the Three Leve z of B of the Evaluative
Analysis ....... ... ......................

Figure 8 The F'ex X Voice Interaction of the
Dynanisn. Analysis. .......... ..... .........

Figure 9 The Age X Voice Interaction of the
Dynamism Analysis. .. . ........

Figure 10 -. Schematic Presentation of the Ne.mnan-IKeuls
Tests on the Order-ed Age Group Dynamism Means.

Figure 11 Schematic Presentation of the Ne(.wman-Keuls
Tests on the Ordered Voice Type Dynamism
dMeans........ . .......... .......

Figure 12 Plot of the Voice Type M'eans at the Two Levels
of A of the Dynamism- Analysis...............

Figure 13 Plot of the Voice Tyve Means at the Three
Levels of B of the Dynam.ism, Analysis........

Figure 14 Plot of th-e oice quality Profile Means at tlhe
Two Levels of A of the Dynamism Analysis....

Figure 15 Plot of the Voice Quality Profile Means at the
Three Levels of T of the Dynamismri Analysis.,

















Chapter I


Voice and rersonali ty

The manner in which h something is said, as well. as

actual content, has long been recognized as extremely important

in determining the message communicated. The American

cowboy showed his belief in the voracity of this statement

with the phrase, "Smile when you say that, Partner." Vocal

variations, as w.:ell as gestural and facial ones, can also be

employed to modify communicated messages. In addition, to

the astute listener, i.nfr:oration .which is not necessarily

communicated intentionally is ava,'.ilable from _araling:iJtic

aspects of the voice. Thus, by his manner of p.ea.kl!g- a

lecturer rnay well infoc'r. his audience of his present mooc

and geneJral personality characteristics as ul.l as the topic

on which he is speaking.

Despite the general agreen!- nt that the voice

expresses both transient ard relatively permanent states of

the individual, it wcs not until capir's article, "Speech

as a Personality Trait'; (1927) appeared that systematic

research in the area Ibearn. Classic studie.. in the area

are these of Pear (1931) and Allpcrt and Cantril (AUport

and Cantril 1934; C'i trial and Allp.ort, 1935). Findings

were that age, sex and total personality s. tches could be

reliably matched with voices by the judges, In genera].,

however, in matching many aspects of personal ty with voices

the inter-judge reliability was much higher than the validity.

This finding has been confirmed by recent research (Kramer,

1964). Kramer (1964) argues that the consistently high

reliability in such judgments is itself a ty.e of validity,

and that the lack of agreement with criter.cion personality

measures should cast doubt on the validity of these measures

rather than on the validity of the voice as an indicator

of individual differences.

In addition to the- need for validity in the

criterion measures of personality employed, Kramer (19S3)

suggests that the personalities of judges is a variable

which should b. investigated in voice and personality

studies. In view of the fact that "..,certain basic

personality differences are a function of biological sex,..

Markel and ,oblin (1965) investigated the effects of this

variable, Differences in personality judgmrents were obtained

on the evaluative dimension of a semantic differential as a

function of sex-of-judge. These differences occurred as a

function of content variations. In the Markel and Roblin study

all voice samples were provided by the same speaker. Employing

this procedure, no sex differences ; ere found -with regard to

the activity and potency dimensions.

Another critical factor in these studies is the

specification of the voices remployed. Market (1969 a)suggests

the use of the terms "structural" and functional" to describe

the methods employed in analyzing speech. lIe notes that

the functional approach consists of various types of rating

techniques in which personality judgments a::e made directly

by judges. These studies have been reviewed by Kramor (1963).

The structural methods are. those which employ some technique to

quantify the speech without attempting to make judgments

about the speaker. Studies employing the latter methods have

been reviewed by Mahl and Schu!ze (1964). Dittmian and "'ynne

(1961), in an attempt to develop a structural methodology,

employed the typical idiographic approach of the linguist

and were unable to obtain reliable ratings of paralinguistic

phenomena. Utilizing a nomothetic approach (comparing the

individual to group rather than focusing on individual

variations through time) Markel (1965) obtained high inter-

judge reliability in the coding of three aspects of

paralinguistic cr non-phonmnic phenomena: pitch, loudness

and tempo.

Children as Judges in Voice and Personality Studies

Kramer (1964) has noted that while there has been

much theorizing about the importance of develoonental variables

in the areaoE speech and personality, few empirical studies

have been conducted. The primary contention in the area has

been that young children ar-e more sensitive to nonverbal

vocal cues than are adults. Dimitrovsky (1964) found sex

and age differences in children's ability to judge emotions

on the basis of voice. In general, however, older children

and femi.ales were more accurate in their judgments. Gates (1927)

in an earlier study, found that rore intelligent and older

children correctly identified the intended emotions of

speakers more frequently. Goodenough's (1957) findings that

an interest in ot-her people is an extremely critical sex--typed

behavior forgirls while boys are encouraged to develop an

interest in objects is important in this regard. Especially

relevant in predicting differential accuracy in judging

personality from the voice is the finding that girls, more

than boys, are encouraged to attend to the expressive

behavior of others and utilize the information obtained

in tha manipulation of people. Thus, particularly when

children are employed as judges, the sex and age of judges

are important variables in the area of voice and personality.

The Voice Quality Profile

MIarke.'s develomment of the Voice Qualitv Profile

(VQP) resulted from his attempts to anE3yze voice qualities

structurally rather than functionally. The advanLage of such

an approach is that the structural aspects of voice quality

can then be related to a variety of both transient and

relatively permanent personality characteristics. As

mentioned above, the voice qualities on which reliability of

ratings has been established are pitch, loudness and tempo.

In order to determine the VOP of a particular speaker, a

group of judges rates one or several voice samples of the

individualin- terms of these three dimensions. The mean of

all ratings obtained on a dimension is the score of that

dimension. The scores for all individual voices rated by

this techniqueare then converted to T-scores, one distribution

of T-scorer; for each dimension. Three profile groups are

then established and each individual is assigned to one on

the basis of the highest T-score obtained. The three VQP

groups are Peak-Pitch (P-P) ; Peak-Loudness (P-L) and Peak-T:c'po


With this technique, the VQP's of psychiatric

patients (Markel, 1969a) and college students (Costanzo,

Markel, and Costanzo, 1969; Vargas, 1968) have been obtained,

Raters have always been graduate or undergraduate university

students. The VQP has been related to ':MPI profiles by

ilarkel (1969a) and to perceived crotion by Costanrzo, Marke.

and Costanzo (1969). Vargas (1963) has related the VOP

to the frequencyi of positive cxperiencing and behaving, the

California Personality Inventory, the T'ennc.-see elf[--Conce:t

Scale, FIRO-L, self ratings of early childhood experiences,

Jourard's Self-Disclosure Questionnaire and judge-rated

self-disclosure. Considering the pattern which emerges from

all three studies, Vargas (1968) hypothesized a two dimensional

trait system to describe the personality differences among

the three VQP groups.

The mean ;:MPI profiles ('-arkel, 1969a) of the

three VQP groups indicate that P-P voices are associated with

personality descriptions which include such adjectives as

"withdrawn" and "introverted." Psychiatric patients with

the P-P VQP are likely to have a poor Tprocnosis and be diagnosed

as psvchotic. The P-L VOP among psychiatric patients is


indicative of a good prognosis and a diagnosis of depression

or psychoneurosis. Personality descriptions of such people

would likely include such adjectives as immaturee" and

"demanding." The P-T VQP is associated with an .MMPI profile

within normal limits, and thus, on the basi:f of the Il--PI

study, a personality description of healthy or average is

suggested. The Costanzo, Markel and Costanzo (1969) study

indicates that among college students P--P voices are more

often associated with grief, P-L with anger or contempt and

P-T with indifference regardless of the emotion intended by

the speaker. This appears to support Markel's findings

(1969a), if one considers that depression is often viewed as

a means of dealing with anger and a grieving person is

characteristically withdrawn. Notice that these are the

findings of the two studies for the P-L and P-P voices,

respectively. The relationship between indifference and

normalacv is less easily understood; however, this may

merely indicate that none of the other emotions was clearly

portrayed. Essentially Vargas' (1968) study is in agreement

with the previous two. lie found that speakers whose VP

was P-T were significantly higher than the P-P and P--L

speakers inself--disclosure, the reported frequency of

positive experiencing and behaving and the number of positive

early childhood experiences reported. All of these variables

are associated with healthy, adjusted people. The Vargas

study also indicates that the P-L VOP is associated with

speakers who feel inadequate and dissatisfied and who are

maladjusted in interpersonal relationships according to various


scales of the Tennessee Self Concept fcalc. Scores on FIRO-B3

indicate that the P-L VQP among college males is charectcristic

of people who wish to control others but w-ho are unwilling

to give in return. FIRO-B scores also indicate significant

differences between the D-P speakers and the P--L and P-T speakers

in that the latter two groups seek to gain control. in

interpersonal relationships.

Thus, it appears that a consistent pattern has

emerged from studies investigating personality and the VQ'0.

