Impression of voice in relation to stereotyped occupational categories

NII-Electronic Library Service

11te/

foPanese

Joi･tt'nal

,!f

11s)/chonomfc

,S'cie"ce

!oel,

voL

2o,

No.

1,

1-lz

Original

Articles

to

Impression

of

voice

in

stereotyped

occupational

Natsuko

YAN･IADA,

Yuji

Sanae

SHIRixsAw

relation

categoriesi)

HAKoDA,

Tomoyasu

NAKAMuRA,

i'x

and

Akiko

KusuHARA

K),ttshuLJitivexsitv'

The

present

stud}, examined

the

relationshjps among

persona]ity

traits,

physical

characteris･

ties,

and occupational categories

derived

from

Iistening

to

hurnan

voices.

Stimuli

were

26

tape-recorded male

vocaL

samples.

In

Study

1,

,l7

undergraduate

andlor

graduate

students

rated

each

of

these

stiinuli

on

22

personaLity

traits

and

11

vocal

characteristics.

I)articipants

were also

given

a

list

of

34

occupational categories

from

which

to

select

the

most adequate occupational category

for

each of

the

26

voices.

In

Stud.v

2,

ninety

junior

collcge students rated

the

vocal samples using

the

twenty

occupational

categeries

which

had

been

most

often

selected

in

Study

1.

Factor

analyses

produced

three

factors

each

for

personality

characteristics,

for

vocal characteristics, ancl

for

occupational categories.

Also,

significant corre]ations among

those

factors

indicated

that

partici-pants

consistently attributed

_particular

_personality

and vocul characteristics

to

specific voices and

assigned

those

voices

to

particular

occupational categories,

Stereotypical

perception

of voices

is

discussed

in

terms

of earwitness

identification.

Key

words :voice,

impression

formatien,

personality

characteristic, vocal characteristic,

tional

category

The

present

stucl}, acldresscs

the

_question

of

whether vuices are

interpreted

in

tenns

of

_personality

characteristics, vocal characteristics. and

oecupa-tional

categories,

Several

studies

have

assessed

the

relationship

between

participants'

perceptions

of

the

facial

characteristics of others and

their

impressions

about

them,

In

_general,

people

seem

to

make

judg-ments about othcr

_pcople's

_{persona]ities,}

nccupations,

intelligence,

and attitudes

by

observing

their

faces.

That

is,

the

outward

appearance

of

the

face

is

used

as

'

_Department

of

Cognitive

I'sychology,

Faculty

of

Letters,

Kyushu

University,

6

19

1,

IIakozaki,

Higaslii-ku,

Fukuoka,

812

8581

1)

IKie

ttre

_gratefur

to

Dr,

Sachie

Nakamizo

at

Kyushu

University

for

his

precious

comments

and

suggestiens

fer

the

present

research.

We

are

also

express

our

appreciation

to

Dr.

Don

Duffy

for

his

critical reading and ehecking

the

later

version

()f

the

present

paper.

Thanks

are

also

te

Dr.

IIiroshi

Arao

for

h{s

suggesti ons

for

physical

measurements of voice stimuli.

an

index

of

the

inward

character

(Goldstein,

Chance,

&

Gilbert,

1984).

Especially

in

the

context of

the

criminal

justice

system, stereotyping on

the

basis

of

facial

features

could

lead

witnesses

to

select

the

wrong

person

in

a

police

lineup

and could affect

jurors'

judgments

of

_guilt

and

innocence

(Yarmey,

1982),

Since

facial

features

elicit similar responses

irom

_participants

Ca

particular

face

oftcn eiTokes

consensual responses

froin

observers), attention

has

mostli,･

been

directed

to

information

processecl

visu-ally

rather

_than

auditorily.

IIowever,

the

re]ationship

between

perceivers'

cognition

of

the

vocal

character-istics

and

their

impressions

have

not

been

examined

much.

Thus,

in

the

_present

study, whether

_people

do

make

judgments

about

others'

_{personalities}

and

occu-pations

only

from

hearing

_the

target

_persons'

voices

was

_questioned.

Also,

which

_personaiity,

vocal,

and

oecupationtt1 category

djmensions

perceivers

tend

to

use when

forming

impressions

about others

from

the

2

The

Japanese

Journal

ofPsvchonomicScience

Vol,

2U,

No.

1

IIayashi

(1978

a) mentioned

that,

when we observe

u certain

face,

we

tend

to

inake

a

certain

kincl

of

personality

judgment

based

on

the

physical

character-istics

of

the

face.

This

forming

of an

impression

ef

personality

from

facial

features

is

a consistent

ten-denc>,.

Cultura]

stereotypes might

be

operativc

in

face

recognition.

I-Iayashi

concluded

that

there

was

a

relation

between

physical

characteristics und

person-ality charucteristics

in

face

recognition.

Bruner

and

Tagiuri

(1954)

and

Cronbach

(I955)

proposed

"implicit

personality",

suggesting

that

when we

judge

another

person's

personality,

our

judgments

are

likely

to

be

biused

in

a

relatively

stable

fashion.

The

content

of

the

basic

dimensions

eperating coinmonly

in

personality

juclgments

of others w･as studied

by

Hayashi

(1978b).

Three

factors

(farniliarity,

social

desirability,

and activity) were extracted.

