Title: Language patterns of repressors and sensitizers in personal and impersonal descriptions
Full Citation
Permanent Link: http://ufdc.ufl.edu/UF00097923/00001
 Material Information
Title: Language patterns of repressors and sensitizers in personal and impersonal descriptions
Physical Description: vi, 109 leaves : ill. ; 28 cm.
Language: English
Creator: Brodsky, Stanley L., 1939-
Publication Date: 1964
Copyright Date: 1964
Subject: Psycholinguistics   ( lcsh )
Meaning (Psychology)   ( lcsh )
Psychology thesis Ph. D   ( lcsh )
Dissertations, Academic -- Psychology -- UF   ( lcsh )
Genre: bibliography   ( marcgt )
non-fiction   ( marcgt )
Bibliography: Bibliography: leaves 91-94.
Additional Physical Form: Also available on World Wide Web
General Note: Manuscript copy.
General Note: Thesis - University of Florida.
General Note: Vita.
 Record Information
Bibliographic ID: UF00097923
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 - 000565665
oclc - 13571645
notis - ACZ2084


This item has the following downloads:

PDF ( 4 MBs ) ( PDF )

Full Text






June, 1964


The author wishes to express his gratitude to

Dr. James C. Dixon, who served as chairman of the supervisory

committee. He has contributed considerably of his time and

energy to the planning and development of this study. His

steadfast support and wise counsel are sincerely appreciated.

Dr. Richard J. Anderson, who aided in the resolution

of statistical problems, and Drs. John J. Wright, Hugh Davis,

and McKenzie Buck of the supervisory committee are thanked

for their many valuable suggestions.

Essential to the data collection process was the

cooperation of the following staffs and departments: The

Mental Health Clinic, the University Housing Staff, the C-3

Department, the Speech Department, and the Psychology Depart-

ment. Others who have generously offered their opinions and

comments are Dr. Louis D. Cohen of the University of Florida

Psychology Department, and Dr. Jean Paul Smith and the staff

of the Clinical Psychology Service, Walter Reed General

Hospital. Also thanked are PFC. Keith Kaminski and PFC.

David Trappenberg who helped with the data computations.

Finally, deep gratitude is extended to the author's

wife, Annette, for her support and encouragement throughout

the evolution of this dissertation.



ACKNOWLEDGMENTS .................... ii

LIST OF TABLES .. . . ...... ......... v


ASPECTS OF LANGUAGE ............ 1

Verb-Adjective Quotient ........... 2

Type-Token Ratio ................ 7

Personal Pronoun Frequency ......... . 11

Qualification and Allness Terms ..... . 15



IV. METHOD ............

Subjects ..........

Instruments ........

Administrative Procedure ..

Statistical Procedure ..

V. RESULTS ...........

VI. DISCUSSION ..........

The Language Measures as
Dependent Variables . .

The Language Measures as
Independent Variables ..

. . .. .. 25

. . . . . 29

. . . . . 29

. . . . . 29

. . . . . 31

. . . . 34

. . . . . 39

. . . . . 69

. . . . . 69

. . . . . 81



VII. SUMMARY ......... ..... .... 89

BIBLIOGRAPHY . . . . . . . . . . 91







BIOGRAPHICAL SKETCH . . . . . . . ... 109




1. Summary of Studies with the Verb-Adjective
Quotient ................... 4

2. Summary of Studies with the Type-Token
Ratio ......................... 8

3. Summary of Studies Involving Personal
Pronoun Usage .................. 12

4. Summary of Studies with Qualification
Terms ...................... 16

5. Summary of Studies with Allness Terms . . .. 17

6. Design of the Experiment, Part 1 ......... 36

7. The Analysis of Variance for Repression-
Sensitization, SCAT Verbal Ability,
and Mode of Description ............ 37

8. Language Scores of R-S Groups and Verbal
Ability Subgroups on Personal and
Impersonal Descriptions ............. 40

9. Mean Language Scores of R-S Groups, of
Verbal Ability Subgroups, and of
Personal and Impersonal Description
Samples..................... 42

10. Analysis of Variance Tables for Six
Language Measures ................ 43

11. Language Data Comparison of Extreme Repressors
(N=15) and Extreme Sensitizers (N=15) in a
Population of 144 Subjects .......... 47

12. Means, Standard Deviations, and Extreme Group
Ranges for Six Language Variables in 144
Subjects ................... 49


13. Comparison of High and Low Language
Usage Subjects by t Tests . . . . . .. 52

14. Significance Levels for High and Low
Language Usage Subjects Comparison . . . 53

15. Summary of Significance Levels for High-
Low Comparison ............... 54

16. Mean MMPI and SCAT Scores of Extremely
High and Extremely Low Groups on
Word Usage and TTR ............... 58

17. Mean MMPI and SCAT Scores of Extremely
High and Extremely Low Groups on
VAQ and Personal Pronouns ........... 59

18. Mean MMPI and SCAT Scores of Extremely
High and Extremely Low Groups on
Qualification Terms and Allness Terms . . .. .60

19. Distribution of Extreme Group Means Falling
Outside the 0.05 Confidence Limits . . ... .63

20. Distribution of Subjects by Source . . ... .96

21. Distribution of Repressors, Neutrals, and
Sensitizers on the R-S Scale ......... 97

22. Distribution of High, Mid, and Low SCAT
Groups with Respect to Repressors,
Neutrals, and Sensitizers . . . . . . 98

23. Distribution of Language Variables by Year
of MMPI Administration: A Reliability
Check Using No. of Words, TTR, VAQ, and
Personal Pronouns ............. .. 100

24. Distribution of Language Variables by Year
of MMrI Administration: A Reliability
Check Using Qualification and Allness
Terms .. . . . . . . . . 102



Man has been concerned with language as an entity

since the time of Plato's inquiries into the origin of

languages. Language in relation to psychological variables

is a relatively contemporary area of exploration, falling

within the past century. Edward Sapir pioneered the explora-

tion of this relationship in the United States when he argued

that people see, hear, and otherwise experience largely

because of their language habits. He stated (1940):

Language . is a self-contained, creative,
symbolic organization which not only refers to
experience largely acquired without its help,
but actually defines experience for us by reason
of its formal completeness and because of our
unconscious projection of its implicit expecta-
tions into the field of experience.

In the theoretical framework of Alfred Korzybski

(1941), thought processes are said to be diseased and faulty

as a direct result of language usage. This theory of the

confining qualities of language is dramatically illustrated

by Aldous Huxley (1939):

It's extraordinary the way the whole quality of
our existence can be changed by altering the words
in which we think and talk about it. We float in
language like icebergs--four-fifths under the
surface and only one-fifth of us projecting into
the open air of immediate, non-linguistic experi-

Psychoanalytic theory conceptualizes differing

language expressions as reflections of dynamic personality

functioning. Focusing particularly on errors in spoken and

written language, Freud (1924) stated ". .. a suppression

of a previous intention to say something is the indispensable

condition for the occurrence of a slip of the tongue."

Fenichel (1945) cites the development of speech and language

as a decisive step in the building of reality testing. He

further states that speech is "a weapon of the ego" in

binding both the external world and inner excitations.

This theorizing involves language as a global communi-

cative process. The interests of the present investigation

are in formal, so-called "miniature" language approaches,

which lend themselves to objective measurement. These are

the Verb-Adjective Quotient (VAQ), the Type-Token Ratio

(TTR), Personal Pronoun frequency (PP), Qualification Term

frequency (Qual), and Allness Term frequency (All).

Verb-Adjective Quotient

The oldest of these is the Verb-Adjective Quotient,

which is defined as the simple arithmetic result of dividing

the total number of verbs by the total number of adjectives

in a language sample. The VAQ is an action-description ratio

and is somewhat similar to the Rorschach introversive-

extrotensive balance since Rorschach movement responses

introversivee) are based on action verbs and color responses

(extratensive) are dependent on use of adjectives.

Marya Mannes (1963) writing in a magazine of book

reviews identified verb and adjective usage as having mascu-

line appeal and feminine appeal, respectively. After first

qualifying her thoughts as an oversimplification, she sug-


S. that the verb dominates the "male" book and
the adjective the "female." Certainly the verb is
action and movement; the adjective, value and mood.
The verb tells what happens, the adjective where
and how and why.

This implication that verb and adjective frequencies

vary in different kinds of writing and other language samples

has been dealt with only minimally in the psychological

research literature. Table 1 summarizes the VAQ research

and it may be observed that most of the studies investigated

VAQ with respect to adjustment or diagnostic category. One

of the exceptions is Boderts (1940) study of Adjective-Verb

Quotients in large samples of dramatic, legal, fictional,

and scientific writing. He found that the relative frequency

of verbs compared to adjectives increased sharply as he com-

puted Science, Fiction, Law, and Drama AVQs in that order.

Osgood and Walker (1959) found the VAQ in suicide notes to

be significantly higher than in social letters and than in

simulated suicide notes. However this latter investigation

did not differentiate between the subjects' adjustments and

the situations in which the language was written as the

focal variable.

The typical study in this area used the VAQ as a

dependent variable with diagnostic groupings as the independent

rgo 4. (N
00 :0 (


H-I ro Cl-

>i >k


43 0 r-l 30 Cf
"C-10 T0 0d

002 42(2

H0 0 0
0rl 00 0

&- 02 HN0 ~
0 Q -l B

2l ; rll C?
a'" "' ^


0 4-1
-H 0-H -

0 -P

PHC 0 00
a A-

SS r^- o 0
fr-I ^ o^) C 3 i
*~HH 0022 * 0
HclrN l\t


HO H 0 0 V 4 E
r -4 CO H- 02 0 02
I 0 0 0 2 3 E 4
00 0.4 00 02 0 0
44E E 44 E-V (
0 0 43 04 0 H 0 0)

)0 H O1 H o 0 oH 02H
0 H3 E ( -H 43 42 I ,
H -q : m >2 n0 ori aH02 4V 0 0 0
HH C.-4 (d 0 0 02 "q 043 > 4H
HI-ir aP 04 o 0 md F E -
p r74
QM pi E tc M O ^ f

n3 (D 43 n)
0 >0 43
t o rj -H r:

0 .F rl 0- (2
r (2 02 m 02 0 0
w. e m c O i-id n^o

430 (2 H 02 (2 0 H 01 40
C -H- a) "4 (D R> (D mpp
(i H 42) 0 0 02 42 0 4 0 440

H 4 H O-* 0H t )0 (D 0 2 0 (2 043 20
?43 0 FA N 1 M -P 1 43 043
42 H H4 H0 0 .2 -H f o 0 0 0 2 () m

0423 Hd 0(2 M r- 0(43 2 H2 '0 -H > -
-H d r H ) 0 0 2 0 d 0O 0} 2c 042 D

43 43 043 V 043a O a ti0 43 N Sh 43 Mi bo3
0 m l a H i - (D H 0 3OHO
04 43 0 02 ( 2 3 > r-H P4 HC > 0 H H
44 0 4 3 > 02 C3 0. 0 0 04 0 2 0 0 H 0b
r04 0 H H 00 HoCoa ( 20102l 0
0-. & / c oOo yid a 00 0~o 0003








0 0

0 0
44 m

o .24
W' Htl

00 V.WAir 0o r-I N
H-I r- H -I HH

0 IA
% o0 0
r-4 rl r-I

H012 (D4

%00 0 0 0'-N 0%
N CO = _t0- -z

0' 0% 0'


4H .r

l O o C-~oH0 I0 t
4f\ o 0 HH H

0 U
0 o
.0 0 02 -1
F-l -ri1 Pi
(D 43 rj O
O ad ; 43
0.43 0
O 4342
042 0
0402 420

004 (D

0 0 043
P-H PA0.0


0D 0 *H



V. .H
1 R

020202 0 0

oco o oo
02 t 4 0r-24 Ia

r-I 0 C P 0O3
0 0 04.0 0 a02Q
0 a 0 2; m 0

00 %O 000


Cd j 0o 90
- 02 040.U






CT 0 t4

Cld4 02




(0 0-
> -H
Hrl 43
43 9 0

X +

04 (D
0 >
sd Q

C 0-44
.! s
4o 43


02 .0 .1
-P (D 0~
> 4-
2 + 0

0 0 0

S 0

01 (D d

2 02

*O O S

>4 04
42 02N
> 02
$ C

fr 0d

H 02 I0
0300 0

> 43^

43 0
-P P4

0 >
cd ro 3
0 0
0j 6

variables. The typical result was higher VAQs in the groups

of greatest maladjustment. The VAQ of spoken language corre-

lated positively with emotional instability in school

children (Busemann, 1925), was significantly higher on both

spoken and written language in hospitalized schizophrenics

than in college freshmen (Fairbanks, 1944; Mann, 194), and

was significantly higher in groups of hospitalized hysterics,

obsessive-compulsives, manics, and paranoid schizophrenics

than in normals in spoken language (Lorenz and Cobb, 1954).

Still another study (Balkan and Masserman, 1940) found

diagnostic group differences, in this case analyzing spoken

Thematic Apperception Test stories; anxiety state patients

had higher VAQs than obsessive-compulsive patients who in

turn had higher VAQs than conversion hysterics. uther studies

(Benton, et al., 1955; Doob, 1958) reported no significant

relationships between Manifest Anxiety Scale groups and VAQ,

and between 28 behavioral measures and VAQ.

Further inspection of Table 1 yields the following


1. There was little replication of even broadly

defined language samples from study to study.

2. Small samples were used in the diagnostic group


3. Modifications of the VAQ were used which make

generalizations about VAQ results uncertain.

This research on the VAQ may be summarized by stating

that many methodological deficits were present and that a

trend appeared in the results suggesting that pathology is

associated with a higher incidence of verb usage relative

to adjective frequency.

Type-Token Ratio

The Type-Token Ratio (TTR) is a measure of formal

language introduced by Wendell Johnson (1944). It is the

ratio of different words (types) to total words (tokens) in

a given language sample. In a 100-word sample with 100

different words the TTR would be 1.00; in a 100-word sample

with the same word repeated 100 times, the TTR would be .01.