The following is a summary of the behavioral correlates of

the three structurally defined VOP's-

(1) P--P withdra,.n and reserved but capable
of maintaining mutuc.ll rewarding inter-
personal relationships with other people

(2) P-L *- angry, demanding and seeking to
control others but not giving in inter--
personal relationships

(3) P-T outgoing and seeking to manipulate
others but also will ing to give to others
and able to maintain mutually rewarding
relationships with them.

Selection of the RPating Technicue

The problem of choosing a rating technique is a

particularly difficult one when children are to be the

raters. They are notorious for the unreliability of their

ratings as well as for their tendency to adopt favorite

response styles regardless of the stimulus properties to be

judged. It is, of course, possible to control for such

difficulties by employing either a ranking or a paired-

comr.orisons! method. However, neither of these approaches

lends itself well to situations such as the present one,

where the to-be- raLrte stimuli are transi.-.nt and where it is

impossible for the child to have more thap one stimulus

before him at a time. Although this is a less serious

drawback for the paired-comparisons approach, where only

two stimuli are considered at once, the time required to

obtain paired-comparison data for twelve stimuli would be

prohibitive in a situation such as the present one where

ratings on two dimensions are required.

The Semantic Differential

The Senantic Differential (Osgood, Suci and

Tannenbaum, 1957) is a rating technique which provides

simultaneous ratings on several dimensions of stimuli.

The instrument was developed and tested extensively by

Osgood et.al. as a technique to be employedi3 in the measurement

of connotative meaning. Semantic differential studies

involve obtaining, from a number of subjects, ratings of

several concepts or stimuli on a series of bipolar adjective

scales. The consistently relevant dimensions of meaning

have been determined by means of factor analyses which

have been computed on the intercorrelations of the adjective

scales. In the typical semantic differential study, the

intercorrelations are calculated across stimuli and subjects.

Such studies have consistently found that most of the common

variance is accounted for by three dimensions or factors:

evaluative (50% 75%), potency (25%--30) and activity

(20%-25%) (Osgood, Suci and Tannenbaum, 1957). Typical

evaluative scales are good-bad and friendly-unfriendly;

strong-weak and hard-soft are typical potency scales, and

the activity dimension consists of scales such as fast-sl.ow

and active-passive.

Previous experiments have employed the semantic

differential to obtain personality ratings on the basis of

voice from adult judges (Markel, Meisels, and lHouck, 1964;

Markel and Roblin, 1965). Markel, Meisels, and Houck (1964)

found that on the basis of voice alone schizophrenics were

judged more potent than nonschizophrenics. Speakers whom

the judges perceived as schizophrenic wecre judged more

active than those perceived as nonschizophrenic. The authors

interpret the failure to obtain differences on the evaluative

dimension of the semantic differenital as attributable to the

constant content enploy',d by all speakers and to the similarity

of voice set. They agree w.itn Trager (1958) in defining

voice set as those aspects of the voice attributable to

societal group, sex, age, state of health and location.

Markel and Roblin (1965) obtained differences on the evaluative

dimension, but not on the activity and potency dimensions

when judges were instructed to rate the personality of one

speaker in three different content conditions, The sex-of-

judge was also found to have a significant effect. The

authors concluded that the evaluative dimension reflects a

general attitude toward the speaker which results from the

relationship between his voice set and what he says. The

significance of sex-of-judge was attributed to sex differences

in responding to inconsistency between voice set and content.

Children as subjects

Studies which have investigated the dimensions of

meaning in children of various ages support the contention

that it is a fairly consistent variable (Small, 1950;

Brandwin, 1965; Lilly, 1965; Darnard, 3966; Di Vesta, 1966).

It has been noted in several studies, however, that the

activity and potency dimensions tend to coalesce when children

are employed as raters (Small, 1958; Brandwin, 1965; Barnard,

1966). Maltz (1963) concluded that the meaning of concepts

to children becomes stable sometime before the fourth grade

in school is reached. Donahoe (1961) investigated the stability

of the three dincnsions separately and concluded that

ratings with respect to acti.\ ity stabilize before, or during,

the first grade in school; evaluative ratings stabilize

prior to the third aznd pcLtency ratings between the third, and

sj::th years in school. However, some caution must be employedd

in interpreting these latter two studies, since both assumed

that the factor loadings of the scales found with adult

subjects would also be obtained with children used as subjects.

The findings of Di Vesta and Dick (1966) with respect to

the reliability of the semantic differential at different

age levels do seem to indicate that a general increase in

the stability of children's semantic differential ratings

occurs between the third and fourth grades. They found

"...a definite increase in reliability of ratings made on

the semantic differential between the third and fourth

grades." Di Vesta and Dick conclude that if twenty children

are employed in the rating procedure, the reliability of the

concept scores obtained should be satisfactory. They

also report that, "All correlations are sufficiently high to

indicate the functional utility of separate factor scores,

based on only two scales for research purposes."

As is typical in studies in which the semantic

differential is used to obtain ratings, it was necessary to

consider several variables in choosing the scales for the

present study. The criteria employed in selecting scales

were as follows:

(1) All scales included were selected on the
basis of their relevance to making ratings
of the personalities of nale speakers.

(2) The scales employed were constructed of
adjectives with which children could be
reasonably expected to be familiar.

(3) Tho scales were s-elected in a manner
which assuied that they loaded consistently
on the dinmenision which they were selected
to represent.

With respect to the third criterion, it was first necessary

to determine which dimensions cf meaning were relevant to

the present study. As was mentioned previously, in most

semantic differential studies, the first factor or dimensions

to emerge have been evaluative, potency, and activity.

Later factors have been given such names as oriented activity,

stability, receptivity, pointedness, and novelty. It has

already been noted that when children are employed as

subjects in semantic differential studies the activity and

potency dimensions merge into one second dimension which

has frequently. boon labeled dynamism. Osgood, Suci, and

Tannoinbaum (1957, pp. 73--74) observe that the activity and


potency factors also coalesce into a dynarmism factor when

judgments are limited to sociopolitical concepts, that is,

people and policies. Since the present study required

children to make ju-dgc'-ens of people exclusively, it was

concluded that evaluative and dynamism were the most critical


Scales were therefore selected for inclusion in

the present study if, in previous studies, they had been

found to load consistently on either the evaluative or the

dynamism factors and to have small loadings on all other

factors. IP order to assure that the children would be

familiar with all adjectives so selected, the word frequencies

established by Thorndike and Lorge (1944) were utilized.

Thorndike and Lorge recommend that words as frequent as

twenty per million be taught to fourth graders, thus children

of this age can be expected to be familiar with words of

this, or greater, frequency. Table 1 lists the six pairs

of bipolar adjectives employed in the present study as wiell

as the frequency per million words of the selected adjectives

according to the Thorndike and Lorge Word List and the major

factor loadings of the scales according to previous studies

which employed children as subjects.

Statement of the Problem

The purpose of the present study was to test

directly the hypotheses emerging form the work with the

VQP. It appears that the VQP groups which have been defined

by a structural methodology can also be defined in functional


rHl r-v t0 r- * *

43 C) 4 J E C
E) > > HQ -, >-

S4) ,-) 4 a a
tr 0 C

C > 4J > U
,- > 0 40 0 0 4J C
0 r -i r 0) H 0 0-
S0 0 0 > 4 *rl > 0
0 0 *1 > > I rO 4 *rl 0O

3 0 0 4 0
': O U K O

< ; 00

C0 C 4
4*- 1 C* * C
- * 4

l o ro > o
0 10 1


4' 0 00
4-' 0 * C 0 C CC
0 r- *-, *n-r 4-qJ

4 Q 00
0 Oi C O O O ,

0 44
oi H.-,.'" -4 --
0 -, rJ 0 0 -, 0 0
00 00 10 0


3O O

I 1- C I

C) + + c
"J .0 0 0 o C, o a
= ci- C. 0 0 1 L 0 0 !-

0 0 0 C

V.O 0
0 H *H I H

L C I9 C CI _
1o C I U

4 0 0( 0(
Hrj 0 M '
0 0 C

0 r o

or: r

terms. Stated simply, if judges were asked to rate the

personalities of speakers in each of the three VQP groups

on the basis of voice quality alone, the personality

descriptions obtained would be expected to differentiate

among the three VQP groups by means of two behavioral or

personality dimensions. The first dimension might be

thought of as reflecting "aggressiveness" and includes those

adjectives in the first phrase of the descriptions of

personality correlates of the three VQC's. The second

dimension includes those adjectives in the second portion

of the descriptions of personality correlates of the

three VQP's and might be summarized as "friendliness."

It was also believed that investigating the effects

of low scores, as well as peak scores on the three qualities,

would contribute significantly to an understanding of the

relationship between voice and personality. Therefore,

this variable was included in the design of the present

study. In order to obtain additional information on the

effects of individual differences among the judges, which

Kramer has suggested as a critical variable, males and

females at three age levels were employed as raters in the

present study.