IIe

con-cluded

that

we mainly use

these

three

dimensions

whcn

we

judge

another

person's

personaliLy

from

facial

features.

Thus,

facial

features

may

function

as

u vis'ible

index

of

behavioral

dispositions,

and

niany

observers within a culture wM seem

to

agrce about

the

"meaning"

of

_particular

facial

configurations

(Goldstein,

Chance,

&

Gilbert,

1984).

If

the

stiinuli

are

voices,

is

it

_possible

to

ascribe variables such as

personality

traits

or

vocal

charac-tcristics

to

the

voice?

According

to

Aronoi,itch

(l976),

there

are currelatiens

between

personality

judgments

of voices and

the

vocal characteristics.

Appurent]y,

_I)e'ople

do

make

_persc)nality

judgments

about

other

people

based

on

vocal

cues.

For

example,

ratcrs

in

the

Aronovitch

study

(]976)

consistently

judged

certain voices as reflecting

variou$

personality

characteristics, ragarclless of whether

the

person

whose voice was

being

rated

possessed

those

charac-terlstlcs.

Much

of

the

research on

voice

idantification

has

invorved

earwitness

identification

studies

investigat-ing

mcmory with rcgard

to

a criminal suspect's voice

(Bull

&

Clifford,

1984

;

Yarmey,

1986

;

Deffenbacher,

Cross,

Handkins,

Chance,

Guld$tein,

Hammersley,

&

Read,

1989;Yarmey,

1991,

1992>.

For

example,

Yarmey

(199D

mentioncd

that

soine

crinies

include

both

visual

and

auditory

inforrnation.

In

crimes

such

as

those

cumniitted

in

the

dark,

or with

the

perpetra-tor

wctaring a mask, or

those

committed over

the

telephone,

the

only

source

ef

identification

evidence

may

be

auditory.

Thus,

it

may

be

necessary･

to

inv･es-t{gate

whether auc]itory characteristics evoke

consen-sttal responses

from

observers.

When

we

hear

the

voice of an unfamiliar

pcrson

over

the

te]ephone,

we may

_gucss

what

type

the

person

is,

ancl

his/her

occupation or social status.

At

that

timc,

we may use

the

_physical

voice

characteris-tic$

such as voice

tone,

_pace,

or

loudness

to

form

a stereotypical

impression

about

the

person.

Impres-sions

derived

from

physical

voice churacteristics seem

to

be

closely related

to

$tereotypical

judgments

about

the

_person

speaking.

Thus,

in

the

present

study,

we

tested

the

hypothesis

that

voices

can serve

as

stimuli

that

trigger

consensual

stereotypica]

responses

in

observers,

Several

appreaches

have

been

made

in

resenrch of

voicc

itsclf.

For

example,

Scherer

(1974)

reported

the

importance

of

_pitch

and

loudness

as

the

most,

power-ful

vocal

dimensiens

affecting

listeners'

judgments.

Also,

",hether

the

use

of

a

_powerful

or

_powerless

style

affects

thc

perccptions

of

the

listeners

was

investigat-ed

(Erickson,

Lind,

Johnson,

&

O'Barr,

1978).

They

found

that

the

_{powerful-powerless}

manipulation

affected

not

only

participants'

perccptioTis

of

the

speakar's

credibility･ ancl attractiveness

but

alse

their

acceptance

of

the

information

cont,ained

in

the

speaker's

testimony.

Additiunally,

there

are

sevcral

studies ()fvocal attractivaness

(e.g.,

Zuckerman,

&

Driver,

1989;

Berry,

1990:Zuckerman,

Hodgins,

&

Mivake,

19YO).

Several

studie's o]'iearwitness

identification

hav･e

investigated

rnainly

the

length

of

the

retcntion

inter-va] after

hearing

the

suspect's

voice,

the

voice

idcntification

of a suspect,

and

t.he

subjective

confidence ef voice

identification.

Observers

presuin-ably use

_physica]

voice characteristics when

the

only

stimulus avaiTable

is

a voice, and

those

characteris-tics

of

the

voice

are

used

us

an

index

ol

the

inward

character.

Thus,

by

using

factors

such

as

the

physical

characteristics ur

impressions

such as `'intal]igcnt" or

"kind",

_]isteners

might

form

a certain

kind

of ster-eotypical

imagc

of

the

_person

(e.g.,

this

person

souncls

like

a

person

working

in

_tbe

field

of

business),

NII-Electronic Library Service

N,

YAMAfm,

Y.

HAKol)A,

T.

N,"K.M,IuT".A,

S.

.

in

the

absence of

that

_person's

_physical

appearance.

Kramer

and

Aronovitch

(197'O)

found

consistent

and

considerable

interrater

reriabilit.v,

which

has

been

also

evident

in

man},

other

studies

(Pear,

]9e,1;

Stagner,

1936;

Sanford,

1942;

Licklider

&

Dv'liller,

195D

that

atternptecl

te

assess

the

validity of

judg-ments of voice.

IL'his

phenomenon,

often referred

to

as voice stcrcotype,

has

bean

considered

t.o

bc

a

confoundirig

variable

_that

interferes

with

a

rater's

ability

to

assess

accurateLy

_personality

Lraits

from

vocal cues.