Thus the TTR is said to be a measure of vocabulary flexi-

bility or variability. It has been computed for all of the

words in a given sample, for overlapping or consecutive sets

of 50 or 100 words, and for words in various grammatical


The first study using the TTR was conducted by

Fairbanks (1944) and represents the typical approach and

use. His subjects were 10 hospitalized schizophrenics,

ranging in age from 20 to 46 and in length of hospitalization

from 1 month to 24 years, and 10 college freshmen chosen for

their high IQs and rapid reading rates. The subjects were

asked to talk about 14 proverbs presented to them, to describe

situations in which they apply, and after they finished this

task to talk about anything they wished until a 3,000 word

language sample was obtained. One of the ways these data were

analyzed was by TTR. Table 2 lists the results of this and

5 other TTR studies. In this study the schizophrenic group

0 ID



-H 4
.^ r00D

000 ,0.0

.0 W

O0 E- c

+OH -


0 4
















0 >

0r OH (

0 cnr0r0


*g43 >
.0 4)
9:cd (D 0

11 4-: .1. 0 4S4
C~2 2 o0.So
x00 0H,0

*2 024 0
CIPI 00*20r

9.1 0
0< 0

. 40 0 0 m

0 02 H 02 p
I 40

i-I o om 9 IY
02'0 0 0Q
HO10 o '0

uaN ^a rl'

00 00000
H r-4 r-I -I -I -I


0'. 0' O
-H H

0d 090
No A S-




-N c: CJ r
(nx cli ( oj (


s0* 00
cdo o= o


*m "4 P4O
H 00 I
H01 0 *-.P.

*2 r-0 04,U
000 HO0
oa 13 o M- ,

a, m 00

01C4 00
'or- HH

0 HD

0 0
t iu
o 3

OHl 0_t

02 3 00

~o ;44.


02* l

0 0 *H



0 0

r k
g >
0 0


0 0
02 0

o) 0

0I 0

(D (

t! &
04 4


0 to
0 s0


'0 04

I o

0 04

P 0

,U 000


had a markedly lower average TTR than the college student

group, with the final TTRs determined by averaging across

thirty 100-word samples and across the 10 subjects in each

group. Furthermore, the distributions within the groups

show almost no overlap.

Similar results were found by Mann (1944), Lorenz and

Cobb (1954), and Osgood and Walker (1959). These studies

indicated that lower TTRs were present in written or spoken

language of schizophrenics, of various neurotic and psychotic

groups, and of suicides than in normalss." Chotlos (1944)

departed from the usual pattern of focusing on the psycho-

pathological and instead studied the TTRs of Iowa school

children aged 6 to 18 on anything the children chose to

write about. He reported highly significant positive corre-

lations between TTR and IQ, and between TTR and chronological

age. In an investigation of dyadicc speech," or the total

speech produced by a two-group, Jaffe (1956) identified

numerous interpersonal and intrapersonal factors that were

associated with low TTRs and high TTRs. For example he noted

that anxious blocking and stereotyped, circumstantial speech

were intrapersonal factors associated with low TTRs.

Criticisms similar to those made of VAQ research may

be made of the TTR studies; small samples tended to be used

and the language samples were broadly defined. Similarly,

also, a tentative generalization may be made; low TTRs tend

to be associated with psychopathology, anxiety, dullness,

and youth.

Personal Pronoun Frequency

Grammatical forms of speech have been a subject for

psychological study since Rowland (1907) investigated "The

psychological experiences connected with the different parts

of speech" in an introspectionistic framework. Of particular

interest in the past 20 years have been personal pronouns as

expressions of personality functioning and seven studies on

this topic are summarized in Table 3. Pathology was associ-

ated with high personal pronoun usage in studies of spoken

language by Fairbanks (1944) and Lorenz and Cobb (1954); in

the first study hospitalized schizophrenics were reported to

use more personal pronouns than college freshmen and, in the

second, different groups of psychiatric patients were reported

to use more total pronouns and the word "I" proportionately

more than normalss." Balkan and Masserman (1940) found that

the least anxious of three psychiatric groups used the fewest

first person pronouns.

These results were contradicted in studies by Mann

(1944) and Conrad and Conrad (1956). Mann reported that in

written life stories college freshmen used more total pronouns

than a schizophrenic group. Conrad and Conrad reported that

high personal pronoun usage was associated with individual

progress and "genuine involvement" in group therapy and with

satisfaction of participants in professional staff meetings.

Doob (195d) noted no significant results in his study of

"ego" pronoun frequency in college students. Thus personal

pronoun usage or its equivalent has been identified as being

41 02
ra :()
0 cli
cd 4 4


0 D(D )
02 0 H

0. 0 cn-dt
K 4 r-1

r- 0% c cr _:
0 -or\J or-_
r-,C cCis0
r-q rq

0 _trj r
0s 0 r'-r't
1rl.0v0 r(-
1 1 r











*p Ec c: S
0 0.
4 H 0 0 1
p 0-rl 00
4 0 >4. C)
0 0 04.'1
H~ ~ 0. 12
(1 P1 4 H
1-1 o C o P
21 H 01 4 00
-1- .4. pp 0 p 0

0a X 43 o O
04 0" & 0

02 >
H- *H 21
14 01 W
P I= n rl Q n ^ C
0 21 H r0 1 4
o ii ^ce o r o E
w ra m -P0 m M'O o fq
00S ado WOWD >M wo
44 q 0 ; i-l Mt 04l ^
21 I 02 41 AP.02 0'a.14 1

o0d O0 m ot~ Cd O r-IC 0 0 p MH f
02100 a 0 bO1 000 0 2 1a H H
"4 0 0 wo or H E E g -CC
0 0 C1 0 21212 0 0 Go m m HH 0-i 00
UKa U = 041 0 03w :4 04=A

00 u^0022 00 r-4 r -H rq
0010 14410 0011 01021 1


gg "o m <
00 0 0 2 C0

a A a ^ ba i-(

U c 0 2 & 1. S
00 c U~'( 0 H




c "i









l 0
0 02


0. Hl 4-)
S> rd
2.l 0 *r?02 02

hO-H 4' C 0 03

-P P4 01 0 00
o od

F3 0 d C0 w 44 M r-4 :

0 o A
43 02

0 0 0 0 r- 02
O.402-i 4302^ 0 0 ?
-P m -14 :0 4-P
0a 0 204 0 P040 C 0 02 0d
ScO c m O o4 0

> 0

*0 0 02 o o0

0 00 0 0 0 0.
. 0i 02 -P 0 -4 0
*3 0 02 0 4
Cq 0 02
0 0m0 00 0

*02.0 00000 0 02 0
E 0. 0 ;2b a i 0
02 0 0

%0 n 00000
o I-I .-4-I- 0

d 0 04 02

4 h D) (D C3

FO p q a 0'
0~ -k) 0N


associated with psychopathology, with some aspects of mental

health, and with neither.

One particular kind of personal pronoun, namely self-

reference, was among five psycholinguistic measures that

Wagner and Williams (1961) used to analyze speech behavior in

groups differing in Achievement Imagery (AI) on the Iowa

Picture Interpretation Test (IPIT) and in defensiveness on

the MMPI. The high and low AI groups were above the 75th

and below the 25th percentiles, respectively. The high and

low defensiveness groups were above and below the median on

the MMPI "K" scale. The subject pool was composed of 170

students in general psychology at the University of Richmond.

Eleven of these fell in the high AI, high "K" group, and

equal numbers in the other three combinations of high and

low AI and "K". The subjects provided the language samples

by talking aloud to a dictaphone on the topics, stating

what they think they should get out of general psychology,

out of all their courses, and out of college. The analysis

of self-references in these samples is shown in Table 3.

Little difference appeared between the high and low AI groups

or between the high and low "K" groups. Two of the other

language measures investigated in this study will be discussed

shortly. The significance of this study is certainly not the

negative results just reported, but the use of a large number

of subjects initially, selection of groups from this pool by

objective, questionnaire methods, and the moderately well-

defined referents for the language sample.

Qualification and Allness Terms

The last two language measures considered here are

Qualification Terms and Allness Terms. Qualification Terms

are words such as approximately and perhaps, which enable

the individual to modify the amount of commitment he makes

in his statement. Allness Terms are extreme and polarized

statements or words, such as always and none-whatsoever.

Studies dealing with these two areas are summarized in

Tables 4 and 5. Balkan and Masserman's (1940) group of

hysterical psychiatric patients produced the largest number

of Allness Terms and fewest Qualification Terms, while the

obsessive-compulsive patient group used the fewest Allness

Terms and the largest number of Qualification Terms. On

both measures the anxiety-state patients were in the mid-

range. The authors interpreted the obsessive-compulsives'

patterns as part of a need ". . to rationalize and elaborate

the many ambivalences and uncertainties reflected in his

fantasies." The hysterical patients' sweeping statements and

minimal qualifications were associated with their charac-

teristic absence of vagueness or ambivalence in their adjust-


Osgood and Walker (1959) identified suicide notes as

having significantly more Allness Terms and Qualification

Terms than the letters of controls. Raimy (1948) reported

more ambivalent, qualifying expressions in unsuccessful

clients at the end of counseling than in successful clients.

The raw scores in Wagner and Williams' (1961) four groups

0 H
0 r- C < II
:j Fqa
0 J D m V 40 0)
4 OFq 0 H
r l V2 ) 00
fr 00oncod 01
) 4o M 3 P
C Hk C-iF+

0 9o~ p r

n I o M ji0j Qn- 0


0 .~
ft 0 U *P 03 -P 4-'
0 bo 40 0 0P
-' C fi C! r-l-rl
2 0 -H < -O*40
1f\:00' 0H 0 HO O

E-i s 0 iA l= d =
C~sO^ ^il ^ C 3

P-IC\J r^ 0 0

0 t>0 00-

a) -PP
0 H 0
Er4 Ca /o3
1 il <0
-P~ 221 *

3m 0 > ?
43-l U :j3 C
V) V2

01 0

0 H

P 01 ;4
> m
on0. o 0

-r H ( 02
4 4P 1 43 10 rd
4) 0 m1 0 000
*D 0C1 00r

rfIAV0A I C% 0M
o%0 r-
vmuir i 0s (\jP

" r\J t-ir
A -;A




0 0

03 0 w U

0 -l 04 0 0MI

I- r- H r-H

o o oa 1-,

rl Hl H

0 30 0 0% 02

N u 0 s 0 4 9 i-l
p 02 rg w d3
PQ z 30 5.4



00 -


























P -p


rl i3

0 0

Ns 0
043 10
0 0 H

0020 0 CDr-
w2N 02020.r-
544002 '00
o caa z,0

0 2\V\ 0 HHH
.-D E, I
0 0 all

-.tW 0'N
So r-4
o40 *0 '

0'\^ 0'<\

H Hs

o _tO -.t
4 H CU









F. O ra ^
0 02 02 HD -
m0 0 m 0 H
rg )T m sCd9-
H 'O x 0 to3:OH3
~ 02d~ 02d













are reported in Tables 4 and 5, but the only significant

difference was between high "K" and low "K" groups in number

of Allness Terms. The more defensive subjects in the latter

study used fewer Allness Terms (p = .07).

The literature on the Qualification and Allness Term

dimensions is smaller than on other language variables dis-

cussed. Furthermore it differs in the absence of adjustmental

poles associated with term usage; only a tendency for defen-

siveness to be associated with use of qualifiers and minimal

use of all-encompassing terms was present. Qualification and

Allness Terms might be generalized and their use be considered

as an expression of the way people perceive, organize and

involve themselves linguistically in their environments.

One of the purposes of reviewing these language

measures was to explore the development, background, and

current meaning of their experimental use. A second purpose

was to focus upon the sources of the language samples in

these studies. The completion of the review now permits

examination of this latter issue.

There has been a wide and largely uncontrolled range

of stimuli used to elicit the language patterns and expres-

sions of subjects. Typical of these stimuli were requests

for the subject to write the story of his life (Fairbanks,

1944), to write about anything he wanted to (Chotlos, 1944),
to create Thematic Apperception Test stories (Balkan and

Masserman, 1940; Benton, et al., 1955), and to report what

he hopes to get out of college (Wagner and Williams, 1961).

In most of these the subjects were free to write or talk

about a wide variety of topics, depending on what set happened

to be in mind. In the Chotlos language sample, for example,

one subject may have written about his anxieties and fears,

while another may have written about skyscraper construction

or oxidation-reduction processes.

This potential, and in some cases likelihood, for a

wide range of subject matter in the language samples develops

special meaning in the context of Boder's (1940) study. He

reported striking differences in one language measure between

extensive samples of four areas of language. The frequency

of verbs in proportion to adjectives was nearly seven times

as great in conversational, dramatic language as in scientific

descriptive language. This implies that in many of the

studies of language variables, the discrepancies between

manipulated clinical groups were due to the subjects pro-

ducing different classes of language rather than different

language patterns on the same class. Where significant

differences were obtained, was this a result of varying

clinical groups' tendencies to talk about different topics,

or was it a result of the groups' dissimilar language patterns

across all topics? One of the objectives of the current

investigation was to vary both personality groups and language

sources in order to explore this question.

Still another dimension involved in personality and

language research is that of verbal ability, and of intelli-

gence. One of the studies reviewed (Chotlos, 1944),


identified IQ as a major contributor to variance in TTRs and

other language measures. Intelligence was deliberately and

inextricably confounded with adjustmental level in two early

studies of TTR and parts of speech (Fairbanks, 1944; Mann,

1944). The effects of intelligence were recognized by Balkan

and Masserman (1940) when they matched their psychiatric

groups by IQ. Similarly, Wagner and Williams (1961) attempted

to rule out verbal ability influences by establishing that

ACE and Cooperative English Test scores were independent of

their personality variables. This implicit recognition of

potential ability effects on language variables has occurred

frequently without specific investigation of the effects in

an adjustmental context. The integration of these effects

with language source and personality variables therefore was

posed as an issue in the current investigation.