It should he noted that the evaluative and dynamism

factors of the semantic differential are quite similar to

th', two dimensions emerging form the studies of personality

and VQP. For example, dimension 1 was characterized by

adjectives such as "reserved" and "withdrawn" at one pole

and "dc-mandinc," "control ing," and "outgoing" at the other pole,

These adjectives appear to --epresent the meanings attached

to the two poles of the dynamism dimension. Dimension 2

of the VQP personality descriptions included the ability to

give and receive pleasure in relationships with people at

one pole and dissatisfaction with human relationships at the

other pole. It seems that ratings of people at the poles

of dimension 2 would be represented by the poles of the

evaluative dimension. For example, friendly unfriendly

is a frequent evaluative scale and in the discussion of

dimension 2 it was referred to as a friendliness dimension.

On the basis of previous studies which investigated

the relationship between personality and the VQP, the

following hypotheses were framed and tested in the present


(1) Speakers rated as HT and TIP on VQP
will be rated as more positively valued
by judges rating personality than speakers
rated as !IL on VQP.

(2) Speakers rated as IIT and HL on V(OP will
be rated as more dynamic by judges rating
personality than speakers rated as HP on

(3) Speakers whose VOP's are similar will
also receive similar ratings on both the
evaluative and the dynam'isr dimensions.

The results of the Dimitrovsky (1964) and Gates

(1927) studies indicate that the above hypotheses are more

]ikely to be confirmed among the female and the older

subjects. Despite the empirical findings, however, the

contention that younger children discriminate anong nonver)bal

vocal cues better than do older children continues to

survive. One reason for this is the fact that the methods

employed to obtain th chiJdre-n's ratings were not extensively


documented for the age groups involved. It was one of the

purposes of this study to investigate the effects of sex

and age in judging personality from the voice.

Chapter II


The Voices

The stimulus voices were obtained from raw data

previously collected by Markel (19G9a). In this study, each

of six raters judged the pitch, loudness and tempo of seventy-

eight voice samples twice. The original voice samples were

obtained from psychiatric patients who were instructed to

read the following sentence, "Many of the reports received

last week showed a large increase in earnings.' On the

basis of these ratings, Markel was thz'n able to divide the

initial seventy-eight subjects into three V)P groups: P-P.

P-L, and P-T. Each voice was assigned to the profile groupv

which represented the voice quality for which the highest

T--score had been obtained. These three groups differed

significantly from each other both in terms of VQP and MIIPI


A possible criticism of Markel's approach is that

it fails to evaluate the effects of extreme ratings on the

three qualities. That is, a voice nay be defined as "peak"

on a quality yet obtain a relatively low absolute rating on

that particular quality. In addition, the possible correlates

of low ratings on the voice qualities are not evaluated.

These considerations, and the need to exploree the effects of


voice qualities with a smaller number of stimulus voices,

led to the decision to select the voices for this study

in the following manner: two high and two low voices for

each quality were included. In order to be classified as

high on quality A, a voice was required to obtain a T-score

at least 1.5 standard deviations above the mean (T-score = 65)

on quality A and between plus and minus 1.0 standard deviations

(T-score = 40 60) on qualities B and C. The same requirements

were employed in establishing other high and low stimulus


In selecting the twelve voices from the original

seventy-eight, an attempt w;as also made to match the high

and low voices for a particular quality with respect to

ratings onther other two qualities. That is, the two High

Pitch (hP) voices were watchedd with the two Low Pitch (L')

voices wiLh respect to loudness and tempo. Since the

limited samDle from which the twelve voices were chosen made

this procedure difficult, the voices were divided into two

sets, with the voices which were more closely matched being

included in Set I. Market (1969h), in an unpublished study,

also obtained pitch, loudness and tempo ratings using a

direct magnitude estimation technique. The correlations

between the original and D.M.E. T-scores are above .85

for all three voice qualities. Table 2 presents the pitch,

loudness, and tempo T--scores of the two sets of six voices

which were obtained in the two rating conditions.

TA-3LE 2

Original and Direct Magnitude Estit-iation T--scores

Voice Set
Type Uumrber

Original T-scores

Pitch Loudness


D.P.E. T-score -
Pitch L.oudness T!-n n

46.09 49.24
50.69 47.47

49.76 54.55
42.42 48.35

72.74 56.33
35.06 50.2.3

70.90 41.26
33.22 40.38

47.00 70.50
41.49 35.06

51.60 67.84
53.44 34.18







49.52 56.78
42.06 50.83

47.25 54.49
40.77 46.71

73.16 56.78
41.10 55.41

78.67 42.58
39.80 39.92

45.31 75.10
41.10 37.01

50.49 65.48
56.32 36.63







------U--~YI-~----I-I ~)---~--mU-~----~-C--- ----~- ------~


Subjects for this study were students from the

University of Florida Laboratory School. Approximately

50% of the children enrolled in the school are children of

professors at the University of Florida. The remainder of

the students are drawn from the city of Gaincsville and

other communities in Alachua County. The professional groups

are highly represented in the parents of the children which

may account for the fact that, as a group, the children

in the Laboratory School score higher on various ability

tests than other children in the state.

Ability test scores were obtained from the school

records in order to balance the experimental groups with

respect to this variable. Tn addition, birthdates were

obtained for all subjects. Six of the experimental groups

made judgments on the evaluative scale and six of them

made dynamism ratings. The elementary school groups, t1.7o

groups of males (ELM) and two groups of females (ELF), ,were

composed of fourth and fifth graders; eighth and ninth

graders were sampled to obtain the junior high males (JRM)

and the junior high females (JRF); and eleventh and twelfth

grade males and females were sampled to obtain the senior

high males (SRH) and females (SRF).

Ability test mean percentile scores and standard

deviations are presented in Table 3. The percentiles were

based on national norms of the California Test of Mental

Maturity for the fourth, fifth and eight grades and of the

School and College Abilities Test for the ninth, eleventh


Iieans and Standard Deviations of
Percentile Placement on Ability Tests
for the Twelve Pubject Groups

Evaluative Dynam sm
Moan N SD Nean N SD

ELi 73,55 20 15.29 71.38 16 21.87

ELF 69.75 20 2..22 77.29 17 11.83

JRPI 72.53 37 18.11 75.60 20 20.77

JRF 73.26 19 27.02 71.15 20 24.40

SRM 71.06 18 20,32 69.10 20 26.94

SRF 66.26 19 26.48 69.73 15 18.46


and twelfth grades. Table 4, presents the means and standard

deviations of the ages of the twelve subject groups. Tables

3 and 4 present data only on subjects actually employed

in the analysis, Ability test scores were not available

for all subjects and age data was not available for one

subject in the SRF group. Thus, in tables 3 and 4 the N

of several groups is less than twenty, Table 5 presents

the data :with regard to discarded subjects in each of the

twelve subject groups. As can be seen from the table,

subjects were discarded from each of the groups on the

basis of ability test percentile scores in order to equate

the groups with respect to this variable.


The semantic differential was aTdninisterod to a1.

children in one of their regular classrooms during the

school day. There were two classrooms for each grade, with

approximately thirty students in each class. One section

of each grade rated all twelve voices on the three evaluative

scales; the remaining section rated the twelve voices on

the three dynamism scales. This approach limited the number

of ratings whj ch any individual subject ma-de to thirty-six:

and limited the possible effects of horeodom on the ratings.

In addition, through such a procedure, the independence of

evaluative and dynamism ratings was assured.

The form of the semantic differential employed

required the children to rate all twelve voices on one scale

before beginnin'i_ the ratings on another. It was felt that


Means and Standard Deviations of
Subject Age in Months for the
Twelve Subject Groups


Mean N

125.10 20 7

124.50 20 5

172.80 20 6

174.75 20 8

208.56 20 8

207.50 20 6







C --' I-----ynamis r--- -~- ~-----~- --















Dy na IS i s

5 20

5 20

0 20

0 20

5 20

5 19









- -------- --- ~YII- ---------~


000C 0 00 [


4 -1

i 0

n C)
U) C-1
rti H

- ,r MI -- I L 'l C N N

co co L co chi r

NN NN 5'-^f

M^ ite C(-r

m -4 on ,-4 Li cO

<4 CIrt o t C )

N ( teN
( M (N I *-("I CN (

mi i m m v) c) '3 c: rf() m () VI m c')
k H

E- "
o 0


4 O 1


C 0.


u- H

this approach would minimize any "halo effects" in the

ratings and would allow the children to form and maintain

the desired set in making the ratings. All instructions

and the twelve stimulus voices were recorded on a tape prior

to the experimental sessions to approximate identical

conditions in all groups. The instructions wore modified

frcm those employed by M-arkel and Mam (1960).

Twelve tapes were recorded in all. Each began

with a "dubbed" recording of the instructions and was

followed by three different, random orders of the twelve

stimulus voices. Approximately ten seconds elapsed between

the voices in each set of twelve in order to allow time for

the voice to be rated. A different random order of the

voices was constructed for each of the thirty-six< experimental

presentations of the voices by reference to a table of random


Another aspect of the presentation in which orderinci

effects 'were important was the inter- and intra.-scale

ordering. Since the three evaluative scales and the three

dynamiism scales were each presented to six groups of

children, it w,-is possible to present each of the six possible

ordcrings of the scales to one of the six groups. To counter-

balance with respect to intr.a--scale ordering, three groups

made the evaluative and dynamism ratirngs with the positive

pole of the- biCoiar scales presented on the left. Three of

the six;: roups making each type of rating rated the voices on

bipolar scales on which the positive Dole was printed on

the right.