Kramer

(1964)

defined

"$tereutyped voices" as voices conveying a stereotype of some

personality

trait

",ithc}ut

the

ste,reotypc neccssarily

having

any validtty;

huwever,

ether

situatiuns

in

which

the

interpretation

of voices ralates

to

a

cal.e-gory

stereotype such as scx

should

alse

be

considered.

Fur

instance,

Aronovitch

(1976)

also mentioned

that,

although

both

niale and

feniale

x･,oices are

stereotyped

iii

pcrsonality

judgments

from

voice,

the

cues upun which

these

stereoryped

judgments

were made ciiffered

between

the

twu

sexes.

Thus,

the

finding

ef

that

study

reflected

the

cultural

stereotype

and

impried

that

modes

of

personality

perception

based

on

voice

cues

are,

in

part,

based

on

these

stereotypes.

One

of

the

stereotypes

that

shapes our

impressions

uf

human

vuices

is

considered

to

be

the

occupationa]

categories assigned

to

a

voiee.

The

question

ad-dressecl

here

is

whether

voices

are

meaningfully

related

to

categories.

Is

a

diffcrent

interpretation

of

a

voice

rnade

when comparing a voice

labeled

by

occupation

with a voice not

Iabeled

by

occupation?

Several

studies

havc

demonstrated

that.

fae{al

stereo-types

have

un

infiuence

on

the

memory of

faces

(Shoemaker,

South,

&

T.owe,

1973

_;

Klatzk},,

Martin,

&

Kanc,

1982;Yarme.v,

1979,

1982).

Hu",ever,

research

demonstrating

_the

existence

of

vocal

stereo-types

for

occupational

categorization of voiccs

has

not

been

donc,

and most especially

the

relation of

such stereotypes

to

personaLity

traits

and

_physical

(vucal)

characteristics (]f

the

voices.

_.

Yarmey

(1991)

mentioned

_that

stereotypes

existed

in

thc

relationship

among

_physical

appearance, voice characteristics, and certafn

personality

charaeteris-tic$.

For

example, according

to

that

same study,

SllllcASAx-,,･1

and iX.

KvsuEd,"E{,i

:

hnpression

of voice

:l

inclividuals

attributed specific

_personality

traits

ancl

vocal characteristics on

the

basis

of stereotypes

regarding

beth

speech

aiid

_personal

appearance.

That

is,

participant,s

systemat{catly associated

cer-tain

pcrsonality

traits

with ccrtain vocal attributcs.

Yarmey

(1991)

showed

thut

there

were signifieant

relationships

bet-'een

a number of

the

personaliLy

traits

and

_particu]ar

vocal characteristics.

Yarmey

(199Z)

employed

three

video

conditions

in

which

{0

participants

saw

and

heard

target

people

produce

an

identical

speech,

{2)

participants

only saw

the

targets

(nonspaaking).

or

(3)

participants

only

heard

the

targets

Cnonvisual).

in

one of

the

tasks,

partici-pants

selected

the

target

person

who

best

represented

a

mass murderer,

a

sexual

assault

fe]on,

and

an

armed

robber

(categorized

co]lective]y

as

'ibad

guys''),

and a clergyman, a medical

doctor

and an engineer

{`'good

guys").

The

finding

suggested

that

participants

systematically associated

particular

traits

of

the

two

classes with

_particular

vocal attrib-utes.

In

that

sense,

the

concept cuncerning "good

guys"

and "bud

guys"

is

censidered

tu

be

strungly

relatccl

to

categorical

labels,

These

labelswill

lead

to

a

better

crassification of voices, since

the

findings

from

research on

fuce

recognitien suggest

that

cate-gorica]

kno-']edge

rnig.

ht

direct,

the

viewer

t.o

particu-larly

inforrnativc

rcgions of

the

face

so

that

more

physical

detail

is

encoded

(Klatzky,

et

al.,

19S2).

For

example,

labeling

a

face

with

a

congruent

occupa-tional

category name at

the

tirne

the

face

is

presented

increased

recognition.

Faces

more easily stereotyped

were

better

recognized.

Thus,

the

rnain

_purpose

uf

the

present

study

was

te

investigate

how

adjectives representing

_personality

traits

and vocal characteristics ure related

to

stereo-typed

categories

by

examining

the

structural

configurations

of

those

threa

factors

influencing

judg-ments of voices and

the

structurcs of

those

factors.

Also,

we

tend

to

have

stereotypical

images

of a

person

of a certain occupation.

That

is,

'`Po]iccmen

are

authoritative"

or

"Salespersons

are

talkat{ve",

Thus,

stereotypical

images

for

the

certain

occupation

seem

to

exist.

Therefore,

are

there

any stereotypical N･,oices

for

lhe

person

of a certain occupation?

4

The

Japanese

Journa]

parsons

in

certain occupations such as university

professors

or

politicians

(e.g.,

Shirasawa,

Hakoda,

I/Iaraguchi,

&

Yamada,

1999)

has

been

dcmonstrated,

the

correlatiuns among vecal characteristics and

personaHty

characteristics

derived

from

voice chan-nel and occupationar catcgories

have

not

been

inves-tigated.

It

was also

cons{dered

important

to

exaniine

corre-lations

among

the

three

sets

of

variables.