The concept of repression-sensitization may have a

special relevance to the question of the relationship between

the experiencing of feelings and miniature language patterns.

Repression-sensitization (R-S) refers to modes of experi-

encing, with repressors defined as people who avoid and deny

experiencing and sensitizers who exaggerate their experi-

encing and feelings.

The notion of repressors and sensitizers arose from

the observation (Postman and Bruner, 1946; McGinnies, 1949;

Bruner, 1957) that a wide range of individual differences

existed in perceptual recognition thresholds of emotionally

tinged stimuli on tachistoscopic presentations. At one

extreme were subjects who characteristically did not recognize

the emotional or threatening stimuli, and these were identi-

fied as repressors. The sensitizer label was attached to

those, at the opposite extreme of the continuum, who recog-

nized the stimuli at shorter exposure times than the other


Other studies produced more clinical and more general

definitions of R-S. Altrocchi and Dickoff (1963) defined

repressors as people who ". . use avoidance, denial, and

repression of potential threat and conflict as primary modes

of adaption and ego-control." They stated that sensitizers



tended to be self-effacing, neurotic extroverts and were

similar to patients that had symptoms of anxiety and depres-


Byrne (1961) stated that repressors have ". .. behav-

ior mechanisms of a predominately avoiding type," while

sensitizers exhibit predominately approaching, intellectu-

alizing, obsessional behaviors. Byrne summarized the results

of a number of studies on R-S and reported that those who

repressed on perceptual tasks ". . also tended to be identi-

fied as repressors on the basis of case history and interview

material, to be classified by psychiatric personnel as inter-

nalizers, to remember success better than failure on a

scrambled sentence task, to forget an anxiety-arousing Blacky

picture, to respond to a sentence completion test with

blocking, avoidance, denial, and cliches, . ." Sensitizers

on the other hand recalled failures, learned affective words

as easily as neutral ones, admitted their inadequacies and

shortcomings easily, and on a word-association test had

short latencies for aggressive and succorant words.

No proposal is made for repression-sensitization as a

pure, unique, or new dimension. Ullman's (195I, 1962)

facilitation-inhibition dimension was defined in virtually

the same terms as R-S. Cristie's (1963) annual review

article on personality structure summarized the strong

evidence for overlap of response-set, social-desirability,

and R-S. Joy (1963) reported that Byrne's R-S MMPI scale

intercorrelated more highly with 50 MMPI and California

Personality Inventory scales than any other scale used. Joy

further noted R-S to correlate +.72 with the MMPI Depression

scale, +.60 with Psychasthenia, +.62 with "F" (validity),

and +.40 with Rokeach's Dogmatism scale. He obtained 20

correlations that were negative and greater than -.50; these

might be summed up as scales of socialibility, well-being,

and good adjustment, with defensiveness and social desira-

bility prominent. Thus, R-S actually involves a broad range

of adjustmental and experiencing variables, cutting across a

number of other personality dimensions.

The R-S dimension itself has been shown to have satis-

factory reliability and moderate validity. Byrne (1961)

studied R-S in 133 college students at the University of

Texas and found a split-half "r" of .88 and a 6-week test-

retest "r" of .68. Byrne also correlated his 156 item MMPI

R-S scale with Ullmants facilitation-inhibition scale of 43

MMPI items, using these college students, and obtained an
"r" of -.76. Ullman (1962) performed the same experiment

with psychiatric patients and found a correlation of -.94.

Altrocchi, Parsons, and Dickoff (1960) reported significantly

lower self-ideal discrepancies in repressors than in sensiti-

zers, and Byrne (1961) reported correlations of +.62 and

+.55 between R-S and self-ideal discrepancies in two separate

Byrne's conclusions, after reviewing the rationale,

reliability, and validity of the repression-sensitization

dimension, were that ". . most of the evidence is positive


and indicates that repression-sensitization is a meaningful

behavior dimension" and that ". . the R-S scale appears to

be a reliable test and, with minor exceptions, the evidence

suggests that it is a measure of defensive behavior."


There were two parts to this investigation of language

variables. The first part sought to explore the relation-

ships between and contributions of personality, ability, and

situational variables to language patterns. There is a large

literature on each of these variables with respect to language

usage, but there has been only a minimal integration of them.

For the purposes of the present study, the following defini-

tions of these variables were made:

Personality was defined by the repression-sensitization

dimension, as scored from Byrne's (1961) revised scale.

Ability was defined by raw scores on the verbal ability

section of the School and College Ability Test (1963).

Situation was defined as the "personal" description

of one's self and the "impersonal" description of a picture

of a room, as expressed in written language.

Language variables were defined as the Type-Token

Ratio, the Verb-Adjective Quotient, and the frequency per

thousand words of Personal Pronouns, Qualification Terms,

and Allness Terms. In addition, the total number of words

used in the description was included as a measure of pro-


No suggestion was made that these definitions were in

any way encompassing or all-inclusive. Rather they were

specific, limited measures of the stated general variables.

In the first part of this study the personality,

ability, and situational variables were the independent

measures and were manipulated by establishing different groups

on repression-sensitization and SCAT verbal ability and by

performing repeated measurements in the situations. The

language measures were the dependent variables and were indi-

vidually studied in relation to the independent variables.

The lack of previous studies integrating these independent

variables led to a series of null hypotheses in which the

language measures were considered as a whole.

The null hypotheses:

1. There are no differences in the language patterns
of repressors, sensitizers, and a middle group.

Rejection of this hypothesis is consistent with

earlier findings that personality dimensions are associated

with language measures. For example, the denial and avoidance

of the repressors might be manifested in low Verb-Adjective

Quotients, high TTRs, many Qualification Terms, and few

Allness Terms. Conversely the exaggerated anxiety of the

sensitizers might be seen in high Verb-Adjective Quotients,

low TTRs, and so on.

2. There are no differences in language patterns used
by subjects grouped according to verbal ability.

The explicit findings of one study (Chotlos, 1944) and

the controls used in others (Balkan and Masserman, 1940;

Wagner and Williams, 1961) suggested possible relationships


between certain language measures and verbal ability.

Rejection of this hypothesis on the TTR is congruent with

Chotlos' findings, and rejection of this hypothesis for

Verb-Adjective Quotient and Qualification and Allness Terms

justifies the controls of verbal ability which were used in

the study of these measures.

3. There are no differences in language patterns used
in personal and impersonal description situations.

Rejection of this hypothesis is consistent with the

marked situational differences reported in the Boder (1940)

study and implied in the Osgood and Walker (1959) study.

4. There are no significant interactions in the
patterns of language found in the independent

The sparcity of literature on this issue resulted in

no directional expectations for rejection of this hypothesis.

Hypothesis number one was further explored in a limited

sense. Extreme repressors and extreme sensitizers were

defined and their language scores compared. The purpose of

this additional comparison was to study the language patterns

of especially deviant subjects on the repression-sensitization

personality dimension.

The second part of the study was concerned with the

language measures as independent variables. While the pro-

cedure followed in the first part of the present study and

in most others was a manipulation of adjustmental groups and

then a studying of language changes, the second aspect of the

present investigation involved a manipulation of language

groupings with an examination of concomitant adjustmental

category changes. In other words the focus was upon the

language measures themselves.

The purpose of redefining the language measures as

independent variables was to explore in a more direct manner

than before the meaning of given language patterns. That is,

there is evidence which suggests that subjects who are mal-

adjusted in certain ways produce lower TTRs, for one instance,

than control subjects. Now does it follow that subjects who

produce lower TTRs than most people are maladjusted in these

certain ways? And, how do they differ from subjects at the

opposite end of the TTR distribution? These are the kinds

of questions that the second part of the study sought to


The procedure was to select the extremes on the language

measures, to compare the extremes with each other on each

individual language measure, and to compare the extremes to

the entire sample. The basis for comparison was chiefly the

Minnesota Multiphasic Personality Inventory scales, with a

single SCAT verbal ability score included. No specific

hypotheses were made because of the exploratory nature of

this part of the investigation.



The subjects were 144 male undergraduate students who

were attending the University of Florida in May of the Spring

Trimester, 1963. All were either living in Tolbert Hall or

were registered for one of the following courses: Intro-

ductory Psychology (PSY 201), Psychology of Adjustment

(PSY 202), Introduction to Speech (SCH 201), or Introductory

English (0-31, 0-32, Eh 133). A plurality of the subjects

came from the English courses, followed by the dormitory

residents, the Psychology students, and the Speech students,

respectively (see Appendix A, Table 20). This population of

undergraduate male students was chosen because of their

availability and because of the extensive standardization of

the repression-sensitization scale on similar populations,

e.g., Byrne (1961).


The Byrne revised 127-item repression-sensitization

scale of the Minnesota Multiphasic Personality Inventory was

used as the R-S measure. Verbal ability was defined as the

raw score of the verbal part of the School and College

Ability Test (SCAT). The MMPI and SCAT are administered

routinely to all entering students at the University of



The impersonal description stimulus was the 2-page

picture of a living room and patio in color on pages 50-51

of 1001 Decorating Ideas (Consco Products, 1963). A 30-inch

square image of this picture was projected by opaque projector

on a white screen in rooms that were sufficiently darkened to

permit easy perception of the image and sufficiently lighted

to permit the subjects to write without discomfort. The

personal stimulus was the request to the subject to "describe

everything about yourself personally."

Six sheets of lined white paper stapled together in

sets of three were prepared for each subject. Personal infor-

mation forms were used which requested the following infor-

mation: name, age, student number, academic classification,

parental education, total parental income, marital status,

and an open-ended statement about the meaning of the task as

the subject saw it.

The standard ten MMPI clinical scales were tabulated

for each subject as well as the three validity scales.

Finally, the actual language measures used were defined

in the following manner:

Number of Words. The total number of words were

tabulated by exact number; hyphenated words were counted as

one word; words accidentally repeated were counted both times

(e.g., "looking in the the room" = five words).

Type-Token Ratios. TTRs were tallied for each full

100 words in each description (TTRs were obtained for words

1-100, 101-200, 201-300, etc.); these were averaged to yield

a single TTR for each subject's impersonal description and a

single TTR for his personal description.

Verb-Adjective Quotients. All verbs in a description

sample were tallied, including auxiliary verbs and parti-

ciples. All adjectives were similarly tallied, including

quantitative, ordinal, and attributive adjectives. The

total verb count was divided by the total adjective count

and the Verb-Adjective Quotient was carried out to two decimal


Personal Pronouns. Sixteen words were defined as

personal pronouns. I, me, my, myself, we, our, you, she, he,

her, his, hers, him, your, they, their were all counted in

any context.

Qualification Terms. Twenty-seven individual terms

and phrases of ambiguity were counted; the most frequently

occurring were almost, about, approximately, apparently,

more-or-less, especially, maybe, perhaps, somewhat, and


Allness Terms. These were defined as extreme and

polarized words. Qualified Allness Terms were not counted;

e.g., that is about all. The following composed the complete

list of Allness Terms: all, never, completely, always,

anything, nothing, nobody, everybody, forever, utter, exactly,

undoubtedly, wholly.

Administrative Procedure

Volunteers were solicited from the speech course and

students wishing to fulfill course requirements for


experimental participation were solicited from the psychology

courses by announcing in written and spoken form the oppor-

tunity to join in "Experiment Life Situation." The dormitory

and English students participated by virtue of attending

their regular floor and class meetings, respectively; at this

time they were informed that an experiment was being conducted

and they were asked to participate. Almost all of the

students in the latter groups did serve as subjects.

The lined paper and the personal information sheets

were passed out with the personal information sheet at the

bottom to give the impression of anonymity. The order of

administration of personal description and the impersonal

description tasks was alternated with successive groups. The

following instructions were given when the impersonal descrip-

tion was requested first.

The purpose of the tasks you will be given is to
study language usage. This is a three part experi-
ment which will last one hour. For the first part
of the experiment, take the top three sheets of
paper you have been given. They should be stapled
together. Now (picture projected on screen) please
look at the picture on the screen and try to describe
everything you see. You may write as much as you
want to, but be sure you fill at least both sides of
one page. Do not list, and remember to describe
everything you see. You will have 25 minutes.

(after 25 minutes)

Now turn to the next set of lined paper. This is
the second part of the experiment. Please describe
everything about yourself personally. Again you may
write as much as you want to, but be sure to fill at
least both sides of one page. Remember, give a
complete personal description of yourself. You will
have 25 minutes.

(after 25 more minutes)

Bow take the personal information sheet and fill it
out completely.

When the personal description was requested first,

similar instructions were given with only the order changed

from those above. In all administrations of the personal

description part, at least one subject asked if the investi-

gator desired a physical description. The standard reply was

"It's up to you."

When all the descriptions had been collected, they

were scored for the language variables by the investigator.

The descriptions were chosen randomly for a second scoring

by the investigator and by a colleague, and a very high

degree of correspondence was found between the two scoring.

Wagner and Williams (1961) also noted high scorer reliability

on some of these measures; they computed rank-order corre-

lation coefficients and obtained the following rho values for

the two sets of judges' scores: Allness Terms = +.85,

Qualification Terms = +.94, and Self-references= +1.00. In

the present study it was felt that the objective, easily

quantified nature of the language variables enhanced scoring


A total of 173 subjects participated in the personal

and impersonal descriptions. Three were eliminated because

they had left one of the descriptions blank or had jotted a

cryptic note on the paper about their feelings of invasion of

privacy. When the Minnesota Multiphasic Fersonality Inventory

datawems obtained from the files of the University Mental

Health Clinic, it was found that no datawere available on

26 subjects. Therefore 144 subjects were left on whom all

necessary information was present. Because the subjects

took the MMfI over a period of three years, with some taking

it immediately before the beginning of the semester in which

the language data was collected, and some three years earlier,

the issue of the reliability of the MMPI was raised. Analysis

of the language data by time of taking the MMPI was performed,

but no definitive answers were found. The analysis is shown

in Appendix B, Tables 23 and 24.