Data An,'.- lvsis

Numerical values of 1-7 were assigned to the seven

categories of the six scales. The dynamic and positively

valued poles of the scales were assigned the numerical

value of one and the opposite poles were assigned the

numerical value of 7. The means of the ratings of the

twelve voices on the three scales were then computed for

each subject. This resulted in the twelve scores obtained

for each subject. One exception to this procedure was

due to the inadvertent commission of the LPT2 voice among the

voices rated on the large-small scale by the ninth grade.

becausee of this oversight, the LP) mean was computed from

only two scales for those ninth graders making ratings on

the dynamisn scales.

The data were analyzed in two analyses of variance.

Each analysis was" conducted as a 2 x 3 x. -2 ro.ca td

measurements design with repeated measurVements ov.;r the

third factor. This design is discussed by liner (1962,

337-347). The two levels of factor A were sex-of-judge, the

three levels of factor B were the three age levels (EL, JR,

and SR) and the twelve levels of factor C were the twelve



Chapter III


Eval.uative Results

Table 6 presents the results of the analysis of

variance of the evaluative ratings. As can be seen from

the table, P tests on the C main effect and the AC and ,3C

interactions '.ire significant, while tes-ts en other main

effects and interactions were .ot. These findings indicate

that the subjects employed in the present sLtu-dy discrimir:nated

among the twvelvec voices on the evaluative dicnmnsion.

Further, che pattern of these di scrin-inations v.-:ied with

both the sex and age of the rater.

Table 7 presents the mean ratings of the twelve

voices by the three age groups and by the males and females.

The AC means are plotted in Figure 1 and the BC .me-ns are

plotted in l'igrure 2. As can be seen in both figures;, the

differences.. t :t'-/c ratinqg of voices of the sarrm. type

were large. P1 so, contrary to ti. findings expected,

were the highly positive ratings of the 1iL voices.

This finding is directly contradictory of hypothesis

1. The initial prediction was that i;H and PP sneakers

would ho rated las more uoositively val.'.ed than "L spek-rs.

Because of this, the planned comr.-arison of the i'T, ,TP

and HL voices ':.'as not corpoleted- The u.ll hvyothesis,



Analysis of Variance of ean-L Eva] ua- j vre atingr

Source lMean S quarter df P

Sex (A) .429 1 .197
Age ()) 1.475 2 .680
AB .376 2 .173
ErBr r B etw een 2.170 114
Voices (C) 44.096 11 46.281**
AC 1.761 11 1,848*
BC 3.885 22 4.078**
ABC 1.226 2?2 .2P6
rror Within .953 1254

signi ficant at .01
*'*s'igni ticant at. .01

CO a)O C) CO

N ~ N 0 ,1

S* " C
u r- r% CO 0 N1
r- C 0" 0 fD

C, L L.) LA L
Q * *r-

) r- *- < I r

C O) C'i V)

0 (Nj C ,- C) C' cL
4 C Lq l H -

EI m m o- c m

5 C)

rl o C C| O LCO
.r u- * . .

O> cm l cI C-

,--4 E
H * -
*H C ci c-i c Hl r
-- H H o on to L-

c (N *. .
ci c- in C N Ci
H Ci C) Li aC Cl

li CI> rC N C| 0

S * . .
ci 0e i 0 O OA

-I -L Q m r- r *
0 '- C) 0

C) C^ L) co -'- -

19 H m 7 m i m

'-*~; I rn o n (^ ~ ~ L'! m rsi m

Negative ly


0-0 Level 1 of A (Males)
O---0 Level 2 of A (Females)


/ '

I \

3.00 \


Positi. velv

1 2 1 2 1 2 1 2 1 2 1 ?

P 1, Ed L L 14i Y Y

The Twelve Voices

Figure 1

The Sex X Voice Interaction
of the Evaluative Analysis




4.0C --+-'(--

O--C- Level 1-of B (EL)
0----. Level 2 -of B (JR)
O--oLevel 3-of B (SR)

I .

'* I



I S i i >i
1 2 1 2 1 2 1 2

1-3 P

1 2

The Twelve Voices

Figure 2

The Age X Voice Interaction
of the Evaluative Analysis

1 2


HO: (Hi + H-2 + HTI + HT2) -2 (HL1 + HL2) > 0, would

obviously be retained at all levels of A and B.

Inspection of Figure 1 indicates that the sex by

voice interaction was primarily a function of differences

between the males and females in the ratings of the high

and low loudness voices. Figure 2 indicates that the age

by voice interaction also reflected differences in the

ratings of the loudness voices and the iHP and LT ratings

as well. Figures 1 and 2 also show that the females and

the elementary group tended to make the most extreme ratings

of the HL and LL voices.

The Newman-Keuls technique (Winer, 1962; 309-330)

was used to test differences among the ordered voice means

at the two levels of A and three levels of B. These te:ts

were conducted in order to evaluate hypothesis 3 which was

that voicesof similar voice type would also receive similar

evaluative and dynamism ratings. The results of these

tests are presented schematically in Figure 3. Voice

means, which are underlined by common lines in this figure,

were not found to be significantly different at the alpha

level indicated. However, because differences among voices

of the samevoice type were obtained, it is difficult to

interpret differences between voices of diff-rent voice
types. Ascan be seen from Figure 3, of the thirty comparisons

made between voices of the same type, all but three were

significant at tihe .03 level. Thus, with respect to

evaluative ratings hypothesis 3 must be rejected.

In order to evaluate the effect of voice type,




*t *


4-) C)
u; 0d
I 0


4 0>

0 0 rH

t04 u: >




C 0 bJ

0 '. *

0 C

., G
*^rl -




CN <

C --



~-4 II

H3 C)

'- c

:1 o\



C) ^
a) 4 1





despite thelarge differences between voices of the same

type, the voice type means were calculated. Table 8 presents

the voice type means for the males and females and for the

three age groups. Figure 4 is a plot of the voice type

means at the two levels of A; and the voice type means at the

three levels of B are presented in Figure 5. Figure 4

shows that the evaluative ratings of both the males and the

females discriminated most between the HL and the LL voice

types. Both sexes rated HL voices toward the positive pole

of the evaluative dimension and LL voices toward the negative

pole. With regard to both the HL and the LL voice types,

the ratings of the females were more extreme than those of

the males. Differences between the high and low pitch and

the high andlow tempo voice types were much smaller, as

were the differences between the male and female ratings

of these four voice types.

As can be seen in Figure 5, the evaluative ratings

of the six voice types display a clear developmental trend.

In general, the EL age group discriminated most among the

voice types, while the SR age group discriminated least.

In all cases, the ratings of the JR age group fell between

those of the EL and SR age groups. Figure 5 also shows

that while the EL age group attached a negative evaluation

to LP and LT speakers and a positive evaluation to HP

speakers, just the reverse was true of the SR age group.

Because the evaluative analysis so clearly showed

that the HT and HP voices were not more positively valued

than the HL voices as originally hypothesized, an examination





E-e O

U )
cV 0 r


CU *

.E- 0

0- 4

S,4- .

*ri U

I, 0



x C
3 E-

4 4


* r-

o 4
0 C)

C >

o A



0 iYl

o a\


Lo 01
CN in

o0 o


f) )

rn -

C) r
1 E
f (1)

o 0 CO
coa co

C r-i 0
0 0 0

\o un u


L' ri co
') co c'

* .
n n

>-C H H

E *H -lH

-1 3 C)

4 CQ n

c----o level 1 of A (males)
o----o level 2 of A (females)

va I ued




3.0 0

va Led


I i"

The Six Voice Types

Figure 4

Plot of the Voice Type Means
at the Two Levels of A
of the Evaluative Analysis


I' T





0--0 Level 1 of B (EL)
0----0 Level 2 of B (JR)
C--- Level 3 of B (SR)

/ \

\ !




i' HP

The Six Voice Types

Figure 5

Plot of the Voice Trype ieans
at the Three Levels of B
of the Evaluative Analysis



of the voices in terms of voice quality profile (VQP) was

made. Table 2 above shows that four voices of each VQP were

included in the present study. The voices included in each

profile group were as follows: P-P: HIP, UIP2, LL2 and LTI;

P-L: LPI, HIL1, !!L2, LT2; P--T: LP2, LLj, lHT1 and lIT2.

Table 9 presents the mean evaluative ratings of the three

VQP's made bythe three age groups and by the males and

females. Figure 6 is a plot of the VQP means at the two

levels of A. The VQP means at the three levels of B are

presented in Figure 7. Figures 6 and 7 show that, although

differencesamong the VQP's were not large, among the males

and the older children there was a tendency for the P-L

profile voices to receive negative ratings. These same

subject groups also tended to rate the P-T voices toward the

positive pole of the evaluative dimension.