The

hypothesis

w･as

that

voices would

be

matched reliably

with occupational categeries.

In

Study

1,

personality

characteristics and vecal

characteristics

were

examinecl

to

extract

factors

underlying

judgments

of

human

voices.

In

Study2,

the

correlations and

thc

structures

o[

_persenality

characteristics, vocal

char-actcristics,

and occupational categories were examined.

That

is,

how

factors

extracted

from

judg-ing

voices with respect

to

stereotypecl occupational categories are associated with vocal characteristics

and

_personality

characterist{c$ was examined.

The

reason

for

examination

of

the

re]atednass

amc)ng

persoriality

characteristics, vocal characterist[cs, and

occupational cutegories was

to

discover

whether

the

findings

from

the

studies about cognitix,e

judgments

from

human

faces

(Shirasawa,

et al.,

1999)

could

be

also

applicable

to

cognitive

juclgments

from

human

volces.

In

the

study

by

Yarmey

{199]),

a

target

vuice

heard

originally over

the

telephone

was

played

among

tape

recordings of six other voices;

the

target

was eithcr

present

or absent

in

the

lineup.

Using

telephone

lines,

Rathborn,

Bull,

and

Clifford

(1981)

found

no.

differences

in

identification

of a

targct

voice

heard

originally over

the

telephone

and

tested

using a

tapecl

]ineup

over

the

telephone

in

contrast

to

identification

of voiccs

hcard

originally over

the

telephone

and

tested

directly

with a

tapecl

lineup

(Yarrney,

1991).

Thus,

in

the

_present

study, voices

_played

on

a

tape

recerding

instead

of stimulus voices

heard

ov･er

the

actua]

telephone

lines

wcre

used as vocal samples

to

be

rated

by

participants.

Furthermore,

C]ifford

and

Davies

(1989)

stated

that

the

duration

of a speech

sample rnakes

Iittle

difference

to

later

accuracy of

identification

if

at

least

one sentence

is

heard

_;

huw-ever,

when

a

sentence

or

conversation

is

the

stimu]us,

of

Psvchononiic

Science

Vol.

20,

No.

1

differences

in

the

actual worcls usecl

in

the

sentence or

the

cenversation

causecl

impressiu'

ns

formed

by

par-ticipants

to

vary with

differences

in

content.

There-fore,

in

the

_present

study, vocal stimuli were

limited

to

the

_phrase

"]N･[oshi-moshi"

(`CHerlo"

_in

_English).

Study

1

Method

Partieipants

Participuntswere47

(3U

women and

17

inen) col}ege students

(from

18

to

31.

_years

old) who volunteercd.

Veeal

Samples

The

voices

of

8L

males

(from

19

to

52

years

old)

were

recorded

on a

digital

audie-recorder.

Voice

samples

were

_gathered

en

the

univer-sity campus and

in

the

city

park

aftcr obtaining consent

from

the

speakers,

Since

male samples need-ed

to

be

speaking

in

a nurmar situatien and since

the

centent of speech

is

important

in

judging

the

speaker's

_persenality,

in

_thc

_present

study

the

content

was restricted

to

-`rLCoshi-Aaoshi.

)vloshi-lv,Ioshi,"

(Hel]o.

Hello.

in

Engrish).

This

short

phrase

is

usually spoken when making and receiving a

tele-phone

call, and anyone of any occupation eommonly

uses

this

_phrase.

Further,

the

_phrase

is

_probably

short

enuugh

to

_produce

ne

_persunulity

biases

associated

w{th

length

of speech and cont.ent.

The

82

origina]

voice saniples wcrc

taken

into

Adobc

Prcmicr

{a

software

for

_picture

and sound composing) using a

Macintosh

computer and were edited

to

reduce

back-ground

noises.

Of

the

82,

26

voices wh{ch seemed

to

be

relatively

clear

and

appropriately

variecl

were

se!ected

and

recorded

on

a

tape

recorder.

The

Z6

x,oices were

mainly

selected

on

the

basis

uf speech characteristics such as "powerful"

ancl `"powerless"

since

those

variables

have

been

shown

to

infiuence

jurors'

sentiments and

juries'

decision$

(Erickson,

et.

al.,

1978).

Also,

the

authors

considered

traits

deriN'ed

from

voices

such

as

L'social

desirabi]ity"

and

"social undesirability" when selecting

the

26

voices.

These

tra{t.s

were chosen since a social

desirability

factor

was extracted

in

research on

faces

(e.g.,

Hayashi,

'1978a;

Hayashi

1978b;

Shirasawa,

et.

a].,

1999),

Raters

_judged

the

voices with a simple "match"

or "not

match"

answer

on

the

basis

of

the

four

traits

NII-Electronic Library Service

N.

YA",iADA,

"t'.

_{I-IA[tuu,･k,}

_T.

_{N,y{,u,iu[(,x,}

S.

_{SmTa,xs,･uL',i}

and

A.

undesiruble).

Also.

the

voiccs wcrc

selected

accord-ing

to

their

clarity

and

noiselessness, rate,

_pitch,

or

LiSt

of

22

varratlon.

KL,sL.'+m]{,y

Impression

of

Table

1

personality

charucteristic

(a)

characteristics

(b)

volce

5

and

11

vocal

Range

of

frequencies

of

the

26

vocal stimuli ",as

from

82

Hz

to

233Hz.