The raw scores on the verbal section of the School

and College Ability Test were obtained from privately

distributed mimeographed papers produced by the Board of

University Examiners, the University of Florida. The raw

scores were chosen as a control over the changing percentile

ranks among University of Florida students on SCAT scores

from year to year. There were 10 subjects for whom no SCAT

scores were available; the twelfth grade placement scores

for this group were converted to percentile scores of

University of Florida freshmen according to the Board of

Examiners' publication, and then the percentile scores were

converted to equivalent SCAT verbal raw scores for the

individual student's year of entry.

Statistical Procedure

Repressors, sensitizers, and a mid-group called

neutrals were identified in the first part of the study of

the 144 subjects. The high 30 on Byrne's revised R-S scale

of the MMPI were the sensitizers; the middle 30 were the

neutrals; the low 30 were the repressors. The range of R-S

scores for sensitizers was 50-81, for neutrals was 27-34,

and for repressors was 2-15. There were 27 subjects in each

of the "in-between" groups.

The exception to this division was the Type-Token

Ratio analysis, in which five subjects were eliminated

because of language samples of less than 100 words on one

of the descriptions. The repression, neutral, and sensiti-

zation groups still contained 30 each, but the "in-between"

groups then had 25 and 24. In addition the range of each of

the R-S groups was expanded by one point on the R-S scale.

The distribution of R-S scores is illustrated in Appendix A,

Table 21.

These 90 subjects who had been classified as repres-

sors, neutrals, or sensitizers were then classified further

by their School and College Ability Test verbal raw scores,

or raw score equivalents. Within each R-S group the 10

subjects with the highest SCAT verbal scores were identified

as the High SCAT group, those with the middle 10 scores as

the Mid group, and those with the lowest 10 scores as the

Low SCAT group. The distributions and ranges of the SCAT

groups were slightly different in each R-S group. As a

result there were a few instances when subjects who were in

different SCAT groups had identical SCAT scores. These

distributions appear in Appendix A, Table 22.


The 90 subjects were thus divided into nine groups of

10 subjects each, three times by R-S scores, and three times

more by SCAT verbal scores. Two measurements, one from the

impersonal description of the room and a second from the

personal description of self, were derived for each subject.

This experimental design is illustrated in Table 6.

Table 6

Design of the Experiment, Part 1

R-S SCAT Verbal Personal Impersonal

High 10 10
Repressors Mid 10 10
Low 10 10

High 10 10
Neutrals Mid 10 10
Low 10 10

High 10 10
Sensitizers Mid 10 10
Low 10 10

90 + 90

180 measurements

Essentially this design is a Lindquist Type III model,

with two sources of variance between subjects and one source

of variance within subjects (Lindquist, 1953). The procedure

for the analysis of the variance is shown in Table 7.

This analysis was performed six times, once for each

of the language measures which served as the dependent


Table 7
The Analysis of Variance for Repression-Sensitization,
SCAT Verbal Ability, and Mode of Description

Source df

(Between subjects 89)

Repression-sensitization (R) 2
SCAT Verbal (V) 2
RxV 4
Subjects within R x V 01

(Within subjects 90)

Personal-impersonal (P) 1
P x R 2
P xV 2
PxR xV 4
P x subjects within R x V 61
Total 179

In this first part of the study the repression-

sensitization dimension was further investigated with the

language variables still serving as the dependent variables.

The 15 highest subjects on the R-S scale were defined as

extreme sensitizers and the 15 lowest of the 141 subjects on

the R-S scale were defined as extreme repressors. The range

of R-S scores for the extreme sensitizers was 59-81 and that

of the extreme repressors was 2-10. These two groups were

compared by t test on the six language measures on both the

personal and impersonal descriptions, a total of 12 t tests.

In the second part of the study the 144 subjects were

studied with the primary focus on the language measures

themselves. The high 30 and the low 30 subjects on each of

the six language measures were selected on the personal

descriptions and then on the impersonal descriptions. The

total number of selections was 12 high groups and 12 low

groups. Of course these groups were not mutually exclusive

and some subjects fell in several groups.

Fifteen "dependent" measures were tallied for these

12 groups; these 15 measures included MMPI scales one through

ten, "L", "F", "K", and R-S, and the SCAT verbal raw scores.

All of the analyses on the MMPI clinical scales were performed

without the "K" correction.

The differences between means on the 14 MMPI scales

and the SCAT score of the 12 high-low sets from the language

variables were tested for significance by t test. Thus 12

times 15 or a total of 180 t tests were performed.

In addition the means and the standard error were

computed for the 15 dependent variables; that is, the MMYI

scales and the SCAT scores. Then confidence limits were

estimated for the 0.05 and U.01 levels of confidence

(Mcnemar, 1955). Each of the 24 scores (two modes of

description times six language measures times two extreme

groups) falling in each of the dependent variables were

evaluated with respect to the confidence levels. Therefore

360 scores, each a mean value for a group, were evaluated.


The results of the analysis of variance section of

Part 1 of the investigation are shown in Tables 8 and 9. In

Table 8 the language measures were accumulated in groups of

10 subjects as proposed in Table 6. Therefore the first sum,

3241, in the personal description column, under the heading

Number of Words, means that the average number of words used

by repressors with high verbal ability was 324.1. It simi-

larly can be seen that the average number of words used in

the impersonal descriptions by the group of sensitizers with

low verbal ability was 281.2.

The Type-Token Ratio and Verb-Adjective Quotient sums

in Table 8 were multiplied by constants for ease of statis-

tical computation. The TTRs were multiplied by 1000 and the

VAQs were multiplied by 100. These constants appear again

in the analysis of variance tables, mean squares column, but

have no meaning or effect with regard to the F-ratios.

Therefore in Table 8, the sum 7197 in the repressor-high

verbal ability-personal description slot represents a mean

TTR of .72; the sum 6228 under TTR in the neutral-low verbal

ability-impersonal slot represents a mean TTR of .62. Under

the VAQ heading, the sum 2870 in the repressor-low verbal

ability-personal slot stands for a mean VAQ of 2.87. The


Table 8
Language Scores of R-S Groups and Verbal Ability Subgroups
on Personal and Impersonal Descriptionsa

Fer- Imper-
SCAT sonal sonal
Number of Words
High 3241 3051
Repressors Mid 2858 2612
Low 3674 3433
High 3031 3097
neutrals Mid 3506 3106
Low 3349 2937
High 2774 2733
Sensitizers Mid 3664 3398
Low 3036 2812

Yer- Imper-
SCAT sonal sonal
Type-Token Ratiob
High 7197 6631
Mid 6877 6150
Low 6960 6470
High 7050 6574
Mid 7050 6309
Low 6957 6228
High 7032 6268
Mid 7126 6316
Low 7083 6342

Verb-Adjective Personal Pronouns
Quotient /l000 Words
High 1903 1049 High 1049 155
Repressors Mid 1918 1137 Mid 1004 199
Low 2870 1119 Low 1172 116
High 2048 1212 High 1063 172
Neutrals Mid 2108 909 Mid 1101 126
Low 2326 1159 Low 1107 178
High 1665 964 High 946 53
Sensitizers Mid 2124 1194 Mid 1159 201
Low 1778 1166 Low 1133 139
Qualification Terms Allness Terms
/1OO0 Words /1000 Words
High 523 357 High 97 86
Repressors Mid 443 307 Mid 103 36
Low 297 289 Low 91 53
High 433 333 High 111 45
Neutrals Mid 363 274 Mid 113
Low 276 306 Low 127 76
High 398 343 High 132 43
Sensitizers Mid 461 321 Mid 167 44
Low 286 287 Low 136 72

aAll scores represent the sum of scores for groups of
10 subjects.
bAll TTR scores listed here were multiplied by a constant
of 1000.
CAll VAQ scores listed here were multiplied by a constant
of 100.


number 964 in the sensitizer-high verbal ability-impersonal

slot stands for a mean VAQ of .96.

Each of the scores listed under Personal Pronouns,

Qualification Terms, and Allness Terms in Table 8 is the sum

of the scores of the 10 subjects falling in the sub-classes.

PP, Qual., and All. are expressed in units per 1000 words;

this was computed by dividing the number of Personal Pronouns,

for example, of each subject in personal and then impersonal

descriptions by the number of words used in that description,

and then multiplying the result by 1000. Therefore under

Personal Pronouns, the 1049 in the repressor-high verbal

ability-personal slot means that the mean for that sub-class

is 104.9 Personal Pronouns per 1000 words used.

The data in Table 8 are summed and averaged in Table 9

in order to show the global relationships on the language

measures between and within the independent variables.

Notable among these within variable differences are those

between the personal and impersonal descriptions; the personal

descriptions yielded more words, more different words, propor-

tionately more Personal Pronouns, Qualification Terms, and

Allness Terms, and a high ratio of verbs to adjectives.

Within the R-S variable, few differences were present. The

primary noteworthy result was a lower production of words and

a lower VAQ by the sensitizers than the other groups. Within

the SCAT verbal ability variable, the High group produced

fewer words, had a lower mean VAQ, lower PP/100U words fre-

quency, and a greater frequency of Qualification Terms. The

Low verbal ability group used more adjectives in proportion

to verbs and used notably fewer Qualification Terms.

Table 9

Mean Language Scores of R-S Groups, of Verbal Ability
Subgroups, and of Personal and Impersonal
Description Samples

PP Qual All
Words TTR VAQ /1000 /1000 /1000

Repressorsa 314.5 .671 1.67 61.6 36.93 7.83
Neutrals 317.1 .669 1.63 62.5 33.03 8.60
Sensitizers 306.9 .669 1.48 60.5 34.93 9.90

High SCATa 298.8 .679 1.47 57.3 39.78 8.60
Mid SCAT 319.1 .664 1.57 62.5 36.15 8.48
Low SCAT 320.7 .667 1.74 60.5 29.02 9.90

Personal 323.7 .704 2.08 106.2 38.67 11.97
Impersonal 301.9 .637 1.10 14.9 31.30 5.59

aN = 30
N = 90

The analysis of variance of these data is presented

in Table 10, by source of the variance, mean squares, and

F-ratios. The only source of variance that was found to

produce consistent significant differences in the language

variables was the personal-impersonal description. All six

of the dependent measures were significantly different

between the personal language sample and the impersonal

language sample beyond the 0.01 level of significance.

The other significant F-ratios were produced by the
verbal ability variable under Qualification Terms. The

number of Qualification Terms used by the three verbal ability

groups was significantly different at the 0.05 level of


Analysis of Variance Tables

Words TTR
Source df -------

(Between Ss
R (R-S)

Ss /R x V

2 1666 .08 .77 .04

2 8937 .44 3905 1.92

4 30032 1.47 3430 1.b9
61 20449 2031

(Within Ss 90)
P (Descr.) 1 21212 8.67* 202944 13.15"'

P x R 2 201 .08 1238 .80
x V 2 2962 1.21 961 .62
P x R x V 4 1230 .50 474 .31
P x Ss/R x V 81 2446 1544

*0.05 level of significance
*-0.01 level of significance
"*0.001 level of significance

,nr Six Language Measures

VAQ PP/1000 Qual/1000 All/1000

5651 .70 58 .10 223 .58 66 1.14
)681 1.32 816 1.42 1800 4.67* 10 .17
163 .89 561 .96 173 .45 38 .66
3092 574 385 58

$258 86.84*** 391347 625.7** 2442 11.7* 1830 30.91*
>278 1.26 34 .05 104 .50 113 1.90
;420 1.09 425 .68 753 3.61* 55 .93
>26d 1.26 300 .48 56 .27 17 .28
989 626 207 59

significance. The same level of significance was found for

the interaction between verbal ability and personal-impersonal


There were no significant results in the repression-

sensitization variable, or in any of the tested interactions,

between R-S and other variables. Examination of the mean

squares for R-S in Table 10 reveals that not only were no

significant differences present, but that very little differ-

ences whatsoever between the R-S groups were found in these

language measures.

The first major null hypothesis was that no differences

between repression-sensitization groups would appear on the

six language measures. The results failed to reject this

hypothesis on any of the measures.

The second major null hypothesis was the prediction

that the SCAT verbal ability groups would not differ on the

six language measures. This hypothesis was rejected only on

the Qualification Term measure, and was not rejected for the

other five measures.

The third null hypothesis was that no differences

would appear between the personal (self) and impersonal

(room) descriptions on the language measures. The results

rejected this hypothesis on all six counts. In each instance

the probability of the personal-impersonal differences

occurring by chance was less than one in a hundred.

The null hypothesis concerning interactions between

the three independent variables was rejected one time in the


24 interaction measures tested. Overall the data failed to

reject this hypothesis, but the one rejection under Qualifi-

cation Terms was consistent with the significant findings of

the interacting variables tested alone.

The further investigation of the repression-sensitization

dimension yielded results that were essentially negative. The

isolation of 15 extreme repressors and 15 extreme sensitizers

from the total subject pool of 144 produced an R-S mean of

7.47 for the former group and an R-S mean of 65.60 for the
latter group. The means, standard deviations, differences

between the means, and t values of these two groups for the

language dependent variables are presented in Table 11. In

addition this table lists the SCAT verbal raw scores for the

two groups; the SCAT means were identical for the two groups

and the standard deviations similar enough to consider that

verbal ability so measured was matched in the groups.

There was a greater difference between the R-S groups

using this sampling of 15 subjects than in the earlier

sampling of 30 subjects per group. Of the 12 mean differ-

ences shown in Table 11, only one is not larger than the

mean differences between the R-S categories in Table 9; that

is Qualification Terms, impersonal description.