Dynamism Results

Table 10 presents the results of the analysis of

variance of the dynamism ratings. As can be seen from the

table, F tests on the B and C main effects and the AC and PC

interactions were significant. These findings indicate that

the pattern of discrimination among the voices on the basis

of the dynamism dimension varied with both the age and

sex of the rater.

Table 11 presents the means obtained from ratings

of the twelve voices made by the three age groups and the

males and females. The AC means are plotted in Figure S and

the BC means are plotted in Figure 9. The figures show that




> 4)






O E-
0 c

0 -


-H r



-p +


C; r- o

Er r 0
4 mm




0-i 0





r -

; C-
^-1 0 .

CO 0 0)

0 0 0
Ol Ca) Co

0'r. r-i

r4- Q

C'N Li


C) 0 0

ot cq mr


0--- Level 1 of A (Males)
0----0 Level 2 of B (Females)







The Three Voice Quality Profiles

Figure 6

Plot of the Voice Quality Profile Means
at the Two Levels of A of the
Evaluative Analysis



&--o Level 1 of B (EL)
.. 0 Level 2 of B (JR)
o-o Level 3 of B (SR)


4.00 ..o- .- .-- -.. ._____________





The Three Voice Quality Profiles

Figure 7

Plot of the Voice Quality Profile '!eans
at the Three Levels of B of the
Evaluative Analysis



Analysis of Variance of Mean Dynamism Ratings

Source Mean Square df F

Sex (A) .299 1 .245

Age (B) 9.065 2 7.424**

A3 .675 2 .553

Error Between 1.221 114

Voices (C) 91.518 11 168.232**

AC 1.060 11 1.949*

BC 3.306 22 6.079**

ABC .809 22 1.487

Error Within .544 1254

-iT377ifanT. at1.0--1--
*significant at .05
**significant at .01

1____ _111____ ___


N co rN r o C)o

E-' c C' ." I)

rr Co n N Cl

i- I Co <

wU) r4N O CO 0o Nr
C) (N r-l CO m CO

Pr C N (N m CN

C 4N N N 4 e- M
C) U' U r- CO Co CO CO CO
*r *
4cJ 0 m r4m 4 4 n

m, C") co N
r4N C) M e N Co

' t' 0 tL L) L Ln
U < i
"-4 E-i In %D t- r- N
o C) r o n 4 '. 0
> C) *. *

Q) Ir E-i
> E.
L) In n -1 cr0L inC
C)Cn' N ct '- ifln N Cl)

E4 i-i (N ClM N Cr Cl

-0 N' r- t oL ri
^: 1 D N N in to

Noc N) C". U' N N
4) *

Cl CC m m m l m

+JE r o C l') NC C
dC) 4 Ce 01 o m m rN

w 4

I*r EJ I) c r-) O O
(N m r-4 r-1 N Ln

Grr: n < r^ m co
> E

C) : C

< < m c
< < M a Q



4.00 --__


Level 1 of A (Males)
Level 2 of A (Females)

I '



1 2 1 2 1 2 1 2 1 2 1 2

1 P

H- Y

The Twelve Voices

Figure 8

The Sex X Voice Interaction
of the Dynamism 7Analysis

o--o Level 1 of B (EL)
o.....o Level 2 of B (JR)
c--co Level 3 of B (SR)

non-dynnai c






\* /

dy/nami c


1 2 1 2 1 2 1 2 1 2 1 2

LP (iL L I.
The Twelve Voices

I3 T

Figure 9

The Aje X Voice Intcraction of th-
Dyrn-i ism Analysis


intra-voice type differences were much smaller than inter-

voice type differences. Inspection of Figure 8 indicates

that the sex by voice type interaction was primarily

attributable to differences between males and females in

the rating cf high and low loudness voices. There was an

overall trend, however, for the females to make ratings,

relative to the male ratings, toward the dynamic pole on

the IIP, 1HL and HT voices and toward the non-dynamic nole

on the ],P, LL and LT voices.

The plot of the age by voice interaction in

Figure 9 shows that sizeable differences occurred among

the age groups with regard to the ratings of several of the

voices. The N\ewnan-Keuls technique (Viner, 1962; 309-310)

was used to test differences among the ordered age grouo

nmans at each of the levels of C. The results of these

tests are presented schematically in Figure 10. Age group

means which share a coinmon underlining were not found to be

significantly different at the .05 level. As can be seen

in Figure 10, significant differences among the age groups

were found at eight of the twelve levels of C. Figure 9

shows that the EL age group tended to rate the HP', LP and

HL voices more toward the dynamic pole than did either of

the other two age groups. This tendency may have accounted

for the significant B main effect. Figure 10 shows that,

of the seven instances in which two of the age groups were

significantly different from the remaining age grour but

not from each other, two involved differences between the

JR age group and the other two. Although such results do

Level 1 ofC
(HP ) -

Level 2 of C B! B2 B_

Level 3 of C B B B2

Level 4 of C Bj._B B2

Level 5 of C B_1 Bi -

Level 6 ofC B] BI B

Level 7 of C B__B2 B1

Level 8 of C B3 B __

Level 9 of C _2. P,- B3

Level 1.0 of C B3 B B2
(lIT2 )

Level 11 of C B J B3

Level 12 of C _BII B3

Figure 10: Schematic Presentation of the Neewman-Keuls
Tests on The Ordered Age Group Dynramism

[Means Which Share a Common Line were n-t Significantly
Different at the .05 alpha Level.]


not make a hypothesis of a developmental trend in the ratings

untenable, neither do they offer strong support to such a


The significant 2C and BC interactions necessitated

evaluating hypothesis 2 by computing the planned comparisons

among the dynamism ratings of the IHP, IL and IT voices at

the two levels of A and the three levels of '. Five t-tests

were employed to test the following directional hypothesis:

HO: (rlL1 + I-L2 + iTi + L7.*T2) -2 (l7P1 + i-2) > 0. The null

hypothesis .'as rejected at all levels of A and beyond the

.001 level, one tailed. Thus, the high loudness and high

tempo voices were rated by all groups as more dynamic than

the high pitch voices.

The Ne.i mn-1reulCs technique innerer 1962; 309-310)

was used to test differences among the ordered voice type

means at the two levels of A and the three levels of B.

The results of these tests are presented schematically in

Figure 11. Voice type means which share a corn~on underlining

were not found to be significantly different at the .01 level.

Figure 11 shoWs that across all subject groups stable

differences were found between high and low loudness and high

and low temno voices. In all cases, the HL and HT voices

were rated as more dynamic that the LL and LT voices. The

findings with regard to HP and LP voices were less consistent,

but there was a general tendency for the LP voices to be

rated as r:ore dynamic than the HP voices.

As a means of evaluating hypothesis 3, which stated

that voices of similar voice type would also receive similar


,-- r l
1.^ c? L^ \



0 -

i -

0o 0


0 }.-
! CO
A 0

E r--I

t cV

C 0C


4Ul C)

0 ^ r-4
H i

C) F -A

-I -P


0 rJ -

0 C U

r- -rH


t0 0




> 0C

evaluative and dynamism ratings, the results of the Newman-

Keuls tests with respect to voices of the same type were

examined. In order to state with more certainty that true

differences between voices of the same type were not

ignored, tests were also computed with alpha set at .10.

Table 12 presents the significance levels obtained for the

comparisons of voices of the same type at the two levels of

A and the three levels of B. As can be seen from the Table,

significant differences between the two HP, the two IHL and

the two IT voices were fairly consistently found across the

levels of A and B. 7At least with regard to the dynamism

ratings of these three voice types, the data of the

present study did not supCprt hypothesis 3. The data in

Table 1.2, however, te:ndea to ccnfirm hypothesis 3 with

regard to the LT and LP voice types and with regard to the

EL age group.

In order to obtain a more reliable estimate of the

contribution of voice type, the voice type means were

calculated. Table 13 presents the voice type means for the

males and females and for the three age groups. Figure 12

is a plot of the voice type means at the two levels of A;

the voice type means at the three levels of B are plotted in

Figure 13. Figure 12 shows that sex differences in the

ratings were small and that both sexes rated HP, LL and LT

voices toward the non-dynamic pole and LP, HL and HT voices

toward the dynamic pole. For both sexes, the most extreme

ratings were assigned to the high and low loudness voices,

CV. C, .
z Z Z






, C.


0 ro

i l


4c )C


O) U


O c
I .4

4 4


O 0

C )

O U)



> --


Z c


CC C: o
2 m .H



(N a

0 (

0 -4






O i-i




0 4

- -
-I +J
r fC

6.0 0


4.00 -

-- Level 1 of A (I.al.es)
0---O Level 2 of A (Females)



1 P

T Y.

The Six Voice Types

Figure 12

Plot of the Voice Type Means
at the Two Levels of A of the
Dynar.ism Analysis


O-- Level 1 of B (EL)
.... 0 Level 2 of B (JR)
O-O Level 3 of B (SR)



' I

' J I




The Six Voice Types

Figure 13

Plot of the Voice Type Means at the
Three Levels of B of the Dynamism Analysis

while the least extreme ratings were assigned to the high

and low pitch voices.