According

to

Utsuki

(1982),

aver'age

frequency

of

human

male voices

is

from

1{}O

Hz

to

150

Hz.

Vocal

stimuli used

in

the

_present

study

seem

to

huve

appropriate

variatiens

of

fraqucnc),

Teve].

Frequency

leve]

of each voice was measured

by

using

SP4"JIN

Custom

(voice

analyzing software).

Alse,

the

weighted

intensities

of

the

tones

of

the

vucal

stimuli measured

bv

a sound

level

met/er varied

between

60

dB

and

75

dBA

(fast-peak)

at

comfort-able

level.

The

sound

level

meter was

_pluced

at

the

position

of

ears

vf

participants.

Various

voices which

had

distinctivc

eharacteris.-tics

rncntioncd above were mixed and

included.

The

26

voiccs wcrc randomly ordered.

Thus,

a

tape

ef stimulus voices used

te

elicit raters'

judgrr)ents

was

prepared.

The

average

duration

of

the

stimulus

voices

was

2.5Z

seconds

Crange

frem

1.:S

seconds

to

4,O

seconds).

Personality

and

Vocal

Characteristics

rating

instrument

A

scale

for

rating

the

26

voices on

personality

and on vocal characteristics

included

22

personality

traits

and

11

vocal characteristics.

In

a

study

by

Shirasawa,

et. al.

(1999),

the

influence

of

personality

iinpressien

and

physiogn(}mic

fe,aturcs

on

the

occupational categorization of

faces

was

inves-tigated

using

22

_personality

traits.

Their

findings

sugg ¢sted

that

personality

charactcristics rcpresentcd occupational categories.

Also,

according

to

studies

by

],Iontepare

and

Zebrowitz-TL'IcArthur

(1987)

and

Yarmey

(1994),

six

voca]

characteristics

(very

high,

very

loud,

tight,

clear,

changing,

and rapid) -rere

rated using

the

7

point

bipular

scales.

In

addition

to

these

scales,

five

new vocal eharacteristics

(not

blur-red, not stuttering, not stiff, not

trembling,

and unmuMed) were eniployed

to

examine more subtle

impressioris

which could

be

exprained

by

the

siX vocal

characteristic

traits

included

in

the

English

version

of

the

scale.

The

authors considerecl

these

new

fivc

vocal characteristics

to

augment

the

original six characteristics used

by

rvTontepare,

et al,

(1987)

and

Yarmey

(1994).

Each

of

the

26

speakers' i,･oices was

(ft)

Personality

Characteristics

lively

favorable

Elctlve safe social

intel]igent

unselfish

humorous

eptlrnlstlc

warm

bright

(b)

Vocal

Characteristics

not

blurred

not stuttermg stiff relaxed

definite

rapid

trustworthv

extrovert,ed

quletconsclenUous

neatslncerecheei'ful stablcmaturedmeekkind not

trembling

very

high

uninufiled verv

ioud

clear

rated

by

the

22

_personality

characteri$tic

adjcctives

and

al

vocal characteristic adjectives on

7'point

scales.

The

22

pers()nality

charncteristic

traits

and

11

vocal characteristic

traits

are

1{sted

in

Table1{u)

and

<b).

Items

fur

_personalit},

traits

and vocnl

charactnris-tics

were randonily ordered.

Procedure

Partieipunt,s

were

instructed

after

lis-tening

to

each stimulus

te

rute

all

of

t.he

adjectival

items

of

both

the

_personality

and

vocal

characterls-tics

using

the

7

point

scale.

A

respunse of

1

was

defined

as "the

stimulus voice

is

nc}t

typical

of

the

trait

at all", while

7

indicated

"the

stimulus voice

is

typical

of

the

trait".

Finarly,

_participants

were

in-structed

to

select one occupation out of

3･1

job

cate-gories

se]ected

frc)m

relatively

_popular

occupations

in

Japan.

']'he

occupat{ons were

those

used

in

the,

study of

the

occupational categorization

for

faces

by

Shi-rasaura, et al.

(1999)

who asked

_participants

Lo

select

the

"ts,pica]''

occupation

for

each

face

prcse,ntcd

on

the

slide

projector.

As

a resutt,

248

occupational names were obtained; of

these

occupations,

the

34

occupntional categories selected

by

at

least

109.(,

of

participant.s

were

taken

for

the

_present

stucly.

Some

6

The

Japanese

Journal

of

Psychono}nlcScience

V'

ol.

20,

No.

1

others

",ere

of

lc)x･v

social

status.

The

authors

thought

that

the

criminal

1]chavior

rnight

be

categorized as an

attitude

related

to

a

certain

$ocia!

aspect,

su

the

name

"sexual

crirninal" ",as

included

{n

the

occupation

list

tu

rclute clcarly

to

a criminal

behavior.

Occupatic)ns

were randomizecl and

_printed

for

_{presentation.}

Rcsults

and

Discussion

T",o

factor

analyses "'ere

performed

to

examine

the

factor

structures

of ratings

of

pcrsonality

traits

and vocal characteristics sepa-rntely.

F(}r

thu

personality

characteristie

adjectival scale,

the

_principal

component

facter

analysis

yielded

three

factors

-,ith eigcnv･alues

_greatcr

than

1.8.