The greater language differences between repressors

and sensitizers in the extreme groups are noted in terms of

mean scores; however these differences were accompanied also

by broad dispersions about the means. Twelve t tests were

performed and one was significant at the 0.06 level. This

Table 11
Language Data Comparison of Extreme Repressors (N=15)
and Extreme Sensitizers (N=15) in a
Population of 144 Subjects

Repressors (N=15) Sensitizers (N=15)
x S.D. X S.D. XR-XS

R-S scores 7.47 2.16 65.60 5.53
SCAT 40.0 9.39 40.0 9.19 0 0

Personal (self) Description
No. of words 326.9 95.2 296.2 165.4 30.7 .85
TTR .704 .039 .707 .046 .003 .26
VAQ 2.26 1.01 1.80 .85 .48 .35
PP/1000 111.8 19.8 101.8 39.6 10.0 .85
Qual/1000 46.4 19.2 34.3 12.5 12.1 1.98*
All/1000 8.33 5.7 11.73 12.1 3.40 .96

Impersonal (room) Description
No. of words 291.3 76.0 267.6 81.7 23.7 .81
TTR .633 .036 .641 .050 .007 .64
VAQ 1.24 .82 1.00 .58 .24 .89
PP/1000 17.27 22.9 13.07 25.9 4.20 .47
Qual/l000 31.73 15.5 32.93 13.2 1.20 .22
All/1000 5.87 6.4 4.40 3.5 1.47 .74

*p = 0.06

was Qualification Terms, personal description. This finding

is amenable to inferences only in a minimal sense, since

finding one of twelve such tests significant is probable by

chance alone. This analysis of the extreme groups is seen

as an additional failure to reject the first null hypothesis.

The second part of the study focused upon the language

measures themselves as the independent variables. The means

and standard deviations for each of the measures were computed

by the arbitrary origin method for the 144 subjects. This

information is presented in Table 12 in the form of separate

means and standard deviations for the personal and impersonal


Comparing these results to the mean scores for the

same variables with the N = 90 R-S groups in Table 9 yielded

a high congruence on 8 of the 12 means. Somewhat lower means

were obtained from the entire subject pool than the sample of

90 subjects on the following four measures: Number of Words,

impersonal description (295.5 : 301.9); Number of Words,

personal description (311.3 : 323.7); Verb-Adjective Quotient,

personal description (1.94 : 2.08); and Qualification Terms,

personal description (35.83 : 38.67). These tendencies of the

R-S groups to write more and to use relatively more verbs to

adjectives and more qualifiers than the subjects as a whole

do not lend themselves to inferences about group composition.

The dispersion measures about the means for the whole subject

pool were at least five times greater than the mean differ-

ences shown above in parentheses and therefore strongly sup-

ported chance variation as an explanation.

0 mro0 %00 ,,0 (O
r s00 0 ,.' H,- ) || (0 IJ H 0" OCJ Cl
bO If\%X U\O 0O 00 00
S r- r 0 0 N


0 r

0 IL Y 0 30 a 0 C Go -t en

S _:co J0 ren M oJ -
bO fn 0 _. :t r-l_
a-a 4 II UO rl HH HM



: C0 %F-

..q> _jtj ,oi o r-r- rco o

H1 OHe
H a.

E- 1 '

Ir'aO fls cn r' IOO L
) i-- HH .00 H-F- iF-N 0

SDo H o o Ho H c3 o F cO 0
Nrd H H

aH H O UHra Cta)-H 0 aa arj a). -r a)Sa

o m 5-a ~ C 0rL
) a) S H HS vi m

0 o 0 K 0 E 0 r-

0 0 cd 4 o d -3 c03 r ^0 V 9

(D S E c

Extreme groups of 30 subjects were selected for each

of the language variables on the personal, and then the

impersonal description. For the TTR variable the high 30

and the low 30 of 139 subjects were the extreme groups as a

result of the elimination of five subjects who did not produce

the 100 words on both personal and impersonal descriptions

which was necessary for TTR scoring. On the five remaining

language measures the high 30 and low 30 subjects were drawn

from the entire subject pool. The ranges of language scores

for each of the extreme groups appears in Table 12.

When these extreme groups were selected, sometimes

there were two or more subjects with the same scores which

fell at the cutoff point. In the case of Allness Terms,

impersonal description there were 43 subjects with scores of

zero. The subjects for the extreme groups in these cases

were randomly chosen; thus the 30 subjects using low Allness

Terms, impersonal description, were not deliberately, if at

all, different from 13 other subjects not so classified on

Allness Term usage.

Examination of the standard deviation (S.D.) column

in Table 12 reveals a consistent finding on all six language

variables. The standard deviations for the personal descrip-

tions were greater than the standard deviations for the imper-

sonal descriptions. The differences were minimal for TTR,

Personal Pronouns, and Qualification Terms, but were marked

for number of Words, VAq, and Allness Terms. A greater varia-

bility in the language in the personal descriptions also was

found in comparisons of the ranges. For example the VAQs

on the impersonal descriptions ranged from .35 to 3.25 while

the personal description range was .24 to 5.79. This broader

personal description range was present on all language


For each language measure the high extreme and the

low extreme groups were compared on the personal and the

impersonal descriptions. The dependent variables were the

ten MMPI clinical scales, the three MMPI validity scales,

the R-S MMPI scale, and the SCAT raw verbal score. None of

the MMPI data included the correction for "K" used in

plotting profiles.

The results of the t tests comparing the high and low

language groups are presented in Table 13. Each entry in

the table represents a t value obtained by the high-low com-

parison in that description situation and for that MMPI or

SCAT scale. These results are summarized in Tables 14 and

15 and the actual mean scores of the high-low groups are

presented in Tables 16, 17, and 18.

Examination of Table 13 reveals that 94 of the 180 ts

were less than one and therefore did not approach signifi-

cance. There were 56 degrees of freedom in each t test and

significance at the 0.10 level was obtained when t was equal

to or greater than 1.67. Similarly the 0.05, 0.01, and

0.001 levels of significance occurred approximately at 2.00,

2.66, and 3.46, respectively.

H 0 C 0 0 r, 0o i\ r-l C-
rI ,. o, 03 oS om ,N m N 0 n om N A
So c o0 0 0 0 41
0 o r0-I I r-I rHl r 4l r

% N"cO N ,A a, -I 0% ** 0%

rl [J-d- C oO'i 0l r^ _ 3 o^ n 30
S0 d r-4 t-\4I If--- 0N
o3 0 2421

o 02 in NN r-I H-H HA

IA(n0 0 0% 2 I 0 CM 0 CM 0 cM co


0I- % 0 IA ol I- r r-I 3o o o 'o ,-
co 0 ,- C\, 0) 0 %0 0O CO (0 2 O-I 0'- H

0 \l f rl ,- m -C\j ( ,-\j I-I --t m r- -t

Sl-l l il

r 0 I Ocr Cy t%(n IA CO % E- C-::- :0 0 0A
00 0

0> HH H H
44 *22 00A O

1 0 00

l02 02024
0 02 0cON2-2O 0'0 *C O rHO 2 CO -. 0 Pr4l E

S. .. .. ooo
l1 E 0000












N 0"





* 3



MlE-11 rO

0 Ir d @ 42 a202. I 0 @)
l Do 3 tM Po N tl C >0 $ V o


* *

*3 *



Table 15

Summary of Significance Levels for High-Low Comparison

Unit Significant Significant Possible
Summarized ts, p=0.10 ts, p=0.05 Significant ts

No. of Words 2 0 30
TTR 3 0 30
VAQ 4 2 30
PP/1000 8 6 30
Qual/1000 13 5 30
All/1000 4 4 30

Room description 18 9 90
Self description 16 8 90

MMPI scales:
K 0 12
F 3 3 12
Hs 3 2 12
D 3 1 12
Hy 3 3 12
Pd 3 1 12
Mf 1 0 12
Pa 3 1 12
Pt 3 2 12
Sc 3 1 12
Ma 1 0 12
Si 1 0 12
L 0 0 12
R-S 0 0 12

MVeral scale 4 3 12

The distribution of significant ts is illustrated in

Table 14 and tabulated in Table 15. It is evident that the

occurrence of significant ts in the Number of Words, TTR,

and VAQ columns is well within the realm of chance. For

example under TTR, the occurrence of three two-tailed ts

significant at the 0.10 level out of a possible 30 is half

the number that would be predicted by chance alone. The

appearance of no significant ts at the 0.05 level in the TTR

comparisons is less than the chance expectations of three

significant ts per language variable. Similar observations

may be made for the number of Words and VAQ variables.

Significant ts for the Personal pronouns and Qualifi-

cation Term variables occurred at a frequency greater than

chance expectation. There were eight significant ts at the

0.10 level and six at the 0.05 level for Personal Pronouns

per 1000 words. Qualification Terms per 1000 words produced

13 and 5 ts significant at the 0.10 and 0.05 levels, respec-


Table 14 indicates that five of the six significant

MMPI scales for PP fell in the "neurotic triad" of the

Hypochondriasis, Depression, and Hysteria scales. The mean

scores, which appear in Table 17, reflect these results; the

low PP usage subjects were strikingly higher than the high

PP subjects on these three scales in the five significant

comparisons (out of a possible total of six significant com-

parisons). In addition the low PP subjects were significantly

higher on both SCAT verbal ability comparisons.


No such concentration of significant results was

found in the Qualification Terms. Rather the significant

differences were scattered on 11 scales, including the SCAT

verbal scale. The frequent users of Qualification Terms

generally were higher than the infrequent group on the

significant MMPI scales and on the SCAT scale.

The high users of Allness Terms were significantly

different from the low users on four MMPI scales at both the

0.10 and 0.05 levels. No consistent pattern was formed

except the occurrence of three of these four significant ts

in the room description. Four significant differences were

judged to be slightly greater than chance expectations at the

0.05 level.

The frequency of significant results in the room versus

self descriptions and on each of the scales is summarized in

Table 15. The frequency differences between the modes of

description were minimal and are not noteworthy. The most

frequent occurrences of the scales at the 0.10 level were on

the "K" MMPI scale and the SCAT verbal scale; at the 0.05

level the most frequent occurrences were on the "F", Hysteria,

and SCAT verbal scales. No significant results whatsoever

appeared on the MMPI "L" and repression-sensitization scales;

scales with only one 0.10 level significant difference were

the Masculinity-Femininity, Hypomania, and Social Introversion

scales. Inspection of the distribution in Table 14 confirms

that most of the significant differences were in the "K",

"F", and first seven clinical scales among the MMPI measures.

A word of caution must be inserted here about the

possible meanings of these results. The number of signifi-

cant ts is unimposing next to the number computed and they

lend themselves only to discussion of trends and suggestions.

The comments just made about the distribution of results

within variables should be interpreted within this framework.

The procedure just described was a comparison of the

extremely high subjects on each language variable to the

extremely low subjects on these same variables, in terms of

MMFI and SCAT scales. The same data, that is, the means of

the extreme groups on these scales, were investigated from

another viewpoint; these means were individually examined

with respect to the means for the entire subject pool. The

comparison of MMPI adjustment and verbal ability of sets of

extreme groups with other extreme groups was minimally

productive, but a comparison of the extreme groups with the

central tendencies of the population was substantially pro-


The means of the 14 MMFI scales, uncorrected for "K",

and the mean of the SCAT verbal scale for the 144 subjects in

the study are listed in the first column of Table 16 and are

repeated in the same position in Tables 17 and 18. The

standard errors of these means are presented in the second

column in Table 16. Each of these standard errors was

multiplied by 1.96 and 2.58 (Mcemar, 1955) to obtain con-

fidence limits about the means at the 0.05 and 0.01 levels,

respectively. Therefore the 0.05 limits about the mean for


--I I H l r- C r- en er ~

0 cl

H O r-4 C\J H 0 r r-4 r-0 i t- cn_:

E m m r- r-m r-r r 0 mr 0 m m
.o - D Go ,o n \I\eCorIcf\ -o,
0 .M 3 ( O l. 0 HC cH 4 H Z c
oo -o o -o o.z

pa j c% -. z J r- (a HH N -4 0 c N
,. r 0-4 ,-v ,-- V ,\--HH.. -4 4 _o

H H HNH Ca Ht c tNr i o -

%0o 0 H 0H "

0 0 0 m- r-H H mH 0 H -- m Or :0
0 0 *
co "o ro %4; ;\ C; C; CA o^ o c o oC C
M r-I -I C i-H ri 1 nU( O

m r-io n 0 0 0 rnm moj\dnrn r.- cM
H 0j rI O m - r l (\I G Cv C\j= r--l30 0 1m
H3 3 ri H r- r-4 c OH HH H NM en1

r- 3 _mNO ovr-tv0oNr-I cr- (M O%3oN H

o% i v a, C\j p)% 3 .t col %-l r-l Zo p^ If\ as
4) H- D t N e HH\ r-r- 0

a0 6 0 a33 -6O 9 4 CA 40 0 0 0 H H)


o ~- ,- ,- t,,- -. ON 0 r' t,,. -
o- z r-, m cr Ioo oj o" o o r-
0 0 H 0 0 0

00 0 a ,-H ,. ,- ,.,- ..
N. r0 roI N- H ro r.- 4 Co r,
i N1 HHHNHHHO r Ni-! Cihr\

.4 PO 0 0 o
0 0

,(1/, O* *
0 o < O r O or- ,, (- C% 0 O ( n-r,,OO-
0 N
a 3 Hi a ,-4l i -I C H- ,-I H- cm \, ..