As can be seen in Figure 13, largest differences

among the age groups occurred in the ratings of the LTP,

IIL, LL and LT voice types. The LP voice type .was rated

toward the dynamic pole by the SP age group and toward the

non-dynamic pole by the JR age group. similarly to the

males and females, the EL age group's rating of the LP

voice type was near the dynamism scale midpoint. All groups

rated the LL and LT voice types as non-dynamic and the IHL

voice type as dynamic. The ratings of these three voice

types did. show something of a developmental trend, in

that differences occurred either between the two clder and

the youngest age groups or the two younger and the oldest

age groups, bxt not between the middle age group and the

other 'two.

It was also of some interest to consider the

dynamism ratings in terms of Markel's original VQP procedure.

Table 2 shows that four voices of each VQP were included in

the present study. The voices included in each profile

group were as follows: P-P: HP1, HP2, LL2 and LTI ; P-L:

LP1, HiL1, HL2, 1LT2; P-T: LP2, LLT, UlT1 and HT2. Table

14 presents the mean dynamism ratings of the three VOQP's

by the three age groups and the males and females. Figure

14 is a plot of the VQP means at the tw.o levels of A; the

VQP means at the three levels of B are presented in Figure 15.

As can be seen i:n Figures 14 and 15, the P--P VOP was

consistently rated as non-rdynaalic, the P-L VO)P was consistently





3 to
0 0





-i m N M
Fic M) C rO al

0 U)

.-I ^
4-4 O
0 Q

>1 e-'





0 >1

cP .

M in

* e

oo n
%1D VL

, CO
N cc

l -I
U) C

co) Co N

- (N Ce

CN r- r-

N Cr-i
r- co 0

* *H -*

-H *H

0 0 0

M 1-) UC

o-o Level i of A (males)
c--o Level 2 of A (females)








The Three Voice Quality Profiles
Figure 14

Plot of the Voice Quality Profile Means
at the Two Levels of A of the
Dynami s;i Ani lysis






Cr-OLevei 1 of B
O-....OLevel 2 of B
O---Level 3 of B




P-P .P-1. --T
The Three Voice Quality Profiles

Figure 15

Piot of the Voice Quality Profile Means
at the Three Levels of B of the
Dynamism Analysis



rated as dynamic and the P-T VO1 was generally rated as

dynamic but less dynamic than the P-L VOW"

Chanter TV


The extent to which the findings of any study

can be generalized is dependent upon the subjects employed,

the variables sampled and the methods used. In the present

study, both males and females were employed as subjects

and three different age levels of children were sampled.

The population from-;n which these children were draw.'n, however,

included_ a high percentage of the children of professional

people and, perhaps. as a result, the children tertde.', as:

a crou, to score 1igch on standard ability escct. Strickere

and Zax (1S66) report tnat the ability to utilize aJl seveT n

points of semantic differential scales is related to

intelligence among children in the fourth and eight grades

but not the twelfth grade. Thus, interpretations of

findings with regard to the EL and JR groups arc limited

by this variable.

Other limiting factors in the present study

include tne _Lchnicaue enmploved to obtain the judgments and

the limited sample of all male voices which the children

rated. The findings' of previous studies with regard to the

evaluative and dvnamiis dimensions indicate that they are

the most critical ones for obtaining judgmr.ents from children,

especially when people are being rated, Previous studies



were also utilized to select individual scales. It should be

noted however, that the results reflect only group stereotypes

and do not provide information about individual subject

responses to the separate speakers. In addition, it is not

possible to make statements about all possible combinations

of pitch, loudness and tempo nor about the possible effects

of other voice qualities.

Sex Differences

The sex-of-judge affected both the dynamism and

evaluative ratings,. In both analyses, the nain effect of

sex was not significant, while the interaction of sex-of-

judge with voice was. In general, this finding was attributable

to the extreme ratings made by the females, especially of

the IIL and LI, voices. Doth males and females judged m.n:

w.ho spoke loudly to be more dynamic and more nositively

valued than ;men ,:hn spoke softly.

Goodenough (.1957) found that, in our culture,

females, more than males, are encouraged to attend to the

expressive behavior of others. That females would be more

able than males to make discrimination of individual

differences on the basis of voice alone, is a logical

extrapolation from these findings. Dimitrovsky (1964)

found this to be the case in a study which required children to

select the intended emotion of speakers on the basis of

voice al.one. The correct-incorrect dichotomy employed by

Dimitrovsky is not clearly applicable to the present study.

However, the hypothesis, with regard to the dcynami sn ratings

of the six voice types was confirmed by the results of the

present study. Thus, one way in which the ratings can be

viewed as correct or incorrect is to consider the following

as correct ratings: (1) dynamic: LP, 1iL and lIT and

(2) non-dynamic: HP, LL and LT. This system is essentially a

restatement of the original hypothesis regarding the dynamAism

ratings. Reinspecting the voice type means reveals that,

although all differences are small, the females' responses are

extreme in the correct direction with regard to the HL, LL,

HT and ILT voice type and the responses of the males and

extreme in the correct direction with regard to the HP and

LP voice types. Thus, it appears that the females are more

sensitive to l.oudn.es and tempo differences while the males

are more sensitive to differe-nces in pitch.

Sex differences in the evaluative ratings of the

voices were much larger. Differences with regard to the

pitch and tempo voice types were sal. while the ratings

of the two lourdness voice tvrype showed a definite interaction-

with the sex factor. The females made more extreme ratings

than did the males of both the HL and the LL voice types.

Age Differences

The ratings of the voices on both the evaluative

and the dynamism dimensions were affected by the age of

the judge. Ir the evaluative analysis, the ace by voice

interaction was significant, while in the dynamism analysis both

the interaction of age with voice and the main ,ffect of

age were significant. Further, the nature oE the age

differences in the evaluative analysis ,was suggestive of a

developmental trend in the ratings.

In practically all cases, the mean evaluative

ratings of the voices by the middle age group (JR) separated

the mean ratings of the voices made by the oldest (SR) and

the youngest (EL) age groups. The few exceptions to this

pattern occurred with respect to voices which elicited small

differences in the ratings of the three age groups. When

the data were converted to voice type means rather than

individual voice means, the developmental pattern was clear

across all voice types. All three age groups evaluatcd

loud (HL) speakers as positively valued and quiet or soft

(LL) speakers as negatively valued. None of the age groups

judged the fast (H!T) speakers to be either positively or

negatively valued. Ho-wver, the youngest children (EL)

rated those who spoke slo..y (LT) as negatively' valued and

the oldest children (SR) rate-d slow spe akers (LT) as positively

valued. Sir.ilarly, while the EL age group rated HP speakers as

positively valued and LP speakers as negatively valued,

just the reverse was true for the SR age group.

It seents possible that the difference obtained

between the EL and SR age groups in the ratings of the !P and

LP voices, might be attributable to the fact that elementary

school children spend much of their time :ith females.

Characteristically, the voices of females arc higher in

pitch than the voices of males and the LE. age group rated I:e

voices as positively valued. Another exulanztion for the

results is that. even the boys in elncmntary school typically

have high pitched voices themselves. Perhaps the negative

evaluation of the LT speakers by the EL age group reflects

the characteristically higher activity level of children of

this age. Older children, on the other hand,are typically

more reflective and perhaps this accounts for the value

they place upon a low tempo in speech.

Perhaps the most striking difference among the

dyncamism ratings of the three age groups was that the EI,

age group responded more consistently to voice type in

judging the voices than did the other two age groups. The

ratings of the former age group differentiated less between

voices of the same voice type. This result seems somewhat

inconsistent with the findings of both Gates (1927) and

Dimitrovsky (1964). In both studies older children were

superior to younger children in judging the intended emotion

of the speaker on the basis of voice alone. As previously

mentioned, a correct-incorrect dichotomy can be extrapolated

from hypothesis 2 of the present study. The system requires that

correct responses be defined as follows: (1) dynamic: LP, H.

and HT and (2) non-dynamic: HP, LL and LT. Comparing the

three age groups with the dichotomy in mind reveals that when

significant differences occur among the age groups, the EL

group's rating of the voices is always in the correct

direction. Thus, in the present study, the youngest children

employed were more often correct in making the dynamism


The Gates (1927) and Dimitrovsky (1964) studies,

as did the present one, required the children to rate the


voices via a primarily verbal methodology. The methodologies

employed in the earlier studies were developed specifically

for that purpose. The semantic differential, the rating method

employed in the present study, in contrast, has previously

been extensively validated for the age groups included.

Thus, it seems likely that younger children did not do as

well as older children in the earlier studies because of an

inability to express the discrimination which they made,

rather than an inability to make the required discrimination.

In fact, Gates (1927) noted that the youngest children in

her study had difficulty understanding the instructions.

In general, the present study offers empirical support to the

theory that young children are particularly sensitive to

nonverbal vocal cues.