Trait

variables

load{ng

on

the

"social

_{desirability"}

_facter

",ere "[avorabled', "saie", and '`unselfish".

_Trait

vari-ables

loading

en

the

"activity''

factor

were "tiveiy",

'`acLive",

_and

i'social",

_while

_variables

_such

_as

L'intelli-gent",

"Lbright'',

or '`stable"

hud

higher

loadings

on

the

''intelligence"

factor,

Those

items

on

the

persenality

characteristic adjcctival scale showed

loadings

higher

than.

50

on

the

threc

factors.

These

three

fHctors,

whose eigenv･alues exceeded

1.8,

were named

i`social

desirability",

`'uctivity",

and

"intelligence"

respective-]y.

These

three

factors

exp]ain

58%

of all

the

vari-ance,

Correlations

between

factor

scores of

the

personality

characteristic

factors

and

the

vocal

char-acteristics

factors

are

listed

in

Table2.

Table3

shows

Pearson

correlations

between

ratings

on

cach

of

thc

personality

traits

and mean ratings on each of

the

three

personality

trait

factors.

In

addition,

three

factors

were extracted after exarnining

the

factor

st.ructure. o'fvocal characteristics of

the

voices.

Vocal

character'istics variables

loading

on

the

"clarity"

fac-tur

svere `'not

blurred"

and "not

_{trenib]irig''.}

_Vocal

c.haructeristic vuriab]es

loading

on

the

"lightness"

factor

were

"very

high"

and

'`not

stiff",

while

vari-abres such us `ttight"

or C`very

_loud"

coinprised

the

"soundness"

_factor.

_Those

vocal characteristics

fac-turs

whusc eigenvalties exceeded

1

.8

were

interpreted

"clarityT', `ilightncss",

and "soundness"

factors,

Those

items

on

the

vocal characteristic scale showed

load-ings

higher

than.

50

on

the

thrce

factors.

These

three

factors

explain

58%

of all

the

variancc

in

the

rating

of vocal characteristics.

Overall,

socially

desirable

variables

iiicluding

"safe'`, "unselfish",

and

"warm"

were

posiLively

related

to

the

clarity

vocal

character-{st{c

yariables

such

as

"unmuMed", "not

stuttering", or "Crelaxecl".

In

[['able

_4,

correlations amung

the

10

vocal characteristics and eaeh of

the

three

factors

of vocal characteristics are shown

(the

variable "slow-rapid" was

dropped

from

the

analysis

because

of

its

low

correlation

with

scores

on

the

vocal

characteris-tic

factors),

Furthermore,

frequencies

of occupations selected

bsJ

_participants

were aver'aged.

Twenty

eccupations

which were selected as

the

st,ereotypical occupations

for

each of

the

26

voices were retained and used

for

the

next

study.

Therefore,

participants

were

_g{ven

a

list

of

34

occupational cutegory nunies used

in

the

study

by

Shirasawa,

et al.

(1999).

Each

of

47

partici-pants

chose one occupation

from

the

Iist

which

hclshe

thought

to

be

the

most suitable

for

the

voice

just

heard.

After

rating

22

_personality

adjectival

items

and

11

vocal adjectival

iteins,

_participants

were

required

to

select

only

one

occupationa]

name

from

the

3･1

occupational

categorical

names.

Altogether,

out of

34

occupational category numes chosen

by

participants

(47

partie.ipantsX26

occupational names

corresponding

to

the

26

v･oices),

ZO

occupational naines wera

selected

at

least

30

times

by

_{participants,}

regardle$s of which stimulu voice wus rated, and

Correlation

matrix

for

Table

2

personality

characteristics and vocal

characteristics

Vocal

CharacteristicsSocial

Personnlity

CharHcteristics

dcsirability

activity

intelligence

claritylightness

soundness

*

_P=.05.

ii

_P<.Ol.

49*86**21

56**45.,86**

53**-.41*

NII-Electronic Library Service

N.}",xpa{,'II)A,

N'.

H,M<oE).,i,

T.N,･IK,ll･ltJICA,

S.Sllll{ASAWAand

A.

Kus-i,vt,y

lmpressiun

of voice

7

Rotatedfactor

matrixTable3

for

_{personalitytraits}

factors

Adjective

social

desi.rli!lility

actlNiltv-111aCtlVlt.Y

intelligence-lack.

of

intelligence

conscientioLIs

safekindfavorable

slnceretrustworthy

IVaMTIunselfishneat

.828280.7g.76,75.737362

19

17

.17

.l39

14

.17-35

.09

12

ext.rovcrted

very

soft

meekactlvelivelvoptimistic cheerful

humQrous

socialmatured

brightintelligent

stable

-.44

07･

.{}3

.09

.42

.37

?,9

.40

44

8385S276757371.64.6106050249

.17

.17-.05

.07

.:s:s

32--.06

-.10...5,3.."

17

.C)8

15

.28

Ol

.03

.10

.06

17

()[1)3743ol

.76.7061.61

Table

4

Rotated

factor

maLrlx

for

vocalcharacteristics

Adjectives

c}arityVocal

Characteristics

lightness

soungl.ness not

tremb]ing

not

b]urred

not

stuttermg

777863

.03.06.38

.02-.07

26

very

high

stiffunmuraed

121042

,78-.74

.50

.20.oz

17

relaxedverv soft

defineteclear

Ol18044S -

l9-.08

.20

,10

.77-.75

.64

63

rapid

.28

.tl

.23

these

occupational names were

included

in

lhe

follow-ing

analysis.