5 Ln Ir r- r- r- rN-H rNn rl-enn
F H -d- %D cle\1A r %l G% OC 0 GO L11- 0 CM

aoo o o
4a P 0 1Ar 1A-.d c ,- ,O ,,O O C -; -C OZ rA

4, 0l H j r 0 r-N r j r- C, (
0 r I

00 D0
(3? 43 P% 0 0 0 r- 0 Lp- (n (n 0 e\ cr\ 0

0- ... m. r .u H N H r.-I r-I r4 J (CA A

a o o Oo
00 0 3

0 00 e -- 00
q 0 00
> 1 -( a m r- 0 ne n coa n f- 0 0

H 0 30H -r-rHHO O ClA l -I O- 0
r -I H N r4 N r-H NHN HH N en


r 04 I-N m w N :P4-N ) cc 00

HNH CU rHN N H H r- C n


0 J OM O 0p%1- CM pr--:O OaO O IX : O IA P%
'o ; 'j; 14, A o0 ; o P- uXA C, -. o
Ha cm 1-4l ( Ml ( -I rl r1 r C1l en m^ r
N 0 -

0 0NCMr-lr-f 000'0
0 In 1A 4 A $ j
fr 0_2 p W oooo

H 0

2r E E-4 0 o-D %0 -,0 0 ,O- 0 OM

Ea4 H O H H H H M O
,0 O, r 0 p 0 e 0 m 00" -- P 0 mr

tO 0 V\ H IA 0 0% %,0 t 0 00^ _z en %0 Cl 00
00 43E r -2: p-oj c oo co- 3 r c

0 r H HHH Ho oH M HH N r- mr
0 0 rH
r 4 H 0 0
.0 0-P 0 cW
d ca s i O r r- o e r-e r- r- p me r- 0
E-4 M 0 H 2) O 0 d- L- M a% D o cr 0 00
S4 CA 0-o.-
0 N -4 02 H HH N en
en E o r oj ren r- r-l r-l r-l oU n

r: 0 0 r-i r\ -l-:t 0 0%30 o r i0 H cm p-r
3..0 H 1. 0 S
t o 0 1) cm C r0 r0 rH H < m fneN
4I 0 E
0 00

j 000
E'0 2J\O^NHCr'r0p-C'-(.pH 00

0 ---t _zt en 06% ( o P4i 4 CA*A
2 0- i- c i- ( i 00
02 c aE 00

c) 0202

"K" were computed, adding to and subtracting from 15.479 the

number .420 x 1.96, or 0.823; the 0.05 limits for "K" were

14.656 16.302. The same operations were carried out on

the other 14 scales for these limits and a similar procedure

for the 0.01 confidence limits.

The MMPI and SCAT mean scores of each of the extreme

groups (N = 30) on the language variables are listed across

the rows in Tables 16, 17, and 18. Those subjects who were

extreme in Number of Words and TTR have their means grouped

in Table 16, the extremes in VAQ and Personal Pronoun usage

in Table 17, and the extremes in frequency of Qualification

and Allness Terms in Table 16. These means are identified

separately for the room and self descriptions. Sixty means

are listed for each of the language variables, a total of

360 means.

After the confidence limits were computed for each of

the population means, the extreme sample means were inspected

to determine whether or not they fell within these 0.05 and

0.01 limits. For example the first extreme sample mean in

the first row of Table 16 is "K" = 15.70 for the 30 subjects

who produced the most words in describing the room. This

falls within the range 14.656-16.302 and does not differ

from the population mean at the 0.05 level of confidence.

The next number in that row, 16.67, does fall outside these

limits, and indeed outside the 0.01 limits, and thus it is

indicated that the sample mean is not a chance deviation

from the population mean. The confidence limits outside of


which the individual sample means fall are indicated by


A total of 114 means were found to fall outside the

0.05 confidence limits of 360 means examined. This included

105 out of 336 MMPI scales means and 9 out of 24 SCAT scale

means. The actual comparison of the sample means may be

made by noting the total sample means listed in the left


Sixty-five of the 360 means examined fell outside the

0.01 confidence limits. This included 5 SCAT verbal scale

sample means and 60 MMPI means. The number of means falling

outside both the 0.05 and 0.01 limits were several times

chance expectations.

Beyond the simple observation that the frequency of

the extreme group means falling as they did was non-chance,

some comments may be made about their distribution among the

language, situational, and subgroup variables. The distribu-

tion of those means that differed from the population means

beyond the 0.05 level of confidence is presented in Table 19,

which is a summary of the results of the confidence limits

testing in Tables 16, 17, and 18. Table 19 categorizes the

MMPI results separately from the SCAT results, and performs

this for the extreme groups (N = 30) on each of the six

language variables and for both modes of description.

The greatest frequency was on the Qualification Terms

variable, in which a total of 22 means out of a possible 56

were outside the 0.05 limits. The greatest number of




cd v





H -0

0 4





























0 0

S0 0

0 0

r-4 r-I 0

CM 0 r-

0 Nm

0 NM -

0 r^ _-

0 'A r-


I= Of .
E-4 0
E- P 4
&< P


























deviations of the sample means from the population means on

this variable was the group of 30 subjects who used the

most Qualification Terms in describing the room; this group

was deviant on 9 of 14 MMFI scales and was higher than the

population on 7 of them. On the other hand the group using

the fewest Qualification Terms on the room description

differed from the population on only one MMPI scale.

This difference did not appear in the self descriptions,

in which both the high and low qualifying groups deviated from

the population on six MMPI scales. There was no overlap on

MMFI scales that were deviant in the two groups.

The second and third highest numbers of deviations

among the language variables were on VAQ and Allness Terms,

respectively. Twenty VAQ means exceeded the scale confidence

limits, and ten of these were in the subgroup using the fewest

verbs in proportion to adjectives in their self descriptions.

The other three VAQ extreme groups produced far less deviant

means than this, and the MMFI discrepancies between sample

and population that predominated in the low VAQs on the self

descriptions, were found to be greater in the high VAQs on

the room descriptions.

Nineteen Allness Term means were outside the MMFI

scale confidence limits and were distributed rather uniformly

on the personal (self) and impersonal (room) descriptions.

The subjects using the largest number of Allness Terms

deviated more from the whole subject pool than did the

subjects using the fewest Allness Terms.

Among the extreme users of Personal Pronouns there

occurred a less pronounced but essentially similar finding

to the Allness Term results. The high users of Personal

Pronouns on both modes of description deviated more from the

population on MMPI scales than did the low Personal Pronoun


The first two language variables listed, Number of

Words used and TTR, had the smallest number of MMPI means

that differed from the population means. There were 15 in

Number of Words and 12 in TTR. Examination of the distribu-

tion of these deviant MMPI means in the TTR classes reveals

that all of the deviant means were among the low TTR subjects.

In both descriptions there were no MMPI scale differences

between the subjects who had high TTRs and the population; on

the other hand there were eight and four means that were

deviant among the low TTRs on the impersonal and personal

descriptions, respectively.

Results in the same direction were found on the Number

of Words variable. Little difference was found comparing

the MMPI scales of the high word producers with the scales

of the subject pool, but much more marked differences were

found when examining the low word producers.

The columns for the MMPI scales in Table 19 were

totaled and only slight discrepancies were present. More

means outside the 0.05 limits were present in the room

descriptions than in the self description, but this was only

55 to 50. Seven more of these means were found in the high


frequency groups than the low frequency groups in the room

descriptions, while ten more means were found in the low

frequency groups than in the high frequency groups in the

self descriptions. The grand total in the MMfI columns

indicates that 105 means of 336 examined were outside the

0.05 confidence limits, as noted earlier.

The examination of the SCAT columns is less complicated

than the MMPI process, for the maximum number in each cell

is one and the minimum number is zero. Again the high

variable in frequency count was Qualification Terms, which

produced three out of a possible four means lying outside

the 0.05 confidence limits about the population SCAT verbal

mean. On the impersonal (room) description the high quali-

fying subjects were on the average higher in verbal ability

than the population and the low qualifiers were lower on this

measure. On the personal description the high qualifiers

were again higher in verbal facility while the low qualifiers

did not differ from the overall mean.

The high TTR producers on both modes of description

had higher mean SCAT scores than the population, while the

lower TTR subjects did not differ from the whole group.

Therefore two of the four TTR groups had deviant SCAT scale


Two of four Personal Pronoun group means on the SCAT

scale differed from the group as a whole. The subjects

using the fewest personal pronouns had higher mean SCAT

scores than the subject pool.

One of four means differed on the VAQ and Allness

Term measures. On the VAQs, the subjects producing higher

VAQs on the room descriptions had higher SCAT scale scores,

and the subjects producing the most Allness Terms on the

room description had lower SCAT scores than the population.

The sixth language variable, Number of Words, produced

no deviant means out of the four possibilities.

Examination of the columns in the SCAT portion of

Table 19 reveals the greater tendency of the room over the

self description situation to bring out verbal ability differ-

ences in the extreme language groups. That is, there were

six extreme group means that were outside the 0.05 SCAT con-

fidence limits on the room description, while there were only

three on the self description.

The distribution of means which were different from

the population at the 0.05 level of confidence was tabulated

for each of the MMPI and SCAT scales. Generally they were

bunched closely, with the exception of the "L" scale in which

only three deviant means were found. Six deviant means out

of the possible total of 24 were obtained for the "K" and

Schizophrenia scales. Seven deviant means were found in the

Depression, Psychopathic deviate, and Masculinity-Femininity

scales. Eight deviant means were obtained for the Hypo-

chondriasis, Social Introversion, and repression-

sensitization scales. Finally, the "F", Hysteria, Paranoia,

Psychasthenia, Hypomania, and SCAT verbal ability scales

contained nine deviant means each.

In concluding the presentation of the statistical

results of the investigation, a further note must be added.

All of the data have been abstracted from the uniquely

organized productions of the participants for the purpose of

objective analysis of the explicitly abstracted measures.

In this approach there is a loss of the manifest contextual

variables as well as the omission of the personal gestalt

inherent in these descriptions. In order to correct somewhat

for this loss, selections from the descriptions are appended

(Appendix C). These selections are identified in terms of

the two descriptive situations, the language variables

illustrated, and personality (R-S), SCAT verbal ability,

age, and academic classifications of the writers.


The Language Measures as

Dependent Variables

Personal and impersonal description. The analyses of

variance clearly identified the mode of description as the

significant source of variance in every language measure

studied. It was significant beyond the 0.01 level on all

six analyses, while the repression-sensitization variable

was significant on none, and the verbal ability measure on

only one, Qualification Terms. No a prior predictions were

made, nor were any unspoken expectations present for these


On some of the language variables the striking

significance of the personal-impersonal description variable

is seen easily post hoc. The markedly higher VAQ on the

self description above that of the room description confirms

the results of Boder's study, in which he reported lower

VAQs in scientific, descriptive writing than in the more

personal texts of drama and fiction. The frequency of

Personal Pronouns seems to have been a natural consequence

of the definition of the description variables; that is, in

the self descriptions it is expected that subjects will use

many Is and mes while in the room description few Is and

many its will appear.



Before considering the significance of the personal-

impersonal differences in relation to some of the other

language variables, it is necessary to review the testing

situation. For both descriptions the subjects were asked

to write for 25 minutes and to cover at least two sides of

a single sheet of lined paper. Now in describing themselves,

the subjects had an enormous repertoire of material upon

which to draw; some discussed their early childhoods, some

wrote of their physical features, some focused on their goals

in life, others described their most immediate and intimate

feelings and anxieties, and so on. On the other hand in

describing the room, the subjects had a much more limited

set of materials to write about. The room contained one

couch, one coffee table, a rug, a painting, flowers, and so

on. The room description was more precisely defined and more

limited than the self description, and as a consequence, many

subjects may have found it more difficult to write for the

required number of pages and for the required time on the

room description.

The data support this interpretation of the situational

demands. Approximately 20 more words were used on the average

in the personal description than in the impersonal descrip-

tion, a difference significant past the 0.01 level. The TTR

comparisons produced a lower TTR in the impersonal descrip-

tion that was significantly less at the 0.001 level than the

personal description. On the personal description the

subjects wrote more with a greater diversity of words.

An observer noting the greatly increased production

of Qualification Terms and the increased number of words on

the personal description might be tempted to hypothesize

that the subjects as a whole were exhibiting a form of defen-

sive behavior. He might reason that the greater threat of

describing themselves was manifested in the subjects'

modifying their statements more and writing more to justify

their self descriptions. He might further press his case

by pointing out that the VAQ, a "known" measure of anxiety

was greater by almost twofold on the personal description.

This possibility is not likely in accounting for the

present results. The VAQ differences have already been dis-

cussed in terms of the stimulus qualities. It is not

unexpected that relatively more adjectives than verbs are

used in describing a room, a situation which implicitly

calls for adjective usage, than in describing oneself. The

differences in the number of words used have been discussed

also in terms of the stimuli. Finally the increase in

Qualification Terms was accompanied by an increase in Allness

Terms used in personal description, which weakens the argu-

ment that increase in Qualification represents defensive

qualifying. Likewise, this argument would have to be rejected

on the basis of the repression-sensitization results. The

lack of significant differences between the R-S groups, in

the face of the wide individual differences in R-S scores,

indicates that defensiveness so measured was not related to

the language variable patterns.

The consistent differences between the personal and

impersonal descriptions on the language measures have exten-

sive implications for the meaning of other studies in language

when the lack of significant differences on the personality

variable is considered. These results raise as an issue the

actual contents and topics of the language samples of the

previous studies. For example, in the Balkan and Masserman

(1940) study in which T.A.T. cards were used to elicit the

language samples, patients identified as anxiety states were

found to have clearly higher mean VAQs than the obsessive-

compulsives and conversion hysterics. Was this due to the

different language style and structure the groups used, or

was it a result of different content production in groups

using the same style and structure? In other words did the

hysterics blandly describe the mountain and other physical

features of card 11 of the T.A.T., while the anxiety states

created highly interpersonal stories with many characters and

much emotion? If this was true, and there are indications

that it was from the authorst reports, then perhaps the

original conclusions about the language of phantasy of the

T.A.T. varying with psychoneurotic groups should be sup-

planted with a statement about the varying contents of the

T.A.T. with psychoneurotic groups.

Lorenz and Cobb acknowledged this division between

structural and topic portions of language with their intro-

ductory statement that ". . within the universality of

structure and inherent stability of the statistical

properties, grammatical divisions, rules of syntax, and

denotative word meaning," there is considerable freedom of

the individual to select, choose, and arrange the form as

well as the content of what he says. They observed that in

their 1000 word speech samples ". . considerable latitude

existed in the choice of content materials." Although they

did not specifically investigate content, the suggestion was

present in their study as well as this one that content

differences were a major factor in the VAQ, TTR, and Pronoun

frequency results.