Voice Type and Voice Quality Profile

One purpose of the present study was to determine

if functional cr personality ratings of voices would

differentiate the voices in a pattern similar to the structural

or voice quality ratings. The evaluative analysis revealed

that although the subjects discriminated among the twelve

voices, it could not be concluded that differences obtained

resulted from differences among the six voice types. This

was true because significant differences were consistently

found between voices of the same voice type. The failure

to obtain consistent differences on the evaluative dimension

among voices of different voice types is in agrenent with

the findings of Markel and Roblin (1965). They found

differences on the evaluative dimension but not the activity

and potency dimensions when voice quality and voice set

were held constant. These differences, however, were

attributable to content variations inconsistent with the

speaker's voice set. Markel, Meisels, and Houck (1964)

found no differences among speakers on the evaluative

dimension when voice set and content were constant. Since

content and voice set were constant in the present study,

the differences obtained cannot be attributed to these


In the Markel and Roblin study all ratings were

made on just one speaker. Ratings in the Markel, Meisels

and Houck study were mean ratings of groups, In the present

study, on the other hand, the individual ratings of several

speakers were employcd. Thus, unlike i-he other studies

discussed above, the present study could be affected by

variations in voice qualities other than pitch, loudness and


An interesting result in the evaluative analysis

was the clear and unexpected rating of the HL voices as

positively valued and of the LL voices as negatively

valued. It seems possible that this finding is related

to the fact that all speakers were males. In our culture

it is difficult for a man to be both dynamic and negatively

valued. Also of interest was the finding that when the

ratings were converted to three VQP means, the P-L VQP was

rated as negatively valued by the males and the SR and JR age

groups. This finding is in agreement with hypothesis 1 which

was not confirmed when the voice type data were examined.

With respect to the dynamism ratings the relationship

between the structural ratings, and the functional ratings

of the voices is more consistent with the original hypotheses.

Although other variables perhaps other voice qualities,

affected the dynamism ratings of the voices, few subject

groups made intra-voice type differentiations among the

LP and LT voice types. In addition, with the exception of

the HP and HT voice tapes, the EL age group did not

significantly differentiate between voices of the same voice


Hypothesis 2, that HL and HT voices would be rated

as more dynamic than II:' voices, was confirmed. Thus, not

only did the dynamism ratings reflect a functional discrimination

among the structural types but also the pattern of discriinations

was that which was predicted fro-m previous studies. That is,

loud and fast speakers were rated as stronger, larger and

faster than speakers whose voices were high in pitch. The

most critical finding with regard to the LL and LT speakers

was that they received ratings on the dynamism dimension

which were just the reverse of the HL and lHT ratings

respectively. In addition, all subject groups rated the LD

voice type as more dynamic than the HP voice type.

When the dynamism ratings wore combined on the

basis of the voice quality profiles, the pattern of ratings

suggested by hypothesis 2 was clear. The P-L and P-T voices

were rated as dynamic, while the P-P voices were rated as

non-dynamic. With the exception of uhe SR age group, the

VQP means indicated that the P.-L speakers were more dynamic

than the P-T speakers. The SR age group differentiated very

little between the P--L and P-T VQP groups.

In general; the subject groups consistently rated

the HP and HL voice types good, fair and important, and

the LL voice type bad, unfair and not important. The HL and

HT voice types were rated strong, fast and large, and the

HP, LL and LT voice types were rated weak, slow and small.

Thus, males who speak loudly or with a high pitched voice

are valued while maleswho speak softly are not, and males

who speak loudly or quickly are seen as Miore dynamic than

males who speak slowly, softly, or have high pitched


Inmolcations for F1ture Resc:rch

An interesting finding of the present study w.as

the powerful effect of !cudness on the ratings. On1 the

basis of these findings, it seems likely that detailed

investigations of loudness alone would be fruitful. In

view of the fact that accurate, irrimediate feedback of

intensity levels can be- obtained mechanically, this variable

could be investigated through many other methodologies.

The data also suggested that investigations of

voice qualities in females would be a valuable contribution

to the area. It was hypothesized that, in our culture, a

dynamic man is also positively valued and that the evaluative

ratings reflected this cultural bias. it seems logical to

expect different patterns to emerge if female speakers are


The ratings of the voices in the present ;study

were- affected by variables other than the voice qualities

included. Since the study controlled for both voice set cnd

content differences, it seems quite likely that the

uncontrolled variables were other voice qualities. Especially

with regard to evaluative ratings, the data of the present

study suggest that these qualities may be extremely important.

There are a variety of reasons for investigating

the relationship between voice and personality through

methods other than the one employed in the present study.

The semantic differential is not reliable for children younger

than eight, yet this study indicates that young children

are sensitive to differences in voice quality. Thus, a

methodology which allowed the use of even younger subjects

would be of value. It has previously been mentioned that

the data of this study iere group stereotypes of the

different voice types. Although data regarding groups is

an important aspect of psychology, a methodology which

provided reliable information with regard to individual

responses to the voice types would be desirable. Finally,

of course, results which are independent of the methodology

employed, are very unlikely to be spurious.

An approach which would deal simultaneously with

several of the limitations of the methodology employed in

the present study is to investigate voice and personality

via an operan;t methodology. For example, loudness might be

investigated by obtaining, from the same person, scver-_a

speech samples identical in all respects except the loudness.


These stimuli could then be employed as subsequent events for

as many operants on which a baseline had previously been

obtained. Thus, in an operant study, changes in the rate of

an operant would define the reinforcing properties of the

different loudness levels. Such a methodology would make

it possible to employ pre-schoolers, and even infants, in

voice and personality studies.

Chapter V


One purpose of this study was to test directly

hypotheses emerging from the work of Markei and his

students on the voice quality profile (VOP). Six voice

types were also defined in order to gain further information

about high and low scores on pitch, loudness and tempo,

the three qualities included in previous V'P studies.

Thus, the six voice typ-e included were as follo-...s: high

pitch (HP) low pitch (LP high loudness (HL), low

lo'.3nness (L.) high tempo (LT) and low tempo (LT) A

second pu;r-pos of the study was to investigate : age and

sex diffec-ences i:n judging personality front the voice.

Two male speakers of each voice type were rated by

a31 subjects. Three e\'vluative scales (e.g. "good-bad") and

three _dy-namism scales (e.g. "large-small") from Osgood's

Se:antic Differential were used to obtain the ratings. The

twelve voices were rated cn the three evaluative scales by

120 males and feonales in the fourth, fift'i, eighth, ninth,

eleventh and twelfth grades. A similar group rated the

twelve voices on the three dynamism scales.

The evaluative and dynarninm ratings were analyzed

in sop-ar-at rcep[ated measurements analyses of variance.

Individual subject scores were the mean ratings made by the


subject of each of the twelve voices. The twelve voices

were the levels of the repeated measurements factor; the

levels of the two other factors were the three age groups

and the two sex groups.

It was predicted that HP and HT speakers would

be rated as more positively valued than III. speakers and that

HL and HT speakers would be rated as more dynamic than HP

speakers. The latter prediction was confirmed by the data

while the former was not. Differences between voices of the

same type led to the conclusion that voice qualities, other

than pitch, loudness and tempo, affected the ratings,

especially those made on the evaluative scales. In general,

men who suoke loudly, or with high pitched voices, were

rated( as valued while those who spoke softly were not; mlen
who spoke loudly or quickly were seen as more dynamic than

speakers w.ho spoke slowly or softly or with high--pitched

voices. An interesting finding was that when the voices

were examined in terms of VQP, rather than voice type, the

data tended to confirm the hypothesis regarding the

evaluative ratings as well as the hypothesis regarding the

dynamism ratings.

The se(x-of-judge affected both the evaluative and

dynamism ratings of the voices. In the evaluative analysis,

the sex differences were attributed to the extreme ratings

of the females, especially of the HL and LL voices. The sex

differences in the dynamism ratings were attributed to the

larger discrimination made by the males between the HP and

LD voices. The females, on the other hand, discriminated

more between I1L and LL voices and I1T and LT voices than did

the males.

Age differences were also found in both the

evaluative and dynamism ratings, There was a consistent

developmental trend in the ratings of the voices on the

evaluative scales. With regard to the ratings of the

voices on the dynamism scales, it was concluded that the

youngest age group was the most accurate.

Further research was suggested with female

speakers, other voice qualities, the loudness voice quality

alone and younger judges.



Instructions to Subjects

Everyone should now have a booklet containing

four sheets. Raise your hand if you don't have a booklet

or .if your booklet doesn't have 4 sheets. (Pause) We

are going to see if you can guess what some ren are like

just by hearing them say one sentence.

Before we try some examples please write the

following information in the spaces provided on pa.e 1 of

your booklet. First your name (Pause); the month, day and

year of your birth (Pause); your sex -- if you are a boy,

put a circle around the ~, if you are a girl, put a circle

around the F (Pause); next fill in the grade you are in

in school (Pause).

Now you are ready to learn how to make your

guesses. You will always indicate what you think a man is

like by making an X on an adjective scale. An adjective

scale is two words which are opposite in meaning like tall

and short. These words are printed on opposite sides of a

page and are separated by 7 dashes or lines. Example A on

page 1 is the adjective scale tall-short (Pause). Notice

that the seven dashes or lines are separated by two dots.

This is to help you place your X in the correct space.