This

_precedure

was

_performed

because,

in

the

free

selection

by

the

_{participants,}

concent/ra-tions

of

certain

occupational

names with

certain

voices

did

not seeni

to

occur.

In

orcler

to

organize

the

occupatioT)al category names used

for

the

analysis, using

frequently

selected occupational nan]es

appear-ed

to

be

more

apprc)priate

than

using all of

the

34

occupational

catcgories.

The

20

occupations

listed

by

participants

at

least

30

times

are

listed

in

Table5.

Factor

scores were calculntad

for

each

of

the

Z6

voice

sainples

(after

ca!culating each

purticipant's

factor

scorcs

for

cach of

the

26

stimulus voiccs,

then

multi-prying

them

by

,t7 and averaging

them

by

dividing

by

8

The

_Japanese

_Journal

of

Tuble

5

List

of

ZO

occupational categories

([requency

used

by

raters)

I'sychonomic

(Seryice

Trade

Occupation)

1.

store

clerk

(sc)

2.

grocer

(36)

3.

taxi

driver

(39)

4.

station

staff

(40

5,

teacher

6,

salesperson

(63)

7.

entertainer

(32)

8.

postman

{38)

9.

newscaster

(38)

(Oustider

Type)

10.

sexual

_pervert

(58)

11,

ss&'indler

(3S>

12.

cult

leader

(35)

13,

niale

_prostitute

(3])

14,

artist

{311)

15,

homcless

_person

U3)

(Authoritative)

16.

inedical

doctor

UO)

17.

Iawyer

(33)

18,

_gangster

(32)

19,

_policeinan

(3Z)

20,

politician

(30)

47,

thus

ca]culating

meun

factor

scores

for

each of

the

26

stimulus

voices on each of

the

three

_personality

characteristic

factors

and on

each

uf

the

three

vocal

charactcristic

fuctors).

These

calculations were

done

to

see which voice

is

associated with which

personal-ity

characteristic

factors

and which vocal

characteris-tic

factors.

Calculating

factor

scores

of vocal stimuli

was

necessary

to

examinc

the

correlations among

facl.ors

extracted.

Study2

Study

1

previded

three

factors

extracted

from

rat-ings

of

_personality

characteristic

traits

and

three

factors

cxtracted

frorn

vocal characteristics.

There

were also sev･eral significant correlations among

those

factors.

In

Study1,

_participants

selected

freely

one

occupational category

from

_the

34

occupation

names

for

each of

the

26

voices

after

rating

the

personality

traits

and vocal characteristics.

Twenty

occupations which

had

been

selected at

least

30

times

by

partiei-pants

were used

for

a rating scule employed

in

Stuc]y

2,

as explained

above.

A

7'point

occupational

Science

Vol.

20,

No,

1

gorical

scale was used

to

examine whether

listeners

associared

particular

traits

or

characteristics

with

particular

uccupational catcgories,

Study

2

examined

whether scure$ on

personality

traits

factors

ancl

those

on vocal characteristics

factors

extructed

in

Study

1

were meaningfully related

to

occupationa] category

factors.

Also,

in

Study2,

which

occupational

cate-gories

wou]d

be

obtained

from

the

vocal stimuli w･ere

examined,

Method

Participants

A

tutal

of

90

junior

college students

(15

males and

75

females,

ages

from

18

to

Z4)

par-ticipated

in

thc

study

to

obtain

course

credit.

Procedure

The

26

voice samplcs wcre

identical

_to

those

of

Study

1.

The

tape

of voices was

played

to

a

group

of

90

coilege student,s.

The

order of

ZC}

occupa-tions

selected

from

Studyl

vvas randun]ized and

presented

in

a

booklet.

Ninety

_participants

rated

the

20

occupationa] categories

for

each ()f

the

26

voices

on

7

_point

scales.

Altegether,

46,8eO

items

of

data

(9e

×

2e

×

26

voices)

were

cntered

for

ana]),zing.

A7

point

scale was employed with

1

:

`tthe

voice

is

not

typical

ef a

person

in

the

occupation on

the

check-list",

and

7:

'`the

voice

is

typical

of a

_person

in

the

occupation on

the

checklist".

Occupational

category

rating

scale

From

Study

1,

frequencies

of eccupations selected

by

participants

were

averaged:

20

occuputions

which were selected as stereotypical

for

each of

the

stimu]us voices were retained.

Twenty

occupational names were selected

at

least

30

times

_by

_{participants.}

_This

selection

was

employed

because

in

the

_pilot

study,

when

purtici-pants

werc

instructed

_to

se]ect

one

suitable

occupa-tion

after

hearing

vocar stimuli,

they

wera

like]},

to

choose certain occupational categorics such as "orace workcr'7 and "univers{ty

studenti' which were very

popular.

This

result mentioned

that

_participants

could not

image

a widc range

of

occupational

cate-gories

in

the

limited

time,

Results

As

in

Study1,

the

iterated

principal

fac-tor

ana]ysis was also

_performed.