The most undefined language referents were those of

Chotlos, in which he told the subjects to ". . write about

anything you want to write about, just make it up as you go

along." His conclusions that higher IQs and higher age

levels were associated with more highly differentiated

language structures may be true, but the data he collected,

with their potentially enormous topic range, cannot sub-

stantiate these conclusions, at least in this form. It is

speculated that the high IQ subjects may have produced

language samples in content spheres quite divergent from

that of the low IQ subjects and that the language analyses

may have reflected this divergence rather than that of

structural language on given topics. Chotlos may have

confused content variability with structural variability.

In this context it is interesting to note that two

studies in which the subjects had a very limited field of

linguistic referents and which used questionnaire-derived


personality dimensions, there were only minimal differences

between the personality groups (Benton, Hartman, and Sarason,

1955; Wagner and Williams, 1961). The topic repertoire in
these studies focused on thoughts about the subjects' majors

and careers; this content delineation is felt to have been

closely associated with the lack of many significant results.

Repression-sensitization. The lack of significance

of the R-S groups on the analyses of variance merits indi-

vidual attention, for most of the personality groups studied

on language measures have produced some significant results

and because the repression-sensitization dimension has con-

sistently differentiated between varying kinds of behavior

in other studies (Altrocchi and Dickoff, 1963; Altrocchi,

Parsons, and Dickoff, 1960; Altrocchi, Shrauger, and McLeod,

1963; Byrne, 1961, 1963; Byrne, Barry, and nelson, 1963; Joy,

1963; Tempone, 1962; and Ullman, 1958, 1962). As noted above,

one possible explanation is the use of questionnaire-derived

personality dimensions rather than behaviorally-derived in

the sense of psychiatric patients.

The kind of subjects who participated in this investi-

gation may have contributed to the R-S results on the language

variables. All of the subjects were college students and

most were probably adequately adjusted since none were at

that time hospitalized for psychiatric reasons. The studies

reporting significant results have generally included overtly

pathological groups of subjects (Balkan and Masserman, 1940;

Fairbanks, 1944; Mann, 1944; Lorenz and Cobb, 1954; and

Osgood and Walker, 1959) while non-significant or marginally

significant results were found in studies using college

students (Benton, et al., 1955; Doob, 1958; Wagner and

Williams, 1961; the present study). However very different

methodologies were used in the two groups of investigations

just cited and conclusions about sample differences as

causal agents would be premature.

Another possibility to account for the absence of a

relationship between R-S groups and language structure is

the nature of R-S itself. While R-S may have meaning with

regard to speed of tachistoscopic perceptions, perhaps it is

insensitive to those elements in personality functioning

that are associated with language structure. The content

of the repressors' and sensitizers' self descriptions may be

used as an example. The investigator observed repressors

whose self descriptions were anxious, concerned, and appar-

ently psychopathological; sensitizers were noted whose self

descriptions were idealized, full of denial, with no admis-

sions of any concerns; and there were neutrals who described

themselves as anxious and others who described themselves as

totally without worry or problems. Thus, self-descriptive

content of the language of repressors and sensitizers need

show no necessary relation to the description of repressors

and sensitizers derived from questionnaires or non-language


The lack of significance on the formal, structural

language measures in differentiating the R-S groups raises


a further question: If matched R-S groups were based on

extremes of defensiveness and over-experiencing of feelings

derived from the subjects' own self descriptions instead of

the MMPI scale, would the structural language patterns have

varied significantly? This question was not explored and

future research might be directed toward its resolution.

An additional issue is the effects of the recency of

the MMPI administration on the reliability of the R-S measure.

The results of the time-of-administration analyses were not

definitive in rejecting or accepting the time interval

between MMFI testing and collection of the language sample

as an influencing factor. However a substantial collection

of the MMPI literature (Welsh and Dahlstrom, 1956) has indi-

cated that the reliability of the commonly used MMPI scales

is adequate over time.

Thus, four possibilities are suggested to explain the

failure of R-S to be reflected in the language patterns:

(1) the use of college students as subjects; (2) the use of

a questionnaire method to select from the subject pool;

(3) a real insensitivity of the R-S scale; and (4) time lag

between obtaining the R-S MMPI data and collecting the

language sample.

Verbal ability. The null hypotheses generated with

regard to verbal ability and the language variables were not

rejected on five of the variables, but on Qualification Terms

significant differences between verbal ability groups and

interaction between verbal ability and personal-impersonal


description were found. The High verbal ability group used

the most Qualification Terms, the Mid ability group the

middle number of Qualification Terms, and the Low ability

group by far the least Qualification Terms. This pattern

was present on both descriptions, but on the room description

the qualifiers used by the Low and Mid groups were close in

number with relatively many more used by the High group; on

the self description the number of Qualification Terms used

by the Mid and High groups were close and both far greater

than that of the Low ability group.

The lack of significant differences between the verbal

ability groupings and TTR did not support the conclusions of

Chotlos and again raises the issue of content of Chotlost

language sample as a TTR influence.

The greater use of qualifiers by the subjects higher

on the SCAT verbal ability measure may well reflect a more

precise and accurate way of writing and of self-expression

by these students. That is, the individual who reports that

the window is "approximately eight feet tall" is responding

in a better defined manner than another person who omits the

"approximately." The word "approximately" in this case adds

the element of uncertainty and extends the situational


In noting the absence and presence of significance

resulting from the verbal ability analyses, it is proper to

consider the base for these analyses. The subjects were

given the SCAT at varying times in the three years preceding


testing for language, and unreliability effects may have

modified the meaning of their SCAT verbal ability scores at

the time that the language sample was obtained. In addition

there were 10 subjects who had equivalents of the SCAT scores

for their verbal ability measures, and possible error lay in

the equating process. However it is felt that these are

sources of only minimal error.

A somewhat greater deterrent to extensive inferences

being drawn from the language and verbal ability data is the

manner of division of the subjects into the groups by verbal

ability. The high 10, low 10, and middle 10 in the R-S

groups were identified as the High, Low and Mid verbal ability

groups. This arbitrary process in which two subjects with

the same SCAT score may have been classified into different

verbal ability groups if one subject were a sensitizer and

the other a repressor, yielded groupings of High, Low, and

Mid ability which were tentative, rough, and at places over-


These guidelines for verbal ability groupings served

to hinder rather than enhance group differences, and the

negative, non-significant results therefore must be accepted

cautiously. On the other hand the positive finding with

regard to Qualification Term incidence is a more certain one

to be so determined in spite of the nature of the ability


Extreme repressors and extreme sensitizers. A similar

finding of the Qualification Term measure as the only

significant language variable occurred in the t test analysis

of the extreme repressors and extreme sensitizers (Table 11).

While it is true that one out of twelve two-tailed ts falling

significant at the 0.06 level is a chance incidence, the

repetition of Qualification Terms as the single significant

measure in two separate analyses with different variables is

likely not due to chance alone. Furthermore the later

analyses with the language variables independently manipulated

reaffirmed that Qualification Terms usage was associated more

than any other language measure studied with MMVI and SCAT


The extreme repressor-extreme sensitizer analysis

itself was of the type the author has criticized in others.

That is, it was a simple, unidimensional comparison of two

groups on languages measures. Nevertheless the absence of

substantial significant results appears as a confirmation

of the findings noted earlier on the R-S dimension.

Comparisons with other studies. The personal-

impersonal description measure has been demonstrated to be

the only one that consistently differentiated language usage.

For this reason the results for the entire subject pool of

144 students on their room and self descriptions, rather than

by R-S and SCAT groups, were chosen to be compared to other

studies in the numerical language measure results. The wide

diversity of situations, the use of spoken rather than written

language, and the reporting of results in other forms than

used in this study, when the numerical results were reported


at all in other studies, makes numerical differences subject

to a broad range of influences. For this reason only studies

that used written language in similar situations and that

reported the language measures in equivalent forms were


The VAQ results in this study and others may be

observed by examining Tables 1 and 12. The sample and situ-

ation which seems closest to the present study was the col-

lection of written life stories in college freshmen by Mann

(1944). The comparison of these results with the personal

description by the college students of all classes in the

current investigation yields close results in VAQ. The Mann

language sample had a mean VAQ of 1.96, while that of the

present personal description sample was 1.94.

Similar comparisons of this study with the Mann study

yield very congruent data. The Mann freshmen language had a

TTR of 0.71 and Personal Pronoun frequency of 14.6%, while

the current study found a mean TTR of 0.705 and a PP frequency

of 11%. Therefore, in spite of the geographical divergence

and the 19-year span between collections of the language

samples, the language used by these college student groups

were highly consistent on the three common structural measures.

The only other VAQ study with raw data reported was

Boder's, in which the descriptive, scientific writing had a

mean VAQ of 1.32 (converted from adjective-verb quotient

scores), the fiction VAQ was 2.82, and the drama VAQ was 8.93.

This lowered VAQ on the impersonally descriptive mode of


course was found in the current study also, with a mean VAQ

of 1.08 for the room description and a VAQ of 1.94 on the

self description.

The remaining studies that were reviewed using the

VAQ, TTR, and Personal Pronouns, and all of the studies

using Qualification Terms and Allness Terms were not amenable

to comparison for one or more of the reasons stated earlier.

A final note may be added in examining the language

data for all of the subjects in Table 12. In every language

variable the standard deviation was greater for the personal

description than for the impersonal description, although

very large differences appeared on just three of the six

variables. This clearly indicates the meaning of the situ-

ational definition in terms of the subjects' variability;

that is, the room description with its more explicitly and

narrowly defined external referents led to greater similarity

in a structural sense than did the self description. It was

contended that this difference reflects the wider range of

content elicited by the personal description situation.

The Language Measures as

Independent Variables

Many of the conclusions drawn from the first part of

this investigation are seemingly contradicted when the

results of the second part of the study are examined. These

contradictions primarily center about the relationship between

adjustmental variables on the MMPI and the language variables

studied. This occurs because the two parts were separate


investigations, although the same basic data and many of the

same classifications were used. In the first part the per-

sonality measure was the independent variable and the

language units the dependent measures; this order was rever-

sed in the second part as the language measures became the

manipulated variables.

The apparent contradiction is between the negative

finding with respect to the personality dimension in the

first part, and the positive findings about this relationship

in the second part. This contradiction dissipates when the

operations involved are considered; the early part led to the

conclusion, among others, that extreme R-S groups do not

produce different formal language patterns, while the later

part led to the inference that extreme language usage groups

do produce R-S (and other MMPI) scores that are different

from the entire subject pool.

t test analyses. In the second part of the study the

extreme groups of the high and low 30 subjects on each

language variable were selected, MMPI and SCAT means were

computed, and the data were analyzed in two ways. First

t tests were performed between the high and low group MMIf

and SCAT means, and the results presented in Tables 13, 14,

and 15. The Personal Pronoun and Qualification Term

variables were the only ones of the six that had more signifi-

cant ts than were expected by chance.

The significant ts on the Personal Pronoun tests formed

a distinct pattern. The low users of Personal pronouns in the


room descriptions were significantly higher than the high PP

users on the "Hs" (0.01 level), "D" (0.1), and "Hy" (0.001)

scales. The low users of Personal Pronouns in the self

descriptions were significantly higher than the high PP users

on the "D" (0.001 level) and "Hy" (0.01) scales.

The "Hs", "D", and "Hy" scales are referred to commonly

as the neurotic triad and their simultaneous elevation in the

low PP users suggests that some dynamic-motivational elements

are present. One interpretation of this is that a reluctance

to use Personal Pronouns is associated with a neurotic denial

of self-involvement. It was also found that the low PP users

had significantly higher verbal ability scores than the high

PP users and therefore no causal or unidimensional inferences

can be drawn confidently from the extreme PP group patterns.

While more Qualification Term ts were significant

than any other language variable, the 13 significant ts were

scattered without any noticeable pattern among 10 MMPI scales

and with a noticeable pattern in the two SCAT tests. Both

of the SCAT t tests were significant, the room description

test falling beyond the 0.05 level and the self description

falling beyond the 0.1 level. On both modes of description

the frequent Qualification Term users were higher on the

verbal ability measure than the infrequent Qualification

Term users. This supports the relationship between SCAT

verbal ability and Qualification Term usage reported in the

first part of the study.

The four remaining language variables, Number of Words,

Type-Token Ratio, Verb-Adjective Quotient, and Allness Terms,

yielded only chance occurrence of significant t tests com-

paring their extremes. However all of these as well as

Personal kronoun frequency and Qualification Terms did fall

at a greater than chance incidence outside the 0.05 confi-

dence limits of the means for the whole subject population.

These more positive findings on the confidence limit analysis

were due to a number of factors.

The discrepancy between the t test and confidence

limits analyses. Obviously there were different goals for

the t tests and the confidence limits analysis. In the former

the purpose was to compare extreme groups, to accentuate

linear differences on the language variables; in the latter

the purpose was to examine the relationships between the

extreme language groups and the population as a whole.

Beyond this there were statistical factors present. In the

t tests there were just 58 degrees of freedom, and this

combined with the wide dispersion of MMFI and SCAT scores

within the extreme language groups to produce relatively

small numbers of significant ts. In the analysis by confi-

dence limits the variance within the extreme groups was not

taken into consideration since the level of confidence was

determined by whether the mean alone of these extreme groups

was within or without the confidence intervals about each

scale mean. Therefore the wide variances that had resulted

in non-significance of some fairly large mean differences

between the extreme groups themselves were not a factor in

testing the confidence limits.

In the confidence limits procedure the standard errors

of the means were determined by dividing the standard devi-

ations by the number of subjects and extracting the square

root. Because the whole subject pool was used, the square

root of 144, or 12, was divided into the standard deviations,

and the division by this relatively large number produced

small enough standard errors to make even those means that

were modestly different from the population means fall

outside the confidence limits. However the extreme group

means were included in the population means, an influence

that could have made the population and extreme group means

closer than the confidence interval setting assumed.