Example B shows the meaning of each of the seven spaces on

the tall-short scale. You right like to read these labels


silently to yourself while I read them aloud. Putting an

X in the space closest to the word tall would mean that you

guessed the man was TVERY TALL, The next space would indicate

that the man was QUITE TALL. You would use this space if

you thought a man was taller than SLIGHTLY TALL but shorter

than VERY TALL. The space in the middle of the tall-short

scale should b used when your guess is that the man is

AVERAGE in height, The first space on the short side of the

tall--short scale should be used if you guess that a man is

SLIGHTLY SHORT, that is, a little shorter than average. The

next space on the scale would indicate that you thought the

man was QUITE SHORT and finally the space closest to the

word short should be used if you guess a man is \L F SHIORT.

Exariple C shows the tall-short scale with a

guess already indicated. When I heard :nan number 1 speak

I guessed that he was just slightly taller than average

so I placed my X in the space next to the center space

on the tallside of the scale. Example D shows my guess

about the height of nan 2. Something about the way he

sounded made me guess that he was QUITE SHORT so I put my

X in the second space from the word short.

Now let's try a few for practice. Before you

hear the voice, you will be told what number it is. Be

sure to put your X for a particular voice on the scale

next to the voice number [Voices 3, 4 and 5 were played

and time w;as allowed for rating.] Are there any questions?


0:; let's try some more. On each of the next three

pages you .will make guesses about 12 voices. On each page

you will mark your guesses on a different scale. Write

down your first impression and be sure that you pat down

an X each time you hear a voice. However, never put more

than 1 X on the same scale. Be sure to put your X's in

the middle of one of the spaces; never put an X on the dots.

Please turn the page and notice which adjective

scale you will using to make guesses about the next 12


The above phrase followed the first two sets of

twelve voicescon each type.

All voices were preceded by an announcement of

the voice number and followed by approximately 10 seconds

of silence,


Samole Instruction Sheet




Example A:

Example B:


Example C:


* ~ v

Exafrl~e I):



3 T L : : : : : :

4 TALL : : :

5 T L : : : : :


N A ,i





H -

H ^









-I -I-" ~


Samp e Rating Sheet

MAN >< H4 < H H K
C; ; < u. C C > -.



> >
r", D 3 5 cc : C: F, > r


5 UNFAIR : : : : : : FAIR

K C2D 5- >
H6 UNFAR : : : : : : FA

K3 C'- L^~ fi; ML M r: T
7 UNFAIR : : : : : FAIR

8 UNFAIR : : : : : FATP

9 UNFAIR : : FAI ,



Allport, G. and Cantril, HI. Judging personality from the
voice. Journal of Social Psychology, 1934, 5,

Barnard, J. W. The effects of anxiety on connctative mcaningq.
Child Development, 1966, 37, 461--472.

Brandwin, M. A. Levels of reference in connotative meaning:
A developmental study. Papers of the lichigan
Academy of Science, Arts and Letters, 19-6, 56,

Cantril, I. and Allport, G. The Psychology of Radio. New
York: Harper, 1935.

Costanzo, F. S., MarkeL, N. N, and Costanzo, P. R. Voice
quality profile and perceived emotion. 'ournal
of Counseling I svcholoqv, 1969, 16, 267;-270.

Dirmitrovsky, L. yhe ability l o intify thCl anotional
meaning of vocal co,'rtjs3sions at successive age
levels. In JDoel R. Dvo/it (Pd.) The C -.L.uni ti
of Emotional M:ea.nin-'. New York: McGraw- i.l, 164.

Dittm-nn, A. T. and Wynne, L. C. Linguistic- techniques and
the analysis of emotionality in interviews, Jou-C. !
of Abnormal and Socia. Psychology, 1961, 63, 2 1-?C04.

Di Vesta, F. J. A normative study of 220 concepts rated on
the ser.'antic diffE--erntial in grades 2 through 7.
Journal of GenCeic s2ycicloov, 1966, 109, 205-229.

Di Vesta, F. J. and Dick, "'. The test-retest reliability: of
children's r.tiJno;s on the semantic differential.
Educati.Onal and ..Pycholocica.l 2Measurement, 1966,
M; ---------- --'----
26, 605- -10.

Donahoe, J. W. Changes in ni eaning as a function of age.
Journa! of Genetic Psychology, 1961, 99, 23-2C.

Gates, G. S. The role of the auditory elei-ent itr the
interpretaJtion of emotion. Psychcl ogical B 'u letin,
1927, 24, 175,

Goodcnough, I. W. Interest ini persons as an aspect of sex
differences in the early years. Genetic Psvchology
M.on.o.graphs, 1j957, 55, 287-323.

Kramier, E. Judgment of personal characteristics and emotions
from nonverbal properties of speech. Psychological
Bulletin, 1963, 60, 408-420.

Kramer,, Personality stereotypes in voice: A reconsideration
of the data. Journal of Social Tsychology, 1.964,
62, 247-251.

Lilly, I. S. A developmental study of the semantic di-fer-.ntial.
Doctoral Dissertation, Princeton University, 1965.

Mahl, G. F. and Schulze, G. Psychological research in the
extralinguistic area. In T. A. Sebeol, A. S. Haves
and 1M. C. Bateson, (Eds.), Aporoaches to Semiotics.
The Hague: Mouton and Comp'anyT'7, '9T", pp. '--TT

Maltz, H. E. Ontogenetic changes in the meaning of concepts
as measured by the semantic differential. Child
Development, 1963, 34, 667-674.

Market, N. N. and lamp, E. P, Connotative meanings of certain
phoneme sequences. Studies in Ling.uistics, 1960,
15, 47-61.

Markel, N. N. The reliability of coding paralanguage-: itch,
loudness and tempo, journall of Verbal Lucrrij r aid
Verbal Behavior, 1965', ,~0?-30. "

Marhkl, N. N. The relationshi'- :t. 'een voice quality profile
and I:,'I profiles in psychiaLric patients, Journal
of Abnormal Psychology, 1969a, 74, 61-66.

Ma,]-el, ". N. Personal communication, 1969hb.

N.and Vouck, J EN Judging
Market, N. N., Maisch!, M. and Pouck, J. E, Judging
personality from voice quality, Journal of
Abnormal and Social Psvcholo y, 19-4-'T''' 5 38-463.

Market, 7. N. and Roblin, G. L. The effect of content and
sex--of.- judge on judgments of personality from
voice. International Journlr of Social Psychiatry,
1965, 11, 295-30T-0.

Osgood, C. E., Suci, G. J. and Tannenbaum, P. H. The
Meas.:ueent of -eaning. Urbana: Univer TSi
or Illiioss Press, 1957.

Pear, T. H. Voice and Personality. London: Chapman and
Hall, TT1931.

S:ipir, F. Speech as a prsonality trait. A o rican Journal
of S.ciolo; v,/ 1927, 32, 892-905.

;Sm'a E,. Ae Prd sex: differences in the semantic strucL-Lre
of children, Doctoral Dissertation, Universit o.f
iichg c'.jan, 1958.

Stricker, and Za:,, Intelligcence and semantic differential
discriminability. Psychological ReporLs, 1966,
18, 775-.778.

Thorndike E L. and Lorge, I. The Teacher's i'ord Book of
30,000 Words. New York: Bureau of-u"-F aT:i"T-s,
Teacher's College, Columbia University, 1944.

Trager, G. L. Paralanguage: A first approximation. Studies
in Linguistics, 1958, 13, 1-12.

Vargas, R. A study of certain personality characteristics
of male college students who report frequent
positive experiencing and behaving. Doctoral
Dissertation, University of Florida, 19E8.

Winner, B, J. Statistical Principles in Experimental Design.
New York: '"cGraw-HflY Book Company, 1962.


Judith Ann Phillis was born in Uniontown,

Pennsylvania, on March 22, 1942. She was graduated from

Susquehanna Township High School, Harrisburg, Pennsylvania,

in June of 1960. In the fall of that year she entered

Carnegie Institute of Technology (Carnegie-Mellon University),

Pittsburgh, Pennsylvania, from which she was graduated with

the degree of Bachelor of Science in June, ]964.

From June, 1964, through August, 1965, Miss Phillis

was employed as a research assistant at the Craig House for

Children in Pittsburgh, Pennsylvania. In September, 1965,

she began graduate study in psychology at the University of

Florida. From June, 1967, until August, 1967, she was

employed as a laboratory technician at the Cormunication

Sciences Laborato-ry, Univrsity of Florida. She received

the degree 1Master of Arts in August, 1967, and since that

time she has pursued her work toward the degree of Doctor

of Philosophy.

This dissertation was prepared under the direction

of the chairman of the candidate's supervisory committee and

has been approved by all members of that committee. It was

submitted to the Dean of the College of Arts and Sciences and

to the Graduate Council, and was approved as partial fulfillment

of the requirements for the degree of Doctor of Philosophy.

December, 1969

Dean, CoL~L Ars and Sciences

Dean, Graduate School

Supervisory Comnittee:


University of Florida Home Page
© 2004 - 2010 University of Florida George A. Smathers Libraries.
All rights reserved.

Acceptable Use, Copyright, and Disclaimer Statement
Last updated October 10, 2010 - - mvs