The

iterated

princi-pal

factor

analysis

_yielded

three

factors

with

eigenvalues

_greater

than

1.9.

A

thrcc-factor

selution

with varimax rotatiun

to

iterate

the

solution

was

NII-Electronic Library Service

N.

YA"･mDA,

}fr.

HAKoDA,

1'.

NAKA"amA,

S.

SiimAsAwA

and

A.

KusuJmRA

:

ITripress{on

uf voice

Table6

Correlations

between

occupational-eategory

factors

(N=90),

personality

charaeteristic

tors

and x,ocal characteristic

factors

(Ar-114)

9

tt...t.t.

tttttttt

tttt

ttttttttt

Occu]]ational

CategoryFactorsI"ersonalitvC'/haracteristicsFactors

'

socialdesirabilityactivityintelligence'VocalCharacteristicsFactorsclaritylightnesssoundness

servicetrade

out${dertype authoritative

.74*rk

.58**-,10

--.24

_.49**

_.39

-.58'"

_.e8

_.7'o*s

.tt/6*rk,73S*,39

I-.64**.14--.52*

.17-.7･1*.-.35

1.9

explained

52.IY6

of

thc

variance.

Frorn

the

solutien of

the

v･arimax rotation,

the

three

factors

were named "seryiee

_trade",

"outsider

_type",

and

'`authoritative".

_Facter

_1,

"service

trade

occupation",

included

salesperson,

taxi

driver,

newscaster, and

entertaincr-thosc whe ineet and

intoract

with

pcople

daily,

usual]y

being

related

to

service

for

the

_public.

Those

variables

loading

on

the

'`service

_t.rade"

occu-pation

factor

showed

loadings

higher

than.

6{}.

The

second

factor,

"outsider

_type

occupation",

included

gang

mernber,

cu]t

leader,

sva,ind]er,

artist,

and

home-Iess

person-occupations

usuallsJ

considered

lo

be

]ess

unsociuble;howev･er,

it

is

doubtful

thHt

"honieless

person"

could

be

treated

as an occupation although "homeless

person"

was

perceived

us u $tereutype representing

a

certain

type

of

pc'.rson

in

society.

The

variables

loading

on `'outsider

_ty･pe"

occupation

factor

showed

toadings

higher

than.

5U.

The

third

factor

was nanied "authoritative occupation",

in-cluded

politician,

lawyer,

policeman,

and

doctor-occuputions usually considered

to

be

mure uficiul and authoritative.

These

vafiables

loading

en

the

`[authoritative"

occupational

factor

also showed

load-ings

higher

than.

50.

Mean

fucter

scores

for

each ef

the

26

voices

were

caiculated

for

each

of

the

three

factors.

U$ing

mean

factor

scores,

Pearson

corrcla-tiens

of

the

three

vocal charact.erist.ic

factors

with

the

three

occupational category

facters

were calculuted

(Table6).

Table6

also

describes

t,he

significant

correrations

batween

thc

three

_personality

character-istics

factors

and

the

three

occupational

categery

factors.

General

I)iseussion

In

the

present

studb･,, whether

there

are certain

meaningful relatjonships ameng

_personality

churac-teristics,

!,ocal characteristics, and occupational

cate-gories

clerived

frorn

listening

to

human

voices was

examined.

As

such,

Pearson

eorrelatiun methods

were

employed

in

order

to

clarify

the

associutions

among

those

factors.

The

results

indicated

that

the

stimulus voices which

had

relativery

highcr

factor

scores

tended

to

be

associated with certain

occupa-tional

categories.

The

principal

findings

of

the

present

studies

cem-prise

the

iollowi'ng;

1)

by

using

factor

analysis of

ratings

a]ong

t'he

scale

of

22

personality

characteris-tics,

three

factors

were

extracted

:

these

were

inter-preted

as "social

desirubility",

'`uctivity",

,and

"intelli-gence".

In

usimilar way,

three

factors

were extracted

from

the

ratirigs along

Lhe

scale of eleven vocal

characteristics.

Thesc

wcre

interpreted

as

`[cLarity", '`lightness",

and "soundnessi'.

₂₎

_According

_to

_the

results

in

Study

1,

_participants

consistentl.i･,

attributed

particu]ar

personality

characteristics and v･ocal char-acteristics

to

particular

voices.

3)

Factor

analysis of ratings

in

the

LU

occupational categuries

_gave

three

factors'

"service

_Lrade",

"outsider

_type",

and

`'author-itative".

4)

The

findings

obtained

in

Study

2

implied

that

_participants

also consistently uttributed

particu-lar

occupational

eategories

to

particular

x,oices,

and

5)

factor

scores

were

ealculated

for

each

of

thc

nine

factors.

There

were several significant cerrelations

between

_personality

traits

scales and occupational

category scales.

rs4ereover,

$everal significant

corre-lations

bctween

vocal characteristics

scalcs

and

occu-putional

uategory seales were obtailled.

The

present

study

clearly

demonstrates

that

partic-ipants

were

lil{ely

to

proccss

pcrceptual

stimuli

(voice

sainp]es)

by

means of some

lcincl

of

ster-eotypical

images

or other conceptual

representa-tiens:that

{s,

sc}eial stereot},pieal

beliefs

cati