Analysis by confidence limits. The confidence limits

data were presented in Tables 16, 17, 18, and 19 and the

frequency of different language extreme group means falling

outside the 0.05 limits was pointed out in the results


The patterns and distributions of the language

variables of the extreme groups that were beyond the 0.05

level of confidence are felt to have the following implica-


1. The subjects using the fewest words differed more
than the high word producers from the average on the MMPI,
and generally in the direction of less pathology and higher

2. The high Type-Token Ratio subjects had sharply
higher verbal ability scores than most subjects. This sup-
ports the results of the Chotlos study and is divergent from

the results on this relationship in the first part of the

3. The low Type-Token Ratio subjects varied notably
from the subject population on many MMPI scales, with a
particular lowering of intellectualization, as indicated by
their lower mean on the MMPI "Mf" scale and a tendency to
greater introversion indicated by their lower MMPI "Si" scale.

4. The subjects with high VAQs in their room descrip-
tions and those with low VAQs on their self descriptions had
many MMPI scales elevated over that of the subject pool.
Thus the highest MMPI scales were found in the subjects who
deviated most from the expectation of the VAQ that propor-
tionately more adjectives would be used in the room descrip-
tion and proportionately more verbs in the self description.

5. The users of few or no Personal Pronouns were
markedly higher on verbal ability than the overall subject
group and tended to have a higher neurotic triad on the MMPI.
These results were also found in the t test analyses. Thus
the low Personal Pronoun users appeared to be more neurotic
and to have superior verbal skills.

6. The high users of Personal Pronouns were lower
generally on the neurotic triad than the mean of the 14
subjects. Those using the most PPs in the room description
were also more defensive, as indicated by a high "K" and
low R-S on the MMPI.

7. The extremely high Qualification Term producers
had higher verbal ability scores than the SCAT means. This
was the most powerful single relationship between any variable
and the SCAT verbal ability measure in terms of the confidence

8. The frequent users of Qualification Terms appeared
to admit more symptoms, were more open, more anxious, greater
intellectualizers, and less defensive than the general subject
population. This was indicated by their higher scores on the
"F", "Mf", "Pt", "K", and R-S scales.

9. The high Allness Term subjects on the room descrip-
tion were more repressive and less anxious than the overall
mean, while the high Allness Term subjects on the self deacrip-
tion were more anxious with a number of elevated MMPI scales.

10. The high Allness Term subjects had lower verbal
ability scores than the entire subject pool.

Of course this list of the ten major implications does

not cover all the extreme language groups and did not focus


on the conformities of some of the extreme groups to the

overall means. For example, the extremely low users of

Allness Terms on the personal descriptions had MMrI and SCAT

means that fell within the 0.05 confidence limits on 14 of

the 15 scales; for practical purposes there were no differ-

ences between the subject pool and this extreme group on the

adjustmental and verbal ability measures.

A general conclusion that may be drawn from this

analysis is that questionnaire-derived personality measures

are indeed related to extreme use of language measures in

the context of the confidence limits procedure. The confi-

dence limits analysis and to a lesser degree the t tests

failed to support the non-relationship between these variables

suggested by the first part and did support the research

literature finding relationships between personality units

and language measures.

This confidence limits analysis found that repression-

sensitization had the median number of deviant extreme group

means, and as a result some of the earlier discussion of

possibilities of R-S as an insensitive scale may be amplified.

In this part of the study it served on the whole neither

better nor worse than the 13 most commonly used MMFI scales.

For the whole investigation there were no indications that

R-S is an especially meaningful behavioral correlate of

language patterns. Neither the defensiveness nor the open

anxiety theoretically associated with R-S extremes were

manifested in language usage. Thus it was felt that caution

should be employed in treating R-S as a single, general

dimension of personality. Repression itself is a broad

concept that can be applied to many areas of behavior, but

repression as measured by Byrne's revised MMrI scale would

seem to be most applicable to the specific fields in which

positive results have been established. Thus a distinction

is drawn between repression as a generalized, clinical entity

and repression as an objectively-scored response to 127

true-false questions.

Finally, of the six language variables studied, Quali-

fication Terms appeared to be most persistently differentiated

and the most persistent differentiator. It was related espe-

cially to verbal ability as measured by the SCAT and it may

prove to be a most useful formal language measure in the



The purpose of this study was to investigate the

contributions of personality, verbal ability, and situational

variables to structural language usage and to explore test

behavior associated with extreme use of this structural

language. The six language measures studied were the Type-

Token Ratio, Verb-Adjective Quotient, Personal Pronoun fre-

quency, Qualification Term frequency, Allness Term frequency,

and Number of Words used.

Written personal descriptions and descriptions of a

projected picture of a room served as the situational stimuli

and language sources. The MMPI repression-sensitization

scale and SCAT verbal scores were the personality and verbal

ability measures. In addition the main 13 MMPI scales were

used in the exploratory part of the study.

The subjects were 144 college students at the Univer-

sity of Florida in May, 1963. They were divided into repres-

sor, neutral, and sensitizer groups and High, Mid, and Low

verbal ability subgroups. In the initial part of the investi-

gation repression-sensitization and SCAT verbal ability were

the sources of variance tested between subjects and the

description situation was tested within subjects. The six

language units served as the dependent variables.



A Lindquist Type III analysis of variance for each of

the language variables indicated that the primary significant

source of variance was the situation, or mode of description.

The personal descriptions produced Type-Token Ratios, Verb-

Adjective Quotients, Personal Pronouns, Qualification Terms,

Allness Terms, and Numbers of Words that were greater than

those of the impersonal descriptions beyond the 0.01 level

of significance. No significant results were found for

repression-sensitization and differences between SCAT measures

were significant only for the Qualification Term variable.

High verbal ability was associated with frequent use of

Qualification Terms beyond the 0.05 level of significance.

In the second part, which was the exploratory section

of the study, the language units were treated as the inde-

pendent variables. Confidence limits testing and analyses

by t tests were performed on the extremely high and extremely

low users of the different language measures and it was found

that the MMPI, SCAT, and modes of description varied in

meaningful ways in these extreme groups.

It was concluded that the nature of the situation

eliciting the language sample was central to the formal aspects

of the language produced and that there is a need for lin-

guistic study of non-content variables to define the external

language referents more precisely.


Altrocchi, J. & Dickoff, Hilda. Ego-control patterns and the
attribution of hostility. Privately circulated manu-
script, 1963.

Altrocchi, J., Parsons, 0. A., & Dickoff, Hilda. Changes in
self-ideal discrepancy in repressors and sensitizers.
J. abnorm. soc. Psychol., 1960, 61, 67-72.

Altrocchi, J., Shrauger, S., & McLeod, Mary Ann. Attribution
of hostility to self and others by expressors, sensi-
tizers, and repressors. Privately circulated manu-
script, 1963.

Balkan, Eva, & Masserman, J. H. The language of phantasy:
III. The language of phantasies of patients with
conversion hysteria, anxiety states, and obsessive-
compulsive neurosis. J. Psychol., 1940, 10, 75-d6.

Bateson, G. Language and psychotherapy; Frieda Fromm-
Reichmann's last project. Psychiatry, 195J, 21, 96-

Benton, A. L., Hartman, C. H., & Sarason, I. G. Some rela-
tions between speech behavior and anxiety level.
J. abnorm. soc. Psychol., 1955, 51, 295-297.

Board of University Examiners, University of Florida.
Results on Florida state-wide twelfth grade testing
program for students registered in the university
college, 1959-1960. Gainesville, Fla.: Privately
circulated mimeographed paper, 1960.

Board of University Examiners, University of Florida. Test
results for the SCAT, 1960, 1961, 1962, 1963.
Gainesville, Fla.: Privately circulated mimeographed
papers, 1960, 1961, 1964, 1963.

Boder, D. P. The Adjective-Verb Quotient. Psychol. Rec.,
1940, 3, 310-343.

Brengelmann, J. C. Expressive movements and abnormal
behavior. In H. S. Eysenck (Ed.), Handbook of
abnormal psychology. New York: Basic Books, 1960.



Brown, R. W., & Lenneberg, E. H. A study in language and
cognition. J. abnorm. soc. Psychol., 1954, 49, 454-

Bruner, J. S. On perceptual readiness. Psychol. Rev., 1957,
64, 123-152.

Busemann, A. Die sprache der jugend als ausdruck der
entwicklungsrhytmik. Germany: Jena, 1925.

Byrne, D. The repression-sensitization scale; rationale,
reliability, and validity. J. Pers., 1961, 29, 334-

Byrne, D. The revised repression-sensitization scale and its
relationship to measures of self-description. Privately
circulated manuscript, 1963.

Byrne, D., Barry, J., & nelson, D. Relation of the revised
repression-sensitization scale to measures of self-
description. Psychol. Rep., 1963, 13, 323-334.

Carroll, J. B., & Casagrande, J. B. The function of language
classifications in behavior. In Eleanor E. Maccoby,
T. M. Newcomb, & E. L. Hartley (Eds.), Readings in
social psychology. (3rd ed.) New York: Henry Holt,

Chotlos, J. W. A statistical and comparative analysis of
individual written language samples. Psychol.
Monogr., 1944, 56, No. 2 (whole No. 255, 77-111.

Christie, R. J. Personality structure. In P. Farnsworth,
Olga McNemar & Q. McNemar (Eds.), Annual review of
psychology. Vol. 14. Palo Alto, Calif.: Annual
Reviews, Inc., 1963.

Conrad, Dorothy, & Conrad, R. The use of personal pronouns
as categories for studying small group interaction.
J. abnorm. soc. Psychol., 1956, 52, 277-279.

Doob, L. W. Behavior and grammatical style. J. abnorm. soc.
Psychol., 195d, 56, 398-401.

Fairbanks, Helen. The quantitative differentiation of samples
of spoken language. Psychol. Monogr., 1944, 56, No. 2
(whole No. 255), 19-40.

Fenichel, U. The psychoanalytic theory of neurosis. New
York: Norton, 1945.

Freud, a. A general introduction to psychoanalysis. New
York: Permabooks, 1953 (original date of publication,

Huxley, A. After many a summer dies the swan. New York:
Avon, 1939.
Jaffe, J. Language of the dyad. Psychiatry, 1958, 21, 249-

Johnson, W. Studies in language behavior. Psychol. Monogr.,
1944, 56, No. 2 (whole No. 255).
Joy, V. L. Repression-sensitization and interpersonal
behavior. Paper read at Amer. Psychol. Ass.,
Philadelphia, September, 1963.

Korzybski, A. Science and sanity. (2nd ed.) Lancaster,
Pa.: The International von-Aristotelian Publishing
Co., 1941.

Lindquist, E. F. Design and analysis of experiments in
pschology and education. Boston: Houghton Mifflin,

Lorenz, Maria. Language behavior in psychoneurotic patients.
Arch. Neurol. Psychiat., 1953, 69, 665-673.

Lorenz, Maria. Language as index to perceptual modes.
J. proj. Tech., 1959, 23, 440-452.

Lorenz, Maria & Cobb, S. Language behavior in psychotic and
neurotic subjects. Arch. Neurol. Psychiat., 1954, 72,

Mann, Mary. The quantitative differentiation of samples of
written language. Psychol. Monogr., 1944, 56, No. 2
(whole No. 255), 4176-

Mannes, Marya. A shelf divided: his and hers. (Bookweek)
Washington Post, Nov. 3, 1963.

McGinnies, E. M. Emotionality and perceptual defense,
Psychol. Rev., 1949, 244-251.

Mcemar, Q. Psychological Statistics. (2nd ed.) New York:
Wiley, 1955.
McQuown, N. Linguistic transcription and specification of
psychiatric interview materials. Psychiatry, 1957,
20, 79-86.
Osgood, C. E., & Walker, Evelyn E. Motivation and language
behavior: A content analysis of suicide notes. J.
abnorm. soc. ssychol., 1959, 59, 58-67.

Pittenger, R. E., & Smith, H. L. Some contributions of
linguistics to psychiatry. Psychiatry, 1957, 20, 61-

Postman, L., & Bruner, J. S. Perception under stress.
Psychol. Rev., 1948, 5, 314-323.

Raimy, V. C. Self-reference in counseling interviews. J.
consult. Psychol., 1948, 12, 155-163.

Rowland, Eleanor. The psychological experiences connected
with the different parts of speech. Psychol. Rev.
Monogr. Suppl., 1907, 1, No. 8.

Ruesch, J. Disturbed communication. New York: Norton,

Sapir, E. Language. In D. Mandlebaum (Ed.), Selected
writings of Edward Sapir. Berkeley, Calif.T: 7 niver.
of Calif. Press, 1940.

Sapir, E. Speech as a personality trait. Amer. J. Sociol.,
1927, 32, 892-905.

Tempone, V. J. Differential thresholds of repressers and
sensitizers as a function of a success and failure
experience. Unpublished doctoral dissertation,
Univer. of Texas, 1962.

Ullman, L. P. Clinical correlates of facilitation and
inhibition of response to emotional stimuli. J.
proj. Tech., 1958, 22, 341-347.

Ullman, L. P. An empirically derived MMPI scale which
measures facilitation-inhibition of recognition of
threatening stimuli. J. clin. Psychol., 1962, 18,

Wagner, R. F., & Williams, J. E. An analysis of speech
behavior in groups differing in achievement imagery
and defensiveness. J. Pers., 1961, 29, 1-10.

Welsh, G. S., & Dahlstrom, W. G. Basic readings on the
MMPI in psychology and medicine. Minneapolis: Univer.
of Minnesota Press, 1956.

Whorf, B. L. Language, thought, and reality. Boston:
Institute of Technology, 1956.

Whorf, B. L. Science and linguistics. In Eleanor E.
Maccoby, T. M. Newcomb, & E. L. Hartley (Eds.),
Readings in Social Psychology. (3rd ed.) New York:
Henry Holt, 1959.

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